Recent Trends and Advances in Artificial Intelligence and Internet of Things [1 ed.] 3030326438, 9783030326432

Table of contents :
Preface
About This Book
Key Features
Contents
About the Editors
1 A 128-bit Tunable True Random Number Generator with Digital Clock Manager
1.1 Introduction
1.1.1 Problem Description
1.1.2 Solution
1.2 Related Work
1.3 Existing Method
1.4 Proposed Model
1.5 Simulation Results
1.6 Conclusion
References
2 Network Monitoring System Using Ping Methodology and GUI
2.1 Introduction
2.2 Existing System
2.3 System Methodology
2.3.1 System Architecture
2.4 Result and Discussion
2.5 Summary
References
3 License Plate Recognition Based on K-Means Clustering Algorithm
3.1 Introduction
3.2 Materials and Methods
3.3 Results and Discussion
3.4 Conclusion
References
4 An Implementation of Bidirectional NOC Router for Reconfigurable Coarse Grained Architecture by Using Vedic Multiplier
4.1 Introduction
4.2 Related Works
4.3 32 × 32 Vedic Multiplier
4.4 CGRA Architecture
4.4.1 Representation of Processing Elements (PE)
4.5 Result and Discussion
4.6 Conclusion
References
5 Breast Cancer Classification Using Tetrolet Transform Based Energy Features and K-Nearest Neighbor Classifier
5.1 Introduction
5.2 Methods and Materials
5.3 Results and Discussion
5.4 Conclusion
References
6 Bayesian Neural Networks of Probabilistic Back Propagation for Scalable Learning on Hyper-Parameters
6.1 Introduction
6.2 Existed Methods
6.2.1 Radial Foundation Purpose Networks
6.3 Proposed Method (BNN-PB)
6.3.1 Computational Requirement
6.3.2 Obtaining Well Calibrated Uncertainty Estimates with Bayesian Neural Networks
6.3.3 Neural Networks Including More Than One Hidden Layer
6.4 Results
6.5 Conclusion
References
7 Extensive Study on Antennae for IoT Applications
7.1 Introduction
7.2 Antennas for IoT Applications
7.2.1 SIW Antenna
7.2.2 RFID Reader Antenna
7.2.3 UNB Miniature Antenna
7.2.4 Dual Band UWB Antenna
7.2.5 Compact Dual Band Antenna
7.2.6 Reconfigurable Patch Antenna
7.3 Conclusion
References
8 A Bi-spectrum Analysis of Uterine Electromyogram Signal Towards the Prediction of Preterm Birth
8.1 Introduction
8.2 Materials and Methods
8.2.1 Data Requisition
8.2.2 SIW Antenna
8.2.3 Pre-processing
8.2.4 Bi-Spectrum Analysis
8.2.5 Classifier
8.3 Results and Discussion
8.4 Conclusion
References
9 Application of Information Science and Technology in Academic Libraries: An Overview
9.1 Introduction
9.2 Informational
9.3 Informational Required
9.4 Information Science and Technology
9.4.1 Basic Components of IST
9.4.2 Implementation of IST
9.5 IST Application in Digital Library
9.6 Effect of IST in Library
9.7 Merits and Demerits of IST
9.8 Organization of IST Based Services
9.8.1 Types of Equipment and Facilities
9.8.2 Service to Users
9.8.3 E-Sources
9.9 Conclusion
References
10 A Stable Routing Algorithm Based on Link Prediction Method for Clustered VANET
10.1 Introduction
10.2 Related Work
10.3 Proposed Framework
10.3.1 System Scenarios
10.3.2 Procedure for Crating Clusters
10.4 Experimental Results
10.5 Conclusion
References
11 Reversible Image Watermarking for Health Informatics Systems Using Distortion Compensation in Wavelet Domain
11.1 Introduction
11.2 Proposed Method
11.3 Experimental Results
11.4 Conclusion
References
12 A Digital Image Encryption Algorithm Based on Bit-Planes and an Improved Logistic Map
12.1 Introduction
12.2 Related Knowledge
12.2.1 Image Bit-Plane
12.2.2 Logistic Map
12.3 Algorithm Descriptions
12.3.1 Encryption Algorithm Description
12.3.2 Decryption Algorithm Description
12.4 Results
12.5 Conclusion
References
13 A TDMA Based Energy Efficient Unequal Clustering Protocol for Wireless Sensor Network Using PSO
13.1 Introduction
13.2 Related Work
13.2.1 LEACH-C
13.2.2 PSO-C
13.2.3 EBUC
13.2.4 IPSO
13.2.5 PSO-ECHS
13.3 Proposed Framework
13.4 Experimental Results
13.5 Conclusion
References
14 An Improved Network Coding Based LEACH Protocol for Energy Effectiveness in Wireless Sensor Networks
14.1 Introduction
14.2 Related Work
14.3 Proposed Framework
14.3.1 LEACH Protocols
14.3.2 I-LEACH Protocol
14.3.3 Node Rank-LEACH Protocol
14.3.4 Node Rank Algorithm
14.4 Network Coding Method
14.4.1 Opportunistic Listening
14.4.2 Opportunistic Coding
14.4.3 Learning Neighbour State
14.5 Experimental Results
14.5.1 Description of the Simulator
14.6 Conclusion
References
15 A Novel FFT Architecture for an Efficient Utilization of OFDM Using Adaptive FFT Method
15.1 Introduction
15.2 Radix-2 FFT Design
15.3 Structure of Single Path Delay Feedback Structure
15.4 MDC with Radix-2 Structure
15.5 Proposed Model of Adaptive FFT
15.6 Results and Discussions
15.7 Results and Discussions
References
16 Priority Based QoS-Aware Medium Access Control Protocol for Mobile Ad-Hoc Networks
16.1 Introduction
16.2 Literature Review
16.3 Proposed Approach
16.4 Evaluation
16.5 Conclusion
References
17 Intend and Accomplishment of Power Utilization Monitoring and Controlling System by Using IoT
17.1 Introduction
17.2 Related Work
17.3 Proposed System
17.3.1 A High Side Current Detecting
17.3.2 Points of Interest
17.3.3 Low Side Current Identifying
17.4 Results and Discussion
17.5 Conclusion
References
18 75 GHz 5G Frequency Spectrum Analysis
18.1 Introduction
18.2 Work Carried
18.3 Model of System
18.4 Results
18.5 OFDM Channel
18.6 Conclusion
References
19 Energy Conservation Strategy for DC Motor Load Applications
19.1 Introduction
19.2 Circuit Representation
19.3 Control Strategy
19.4 Simulation & Results
19.4.1 Fully Controlled Rectifier Block With Resistive Load
19.4.2 Fully Controlled Rectifier Block with Dc Motor Load
19.5 Experimental Setup and Hardware Results
19.5.1 Output Without Load
19.5.2 Output with Load
19.6 Conclusion
References
20 End-to-End Delay Analyses via LER in Wireless Sensor Networks
20.1 Introduction
20.2 Related Work
20.3 RRBNs Encoding and Decoding Method
20.4 End-to-End Delay in WSN
20.4.1 Delay Analysis of Multi-hop Networks
20.4.2 WSN Low Energy Routing Direction
20.5 Validation Results in WSN
20.6 Conclusion
References
21 Multi Band Antenna System for Quality Evaluation Application of Apple Fruit
21.1 Introduction
21.2 Dielectric Properties for Quality Assessment
21.2.1 Grading of Apple Fruit
21.3 Antenna Design and Geometry
21.3.1 Prototype Design of 2 × 2 Antenna Array
21.4 Antenna Sensing Technique
21.5 Stability Analysis of Antenna System
21.6 Data Transmission Using IOT
21.7 Evaluation of Apple Sample
21.8 Conclusion
References
22 Effective Utilization of Image Information Using Data Mining Technique
22.1 Introduction
22.1.1 Steps of Image Mining
22.2 Preprocessing Steps of Data Mining
22.2.1 Image Extraction
22.2.2 Relational Database Versus Image Database
22.3 Information Retrieval System
22.3.1 Image Mining Algorithm Steps
22.3.2 Creation of Index on Image Data Base
22.4 Application of Data Mining
22.4.1 Video Data Mining Shot Detection
22.4.2 Creation of Histogram on Images
22.4.3 Experimental Results
22.5 Conclusion
References
23 Particle Swarm Optimization Algorithm Based PID Controller for the Control of the Automatic Generation Control
23.1 Introduction
23.2 Materials and Methods
23.3 Particle Swarm Optimization (PSO)
23.4 Simulink Model of AGC with PSO Algorithm
23.5 Simulation Results
23.6 Conclusion
References
24 Proposed Improving Self-management Support System for Chronic Care Model (Heart Diseases)
24.1 Introduction
24.2 Related Work
24.3 The Proposed Method
24.4 Conclusions
References
25 DWINE Your Fear—Defensive Device for Women in Need
25.1 Introduction
25.2 Literature Review
25.3 Drawbacks in the Current Systems
25.4 Proposed Idea
25.5 Architecture Diagram
25.6 Scenario
25.7 Implementation
25.7.1 Hardware Components Used
25.7.2 Main Distinct Modules
25.7.3 Inputs Given
25.7.4 Outputs Obtained
25.7.5 Framework Challenges
25.8 Result
25.9 Conclusion and Future Work
References
26 Microstrip Patch Antenna for Peripheral Arterary Disease Diagnosis
26.1 Introduction
26.2 Proposed Antenna System
26.3 Antenna System Analysis Without Blood Fluid Sample
26.4 Antenna System Analysis During Blood Flow
26.5 Antenna System Analysis with Blood Accumulation
26.6 Conclusion
References
27 Wireless EAR EEG Signal Analysis with Stationary Wavelet Transform for Co Channel Interference in Schizophrenia Diagnosis
27.1 Introduction
27.2 Related Work
27.3 Methodology
27.4 Co-channel Interference in WSN for Dynamic Signal Transmission
27.4.1 Dynamic Signal Transmission in WSN
27.5 Conclusion
References
28 Advance Approach for Effective EEG Artefacts Removal
28.1 Introduction
28.2 Related Work
28.3 Proposed System
28.3.1 Implementation Algorithm The Proposed EEG Motion Artifact Removal Algorithm is as Follows
28.4 Results and Discussion
28.5 Conclusion
References
29 Security in Internet of Things
29.1 Introduction
29.2 IoT Layered Architecture
29.2.1 Sensor Connectivity Layer
29.2.2 Gateway Network Layer
29.2.3 Management Layer
29.2.4 Application Layer
29.3 Security Issues
29.4 Solutions
29.5 Conclusion
References
30 A Hybrid TLBO Algorithm by Quadratic Approximation for Function Optimization and Its Application
30.1 Introduction
30.2 Related Work
30.3 Details of Basic TLBO and QA
30.3.1 Teaching Learning Based Optimization
30.3.2 Quadratic Approximation (QA)
30.4 The Hybrid TLBO Algorithm
30.4.1 Adaptive Teaching Factor
30.5 Results and Discussion
30.5.1 Comparison Results for 10 Dimensional Test Functions
30.5.2 Comparison Results for 30 Dimensional Test Functions
30.5.3 Comparison Results for 50 Dimensional Test Functions
30.6 Application to Real Life Problems
30.6.1 Spread Spectrum Radar Polyphase Code Design Problem
30.7 Conclusion
References
31 Home Automation Using IoT
31.1 Introduction
31.2 Internet of Things
31.2.1 Sensors/Electronic Devices
31.2.2 Data Processing
31.2.3 Cloud-Based System
31.3 Embedded System
31.3.1 Microcontroller
31.3.2 Sensor
31.4 Automation
31.5 IoT Devices and Applications
31.5.1 Application of IoT Devices
31.6 Home Automation
31.7 Embedded System for Home Automation
31.7.1 Hardware Components
31.7.2 Software Requirement
31.8 Home Automation Using IoT
31.9 Advantages of IoT for Home Automation
31.10 Discussion and Recommendations
References
32 Artificial Intelligence: State of the Art
32.1 Introduction
32.1.1 What Is It?
32.1.2 A Short History of AI
32.1.3 The Turing Test
32.2 Applications of AI
32.2.1 AI, Machine Learning and Deep Learning
32.3 Solving Problems by Searching
32.3.1 Uninformed Search Techniques
32.3.2 Bidirectional Search
32.3.3 Informed or Heuristic Search Techniques
32.4 Adversarial Search
32.4.1 Min–Max
32.5 Knowledge Representation, Reasoning and Problem Solving
32.5.1 Propositional Logic (PL)
32.5.2 First Order Predicate Logic
32.5.3 Rule Based Systems
32.5.4 Semantic Nets
32.5.5 Planning Agents
32.6 Reasoning Using Statistics
32.6.1 Joint Probability
32.6.2 Conditional Probability
32.6.3 Chain Rule
32.6.4 Bayes' Theorem
32.6.5 Bayes' Net
32.7 Machine Learning
32.7.1 Supervised Learning
32.7.2 Unsupervised Learning
32.7.3 Reinforcement Learning
32.8 Introduction to ANN
32.8.1 Unit Step Function (Heaviside Step Function)
32.8.2 Logistic Activation Function
32.8.3 Nice Property of Sigmoid Function (Fig. 32.31)
32.8.4 Loss Functions
32.9 Gradient Descent
32.10 Natural Language Understanding
32.11 Conclusion
References
33 Logarithm Similarity Measure Based Automatic Esophageal Cancer Detection Using Discrete Wavelet Transform
33.1 Introduction
33.2 Proposed Esophageal Cancer Detection Scheme
33.3 Data Set
33.4 Discrete Cosine Transformation
33.5 Discrete Wavelet Transform
33.6 Feature Extraction
33.7 Principal Component Analysis
33.8 Linear Discriminant Analysis
33.9 Similarity Measure
33.10 Euclidean Based Similarity Measure
33.11 Logarithm Similarity Measure
33.12 Results
33.12.1 Time and Recognition Rate Taken by DWT and DCT
33.12.2 Comparison Among All Channels
33.12.3 Recognition Rate at Various Feature Extraction Methods
33.13 Conclusion
References
34 Ai Chatbots: Transforming the Digital World
34.1 Introduction
34.2 Chatbot
34.2.1 History of Chat Bots
34.3 Eliza: The First Chatbot
34.4 Alice the Smater Chatbot
34.5 Rise and Evolution of Chatbots
34.5.1 Growth in the Usage of Internet
34.5.2 Recent Advancement in Technology
34.6 Components of a Chat Bot
34.6.1 Natural Language Processing (NLP)
34.6.2 Dialog Manager
34.6.3 Content
34.7 Architectural Model of Chatbot
34.7.1 Generative Model
34.7.2 Retrieval Based Model
34.8 Generation Mechanism of Response by Chat Bots
34.8.1 Artificial Intelligence Modelling Language (AIML)
34.8.2 Pattern Based Heuristics
34.8.3 Intent Classification Based on Machine Learning
34.9 Types of Chat Bots
34.10 Working Mechanism of Chatbots
34.10.1 Pattern Matchers
34.10.2 Algorithms
34.11 Natural Language Processing (NLP) for Chatbot
34.12 Trending Artificial Intelligence Platforms
34.13 Conversational User Interfaces
34.13.1 Basic Bots
34.13.2 Text Based Assistants
34.13.3 Voice Based Assistants
34.14 Bricks of Bot Building
34.15 Design Principles of Chatbot
34.16 Designing Chat and Voice Bots
34.17 Benefits of Chat Bots
34.18 Chatbots: Offering a Boom to Business
34.19 Programming Languages
34.20 Dialog Flow Chatbot Framework
34.21 Building a Chat Bot with Python
34.22 Chatbot in Finance
34.23 Chat Bot in Healthcare
34.24 Conclusion
References
35 Applications of Smart Devices
35.1 Introduction
35.1.1 This Chapter Explains Why there is a Need to Study How Smart Farming is Transforming Agriculture. Why Should the Farmers Make a Shift from Traditional Methods of Farming and Adopt IOT in Farming. The 5 Key Aspects IOT Can Transform Agriculture Are Described Below
35.2 What Essential Things the Farmers Should Take into Consideration Before Adopting the Smart Farming Solutions
35.3 Components of Smart Farming
35.3.1 Management Information Systems
35.3.2 Devices
35.3.3 Application of Smart Devices in Farming
35.4 Smart Farming in the Indian Agriculture Industry Perspective
35.4.1 Introduction: Agriculture in India
35.4.2 ‘DIGITAL INDIA’ Campaign and BIG Data Bringing Technological Revolution in Indian Agriculture
35.4.3 Satsure
35.4.4 Cropin
35.5 Challenges of Smart/Precision Farming
35.5.1 Right Resources
35.6 Limitations of Smart/Precision Farming
35.7 Conclusion
References
36 Fundamental Concepts of Convolutional Neural Network
36.1 Introduction
36.2 Foundation of Convolutional Neural Network
36.3 Concepts of Convolutional Neural Network
36.3.1 Network Layers
36.3.2 Loss Functions
36.4 Training Process of Convolutional Neural Network
36.4.1 Data Pre-processing and Data Augmentation
36.4.2 Parameter Initialization
36.4.3 Regularization to CNN
36.4.4 Optimizer Selection
36.5 Recent Advancement in CNN Architectures
36.5.1 Image Classification
36.5.2 Object Detection
36.5.3 Image Segmentation
36.6 Applications Areas of CNNs
36.6.1 Image Classification
36.6.2 Text Recognition
36.6.3 Action Recognition
36.6.4 Image Caption Generation
36.6.5 Medical Image Analysis
36.6.6 Security and Surveillance
36.6.7 Automatic Colorization of Image and Style Transfer
36.6.8 Satellite Imagery
36.7 Conclusion
References
37 Router Problems of Networking in Cloud Using SIEM
37.1 Introduction
37.2 Working of SIEM
37.3 Architecture of SIEM
37.4 Accessing Information on Cloud
37.4.1 Concept of Public and Private Network
37.5 Introduction to DOS Attack
37.6 How D-DOS Victims Report Cost in Different Categories
37.7 Major Case Studies Related to D-DOS Attack
37.8 Preventions of D-Dos Attack
37.9 Routing and Network Concept in Stem
37.10 Major Risk on Cloud
37.11 Prevention
37.12 Conclusion and Future Scope
References
38 An Energy Efficient Clustered Routing Protocols for Wireless Sensor Networks
38.1 Introduction
38.2 Energy Aware Routing in WSN
38.3 Cluster-Based Routing Protocols of WSN
38.3.1 Classical Cluster-Based Routing Protocols
38.3.2 Heuristic-Based Clustering Protocols in WSN
38.4 Conclusion
References
39 Analysis of Different Detection and Mitigation Algorithm of DDoS Attack in Software-Defined Internet of Things Framework: A Review
39.1 Introduction
39.2 Architecture and Application of IoT
39.2.1 IoT Architecture
39.2.2 Applications of IoT
39.2.3 Issues and Challenges in IoT
39.3 Denial of Service and Distributed Denial of Service Attack
39.3.1 Denial of Service Attack
39.3.2 Distributed Denial of Service Attack
39.3.3 Some Solutions to DoS Attacks
39.4 Detection and Mitigation Techniques of DDoS Attack
39.5 Conclusion
References

Citation preview

Intelligent Systems Reference Library 172

Valentina E. Balas Raghvendra Kumar Rajshree Srivastava   Editors

Recent Trends and Advances in Artificial Intelligence and Internet of Things

Intelligent Systems Reference Library Volume 172

Series Editors Janusz Kacprzyk, Polish Academy of Sciences, Warsaw, Poland Lakhmi C. Jain, Faculty of Engineering and Information Technology, Centre for Artificial Intelligence, University of Technology, Sydney, NSW, Australia; Faculty of Science, Technology and Mathematics, University of Canberra, Canberra, ACT, Australia; KES International, Shoreham-by-Sea, UK; Liverpool Hope University, Liverpool, UK

The aim of this series is to publish a Reference Library, including novel advances and developments in all aspects of Intelligent Systems in an easily accessible and well structured form. The series includes reference works, handbooks, compendia, textbooks, well-structured monographs, dictionaries, and encyclopedias. It contains well integrated knowledge and current information in the field of Intelligent Systems. The series covers the theory, applications, and design methods of Intelligent Systems. Virtually all disciplines such as engineering, computer science, avionics, business, e-commerce, environment, healthcare, physics and life science are included. The list of topics spans all the areas of modern intelligent systems such as: Ambient intelligence, Computational intelligence, Social intelligence, Computational neuroscience, Artificial life, Virtual society, Cognitive systems, DNA and immunity-based systems, e-Learning and teaching, Human-centred computing and Machine ethics, Intelligent control, Intelligent data analysis, Knowledge-based paradigms, Knowledge management, Intelligent agents, Intelligent decision making, Intelligent network security, Interactive entertainment, Learning paradigms, Recommender systems, Robotics and Mechatronics including human-machine teaming, Self-organizing and adaptive systems, Soft computing including Neural systems, Fuzzy systems, Evolutionary computing and the Fusion of these paradigms, Perception and Vision, Web intelligence and Multimedia. ** Indexing: The books of this series are submitted to ISI Web of Science, SCOPUS, DBLP and Springerlink.

More information about this series at http://www.springer.com/series/8578

Valentina E. Balas Raghvendra Kumar Rajshree Srivastava •

•

Editors

Recent Trends and Advances in Artificial Intelligence and Internet of Things

123

Editors Valentina E. Balas Department of Automatics and Applied Software Aurel Vlaicu University of Arad Arad, Romania

Raghvendra Kumar Department of Computer Science and Engineering LNCT Group of Colleges Jabalpur, Madhya Pradesh, India

Rajshree Srivastava Department of Computer Science and Engineering DIT University Dehradun, Uttarakhand, India

ISSN 1868-4394 ISSN 1868-4408 (electronic) Intelligent Systems Reference Library ISBN 978-3-030-32643-2 ISBN 978-3-030-32644-9 (eBook) https://doi.org/10.1007/978-3-030-32644-9 © Springer Nature Switzerland AG 2020 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Preface

There is an increase in number of physical objects that are being connected to the Internet at an unprecedented rate realizing the idea of the Internet of things (IoT). A recent report states that “IoT smart objects are expected to reach 212 billion entities deployed globally by the end of 2020.” Similarly, while the number of connected devices already exceeds the number of humans on the planet by over two times, for most enterprises, simply connecting their systems and devices remains the first priority. A recent report state that, “The overall Internet of Things market is projected to be worth more than one billion U.S. dollars annually from 2017 onwards.” As a result, data production at this stage will be 44 times greater than that in 2009, indicating a rapid increase in the volume, velocity, and variety of data. However, there is highly useful information and so many potential values hidden in the huge volume of IoT-based sensor data. IoT-based sensor data has gained much attention from researchers in health care, bioinformatics, information sciences, and policy- and decision-makers in governments and enterprises. Nowadays, AI methods play a significant role in various environments including business monitoring, healthcare applications, production development, research and development, share market prediction, industrial applications, social network analysis, weather analysis, and environmental monitoring. The IoT and artificial intelligence (AI) will play a vital role in numerous ways in the future. There are multiple forces which are driving the growing need for both technologies and more and more industries; governments, engineers, scientists, and technologists have started to implement it in manifold circumstances. The potential opportunities and benefits of both AI and IoT can be practiced when they are combined, both at the device end as well as at the server. For example, AI combined with machine learning can study from the data to analyze and predict the future actions in advance, such as order replacements in marketing and failure of equipment in an industry just in time. Moreover, AI can be used with machine learning in smart homes to make a truly grand smart home experience. Similarly, AI methods with IoT can be used to analyze the human behavior via Bluetooth signals, motion sensors, or facial recognition technology and to make the corresponding changes in

v

vi

Preface

lighting and room temperatures. This book aims to gather recent research works in emerging artificial intelligence and IoT methods. This book will be competitive enough for the books in machine learning, AI, IoT, application of machine learning, etc., as it will include new emerging work in this domain which will be beneficial to the professors, researchers, scientists, and students who are working in this domain. Moreover, the book will provide insight on the future development of AI and IoT fields that can be used as the decision support for management. The books will be covering edited chapters belong to the IoT and Big Data technologies. The book is organized into 39 chapters; in this, Chap. 1 discussed a true random number generator (TRNG) used for cryptography application which is proposed. The current work relies upon ring oscillators. The proposed work relies upon standard of beat frequency detection (BFD). To the deficiencies and jitter from the oscillators being the hot spot for the arbitrariness, we proposed an enhanced BFD-TRNG setup fitting for FPGA-based applications. This work is finished by utilizing Xilinx programming. Chapter 2 showed root nodes of the network; if any fault occurs in child node, it also affects the root node and shows red color sign to intimate there is fault inside this particular root node, and we can go inside the root node and monitor the child nodes through sub-nodes. This overcomes all disadvantages of existing system which can work on single system only, but proposed system can able to monitor distributed network. Chapter 3 discussed LPR tracking system using k-means (KM) clustering algorithm, and optical character recognition (OCR) technique is discussed. LPR system includes pre-processing using median filter, KM segmentation, and binarization of KM-segmented image; characters are segmented by the license plate region, and finally, characters are recognized by OCR technique. The LPR system is tested by different license plate images in different lighting conditions. The experimental research shows the better performance of the LPR system. Chapter 4 discussed a new bidirectional NoC router with and without contention projected, which provides small area and increased speed. Bidirectional router is used to route the information from source channel to every destination channel. Therefore, it overcomes dispute state and path malfunction troubles. If some path failure may occur, directly it will take other path during the switch allocator. The projected architecture is used to improve the speed of the interconnection link. Simulation result is achieved by ModelSim6.3c, and synthesis is approved out by Xilinx 12.4. Chapter 5 studied a technique for breast cancer classification in digitized mammogram which is put forth employing tetrolet transform-based energy features and k-nearest neighbor (KNN) classifier. The breast mammogram images of benign and malignant category are decomposed into sub-band coefficients using tetrolet transform, and the energy features are extracted. These extracted features are given as input to the KNN classifier. Results show better classification accuracy in the breast cancer images using tetrolet transform-based energy features and KNN classifier.

Preface

vii

Chapter 6 presented a novel versatile strategy for learning Bayesian neural systems, got back to probabilistic engendering (PBP). Like traditional back spread, PBP works by figuring a forward engendering of probabilities through the system and afterward completing a retrogressive calculation of inclinations. A progression of analyses on ten true datasets demonstrates that PBP is essentially quicker than different methods, while offering aggressive prescient capacities. Our examination additionally demonstrates that PBP-BNN gives precise appraisals of the back change on the system weights. Chapter 7 presented an extensive study on various antennae and their performance of IND. Chapter 8 discussed two neural network classifiers applied to the bi-spectrum feature obtained from the uterine EMG signal. The bi-spectrum analysis was done after pre-processing the signal. Three pre-processing methods were tried to improve the performance. The best classification accuracy of 99.89% was obtained with Elman neural network classifier when pre-processed with three-level wavelet (db4) decomposition. The sensitivity and specificity were found to be 100 and 99.77%, respectively. Chapter 9 dealt with the rapid evolution of IST and implementation of its services in the academic libraries. Presently, all academic libraries are established to provide IST-based digital library. IST includes services such as electronic sources to enhance the info required as per the needs of user’s effective manner. Chapter 10 discussed intelligent transport system (ITS) which is self-controlled, wheeled, and stimulating class of MANET. They use RIVLP to improve clustering. Chapter 11 practiced numeral wavelet transforms. The formed distortion is recompensed in the succeeding repetition as soon as a constant is altered in single repetition, and this has been formulated using a novel methodology. Condensed alteration proportion is produced using distortion compensation technique. Upon four varieties of medical images comprising MRI of the brain, cardiac MRI, MRI of breast, and intestinal polyp images, the anticipated scheme is verified. Through a single-stage wavelet transform, the extreme capability of 1.5 BPP is attained. With reference to capability and alteration, the anticipated scheme is greater to the state-of-the-art mechanisms which are validated using investigational outcomes. Chapter 12 acquired the cipher text image, twofold matrices are joined into image matrix which is of eight bits. One-time pad characteristic is the implication of this procedure. In the terminologies of the histogram, to evaluate the safekeeping of image encryption, plaintext sensitivity, data entropy, and pixel relationship index, MATLAB simulation investigates are added. Validating that the procedure compromises upright encryption, investigational outcomes display that the number of pixel changes ratio (NPCR) is superior to ninety percentages; then, the data entropy of the cipher text image extents 7.99. Chapter 13 elevated inter-cluster communication traffic load resulting in hop spot difficulty. By selecting an extra CH named as surrogate cluster head (SCH) in order to restore network connectivity which is interrupted because of failure of MCH, this method uses TDMA to allot time slots.

viii

Preface

Chapter 14 proposed a method works on rank, outperforms random process selection leading to unforeseen fail for CH in other LEACH versions, and results in good performance. Chapter 15 proposed a novel FFT architecture which adaptively switches to the required FFT structures such as R2SDF and R2MDC, based on the SNR values for high-speed and low-power applications. Chapter 16 proposed a new priority-based QoS design for the mobile ad hoc networks (MANETs) to achieve the better performance in terms of efficient bandwidth utilization and less collision rate with prioritized data transmission among the nodes. Chapter 17 showed the functionality of our user-friendly power dimension system for various usage instances that exist. Utilizing two independent existing dimension networks enables to examine the timing connection of exclusive RF interaction. In addition, a projection is offered on the anticipated battery lifetime of a Wi-Fi-based information procurement system. Chapter 18 structured the execution by assessing typical customer throughput, regular cell throughput, cell-edge customer throughput, top customer throughput, and repulsive reason for detainment. The results express the vital practically changes up to 95% for 28 GHz and 180% for 75 GHz ace with respect to 2.14 GHz. Our proposed work correspondingly exhibiting that the 28 and 75 GHz go over band can desert on 80 and 185% of colossal change in UE throughput self-governingly when veered from right now LTE advance repeat band. Chapter 19 dealt with the energy conservation for a separately excited DC motor by applying speed control technique. The experimental setup is established for the system, and the speed control is carried out through variable armature input voltage using firing angle control. Chapter 20 analyzed the IEEE 805.15.4 MAC with changing routing direction. Moreover, considering the CSMA/CA MAC with Redundant Radix Based Number size communication (RRBNs) system. Added to multi-hop networks with novel based routing schemes of Low Energy Routing (LER) direction. In particular, various load conditions of network determine different performance regarding the delay, energy consumption, and reliability of communication links. Finally, determine the network equations based on Markov chain model and communicating the data via RRBNs communication system and also finding the solution at critical condition for various load distributions of the WSNs. Chapter 21 discussed analysis of device stability with stability factor and the input and output impedance matching with Smith Chart and Tuning which are analyzed and validated with the simulation of S parameters using advanced design system (ADS) simulation tools. The designed antenna array evaluates the grade of apple fruit and reports the data to the consumer through IoT. Chapter 22 studied such multimedia information retrieval, productive storage, and organization of available information in focus. This paper discussed how effectively can handle the image data.

Preface

ix

Chapter 23 applied the particle swarm optimization (PSO) algorithm toward the realization of fine PID controllers, especially in contexts involving two area load frequency controls. From the findings, the investigation demonstrates that the selected controller improves the performance of targeted systems and also enhances operations in AGC supplies, a trend confirmed by the resultant sensible dynamic response. Also, Simulink and MATLAB are used to investigate the two areas’ performance control. Similarly, the study employed K–800–23.5–0.0034 or the Al-Dura power plant form. Chapter 24 discussed one of the important CCM model parts, which is self-management support, which will be improved in order to make the model more efficient in the processing. Chapter 25 discussed wearable device which records the adrenaline and oxytocin levels of the body using sensors which is suggested. The heartbeat rate is also taken into consideration to detect if the girl is in any danger. Pictures of the surroundings are also clicked using an in-built camera to have a proper evidence. Chapter 26 diagnosed peripheral artery disease based on frequency from 0.7 to 4.8 GHz which the analysis is proposed. The patch antenna keeps in close proximity over the nerve skin surface. The signal from the antenna acquires and processes with signal processing toolbox to determine blood fluid dynamics for peripheral artery disease diagnosis. Chapter 27 studied the EEG pattern, which acquired for normal and schizophrenia person while watching different videos, namely funny video and horror video. The EEG signal acquires during movie watching task and transmits EEG to the base station through wireless sensor network for the wavelet analysis and classification to evaluate the efficiency of data transmission in various routing algorithms, such as AODV and DSR and co-channel interference of spread spectrum modulation address. Chapter 28 proposed a new technique for removing the artifacts from the EEG signal which uses kurtosis based on difference of Gaussian and super-Gaussian signal and spatially constrained ICA (SCICA) and Daubechies wavelet techniques. Threshold plays an important role in separating the artifacts from the non-artifact EEG. Otsu’s threshold has been adopted as the thresholding method in this chapter. Chapter 29 explained the architecture of IoT and along with general introduction of what IoT is with some examples and continue on to security problems IoT is facing, challenges, and its solutions. Chapter 30, existing work a modified adaptive based teaching factor is suggested for the basic TLBO algorithm. Also, a novel hybrid approach is proposed that combines the teaching learning base optimization (TLBO) algorithm and quadratic approximation (QA). The QA is applied to improve the global as well as local search capability of the method that also represents the characters of “Teacher Refresh.” For the performance investigation, the suggested algorithm is involved to solve twenty classical optimization functions and one real-life optimization problem, and the performances are differentiated with different state-of-the-art methods in terms of numerical results of the solution.

x

Preface

Chapter 31 discussed some basic ideas related to Internet and working of Internet, IoT and its architecture, embedded system, and automation which have been discussed. Further, IoT devices and application, home automation, and embedded system’s requirements for designing the home automation have been discussed in detail. At the end, home automation using IoT and advantages of IoT for the home automation system have been reported. Chapter 32 defined AI and its relationship with machine learning and deep learning followed by a brief timeline of the evaluation of AI, advantages and challenges of AI in today’s world, and then discussed the three fundamental techniques—problem solving, knowledge and reasoning, machine learning, artificial neural networks and natural language processing (NLP) are presented. Chapter 33 showed the purpose to make a computer-aided diagnosis system which can easily detect the cancerous portion. Here two video samples are taken for the work, one is of normal esophageal and another is of cancerous. The videos are split into number of image samples; from them a few images are considered as training samples and rest of the images are taken as testing images. The proposed framework is followed by the application of discrete wavelet transform for image transformation and principal component analysis for the feature extraction, and finally, the comparison between the testing and training images is achieved using logarithm similarity measure. The outcomes demonstrate an accuracy of more than 87%. The accuracy results might be high, if the database should have sufficient and accurate in respect of resolution of image samples. This result is high enough than some benchmark and well-known frameworks. Outcome obtained proves the experiment to be highly efficient and requires a very less amount of time of operation, thereby making it extremely useful in the diagnosis of esophageal cancer. Chapter 34 described the history of chatbots along with their development platforms. For the better understanding, we have created conversations with various available chatbot platforms and also tried to present the procedure of a chatbot creation. Chapter 35 tried to explore how IoT-connected devices and automation shall bring about a revolution in the field of agriculture and tremendously improve nearly every facet of it. IoT in agriculture is the amalgamation of information technology, telecommunications, and sensor technology. Agriculture has been a neglected field in India as far as automation and technological use and applicability are concerned due to lack of funds for technological expansion and limited technical expertise for implementation of the available technologies. Chapter 36 elaborated the discussion of all the basic components of CNN. It also gives a general view of foundation of CNN, recent advancements of CNN, and some major application areas. Chapter 37 described the solution to some of major routing problems during an attack. It contains statistical data as well as tools and techniques of major attacks performed on cloud. It also contains preventions from DOS attack on cloud server. The concept of accessing the data in cloud with the help of public and private key networks is also explained with this survey report.

Preface

xi

Chapter 38 illustrated a survey of clustering hierarchical routing protocols along with clustering protocols based on optimization algorithms with possible future directions. Chapter 39 presented the overview and applications of IoT and SDx paradigm, brief description of distributed denial-of-service attack as well as the different detection and mitigation algorithms. The study showed that security has extremely important for the reliable communication. There have been several influences from our family and friends who have sacrificed lot of their time and attention to ensure that we are kept motivated to complete this crucial project. The editors are thankful to all the members of Springer India Private Limited, especially Prof. (Dr.) Lakhmi C. Jain and Aninda Bose, for the given opportunities to edit this book. Arad, Romania Jabalpur, India Dehradun, India

Valentina E. Balas Raghvendra Kumar Rajshree Srivastava

About This Book

The Internet of things (IoT) and sensors have the ability to harness large volumes of data, while artificial intelligence (AI) can learn patterns in the data to automate tasks for a variety of business benefits The Internet of things (IoT) is a term that has been introduced in recent years to define objects that are able to connect and transfer data via the Internet. “Thing” refers to a device which is connected to the Internet and transfers the device information to other devices. AI is playing a starring role in IoT because of its ability to quickly extract insights from data. Machine learning, an AI technology, brings the ability to automatically identify patterns and detect anomalies in the data that smart sensors and devices generate information such as temperature, pressure, humidity, air quality, vibration, and sound. It is also found that machine learning can have significant advantages over traditional business intelligence tools for analyzing IoT data, including being able to make operational predictions up to 20 times earlier and with greater accuracy than threshold-based monitoring systems and other AI technologies such as speech recognition and computer vision can help extract insight from data that used to require human review. The powerful combination of AI and IoT technologies is helping to avoid unplanned downtime, increase operating efficiency, enable new products and services, and enhance risk management. This book will cover all the emerging trends of artificial intelligence and IoT.

xiii

Key Features

1. 2. 3. 4.

This book will provide new directions of research in the field of AI. This book will provide new directions of research in the field of IoT. It could be beneficial to the industries also to develop software and tools. It will give them new direction in the area of research.

xv

Contents

1

2

3

A 128-bit Tunable True Random Number Generator with Digital Clock Manager . . . . . . . . . . . . . . . . . . . . . . B. Mounika, Vaseem Ahmed Qureshi and Amgoth Srinivas 1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1.1 Problem Description . . . . . . . . . . . . . . . . 1.1.2 Solution . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Existing Method . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4 Proposed Model . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . 1.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Network Monitoring System Using Ping Methodology and GUI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . J. Dhillipan, N. Vijayalakshmi and S. Suriya 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Existing System . . . . . . . . . . . . . . . . . . . . . . . . 2.3 System Methodology . . . . . . . . . . . . . . . . . . . . . 2.3.1 System Architecture . . . . . . . . . . . . . . 2.4 Result and Discussion . . . . . . . . . . . . . . . . . . . . 2.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

........

1

. . . . . . . . .

1 2 2 3 4 5 6 10 10

..........

13

. . . . . . .

. . . . . . .

13 15 15 16 18 20 21

........

23

........ ........ ........

23 24 26

. . . . . . .

License Plate Recognition Based on K-Means Clustering Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. R. Viju and Radha 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . 3.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . .

. . . . . . . . .

. . . . . . .

. . . . . . . . .

. . . . . . .

. . . . . . . . .

. . . . . . .

. . . . . . . . .

. . . . . . .

. . . . . . . . .

. . . . . . .

. . . . . . . . .

. . . . . . .

xvii

xviii

Contents

3.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

5

6

7

An Implementation of Bidirectional NOC Router for Reconfigurable Coarse Grained Architecture by Using Vedic Multiplier . . . . . . . . . . . . . . . . . . . . . . . . . . Yazhinian Sougoumar and Tamilselvan 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Related Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 32  32 Vedic Multiplier . . . . . . . . . . . . . . . . . . . . . 4.4 CGRA Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.1 Representation of Processing Elements (PE) . 4.5 Result and Discussion . . . . . . . . . . . . . . . . . . . . . . . . 4.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

28 28

......

31

. . . . . . . .

. . . . . . . .

31 32 33 34 35 37 37 37

......

39

. . . . .

. . . . .

39 41 43 45 46

...

47

. . . . .

. . . . .

48 48 48 51 52

...

52

. . . .

. . . .

53 54 55 56

Extensive Study on Antennae for IoT Applications . . . . . . . . . . . . . T. Jayanthi, N. Sai Akhila and G. Pravallika 7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2 Antennas for IoT Applications . . . . . . . . . . . . . . . . . . . . . . . .

59

Breast Cancer Classification Using Tetrolet Transform Based Energy Features and K-Nearest Neighbor Classifier A. Amjath Ali, Suman Mishra and Bhasker Dappuri 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Methods and Materials . . . . . . . . . . . . . . . . . . . . . . . 5.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . 5.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

Bayesian Neural Networks of Probabilistic Back Propagation for Scalable Learning on Hyper-Parameters . . . . . . . . . . . . . . . K. Thirupal Reddy and T. Swarnalatha 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Existed Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.1 Radial Foundation Purpose Networks . . . . . . . . . 6.3 Proposed Method (BNN-PB) . . . . . . . . . . . . . . . . . . . . . . 6.3.1 Computational Requirement . . . . . . . . . . . . . . . 6.3.2 Obtaining Well Calibrated Uncertainty Estimates with Bayesian Neural Networks . . . . . . . . . . . . . 6.3.3 Neural Networks Including More Than One Hidden Layer . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . .

. . . .

59 61

Contents

7.2.1 SIW Antenna . . . . . . . . . . . . . 7.2.2 RFID Reader Antenna . . . . . . 7.2.3 UNB Miniature Antenna . . . . . 7.2.4 Dual Band UWB Antenna . . . . 7.2.5 Compact Dual Band Antenna . 7.2.6 Reconfigurable Patch Antenna . 7.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

9

xix

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

A Bi-spectrum Analysis of Uterine Electromyogram Signal Towards the Prediction of Preterm Birth . . . . . . . . . . . . . . . . Kamalraj Subramaniam, P. Shaniba Asmi and Nisheena V. Iqbal 8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . 8.2.1 Data Requisition . . . . . . . . . . . . . . . . . . . . . . . 8.2.2 SIW Antenna . . . . . . . . . . . . . . . . . . . . . . . . . 8.2.3 Pre-processing . . . . . . . . . . . . . . . . . . . . . . . . 8.2.4 Bi-Spectrum Analysis . . . . . . . . . . . . . . . . . . . 8.2.5 Classifier . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . 8.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . .

61 62 62 63 64 65 66 66

....

67

. . . . . . . . . .

. . . . . . . . . .

67 68 68 68 69 70 71 71 73 73

..

75

. . . . . . . . . . . . . . .

75 76 76 77 77 78 78 78 79 80 80 81 82 82 83

. . . . . . . .

. . . . . . . .

. . . . . . . . . .

Application of Information Science and Technology in Academic Libraries: An Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . S. Velmurugan and G. P. Ramesh 9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2 Informational . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.3 Informational Required . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.4 Information Science and Technology . . . . . . . . . . . . . . . . . 9.4.1 Basic Components of IST . . . . . . . . . . . . . . . . . . 9.4.2 Implementation of IST . . . . . . . . . . . . . . . . . . . . 9.5 IST Application in Digital Library . . . . . . . . . . . . . . . . . . . 9.6 Effect of IST in Library . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.7 Merits and Demerits of IST . . . . . . . . . . . . . . . . . . . . . . . . 9.8 Organization of IST Based Services . . . . . . . . . . . . . . . . . . 9.8.1 Types of Equipment and Facilities . . . . . . . . . . . . 9.8.2 Service to Users . . . . . . . . . . . . . . . . . . . . . . . . . 9.8.3 E-Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . .

. . . . . . . . . .

. . . . . . . . . . . . . . .

xx

Contents

10 A Stable Routing Algorithm Based on Link Prediction Method for Clustered VANET . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bhasker Dappuri, Malothu Amru and Allam Mahesh Venkatanaga 10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.3 Proposed Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.3.1 System Scenarios . . . . . . . . . . . . . . . . . . . . . . . 10.3.2 Procedure for Crating Clusters . . . . . . . . . . . . . . 10.4 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

...

85

. . . . . . . .

. . . . . . . .

85 86 87 87 89 89 94 94

..

97

11 Reversible Image Watermarking for Health Informatics Systems Using Distortion Compensation in Wavelet Domain . . . . . . . . . . Swathi Guntupalli, M. Sreevani and M. Raja 11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Proposed Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.3 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 A Digital Image Encryption Algorithm Based on Bit-Planes and an Improved Logistic Map . . . . . . . . . . . . . . . . . . . . . . . Mohammad Jabirulah, Amgoth Srinivas and Panduga Kavitha 12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.2 Related Knowledge . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.2.1 Image Bit-Plane . . . . . . . . . . . . . . . . . . . . . . 12.2.2 Logistic Map . . . . . . . . . . . . . . . . . . . . . . . . 12.3 Algorithm Descriptions . . . . . . . . . . . . . . . . . . . . . . . . 12.3.1 Encryption Algorithm Description . . . . . . . . . 12.3.2 Decryption Algorithm Description . . . . . . . . . 12.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . .

. . . . .

98 99 103 105 106

.....

109

. . . . . . . . . .

. . . . . . . . . .

109 110 110 111 114 114 115 116 117 118

...

119

. . . . . .

119 120 120 120 121 121

. . . . . . . . . .

13 A TDMA Based Energy Efficient Unequal Clustering Protocol for Wireless Sensor Network Using PSO . . . . . . . . . . . . . . . . . Biroju Papachary, Allam Mahesh Venkatanaga and G. Kalpana 13.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.2.1 LEACH-C . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.2.2 PSO-C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.2.3 EBUC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.2.4 IPSO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . .

. . . . .

. . . . . . . . . .

. . . . . .

. . . . . .

Contents

13.2.5 PSO-ECHS 13.3 Proposed Framework . 13.4 Experimental Results . 13.5 Conclusion . . . . . . . . References . . . . . . . . . . . . . .

xxi

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

121 121 123 123 124

.

125

. . . . . . . . . . . . . . .

125 126 127 127 127 128 129 131 132 133 133 134 134 135 135

..

137

. . . . . . . .

. . . . . . . .

137 138 139 139 140 141 142 142

............

145

. . . .

145 146 147 149

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

14 An Improved Network Coding Based LEACH Protocol for Energy Effectiveness in Wireless Sensor Networks . . . . . . . . . Malothu Amru, Mohammad Jabirullah and Asuri Chaitanya Krishna 14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.3 Proposed Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.3.1 LEACH Protocols . . . . . . . . . . . . . . . . . . . . . . . . . 14.3.2 I-LEACH Protocol . . . . . . . . . . . . . . . . . . . . . . . . 14.3.3 Node Rank-LEACH Protocol . . . . . . . . . . . . . . . . 14.3.4 Node Rank Algorithm . . . . . . . . . . . . . . . . . . . . . . 14.4 Network Coding Method . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.4.1 Opportunistic Listening . . . . . . . . . . . . . . . . . . . . . 14.4.2 Opportunistic Coding . . . . . . . . . . . . . . . . . . . . . . 14.4.3 Learning Neighbour State . . . . . . . . . . . . . . . . . . . 14.5 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.5.1 Description of the Simulator . . . . . . . . . . . . . . . . . 14.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 A Novel FFT Architecture for an Efficient Utilization of OFDM Using Adaptive FFT Method . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Manimaran and Aby K. Thomas 15.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.2 Radix-2 FFT Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.3 Structure of Single Path Delay Feedback Structure . . . . . . . 15.4 MDC with Radix-2 Structure . . . . . . . . . . . . . . . . . . . . . . . 15.5 Proposed Model of Adaptive FFT . . . . . . . . . . . . . . . . . . . 15.6 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . 15.7 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Priority Based QoS-Aware Medium Access Control for Mobile Ad-Hoc Networks . . . . . . . . . . . . . . . . . Y. Neeraja and V. Sumalatha 16.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 16.2 Literature Review . . . . . . . . . . . . . . . . . . . . . 16.3 Proposed Approach . . . . . . . . . . . . . . . . . . . . 16.4 Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . .

Protocol

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

xxii

Contents

16.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 17 Intend and Accomplishment of Power Utilization Monitoring and Controlling System by Using IoT . . . . . . . . . . . . . . . . . . . Sk. Md. Afroz Hussain, T. Satya Narayana and Subramanyachari 17.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.3 Proposed System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.3.1 A High Side Current Detecting . . . . . . . . . . . . 17.3.2 Points of Interest . . . . . . . . . . . . . . . . . . . . . . 17.3.3 Low Side Current Identifying . . . . . . . . . . . . . 17.4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . 17.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 75 GHz 5G Frequency Spectrum Analysis Sireesha Pendem and G. P. Ramesh 18.1 Introduction . . . . . . . . . . . . . . . . . . 18.2 Work Carried . . . . . . . . . . . . . . . . . 18.3 Model of System . . . . . . . . . . . . . . 18.4 Results . . . . . . . . . . . . . . . . . . . . . . 18.5 OFDM Channel . . . . . . . . . . . . . . . 18.6 Conclusion . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . .

....

155

. . . . . . . . .

. . . . . . . . .

155 156 157 159 159 159 159 162 163

...................

165

. . . . . . .

. . . . . . .

166 167 168 169 175 175 176

...

177

. . . .

. . . .

177 179 180 180

...

180

. . . . . .

182 182 182 185 185 186

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

19 Energy Conservation Strategy for DC Motor Load Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . U. Hari Priya, P. Jyothi, V. V. S. S. Phanipavan, K. Deepa and Anjana Jain 19.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.2 Circuit Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.3 Control Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.4 Simulation & Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.4.1 Fully Controlled Rectifier Block With Resistive Load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.4.2 Fully Controlled Rectifier Block with Dc Motor Load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.5 Experimental Setup and Hardware Results . . . . . . . . . . . . 19.5.1 Output Without Load . . . . . . . . . . . . . . . . . . . . 19.5.2 Output with Load . . . . . . . . . . . . . . . . . . . . . . . 19.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . .

. . . . . . . . .

. . . . . . .

. . . .

. . . . . .

. . . . . .

Contents

20 End-to-End Delay Analyses via LER in Wireless Sensor Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . K. Ramesh and V. Kannan 20.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20.2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20.3 RRBNs Encoding and Decoding Method . . . . . . . . 20.4 End-to-End Delay in WSN . . . . . . . . . . . . . . . . . . 20.4.1 Delay Analysis of Multi-hop Networks . . 20.4.2 WSN Low Energy Routing Direction . . . . 20.5 Validation Results in WSN . . . . . . . . . . . . . . . . . . 20.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xxiii

........

187

. . . . . . . . .

. . . . . . . . .

187 188 189 190 191 193 193 196 197

..

199

. . . . . . . . . . .

. . . . . . . . . . .

200 200 200 201 201 203 204 204 205 206 206

....

207

. . . . . . . . . . . .

207 208 208 208 209 209 211 211 211 211 212 212

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

21 Multi Band Antenna System for Quality Evaluation Application of Apple Fruit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Angeline M. Flashy and G. P. Ramesh 21.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.2 Dielectric Properties for Quality Assessment . . . . . . . . . . . . 21.2.1 Grading of Apple Fruit . . . . . . . . . . . . . . . . . . . . 21.3 Antenna Design and Geometry . . . . . . . . . . . . . . . . . . . . . 21.3.1 Prototype Design of 2  2 Antenna Array . . . . . . 21.4 Antenna Sensing Technique . . . . . . . . . . . . . . . . . . . . . . . . 21.5 Stability Analysis of Antenna System . . . . . . . . . . . . . . . . . 21.6 Data Transmission Using IOT . . . . . . . . . . . . . . . . . . . . . . 21.7 Evaluation of Apple Sample . . . . . . . . . . . . . . . . . . . . . . . 21.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Effective Utilization of Image Information Using Data Mining Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Saravanan, Dennis Joseph and S. Vaithyasubramanian 22.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.1.1 Steps of Image Mining . . . . . . . . . . . . . . . . . . 22.2 Preprocessing Steps of Data Mining . . . . . . . . . . . . . . . . 22.2.1 Image Extraction . . . . . . . . . . . . . . . . . . . . . . 22.2.2 Relational Database Versus Image Database . . . 22.3 Information Retrieval System . . . . . . . . . . . . . . . . . . . . . 22.3.1 Image Mining Algorithm Steps . . . . . . . . . . . . 22.3.2 Creation of Index on Image Data Base . . . . . . 22.4 Application of Data Mining . . . . . . . . . . . . . . . . . . . . . . 22.4.1 Video Data Mining Shot Detection . . . . . . . . . 22.4.2 Creation of Histogram on Images . . . . . . . . . . 22.4.3 Experimental Results . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

xxiv

Contents

22.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215 23 Particle Swarm Optimization Algorithm Based PID Controller for the Control of the Automatic Generation Control . . . . . . . . Ali Abdyasser Kadhum, Thaeer Mueen Sahib and Mohsın Mousa Mohammed Ali 23.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23.2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . 23.3 Particle Swarm Optimization (PSO) . . . . . . . . . . . . . . . . . 23.4 Simulink Model of AGC with PSO Algorithm . . . . . . . . . 23.5 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Proposed Improving Self-management Support System for Chronic Care Model (Heart Diseases) . . . . . . . . . . . . . . Jammel Mona, Mohammad Dosh and Wafaa Kamel Al-Jibory 24.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.3 The Proposed Method . . . . . . . . . . . . . . . . . . . . . . . . 24.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 DWINE Your Fear—Defensive Device for Women in Need A. B. Sarada Pyngas, B. Ruchitha Chowdary and R. Kavitha 25.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . 25.3 Drawbacks in the Current Systems . . . . . . . . . . . . . . . 25.4 Proposed Idea . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25.5 Architecture Diagram . . . . . . . . . . . . . . . . . . . . . . . . 25.6 Scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25.7 Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25.7.1 Hardware Components Used . . . . . . . . . . . . 25.7.2 Main Distinct Modules . . . . . . . . . . . . . . . . 25.7.3 Inputs Given . . . . . . . . . . . . . . . . . . . . . . . 25.7.4 Outputs Obtained . . . . . . . . . . . . . . . . . . . . 25.7.5 Framework Challenges . . . . . . . . . . . . . . . . 25.8 Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25.9 Conclusion and Future Work . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

...

217

. . . . . . .

. . . . . . .

217 219 222 222 223 224 224

......

227

. . . . .

. . . . .

227 229 230 230 231

......

233

. . . . . . . . . . . . . . .

234 234 234 235 235 235 237 237 237 238 238 238 238 239 239

. . . . .

. . . . . . . . . . . . . . .

. . . . .

. . . . . . . . . . . . . . .

. . . . .

. . . . . . . . . . . . . . .

. . . . . . .

. . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

Contents

xxv

26 Microstrip Patch Antenna for Peripheral Arterary Disease Diagnosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G. P. Ramesh 26.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26.2 Proposed Antenna System . . . . . . . . . . . . . . . . . . . . . . 26.3 Antenna System Analysis Without Blood Fluid Sample 26.4 Antenna System Analysis During Blood Flow . . . . . . . 26.5 Antenna System Analysis with Blood Accumulation . . . 26.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

.....

241

. . . . . . .

. . . . . . .

241 242 244 244 246 250 250

....

253

.... .... ....

253 255 256

. . . .

. . . .

258 258 263 264

....

267

.... .... ....

267 269 271

. . . .

. . . .

271 275 277 278

..................

279

. . . . . .

279 281 281 281 284 285

27 Wireless EAR EEG Signal Analysis with Stationary Wavelet Transform for Co Channel Interference in Schizophrenia Diagnosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Nithya and G. P. Ramesh 27.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27.2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27.3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27.4 Co-channel Interference in WSN for Dynamic Signal Transmission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27.4.1 Dynamic Signal Transmission in WSN . . . . . . 27.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 Advance Approach for Effective EEG Artefacts Removal . . . . Rudra Bhanu Satpathy and G. P. Ramesh 28.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28.2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28.3 Proposed System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28.3.1 Implementation Algorithm The Proposed EEG Motion Artifact Removal Algorithm is as Follows . . . . . . . . . . . . . . . . . . . . . . . . . 28.4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . 28.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Security in Internet of Things . . . . . . . . . . . Shivam Kolhe, Sonia Nagpal and Jesal Desai 29.1 Introduction . . . . . . . . . . . . . . . . . . . 29.2 IoT Layered Architecture . . . . . . . . . . 29.2.1 Sensor Connectivity Layer . 29.2.2 Gateway Network Layer . . . 29.2.3 Management Layer . . . . . . . 29.2.4 Application Layer . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . . .

. . . . . .

. . . . . . .

. . . .

. . . .

. . . . . .

. . . . . . .

. . . .

. . . .

. . . . . .

. . . . . .

xxvi

Contents

29.3 Security Issues 29.4 Solutions . . . . 29.5 Conclusion . . . References . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

30 A Hybrid TLBO Algorithm by Quadratic Approximation for Function Optimization and Its Application . . . . . . . . . . . . . Sukanta Nama, Apu Kumar Saha and Sushmita Sharma 30.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30.2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30.3 Details of Basic TLBO and QA . . . . . . . . . . . . . . . . . . . . 30.3.1 Teaching Learning Based Optimization . . . . . . . 30.3.2 Quadratic Approximation (QA) . . . . . . . . . . . . . 30.4 The Hybrid TLBO Algorithm . . . . . . . . . . . . . . . . . . . . . 30.4.1 Adaptive Teaching Factor . . . . . . . . . . . . . . . . . 30.5 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . 30.5.1 Comparison Results for 10 Dimensional Test Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30.5.2 Comparison Results for 30 Dimensional Test Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30.5.3 Comparison Results for 50 Dimensional Test Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30.6 Application to Real Life Problems . . . . . . . . . . . . . . . . . . 30.6.1 Spread Spectrum Radar Polyphase Code Design Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . .

286 289 290 290

...

291

. . . . . . . .

. . . . . . . .

292 293 308 308 311 311 311 315

...

315

...

319

... ...

325 325

... ... ...

327 335 337

. . . .

31 Home Automation Using IoT . . . . . . . . . . . . . . . . . . . . . . . . . . . Shahzadi Tayyaba, Salman Ayub Khan, Muhammad Waseem Ashraf and Valentina E. Balas 31.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.2 Internet of Things . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.2.1 Sensors/Electronic Devices . . . . . . . . . . . . . . . . . 31.2.2 Data Processing . . . . . . . . . . . . . . . . . . . . . . . . . 31.2.3 Cloud-Based System . . . . . . . . . . . . . . . . . . . . . . 31.3 Embedded System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.3.1 Microcontroller . . . . . . . . . . . . . . . . . . . . . . . . . . 31.3.2 Sensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.4 Automation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.5 IoT Devices and Applications . . . . . . . . . . . . . . . . . . . . . . 31.5.1 Application of IoT Devices . . . . . . . . . . . . . . . . . 31.6 Home Automation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.7 Embedded System for Home Automation . . . . . . . . . . . . . .

. . . .

. . . . . . . .

..

343

. . . . . . . . . . . . .

344 346 349 349 349 351 353 353 354 356 358 364 367

. . . . . . . . . . . . .

Contents

31.7.1 Hardware Components . . . . . . . 31.7.2 Software Requirement . . . . . . . . 31.8 Home Automation Using IoT . . . . . . . . . 31.9 Advantages of IoT for Home Automation . 31.10 Discussion and Recommendations . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xxvii

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

32 Artificial Intelligence: State of the Art . . . . . . . . . . . . . . . . . . . Bhaskar Mondal 32.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.1.1 What Is It? . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.1.2 A Short History of AI . . . . . . . . . . . . . . . . . . . . 32.1.3 The Turing Test . . . . . . . . . . . . . . . . . . . . . . . . 32.2 Applications of AI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.2.1 AI, Machine Learning and Deep Learning . . . . . 32.3 Solving Problems by Searching . . . . . . . . . . . . . . . . . . . . 32.3.1 Uninformed Search Techniques . . . . . . . . . . . . . 32.3.2 Bidirectional Search . . . . . . . . . . . . . . . . . . . . . 32.3.3 Informed or Heuristic Search Techniques . . . . . . 32.4 Adversarial Search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.4.1 Min–Max . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.5 Knowledge Representation, Reasoning and Problem Solving . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.5.1 Propositional Logic (PL) . . . . . . . . . . . . . . . . . . 32.5.2 First Order Predicate Logic . . . . . . . . . . . . . . . . 32.5.3 Rule Based Systems . . . . . . . . . . . . . . . . . . . . . 32.5.4 Semantic Nets . . . . . . . . . . . . . . . . . . . . . . . . . 32.5.5 Planning Agents . . . . . . . . . . . . . . . . . . . . . . . . 32.6 Reasoning Using Statistics . . . . . . . . . . . . . . . . . . . . . . . . 32.6.1 Joint Probability . . . . . . . . . . . . . . . . . . . . . . . . 32.6.2 Conditional Probability . . . . . . . . . . . . . . . . . . . 32.6.3 Chain Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.6.4 Bayes’ Theorem . . . . . . . . . . . . . . . . . . . . . . . . 32.6.5 Bayes’ Net . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.7 Machine Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.7.1 Supervised Learning . . . . . . . . . . . . . . . . . . . . . 32.7.2 Unsupervised Learning . . . . . . . . . . . . . . . . . . . 32.7.3 Reinforcement Learning . . . . . . . . . . . . . . . . . . 32.8 Introduction to ANN . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.8.1 Unit Step Function (Heaviside Step Function) . . 32.8.2 Logistic Activation Function . . . . . . . . . . . . . . . 32.8.3 Nice Property of Sigmoid Function (Fig. 32.31) . 32.8.4 Loss Functions . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . .

369 373 375 381 382 384

...

389

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

390 390 391 393 393 395 395 396 399 399 401 401

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

403 404 405 406 406 407 407 407 408 408 408 409 410 410 414 416 416 417 419 421 421

. . . . . .

. . . . . .

xxviii

Contents

32.9 Gradient Descent . . . . . . . . . . . 32.10 Natural Language Understanding 32.11 Conclusion . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

421 422 424 424

....

427

. . . . . . . . . . . .

. . . . . . . . . . . .

428 431 431 434 435 437 440 441 442 443 443 444

.... ....

444 444

.... .... ....

447 450 451

...........

455

. . . . . . . . . . . .

455 456 456 457 457 458 458 459 459 459 459 460

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

33 Logarithm Similarity Measure Based Automatic Esophageal Cancer Detection Using Discrete Wavelet Transform . . . . . . . Sayan Chatterjee, Mainak Biswas, Debasis Maji, Birendra Krishna Ghosh and Rajat Kumar Mandal 33.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33.2 Proposed Esophageal Cancer Detection Scheme . . . . . . . 33.3 Data Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33.4 Discrete Cosine Transformation . . . . . . . . . . . . . . . . . . . 33.5 Discrete Wavelet Transform . . . . . . . . . . . . . . . . . . . . . 33.6 Feature Extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33.7 Principal Component Analysis . . . . . . . . . . . . . . . . . . . . 33.8 Linear Discriminant Analysis . . . . . . . . . . . . . . . . . . . . . 33.9 Similarity Measure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33.10 Euclidean Based Similarity Measure . . . . . . . . . . . . . . . 33.11 Logarithm Similarity Measure . . . . . . . . . . . . . . . . . . . . 33.12 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33.12.1 Time and Recognition Rate Taken by DWT and DCT . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33.12.2 Comparison Among All Channels . . . . . . . . . . 33.12.3 Recognition Rate at Various Feature Extraction Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33.13 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 Ai Chatbots: Transforming the Digital World . . . . . Shweta Paliwal, Vishal Bharti and Amit Kumar Mishra 34.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 34.2 Chatbot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34.2.1 History of Chat Bots . . . . . . . . . . . . . 34.3 Eliza: The First Chatbot . . . . . . . . . . . . . . . . . 34.4 Alice the Smater Chatbot . . . . . . . . . . . . . . . . . 34.5 Rise and Evolution of Chatbots . . . . . . . . . . . . 34.5.1 Growth in the Usage of Internet . . . . 34.5.2 Recent Advancement in Technology . 34.6 Components of a Chat Bot . . . . . . . . . . . . . . . 34.6.1 Natural Language Processing (NLP) . 34.6.2 Dialog Manager . . . . . . . . . . . . . . . . 34.6.3 Content . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . .

. . . . . . . . . . . .

. . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

Contents

34.7

Architectural Model of Chatbot . . . . . . . . . . . . . . . . . . . . 34.7.1 Generative Model . . . . . . . . . . . . . . . . . . . . . . . 34.7.2 Retrieval Based Model . . . . . . . . . . . . . . . . . . . 34.8 Generation Mechanism of Response by Chat Bots . . . . . . 34.8.1 Artificial Intelligence Modelling Language (AIML) . . . . . . . . . . . . . . . . . . . . . . 34.8.2 Pattern Based Heuristics . . . . . . . . . . . . . . . . . . 34.8.3 Intent Classification Based on Machine Learning 34.9 Types of Chat Bots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34.10 Working Mechanism of Chatbots . . . . . . . . . . . . . . . . . . . 34.10.1 Pattern Matchers . . . . . . . . . . . . . . . . . . . . . . . . 34.10.2 Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34.11 Natural Language Processing (NLP) for Chatbot . . . . . . . . 34.12 Trending Artificial Intelligence Platforms . . . . . . . . . . . . . 34.13 Conversational User Interfaces . . . . . . . . . . . . . . . . . . . . . 34.13.1 Basic Bots . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34.13.2 Text Based Assistants . . . . . . . . . . . . . . . . . . . . 34.13.3 Voice Based Assistants . . . . . . . . . . . . . . . . . . . 34.14 Bricks of Bot Building . . . . . . . . . . . . . . . . . . . . . . . . . . 34.15 Design Principles of Chatbot . . . . . . . . . . . . . . . . . . . . . . 34.16 Designing Chat and Voice Bots . . . . . . . . . . . . . . . . . . . . 34.17 Benefits of Chat Bots . . . . . . . . . . . . . . . . . . . . . . . . . . . 34.18 Chatbots: Offering a Boom to Business . . . . . . . . . . . . . . 34.19 Programming Languages . . . . . . . . . . . . . . . . . . . . . . . . . 34.20 Dialog Flow Chatbot Framework . . . . . . . . . . . . . . . . . . . 34.21 Building a Chat Bot with Python . . . . . . . . . . . . . . . . . . . 34.22 Chatbot in Finance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34.23 Chat Bot in Healthcare . . . . . . . . . . . . . . . . . . . . . . . . . . 34.24 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xxix

. . . .

. . . .

. . . .

460 460 460 460

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

461 462 462 462 463 464 464 465 465 468 469 469 469 469 470 470 471 472 473 473 475 477 479 480 481

..

483

..

483

..

485

..

486

35 Applications of Smart Devices . . . . . . . . . . . . . . . . . . . . . . . . . . Prabhsimar Kaur and Vishal Bharti 35.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35.1.1 This Chapter Explains Why there is a Need to Study How Smart Farming is Transforming Agriculture. Why Should the Farmers Make a Shift from Traditional Methods of Farming and Adopt IOT in Farming. The 5 Key Aspects IOT Can Transform Agriculture Are Described Below . . . . 35.2 What Essential Things the Farmers Should Take into Consideration Before Adopting the Smart Farming Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xxx

Contents

35.3

Components of Smart Farming . . . . . . . . . . . . . . . . . 35.3.1 Management Information Systems . . . . . . . . 35.3.2 Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . 35.3.3 Application of Smart Devices in Farming . . . 35.4 Smart Farming in the Indian Agriculture Industry Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35.4.1 Introduction: Agriculture in India . . . . . . . . 35.4.2 ‘DIGITAL INDIA’ Campaign and BIG Data Bringing Technological Revolution in Indian Agriculture . . . . . . . . . . . . . . . . . . . . . . . . . 35.4.3 Satsure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35.4.4 Cropin . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35.5 Challenges of Smart/Precision Farming . . . . . . . . . . . 35.5.1 Right Resources . . . . . . . . . . . . . . . . . . . . . 35.6 Limitations of Smart/Precision Farming . . . . . . . . . . . 35.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . .

487 487 488 488

...... ......

511 511

. . . . . . . .

. . . . . . . .

512 512 512 513 513 514 515 515

..

519

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

519 521 522 523 533 535 536 537 540 545 551 551 556 559 563 563 563 563 563 564 564

.. ..

564 564

. . . .

. . . .

. . . . . . . .

. . . .

. . . . . . . .

. . . .

. . . . . . . .

36 Fundamental Concepts of Convolutional Neural Network . . . . . Anirudha Ghosh, Abu Sufian, Farhana Sultana, Amlan Chakrabarti and Debashis De 36.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36.2 Foundation of Convolutional Neural Network . . . . . . . . . . . 36.3 Concepts of Convolutional Neural Network . . . . . . . . . . . . 36.3.1 Network Layers . . . . . . . . . . . . . . . . . . . . . . . . . 36.3.2 Loss Functions . . . . . . . . . . . . . . . . . . . . . . . . . . 36.4 Training Process of Convolutional Neural Network . . . . . . . 36.4.1 Data Pre-processing and Data Augmentation . . . . 36.4.2 Parameter Initialization . . . . . . . . . . . . . . . . . . . . 36.4.3 Regularization to CNN . . . . . . . . . . . . . . . . . . . . 36.4.4 Optimizer Selection . . . . . . . . . . . . . . . . . . . . . . 36.5 Recent Advancement in CNN Architectures . . . . . . . . . . . . 36.5.1 Image Classification . . . . . . . . . . . . . . . . . . . . . . 36.5.2 Object Detection . . . . . . . . . . . . . . . . . . . . . . . . . 36.5.3 Image Segmentation . . . . . . . . . . . . . . . . . . . . . . 36.6 Applications Areas of CNNs . . . . . . . . . . . . . . . . . . . . . . . 36.6.1 Image Classification . . . . . . . . . . . . . . . . . . . . . . 36.6.2 Text Recognition . . . . . . . . . . . . . . . . . . . . . . . . 36.6.3 Action Recognition . . . . . . . . . . . . . . . . . . . . . . . 36.6.4 Image Caption Generation . . . . . . . . . . . . . . . . . . 36.6.5 Medical Image Analysis . . . . . . . . . . . . . . . . . . . 36.6.6 Security and Surveillance . . . . . . . . . . . . . . . . . . 36.6.7 Automatic Colorization of Image and Style Transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36.6.8 Satellite Imagery . . . . . . . . . . . . . . . . . . . . . . . . .

. . . .

. . . . . . . .

Contents

xxxi

36.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565 ...

569

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

569 570 570 571 571 571 572 573 574 576 577 578 579 579

....

581

. . . . . . .

. . . . . . .

581 583 583 585 588 593 594

..

597

. . . . . . . . .

597 598 599 600 601 602 602 603 604

37 Router Problems of Networking in Cloud Using SIEM . . . . . . Rajshree Srivastava 37.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37.2 Working of SIEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37.3 Architecture of SIEM . . . . . . . . . . . . . . . . . . . . . . . . . . . 37.4 Accessing Information on Cloud . . . . . . . . . . . . . . . . . . . 37.4.1 Concept of Public and Private Network . . . . . . . 37.5 Introduction to DOS Attack . . . . . . . . . . . . . . . . . . . . . . . 37.6 How D-DOS Victims Report Cost in Different Categories . 37.7 Major Case Studies Related to D-DOS Attack . . . . . . . . . 37.8 Preventions of D-Dos Attack . . . . . . . . . . . . . . . . . . . . . . 37.9 Routing and Network Concept in Stem . . . . . . . . . . . . . . 37.10 Major Risk on Cloud . . . . . . . . . . . . . . . . . . . . . . . . . . . 37.11 Prevention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37.12 Conclusion and Future Scope . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 An Energy Efficient Clustered Routing Protocols for Wireless Sensor Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nitin Mittal and Rajshree Srivastava 38.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38.2 Energy Aware Routing in WSN . . . . . . . . . . . . . . . . . . . 38.3 Cluster-Based Routing Protocols of WSN . . . . . . . . . . . . 38.3.1 Classical Cluster-Based Routing Protocols . . . . 38.3.2 Heuristic-Based Clustering Protocols in WSN . 38.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . .

39 Analysis of Different Detection and Mitigation Algorithm of DDoS Attack in Software-Defined Internet of Things Framework: A Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Naveen Kumar, Nitin Mittal, Palak Thakur and Rajshree Srivastava 39.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39.2 Architecture and Application of IoT . . . . . . . . . . . . . . . . . . 39.2.1 IoT Architecture . . . . . . . . . . . . . . . . . . . . . . . . . 39.2.2 Applications of IoT . . . . . . . . . . . . . . . . . . . . . . 39.2.3 Issues and Challenges in IoT . . . . . . . . . . . . . . . . 39.3 Denial of Service and Distributed Denial of Service Attack . 39.3.1 Denial of Service Attack . . . . . . . . . . . . . . . . . . . 39.3.2 Distributed Denial of Service Attack . . . . . . . . . . 39.3.3 Some Solutions to DoS Attacks . . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . .

. . . . . . . . .

xxxii

Contents

39.4 Detection and Mitigation Techniques of DDoS Attack . . . . . . 604 39.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 606 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 606

About the Editors

Valentina E. Balas, Ph.D., is currently Full Professor in the Department of Automatics and Applied Software at the Faculty of Engineering, “Aurel Vlaicu” University of Arad, Romania. She holds a Ph.D. in Applied Electronics and Telecommunications from Polytechnic University of Timisoara. Dr. Balas is author of more than 270 research papers in refereed journals and international conferences. Her research interests are in intelligent systems, fuzzy control, soft computing, smart sensors, information fusion, modeling and simulation. She is the Editor-in-Chief to International Journal of Advanced Intelligence Paradigms (IJAIP) and to International Journal of Computational Systems Engineering (IJCSysE), Editorial Board member of several national and international journals, and is evaluator expert for national and international projects. She served as General Chair of the International Workshop Soft Computing and Applications in seven editions 2005– 2016 held in Romania and Hungary. Dr. Balas participated in many international conferences as Organizer, Session Chair, and member in International Program Committee. Now she is working in a national project with EU-funding support: BioCell-NanoART = Novel Bio-inspired Cellular Nano-Architectures—For Digital Integrated Circuits, 2M Euro from National Authority for Scientific Research and Innovation. She is a member of EUSFLAT, ACM and a Senior Member IEEE, member in TC—Fuzzy Systems (IEEE CIS), member in TC—Emergent Technologies (IEEE CIS), member in TC—Soft Computing (IEEE SMCS). Dr. Balas was Vice-President (Awards) of IFSA International Fuzzy Systems Association Council (2013–2015) and is a Joint Secretary of the Governing Council of Forum for Interdisciplinary Mathematics (FIM),—A Multidisciplinary Academic Body, India. Raghvendra Kumar is working as Asst. Professor in Computer Science and Engineering Department at LNCT Group of College Jabalpur, M.P., India, and serving as Director of IT and Data Science Department, Vietnam Center of Research in Economics, Management, Environment (VCREME)—Branch VCREME One Member Company Limited, Vietnam. He received B.Tech. in Computer Science and Engineering from SRM University Chennai, Tamil Nadu, India, M.Tech. in xxxiii

xxxiv

About the Editors

Computer Science and Engineering from KIIT University, Bhubaneswar, Odisha, India, and Ph.D. in Computer Science and Engineering from Jodhpur National University, Jodhpur, Rajasthan, India. He serves as Series Editor of Internet of Everything (IOE): Security and Privacy Paradigm publishes by CRC press, Taylor & Francis Group, USA, and Bio-Medical Engineering: Techniques and Applications, Publishes by Apple Academic Press, CRC Press, Taylor & Francis Group, USA. He also serves as Acquisition Editor for Computer Science by Apple Academic Press, CRC Press, Taylor & Francis Group, USA. He has published number of research papers in international journal (SCI/SCIE/ESCI/Scopus) and conferences including IEEE and Springer as well as serves as organizing chair (RICE-2019), volume editor (RICE-2018), keynote speaker, session chair, co-chair, publicity chair, publication chair (NGCT-2017), advisory board, Technical Program Committee members in many international and national conferences, and serves as guest editors in many special issues from reputed journals (Indexed By: Scopus, ESCI). He also published 11 chapters in edited book published by IGI Global, Springer, and Elsevier. He also received best paper award in IEEE Conference 2013 and Young Achiever Award-2016 by IEAE Association for his research work in the field of distributed database. His research areas are computer networks, data mining, cloud computing and secure multiparty computations, theory of computer science and design of algorithms. He authored and edited 17 computer science books in field of Internet of things, data mining, biomedical engineering, Big Data, robotics, graph theory, and turing machine by IGI Global Publication, USA, IOS Press Netherland, Springer, Elsevier, CRC Press, USA, S. Chand Publication and Laxmi Publication. He is Managing Editor in International Journal of Machine Learning and Networked Collaborative Engineering (IJMLNCE) ISSN 2581-3242. Rajshree Srivastava is Assistant Professor in DIT University Dehradun, in the Department of Computer Science and Engineering. She has completed her M.Tech. from JIIT Noida (Deemed to be University) in CSE-IS and B.Tech. from RTU in Computer Science and Engineering. She is a lifetime member of (IEAE), member of IEEE, CSI, ACM, ACM-W, IAENG, Internet of things. Her areas of research are in machine learning, Big Data, biomedical, privacy security. She has published chapters; Scopus indexed papers and many in IEEE/Springer conferences. Currently, she is a TPC member of ICETIT-2019, ICETIT-2019, INDIACOM 2019, ICIC 2018, ICETMSD-2018, ICACE 2018, WECON 2018, ICAST-2018, RTESD-2018, any more IEEE/Scopus indexed conferences. She has also given talk in IIT Kharagpur on the event of Vishleshan. Currently, she is also session chair holder of PDGC 2018 and ICETIT 2019, and reviewer of the journal entitled International Journal of Handheld Computing Research (IJHCR), IGI Global Publication. She has guided many undergraduate projects. She has attended various FDP, short-term courses, and workshops from IITs and NITs.

Chapter 1

A 128-bit Tunable True Random Number Generator with Digital Clock Manager B. Mounika, Vaseem Ahmed Qureshi and Amgoth Srinivas

Abstract This paper introduces a 128 bit Tunable genuine Random number generator for cryptographic applications. In this, Digital clock supervisor (DCM), Dynamic incomplete reconfiguration (DPR), Beat Frequency Detection (BFD)—TRNG (True irregular number generator) methods were utilized. Cryptographic computations can be executed on Field Programmable Gate Arrays (FPGAs). In this work a True Random Number Generator (TRNG) used for cryptography application is proposed. The current work relies upon ring oscillators. The proposed work relies upon standard of Beat Frequency Detection (BFD). To the deficiencies and Jitter from the oscillators being the hotspot for the arbitrariness, we proposed an enhanced BFD—TRNG setup fitting for FPGA based applications. This work is finished by utilizing Xilinx programming. Keywords Dynamic partial reconfiguration · FPGA · True random number generator · Equipment trojan

1.1 Introduction In Cryptographic applications require irregular numbers to work. There are various sporadic number age plans, and Random Number Generators (RNGs) are successfully used as IT security things. The discretionary numbers created for veritable arbitrary; else they can essentially injure the security framework. They should not to be self-evident. They should be dependably scattered on a given range and selfadministering of one another. Accordingly there is a need for a perfect RNG that B. Mounika (B) · V. A. Qureshi · A. Srinivas Department of ECE, CMR Engineering College, Hyderabad, India e-mail: [email protected] V. A. Qureshi e-mail: [email protected] A. Srinivas e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_1

1

2

B. Mounika et al.

fulfils these objectives, disregarding the manner in which that its improvement fuses powerfully numerical examination [1]. Discretionary number generators can be made utilizing Field Programmable Gate Arrays (FPGAs). The ported plans on the FPGAs can be utilized as a touch of the space applications. There are a few prerequisites that must be viewed as when the contraption is to be set in space. These are radiation impacts, correspondingly as the nearness cycle of the packaging work so as to have a dismal framework which improves the dependability of the whole device.

1.1.1 Problem Description The Random number generator must be arranged with a better than average cryptographic quality and it ought to similarly be seen as that it is being created for a space application. Cryptographic quality is accomplished by arbitrary numbers that fulfil prerequisites of cryptographic calculations. A FPGA-based structure must be executed. The commotion source which is the essential arbitrary hotspot for any key generator is to be actualized in the FPGA and a discretionary structure using a different equipment board [2]. The most significant nature of factual freedom ought to likewise be checked while executing the arrangement. There are diverse sorts of irregular number generators, for instance, real, deterministic, etc. They have picked a True Random Number Generator, TRNG. A TRNG is a physical gadget that ensures fairminded bits and measurable autonomy. Such a structure also ensures high throughput to district extent [8]. It like manner makes a dependable piece rate [4]. At the point when the sporadic gathering is delivered, it is presented to genuine tests to test its quality. Generators for example, genuine, deterministic and so on. They have chosen a TRNG. A TRNG is a physical gadget that guarantees unprejudiced bits and factual autonomy. Such a plan likewise guarantees high throughput to region proportion [8]. It likewise delivers a solid piece rate [4]. When the arbitrary succession is created, it is exposed to measurable tests to test its quality.

1.1.2 Solution In this work, a completely advanced TRNG circuit dependent on standard computerized rationale is exhibited in which the unobtrusive recurrence contrast between two indistinguishable free-running DCMs is estimated utilizing standard advanced rationale.

1 A 128-bit Tunable True Random Number Generator with Digital …

3

1.2 Related Work Informational This paper presents a straightforward reconfigurable framework and spotlights on the benefits of the most current unique incomplete reconfiguration configuration stream. This paper examines a small number of parts choosing, pseudo random number generators as well as testing arbitrary. Such generators output be used in a different cryptographic applications [5]. Such applications have need to meet more grounded basics than for various applications. Specially, their output must be erratic not including learning of the sources of info. A few criteria for describing and choosing fitting generators are talked about in this documentation. The tests might be valuable as an initial phase in deciding if a generator is appropriate for a specific cryptographic application. This paper [11] is a commitment to the hypothesis of genuine arbitrary number generators dependent on inspecting stage jitter in oscillator rings. In the wake of talking about a few misguided judgments and obviously unfavourable hindrances, they offer a general model which, under mellow suppositions, will produce provably irregular bits with some resilience to antagonistic control and running in the megabit-every second range. A key thought all through the paper is the fill rate, which estimates the portion of the time area in which the simple yield flag is seemingly arbitrary [6]. Our investigation demonstrates that an exponential increment in the quantity of oscillators is required to acquire a consistent consider improvement the fill rate. However, they defeat this issue by presenting a post handling step which comprises of a use of a fitting versatile capacity. These enable the fashioner to separate irregular examples just from a flag with just moderate fill rate and, in this manner, numerous less oscillators than in different structures [12]. Last, they create flaw assault models and they utilize the properties of flexible capacities to withstand such assaults. Exact age of genuinely irregular bits is pivotal to present day cryptography. Our undertaking objective was to decide if the jitter made utilizing reconfigurable rationale could be utilized to make a genuine arbitrary number generator. Our examination proposes that the nondeterministic parts of the jitter, whenever controlled effectively, can create really arbitrary bits. They concocted two conceivable structures to use this jitter to guarantee a precise extraction of bits of nondeterministic wonders. The plan of a gadget that ought to have a basic and dependable execution and that, under just undeniable conditions, ought to produce a genuine arbitrary twofold succession is characterized. A few traps are utilized to smother predisposition and relationship so the ideal measurable properties are acquired without utilizing any pseudorandom change. The proposed plan is very much spoken to by an investigative model that portrays the framework conduct both under ordinary conditions and when distinctive disappointments happen. Inside the model, it is demonstrated that the framework is vigorous to changes in the circuit parameters [13, 14]. Besides, a test system can be characterized to check the right activity of the generator without playing out any factual examination of its yield. This paper shows another TRNG in light of a simple PLL actualized in an advanced Altera

4

B. Mounika et al.

FPLD. Beginning with an investigation of the one accessible on chip wellspring of arbitrariness—the PLL integrated low jitter clock flag, another basic and dependable technique for genuine irregularity extraction is proposed [7]. Essential suppositions about factual properties of jitter flag are affirmed by testing of mean estimation of the TRNG yield flag [9, 10]. The nature of created genuine arbitrary numbers is affirmed by breezing through standard NIST factual tests. The depicted TRNG is custom fitted for implanted System-On-a-Programmable Chip (SOPC) cryptographic applications and can give a decent quality genuine arbitrary piece stream with throughput of a few several kilobits for every second.

1.3 Existing Method The counter yield (N in Fig. 1.1) increases each ROSC period till it achieves the strike recurrence interim after ally is tested and reset. For enhanced representation, how about we consider a precedent in which the normal recurrence distinction between the ROSC pair is 1% and the greatest recurrence contrast because of arbitrary jitter is 0.01%. Under this condition, the normal counter yield is 100 while the most extreme and least checks are 101 and 99, separately. In this situation, we can take the least noteworthy piece (LSB) of the yield consider the TRNG yield. Presently assume the normal recurrence distinction is decreased to 0.5% by altering the recurrence contrast, while the arbitrary jitter continues as before

Fig. 1.1 Ring oscillator based TRNG circuit

1 A 128-bit Tunable True Random Number Generator with Digital …

5

at 0.01%. At that point, the yield check will vary somewhere in the range of 196 and 204, subsequently giving up to three arbitrary bits (first, second, and third LSBs) per yield tally and in the meantime expanding the arbitrariness of the LSB. By arranging the frequencies significantly closer to utilizing the fine grain cutting ckts, create increasingly irregular bits from bigger tally to detriment more drawn out testing time.

1.4 Proposed Model A recurrence investigation of the neighbourhood arbitrary commotion underlines different sub-groupings regarding the noise recurrence. The level band repetitive sound is the arbitrary unprejudiced uncorrelated commotion source that is the most reasonable for haphazardness age. It comes for the most part from warm commotion, for example the irregular developments of the present transporters, for instance over a PN intersection or channel of transistor. Other Sort of noise, the 1 F commotion (otherwise called the Flicker commotion) is exploitable, however connected because of its recurrence reliance. This recurrence reliance is considerably more eminent for the 1/F2 Brownian commotion, making it barely utilize for arbitrariness age. Worldwide deterministic noise sources allude to the non-arbitrary commotion sources which influence similarly every segment of a circuit, as: control supply noise, natural changes (temperature, electromagnetic transmissions). These commotion sources are hazardous & undesirable in TRNG structure for some reasons. They can be anticipated and controlled giving an indirect access to cryptographic assaults. They can likewise command the neighbourhood irregular sources making their estimation troublesome (Fig. 1.2). The BFD-TRNG circuit is a completely computerized TRNG, which depends on jitter extraction by the Beat Frequency Detection (BFD) instrument. Multiphase BFD-TRNG execution is a straightforward augmentation of the single stage design with the expansion of insignificant measure of equipment. The proposed assault is powerful for multiphase execution also. The structure and working of the (single stage) BFD-TRNG

Fig. 1.2 Beat frequency detection and a reference clock scheme

6

B. Mounika et al.

Fig. 1.3 Digital clock manager based tuneable BFD–TRNG

1. The circuit includes 2 semi indistinguishable DCM modules (DCMA/DCMB), with elative improvement as well as situation. Because of inborn physical arbitrariness starting from procedure variety impacts, one of the oscillators (state, DCMA) wavers somewhat quicker than the other oscillator (DCMB) (Fig. 1.3). 2. The yield of the DCMs is utilized to test the yield of another, utilizing a DFF. Without loss of all-inclusive statement, accept the yield of ROSCA is sustained to the D-contribution of the DFF, while the yield of ROSCB is associated with the clock contribution of the DFF. In the DCM hardware the jitter displayed of haphazardness in the projected plan [3]. 3. The output of the counter is calculated by a clock to attain. Furthermore, we have a basic post-handling unit utilizing a VNC to take out any biasing in the created irregular bits. VNC is a notable low overhead plan to dispense with predisposition from an irregular bit stream. In this plan, any information bit “00” or “11” design is wiped out; something else, if the information bit design is “01” or “10”, just the primary piece is held. The last three LSBs of the produced arbitrary number are gone through the VNC. The VNC improves the factual characteristics at the expense of slight decline in throughput. 4. At the certain range the oscillator passes the flag faster. Due to the jitter irregularity called as “beat Frequency intervals”. 5. The inspected reaction is then serialized (more often than not up to three LSBs are found to indicate great irregularity properties with no remedy factors connected), to get the arbitrary bit stream. The structure of tuning circuit is shown in the Fig. 1.4. From the address generation module the 6 bit address is given to the input BRAM. Then the 128 data input bit is given to the DCM-DRF controller.

1.5 Simulation Results Simulation output of proposed system is shown from Figs. 1.5 to 1.6. Figures 1.7 and 1.8 shows the overall flow chart and functions used in the method (Tables 1.1 and 1.2).

1 A 128-bit Tunable True Random Number Generator with Digital …

Fig. 1.4 Architecture of tuning circuitry

Fig. 1.5 Simulation waveforms

Fig. 1.6 RTL schematic

7

8

Fig. 1.7 Flow of the proposed method

B. Mounika et al.

1 A 128-bit Tunable True Random Number Generator with Digital …

9

module TRNG(add,clk,reset,drp,out,en); module BRAM(add,clk,data,drp); module DRP_CONTROL (clk_in, drp,reset, dcm_a,dcm_b,data_in,data_out); module DCM_A(in,en,out,dcm); module DCM_B(dcm,en,out); module DFF(in,clk,out); module COUNTER(in,clk,out); module POST(a,b);

Fig. 1.8 Functions used in proposed method Table 1.1 Proposed results summary Proposed results summery Area

28%

Power

32.83 mw

Delay

0.907 ns

Table 1.2 Comparison between ROSC and DCM based TRNGs Method

Area (%)

Power (mw)

Delay (ns)

Existing

62

32.48

5.6

Proposed

28

32.83

0.907

Fig. 1.9 Graphical representation of existing and proposed result parameters

10

B. Mounika et al.

Figure 1.9 illustrates the graphical representation of result parameters of existing and proposed method.

1.6 Conclusion The proposed 128bit Tuneable genuine irregular number age was reproduced on Xilinx apparatus. The Completion of Comparison of existing and proposed parameter of the genuine arbitrary number age improves the time delay, recurrence, improves the arbitrariness quality. The TRNG uses this tunability highlight for choosing the level of irregularity, consequently generous a large level of adaptability for different functions. Planned framework is better decision when contrasted with existing framework.

References 1. Bhasin, S., Danger, J.L., Guilley, S., Ngo, X.T., Sauvage, L.: Hardware Trojan horses in cryptographic IP cores. In: 2013 Workshop on Fault Diagnosis and Tolerance in Cryptography, IEEE, pp. 15–29 (2013) 2. Cherkaoui, A., Fischer, V., Aubert, A., Fesquet, L.: A self-timed ring based true random number generator. In: 2013 IEEE 19th International Symposium on Asynchronous Circuits and Systems, IEEE, pp. 99–106 (2013) 3. Dichtl, M., Goli´c, J.D.: High-speed true random number generation with logic gates only. In: International Workshop on Cryptographic Hardware and Embedded Systems, pp. 45–62. Springer, Heidelberg (2007) 4. Geary, R.C.: The frequency distribution of the quotient of two normal variates. J. Roy. Stat. Soc. 93(3), 442–446 (1930) 5. Jin, Y., Makris, Y.: Hardware Trojans in wireless cryptographic ICs. IEEE Des. Test Comput. 27(1), 26–35 (2010) 6. Johnson, A.P., Saha, S., Chakraborty, R.S., Mukhopadhyay, D., Gören, S.: Fault attack on AES via hardware Trojan insertion by dynamic partial reconfiguration of FPGA over ethernet. In: Proceedings of the 9th Workshop on Embedded Systems Security, p. 1. ACM (2014) 7. Lie, W., Feng-Yan, W.: Dynamic partial reconfiguration in FPGAs. In: 2009 Third International Symposium on Intelligent Information Technology Application, IEEE, vol. 2, pp. 445–448 (2009) 8. Paillier, P., Verbauwhede, I. (eds.): Vol. 4727 of Lecture Notes in Computer Science, pp. 45–62. Springer, Heidelberg (2007) 9. Ramesh, G.P., Kumar, N.M.: Radiometric analysis of Ankle Edema via RZF antenna for biomedical applications. Wireless Pers. Commun. 102(2), 1785–1798 (2018) 10. Ramesh, G.P., Rajan, A.: RF energy harvesting systems for low power applications. Int. J. Technol. Eng. Sci., 1085–1091 (2013) 11. Rukhin, A., Soto, J., Nechvatal, J., Smid, M., Barker, E.: A statistical test suite for random and pseudorandom number generators for cryptographic applications. Booz-Allen and Hamilton Inc Mclean Va (2001) 12. Sunar, B., Martin, W.J., Stinson, D.R.: A provably secure true random number generator with built-in tolerance to active attacks. IEEE Trans. Comput. 56(1), 109–119 (2006)

1 A 128-bit Tunable True Random Number Generator with Digital …

11

13. Tang, Q., Kim, B., Lao, Y., Parhi, K.K., Kim, C.H.: True random number generator circuits based on single-and multi-phase beat frequency detection. In: Proceedings of the IEEE 2014 Custom Integrated Circuits Conference, IEEE, pp. 1–4 (2014) 14. Tehranipoor, M., Koushanfar, F.: A survey of hardware trojan taxonomy and detection. IEEE Des. Test Comput. 27(1), 10–25 (2010)

Chapter 2

Network Monitoring System Using Ping Methodology and GUI J. Dhillipan, N. Vijayalakshmi and S. Suriya

Abstract Network Monitoring is a dynamic web application which is used to monitor the devices among distributed network. This application mainly used in Organisation which widely spread in different locations, by registering all devices of the organisation to this application they can be monitored from the organisation’s headquarters or anywhere else. Using network monitoring which gives detailed information about the device which is defect or disconnected from particular network. Here Searching also done throughout the network using Network IP address, searching can be done on tree hierarchy level. It monitoring starts which displays the overall view of network through graphical user interface here the user can interact with it. Monitoring done on tree level hierarchy which shows root nodes of the network if any fault occurs in child node it also affects the root node and shows red colour sign to intimate there is fault inside this particular root node we can go inside the root node and monitor the child nodes through sub nodes. This overcomes all disadvantages of existing system which can work on single system only but proposed system can able to monitor distributed network. Keywords Ping methodology · GUI · Network monitoring

2.1 Introduction System Monitoring is the utilization of a framework that always screens a PC arrange for moderate or fizzling segments and that informs the system chairman, a system observing framework screens the system for issues caused by over-burden as well as smashed servers, organize associations or different gadgets. This likewise gives the J. Dhillipan (B) · N. Vijayalakshmi · S. Suriya Department of Computer Applications, SRM IST, Ramapuram Campus, Tamil Nadu, India e-mail: [email protected] N. Vijayalakshmi e-mail: [email protected] S. Suriya e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_2

13

14

J. Dhillipan et al.

data important to arrange the executives [1]. The motivation behind system checking is the gathering of valuable data from different parts of the system with the goal that the system can be overseen and controlled utilizing the gathered data. A large portion of the system gadgets are situated in remote areas. These gadgets don’t for the most part have specifically associated terminals with the goal that organize the board application can’t screen their statuses effectively. In this manner, arrange observing procedures are created to permit organize the board applications to check the conditions of their system gadgets [2]. As increasingly more system gadgets are utilized to manufacture greater systems, arrange checking strategies are extended to observing systems overall. As more individuals impart utilizing systems, systems have turned out to be greater and increasingly mind boggling. The multiplication of the web has expanded the pace of system developments. At this period of huge and complex systems, organize observing applications need to utilize compelling methods for checking the status of their systems with the goal that arrange the executives applications can completely control their system and give practical and brilliant systems administration administrations to the clients [3]. It is imperative to recognize what the objectives to accomplish in system checking are. By knowing the objectives of system checking, arrange observing application can pick among system checking methods that will best enable them to screen their system System Monitoring Systems are fundamental in running the mind boggling PC systems of today. They guarantee all shortcomings on the system are known and help the system administrator in fixing these issues. There are issues with the run of the mill NMSs utilized for observing systems, both with the difficulty of configuration and the age of these frameworks [4]. The well known NMSs being used today are either themselves old or stretched out from old frameworks structured 10 years back for observing a lot less complex and littler systems [5]. There is an existing system which can be run on single independent system but the proposed one overcomes that and run on distributed system, this provides different view other than different monitoring system, which consists of monitoring in the form of graphical user interface. The old systems can able to monitor organisation in particular area because which runs on single independent system but proposed system overcomes the disadvantage and it can able to run on distributed system [6]. This consists of device management, searching devices location with its IP and monitoring. The monitoring module plays an major role, by fetching device information from the back end regarding the devices and intimate us with the necessary device information, the defected device or disconnected device or low speed device connected to the network can be shown with red alarm signal this done with the help of svg circles and jquery. System Monitoring framework, which is utilized to enhance organize security and furthermore to enhance arrange uptime and in this manner give access to business applications and furthermore to oversee traffic authority over system. The systems administration process makes interdependency between PCs, servers, organize administrations and clients. Association that neglects to screen the soundness of their system may encounter bring down efficiency, diminished between departmental correspondences and extreme interruptions in client benefit. System checking gives

2 Network Monitoring System Using Ping Methodology and GUI

15

oversight to the complete operational system frameworks through the execution of a lot of specific programming [7]. The strength of system framework is dictated by its general adequacy. The product ought to have the capacity to consistently coordinate with an assortment of switches, IP switches, and firewalls. This quality observing package ought to give clear, visual dashboard portrayal of all checking capacities. This gives an exhaustive diagram of the total system usefulness and foundation all inside a solitary screen condition. The prevalent NMSs being used to day are either themselves old or stretched out from old frameworks planned 10 years prior for observing a lot easier and littler systems.

2.2 Existing System The existing network monitoring system which can able to run on independent system which also consists of device management menu with in that it consists of Add/Register Device, Deleting Device. Next menu consists of search option by clicking it which asks for IP address by providing this searching can be done and results information about devices as Device Name, Device IP address, Device Mac address, location and type of location and so on. Next one is Monitoring menu which consists of two things they are start and stop. Here stop submenu disables until start clicked and start disabled until stop clicked. When starting monitor which shows the status of devices and its information as Time when monitoring starts, Device IP indicates uniqueness, Device type indicates what type of device it monitoring, Location where the monitoring device located, Status which indicate whether it is connected or not, Link speed. The major disadvantage is which not suited for today’s distributed network.

2.3 System Methodology The Proposed network monitoring system which can able to run on distributed systems and it overcomes major disadvantage of Existing system, Comparing to other network monitoring it produces different view with graphical user interface, user can able to interact with it. First there is Login screen which maintains security; there are three users administrator who have all rights. Network controller who can able to access particular organisations devices, users can able to just monitoring only. Existing system consists of Device Management, Search Device, Monitoring. When starting Monitoring process it displays map with colour circles, the colour circle constructed using svg (Scalable Vector Graphics) and jquery if the circle turns to red it indicates some fault in that path by clicking into the circle it retrieves to sub nodes of that selected root node by redirecting finally it gives the expected result where the device fault occurs, whether it is connected or not and so on. Thus it is very useful for today’s distributed network system.

16

J. Dhillipan et al.

2.3.1 System Architecture 2.3.1.1

Admin Login

There In admin Login only admin previllage users can able to login, concern admin people have full authority to monitor the whole Organisation and its branches spread over. In admin role first loged in and need to select the Organisation which is need to be monitored. After that admin need to get single Organisation authentication and moved on to the next step i.e., manipulating the datas which are adding new device, deleting existing device, modifying datas of the device, searching device and also Monitoring the device (Fig. 2.1).

Login

Admin login

Select Organisation name

N/w controller login

Single Organisations authentication

Authorized User login

Registering Device

Edit/view Device

Select Type Of location

Delete Device

Search Device

Monitoring Device

Detect the Fault device Device Management

Report it with Detailed Device information

Fig. 2.1 System architecture

2 Network Monitoring System Using Ping Methodology and GUI

2.3.1.2

17

Network Controller

In this type of login the user can able to monitor single Organisation’s network only, with concern Organisation’s authentication and move on to the next steps as Add, modify, delete, view, Monitor.

2.3.1.3

Authorised User

In this type User cannot able to access the device management database so no previllage to access Device informations like add, view, modify and delete. Certain user can able to monitor the devices belonging to particular Organisation.

2.3.1.4

Register/Add Device

Only particular network controller can able to register or add device to the particular network. Here Location of the device will automatically set to particular region of network controller logged in.

2.3.1.5

Register/Add Device Consists of Following Information

Register/Add Device consists of following information Device Name: Which represents the name of the device Device ip: An IP address is the location of a given computer or other network device on an IP network Which represents the unique ip Device Mac address: A MAC address is kind of a serial number for network devices like ethernet cards—the first half of the MAC address tells what brand/model the card is, and the second half is a unique identifier specific to that card. Location type: This represents the category of location to which the device belongs to Device type: Which represents the category of device to which location it belongs to, it will automatically displays the available types of devices in selected location type. Username: To identify the user of particular device. Password: To prevent unauthorized access and provide security. Edit Device: Only network controller and admin can use this Edit Device. Network Controller can able to edit particular Organisation’s device information only. Admin can able

18

J. Dhillipan et al.

Fig. 2.2 Information about system

to edit all locations device because admin has full authority. First they need to select device type and it results some device which belongs to particular type then they can able to edit and save View device: This is used to view all information regarding the device Delete Device: Network controller and admin can able to delete the device information. Monitoring Device: User admin can able to monitor the device throughout the network. User network controller can able to monitor the particular organisation’s device only. Authorized users can able to monitor the network to which location they belongs to. Which is too detect the defect of device where it occurs and its information. Search Device: Search can be done with the help of Device ip address. By clicking search it displays an input box which asks for IP address to search By providing ip address it shows details of the particular device belongs to IP address, it shows error message box if IP is not valid.

2.4 Result and Discussion Figure 2.2 explains the page which is used to add information about the device with device name, device ip, device MAC address, type of device which belongs, type of location which belongs, username and password of the concern device. Figure 2.3

2 Network Monitoring System Using Ping Methodology and GUI

Fig. 2.3 Fetching device information

Fig. 2.4 Status of the device

19

20

J. Dhillipan et al.

Fig. 2.5 Object locating the fault

describes the device information with its ip address and fetch all information about the device that all present in the registered device information. Figure 2.4 which describes the running status of the device with its link speed and connectivity i.e., status of the device that is connected or not and its type, ip address and time stamp. Figure 2.5 which shows the locality of device fault and capture when the device disconnection. Figure 2.6 is a tree hierarchy of single area monitoring status, it explains that kovai is an parent with childrens as schools and colleges and that two of them having further child and devices connected over there which shows colour representation as green for connected status and Red for disconnected status.

2.5 Summary Network monitoring system is proposed to monitor the distributed system over the network and report the status of the device through the graphical user interface (GUI). Device management like register device, delete device, search device are also possible here to maintain the whole device information in the database. Using ping

2 Network Monitoring System Using Ping Methodology and GUI

21

Fig. 2.6 Tree hierarchy

technology particular timestamp is maintained to stores the status of device into the database. The proposed system uses NMS to retrieves the data’s from there and show it in GUI for easy reference.

References 1. Tan, H., Huang, Z., Wu, M.: Development of a customizable real-time monitoring system for networked control systems. In: 2018 37th Chinese Control Conference (CCC), pp. 6350–6355 (2018) 2. Wang, B., Xu, H., Wang, Y., Wu, G., Liu, L.: Distributed sensor diagnosis for wire fault of complex topology wired networks based on chaos-TDR. In: 2016 Progress in Electromagnetic Research Symposium (PIERS), p. 2152 (2016) 3. Duan, L., Wang, F., Guo, R., Gai, R.: A fault diagnosis method for information systems based on weighted fault diagnosis tree. In: 2017 IEEE 19th International Conference on e-Health Networking, Applications and Services (Healthcom), pp. 1–6. IEEE (2017) 4. Chi, Y., Yang, Y., Xu, P., Li, G., Li, S.: Design and implementation of monitoring data storage and processing scheme based on distributed computing. In: 2018 IEEE 3rd International Conference on Big Data Analysis (ICBDA), pp. 206–211 (2018) 5. Teymoori, P., Sohraby, K., Kim, K.: A fair and efficient resource allocation scheme for multiserver distributed systems and networks. IEEE Trans. Mob. Comput. 15(9), 2137–2150 (2015)

22

J. Dhillipan et al.

6. Ramesh, G.P., Rajan, A.: SRAM based random number generator for non-repeating pattern generation. In: Applied Mechanics and Materials, vol. 573, pp. 181–186. Trans Tech Publications (2014) 7. Ramesh, G.P., Aravind, C.V., Rajparthiban, R., Soysa, N.: Body area network through wireless technology. Int. J. Comput. Sci. Eng. Commun. 2(1), 129–134 (2014)

Chapter 3

License Plate Recognition Based on K-Means Clustering Algorithm V. R. Viju and Radha

Abstract The stolen vehicles are tracked by License Plate Recognition (LPR) system. In image processing technique LPR is used to identify vehicles by their license plates. LPR used in traffic and other various security applications. In this work, LPR tracking system using K-Means (KM) clustering algorithm and Optical Character Recognition (OCR) technique is discussed. LPR system includes pre-processing using median filter, KM segmentation, binarization of KM segmented image; characters are segmented by the license plate region and finally, characters are recognized by OCR technique. The LPR system is tested by different license plate images in different lighting conditions. The experimental research shows the better performance of the LPR system. Keywords LPR · KM clustering · Binarization · OCR

3.1 Introduction Vehicle system hacking based on LPR is discussed in [1]. The input license plate image is converted into gray scale then converted into binary image. Then the characters are extracted. K-Nearest Neighbor (KNN) is used for character recognition. Symmetry features for license plate classification is discussed in [2]. The license plate mage is explored by gradient vector flow for defining the symmetry features. The features are extracted by statistical features. SVM classifier is used for classification. Detection and recognition of LPR characters in Indian vehicles is discussed in [3]. Initially the license plate image is preprocessed to remove illumination condition. Features are extracted by boundary analysis of character. The character segmentation is made by using horizontal and vertical projection of extracted license plate. KNN is used for the character recognition. System design for LPR by using edge detection and Convolution Neural Network (CNN) is discussed in [4]. The license plate is V. R. Viju (B) · Radha Research Department of Computer Science, SDNB Vaishnav College for Women, Chrompet, Chennai, Tamil Nadu, India e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_3

23

24

V. R. Viju and Radha

preprocessed and the edges are detected by filtering the region. Robust distance to borders is used to verify the shape. Morphological operations are used to extract the features. CNN is used for the character recognition. An efficient LPR system using CNN is discussed in [5]. The vehicle license plates are detected. The license plates are retrieved to reduce false positive rate for license plate detection. CNN is applied for the recognition of unclear and blurred image. LPR system based on intrinsic image decomposition algorithm is discussed in [6]. The license plate location is identified. The vehicle license plate image features are extracted by scale invariant feature transform. Finally the matched photos in the database may be suspect vehicle in the ring system. LPR application using Extreme Learning Machines (ELM) is discussed in [7]. The vehicle license plate image features are extracted by histogram of oriented gradients. The recognition of license plate is made by ELM. Research and system design of intelligent LPR algorithm is discussed in [8]. The input license plate images are extracted by three frame difference and background subtraction method. These key frames where treated by edge detection and morphology. License plate pixels identify the location of the license plate. Character projection method used for LPR extraction. The character features are extracted by back propagation method. LPR detection of Myanmar vehicle images which is captured from the different environmental conditions is discussed in [9]. The input license plate image is converted into gray scale. Otsu’s threshold method is used for preprocessing. The edge histogram is detected. The horizontal and vertical dilation is made for license plate images. Binary regions are computed by labeling method. Skew angle is detected. Boundary extraction is made for actual number plate. Finally the license plate is detected. Character recognition for Chinese plate based on CNN is discussed in [10]. The input license plate images are preprocessed by salt and pepper noise to remove noise. The character location is segmented by affine transformation. The characters are recognized by CNN. Point weighting and template matching system based automatic license plate detection is discussed in [11]. The morphological method is used for extracting the license plate image. The location of the license plate image is identified and segmented. The characters in the license plate regions are recognized. Development of intelligent transport system for Philippine LPR is discussed in [12]. The shadow of the license plate image is improved by Bernsen algorithm. Then gray scale image is converted into binary image. Characters are segmented and Hough transform is used for the tilt correction. The horizontal and vertical boundaries are obtained for character recognition. The features are extracted by dual tree complex wavelet transform and artificial neural network is used for classification.

3.2 Materials and Methods Figure 3.1 indicates the workflow of LPR system. The implementation of LPR system includes three stages they are pre-processing, KM segmentation and OCR.

3 License Plate Recognition Based on K-Means Clustering Algorithm

25

K-Means Clustering The clustering problems are solved by KM. KM is a simplest unsupervised learning algorithm. KM clustering is also used in the video analysis [13]. It is a simplest and easy way to classify the datasets in through a large number of clusters. KM algorithm is used for specific number of disjoint clusters. The algorithm is aim at minimizing objective function for square error function. The objective function is defined by, K =

m l    (k)  z − vk 2 p

(3.1)

k=1 p=1

 2  is a distance between a data point z (k) and the cluster centre vk , is where z (k) p − vk p an indicator for distance m data points from the individual cluster centers. The license plate images are segmented by KM algorithm. It produces centroids of k clusters to label new data.

Input car license image

RGB to Gray conversion

Median filtering

KM clustering

Binarization of the image

Segmenting characters

OCR of license plate

Output Fig. 3.1 LPR system—workflow

26

V. R. Viju and Radha

Binarization It is the process of converting pixel into a binary image. Binarization is also used in nature scene text [14] and degraded documents [15]. The binary image is produced after segmentation technique. The assigning of each pixel into two or more classes in the image is known as segmentation. The result with more than two classes is referred as binary image. Binary image is essential for digitalizing and segmentation. In this work, binarization performed by the operation with KM clustering algorithm and converts the pixels into binary image. OCR Technique OCR is the recognition of printed characters in an image. OCR technique also recognizes the characters of historical medical text [16] and it also used in wavelet decomposition [17]. The scanning of characters in an image and it is commonly used in data processing. OCR technique analyze the scanned image and it speed up the recognition process. The OCR technique is used for the recognizing the characters in the segmented image and identify the license plate number.

3.3 Results and Discussion Performance of LPR system is implemented by license plate images. The image database contains 30 license plate images. Figure 3.2 shows some of the sample images in the LPR database.

Fig. 3.2 Sample license plate images in LPR system

3 License Plate Recognition Based on K-Means Clustering Algorithm

(a) Input image

27

(b) Gray image

(C) Filtered image

(d) KM segmented image

(e) Binary image

(f) Edge detected image

(g) OCR recognized image Fig. 3.3 Experimental results obtained from KM segmentation and OCR technique for LPR system

28

V. R. Viju and Radha

The input license plate image is converted to gray scale image. The median filter is used to illumination variations. Then KM segmentation is used to segment the image. The pixel is of the segmented image is converted into binary image. The edges of the characters in the binary image are detected by prewitt filter. The OCR technique is used to recognize the characters in the license plate images. The performance and results of the LPR system is shown in Fig. 3.3.

3.4 Conclusion An efficient method for LPR system for license plate images is discussed. It uses KM clustering algorithm for segmentation and OCR technique for character identification. KM clustering algorithm is used for the segmentation of license plate images. The segmented image pixels are converted into binary image. The edge of the characters in the image is detected. The OCR technique is used to identify the letters and numbers in the image.

References 1. Thaiparnit, S., Khuadthong, N., Chumuang, N., Ketcham, M.: Tracking vehicles system based on license plate recognition. In: 2018 18th International Symposium on Communications and Information Technologies (ISCIT), IEEE, pp. 220–225 (2018). https://doi.org/10.1109/iscit. 2018.8588008 2. Raghunandan, K.S., Shivakumara, P., Padmanabhan, L., Kumar, G.H., Lu, T., Pal, U.: Symmetry features for license plate classification. CAAI Trans. Intel. Technol. 3(3), 176–183 (2018). https://doi.org/10.1049/trit.2018.1016 3. Ingole, S.K., Gundre, S.B.: Characters feature based Indian Vehicle license plate detection and recognition. In: 2017 International Conference on Intelligent Computing and Control (I2C2) IEEE, pp. 1–5 (2017). https://doi.org/10.1109/i2c2.2017.8321953 4. Dhar, P., Guha, S., Biswas, T., Abedin, M.Z.: A system design for license plate recognition by using edge detection and convolution neural network. In: 2018 International Conference on Computer, Communication, Chemical, Material and Electronic Engineering (IC4ME2), IEEE, pp. 1–4 (2018). https://doi.org/10.1109/ic4me2.2018.8465630 5. Lin, C.H., Lin, Y.S., Liu, W.C.: An efficient license plate recognition system using convolution neural networks. In: 2018 IEEE International Conference on Applied System Invention (ICASI), IEEE, pp. 224–227 (2018). https://doi.org/10.1109/icasi.2018.8394573 6. Li, H., Zhou, C., Xue, W., Guo, Y.: License plate recognition based on intrinsic image decomposition algorithm. In: 2014 9th International Conference on Computer Science & Education, IEEE, pp. 512–515 (2014). https://doi.org/10.1109/iccse.2014.6926514 7. Subhadhira, S., Juithonglang, U., Sakulkoo, P., Horata, P.: License plate recognition application using extreme learning machines. In: 2014 Third ICT International Student Project Conference (ICT-ISPC), IEEE, pp. 103–106 (2014). https://doi.org/10.1109/ict-ispc.2014.6923228 8. Feng, Y., Li, S., Pang, T.: Research and system design of intelligent license plate recognition algorithm. In: 2018 37th Chinese Control Conference (CCC) IEEE, pp. 9209–9213 (2018). https://doi.org/10.23919/chicc.2018.8483282

3 License Plate Recognition Based on K-Means Clustering Algorithm

29

9. Khin, O., Phothisonothai, M., Choomchuay, S.: License plate detection of Myanmar vehicle images captured from the dissimilar environmental conditions. In: 2017 International Conference on Advanced Computing and Applications (ACOMP), IEEE, pp. 127–132 (2017). https:// doi.org/10.1109/acomp.2017.31 10. Wu, P., Huang, Z., Li, D.: Research on the character recognition for Chinese license plate based on CNN. In: 2017 3rd IEEE International Conference on Computer and Communications (ICCC), IEEE, pp. 1652–1656 (2017). https://doi.org/10.1109/compcomm.2017.8322820 11. Dastjerdi, H.V., Rostam, V., Kheiri, F.: Automatic license plate detection system based on the point weighting and template matching. In: 2015 7th Conference on Information and Knowledge Technology (IKT), IEEE, pp. 1–5 (2015). https://doi.org/10.1109/ikt.2015.7288783 12. Dalida, J.P.D., Galiza, A.J.N., Godoy, A.G.O., Nakaegawa, M.Q., Vallester, J.L.M., dela Cruz, A.R.: Development of intelligent transportation system for Philippine license plate recognition. In: 2016 IEEE Region 10 Conference (TENCON), IEEE, pp. 3762–3766 (2016). https://doi. org/10.1109/tencon.2016.7848764 13. Chen, T.W., Chien, S.Y.: Bandwidth adaptive hardware architecture of K-means clustering for video analysis. IEEE Trans. Very Large Scale Integr. (VLSI) Syst. 18(6), 957–966 (2010). https://doi.org/10.1109/tvlsi.2009.2017543 14. Thompson, P., McNaught, J., Ananiadou, S.: Customised OCR correction for historical medical text. In: 2015 Digital Heritage, vol. 1, pp. 35–42, IEEE (2015) 15. Su, B., Lu, S., Tan, C.L.: Robust document image binarization technique for degraded document images. IEEE Trans. Image Process. 22(4), 1408–1417 (2013). https://doi.org/10.1109/TIP. 2012.2231089 16. Mishra, A., Alahari, K., Jawahar, C.V.: An MRF model for binarization of natural scene text. In: 2011 International Conference on Document Analysis and Recognition, IEEE, pp. 11–16 (2011). https://doi.org/10.1109/icdar.2011.12 17. Ramesh, G.P., Malini, M., Professor, P.G.: An efficacious method of cup to disc ratio calculation for glaucoma diagnosis using super pixel. Int. J. Comput. Sci. Eng. Commun. 2(3) (2014)

Chapter 4

An Implementation of Bidirectional NOC Router for Reconfigurable Coarse Grained Architecture by Using Vedic Multiplier Yazhinian Sougoumar and Tamilselvan Abstract In Soc approaches Noc router performs an essential task. But in SoC chip, the routing operation performance is very complex. Since single Integrated Circuit carries millions of chips, to which each chip includes of millions of transistors. Therefore, NoC router is intended to facilitate effective routing operation in the SoC board. Crossbar Switches, arbiters, buffers, a routing logic, and Network Interconnects are incorporated in NoC router. Priority based Round Robin Arbiter (RRA) is used to design traditional unidirectional router. In unidirectional router area utilization has been consumed and it determines priority by the use of delay factor, which guarantee from different input channels. Moreover if any path failure appears, router cannot route the information across additional output channel. To conquer this difficulty, a new bidirectional NoC router with and without contention is projected, that provides small area and increased speed. Bidirectional router used to route the information from source channel to every destination channel. Therefore it overcomes dispute state and path malfunction troubles. If some path failure may occurs, directly it will take other path during the switch allocator. The projected architecture used to improve the speed of the interconnection link. Simulation result achieved by ModelSim6.3c and synthesis is approved out by Xilinx 12.4. Keywords SoC · Bidirectional NoC · Round robin arbiter (RRA)

4.1 Introduction The word ‘Vedic’ is ambitious from word ‘Veda’ which is antique warehouse of all familiarity. To overcome extensive calculation time by minimizing the delay of an function that can performed by vedic mathematics. In arithmetic operation multiplication plays a vital role and multiplication is a dominant factor in Digital Signal Processing Applications, which is used to increase the speed of the processing elements. Multipliers mainly used to provide high performance structures such as FIR Y. Sougoumar (B) · Tamilselvan Pondicherry Engineering College, Pondicherry, India e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_4

31

32

Y. Sougoumar and Tamilselvan

filter Microprocessors, Digital Signal Processors, etc. Arithmetic operation basically used to achieve higher throughput in real and imaginary time signal processing applications. Vedic multiplier is the quick and flat power multiplier. This multiplier also reduces the complexity, execution time of the system. VLSI developers mostly like the design of System on Chip that depends on extensive IP core reprocess. System on Chip takes interconnection architecture and interfaces to peripheral devices. In large number of IP cores integrated by NoC interconnection methods and then the router is homogeneous or heterogeneous. In SoC the router contains same buffer size to every channel that router called homogeneous router. SoC mainly used to combine various types of Processor cores and data memory units of various sizes can be used to improve the performance and flexibility of the system. Reconfiguration allocates high performance at low power and it mainly used in embedded system and mobile applications. Reconfiguration architecture mainly used to combine the software with high performance hardware [1]. FPGA also carry partial reconfiguration and increased number of transistors incorporated into a single device. Integrated technique has been used to fabricate the system on a single chip. In proposed method is the implementation of bidirectional NoC router without as well as with contention for CGR design by using Vedic multiplier that is used to provide reduction in power, area and delay applications.

4.2 Related Works Qian et al. [2] proposed a flit-level get faster approaches empower flit surrounded by the similar packet to be transmitted concurrently. Writing and well reading 2 flits as of the identical virtual channel at similar time, which will be depend on input buffer design. Lan et al. [3] explained bidirectional NoC design with dynamic self-reconfigurable channel. Transmit flits are dynamically self-reconfigured that will used in Bi NoC for communication channels. This system provides enhancement in bandwidth utilization and its performance. Yao et al. [4] proposed a low-power dependable CGR design processor and its irradiation examinations. Numerous fields of computing as well as information intensive approaches namely baseband communication, cipher processing and computer vision have been appealed in coarse grained reconfigurable architectures. Kim et al. [5] projected tremendous throughput information data mapping for CGR designs. Main aim of the project is to translate the user programmable redundancy that consumes the energy and it will increase the vulnerability to single event effects and this scheme limits power of the system. Tran and Baas [6] proposed attaining huge performance Noc with shared-buffer routers. Buffers committed their output and input ports for provisionally accumulating packets. Buffers, unfortunately consume more area and power of the system. But, not all input ports of routers hold arriving packets required to be transmitted

4 An Implementation of Bidirectional NOC Router …

33

concurrently. Hence, a big number of buffer queue in system are unoccupied and additional queues are frequently active. Killian et al. [7] proposed Smart durable Noc and it is depend on original fault detection methods appropriate for dynamic NoCs. This technique offers networked detection of information packet and adaptive routing technique errors. Both the methods are clever to differentiate stable and brief errors as well as restrict exactly the location of the error blocks in NoC routers, whereas protecting the output and time period. Kim and Mahapatra [8] explained to achieve elevated performance and flexibility coarse-grained reconfigurable design has been used. Dynamic reconfiguration in each cycle can achieve by configuration cache. But, dynamic reconfiguration provides more power consumption and power utilization has been reduced by configuration cache. Tuwanuti and Thongbai [9] proposed completion of vedic multiplier method on multicore processor. Vedic multiplier used to calculate popular and universal acceptation in tutor school. In science and IT field vedic can be used. In research filed vedic arithmetic is used to speed up efficient of calculation and then long digits are splitted into sub blocks. Toro et al. [10] proposed division operation based on Vedic mathematics. Basic arithmetic operations are addition, subtraction and multiplication that should be invented by vedic mathematics for different approaches such as RSA encryption and decryption algorithm. But, proposed method focuses on division operation that used in image processing, networking, signal processing, etc. Oh et al. [11] explained proficient implementation of Stream Graphs on CGR structures. In arithmetic operation CGRAs offer energy efficiency and it accelerated in DSP and multimedia applications. Because this method allocutions two problem, they are more-buffer trouble and control overhead trouble, in result to reduces the execution of stream graphs with low cost. Cao et al. [12] proposed the scheme for multimedia applications namely context management for CGR structures. Implementation of computing-intensive approaches acceptable from coarse grained reconfigurable design. Memory overhead has been reduced by context cache without configuration performance degeneration and proposed context cache used to get better the configuration performance of the base Coarse Grained Reconfigurable Architecture.

4.3 32 × 32 Vedic Multiplier Vedic mathematics mainly used in estimation time used to reduce the time delay required for the function to be implemented. Vedic multiplier contains many bits of data such as 2-bit, 4-bit ….n-bits. This proposed method 32-bit Vedic multiplier has been used. Figure 4.1 illustrates the block diagram of 32 bit Vedic multiplier. 32 bit Vedic multiplier takes 32 bit input and that input send to 16 bit Vedic multiplier. Vedic multiplier performs mathematical operation and then input will

34

Y. Sougoumar and Tamilselvan

Fig. 4.1 32 × 32 vedic multiplier

give to 32 bit ripple carry adder that can do some addition operation. From the adder 64 bits sum and 1 bit carry out has been received.

4.4 CGRA Architecture Granularity is used to classify the Reconfigurable structures and it shows which design is best for reconfiguration. Fine-grained, coarse-grained, medium-grained and mixed-grained are classified by Reconfiguration structural design. Bit level hardware reconfigurable system called fine-grained structural design and word level is called coarse-grained reconfigurable hardware. Fine grained design are difficult to design and but more flexible than coarse grained hardware. However, coarse grained designs are faster and more efficient and CGRA scheme will improve the performance and decrease the power utilization of the reconfigurable organization is shown in Fig. 4.2.

4 An Implementation of Bidirectional NOC Router …

35

Fig. 4.2 General CGRA architecture

4.4.1 Representation of Processing Elements (PE) The CGRA scheme has been intended by the array of 4 bit PEs. Opcode is used to perform array processing elements and that are1. Encryption, 2. Decryption, 3. Vedic Multiplier, 4. FIR filters, 5. Comparator, 6. FFT processor, 7. Encoder, 8. Decoder, 9. Convolution, 10. Factorial, 11. Cyclic redundancy check, 12. ALU, 13. Modulation, 14. Demodulation. These are processing portions used in the CGRA architecture. NoC router performs a very important responsibility in SoC based approaches. Usually routing process is complex to complete within the Soc Chip. NoC router takes following mechanism, NI, buffer, routing logic, arbiters, and crossbar switches. Two types of routers are unidirectional router and bidirectional router (Fig. 4.3). In this routing path failure may occur, deceased lock difficulty and live lock difficulty. Routing logic is used to generate the route between the origin and the end. Logic of routing mainly used to avoid deadlock, live lock and starvation difficulty. Cyclic need between hubs expecting contact to the group of sources is called as deadlock. Procedure of moving the packet to the system with no development

36

Y. Sougoumar and Tamilselvan

In-out control

Channel_req

arb_req Arbiter

Channel B (I/II/II/IV/V) In-out select

Routing Module

Channel A(II)

Channel B (I/II/II/IV/V)

Routing Module

Routing Module

Channel A(III)

Channel B (I/II/II/IV/V)

Routing Module

Routing Module

Channel A(IV)

Input SRAM

In-out port

Input SRAM

In-out select

Cross bar

In-out port

In-out select

Channel B (I/II/II/IV/V) In-out select

In-out port

Input SRAM

In-out port

Input SRAM

In-out select

Input SRAM

In-out select

Routing Module

Routing Module

Channel A(V)

Input SRAM

In-out port

In-out port

In-out select

Channel B (I/II/III/IV/V) In-out select

In-out port

Input SRAM

Input SRAM

In-out select

Routing Module

Routing Module

Input SRAM

In-out port

In-out port

Routing Module

Round Robin

Channel_req

In-out select

In-out control

Channel A (I)

Input SRAM

Output req

Channel Control Module

In-out port

Input req

Fig. 4.3 Architecture of Bi-directional NoC router

towards their goal is called as live lock. When the output channel is apportioned to additional bundle is called as starvation problem, at that time packet will requesting buffer. Three criteria’s are there in routing algorithm that are, (a) how the path defined, (b) where the routing choices are taken, (c) length of path, First in First out (FIFO) buffers, crossbar switches and Round Robin arbiter are included in Unidirectional router. Data can be accessed based on that priority in Arbiter. In the design high preference information will be steered at first. Normally buffer is a temporary storage mechanism. Packet or data temporally stored by the use of FIFO buffer.

4 An Implementation of Bidirectional NOC Router …

37

Table 4.1 Comparison between modified Russian peasant multiplier and proposed vedic multiplier Parameters noticed

Modified Russian peasant multiplier

Proposed vedic multiplier

Look up Tables (LUTs)

688

719

Number of occupied slices

395

64

Delay (ns)

6.21

5.143

Power (W)

0.307

0.22

4.5 Result and Discussion The organization of coarse grained reconfigurable design and fine grained reconfigurable design was intended by using Verilog HDL. The simulation results of the CGRA scheme has been achieved with the help of Modelsim XE 6.3C and those the results has been synthesized and determined by using Xilinx ISE 10.1. Each processing element restrictions are estimated independently and that is illustrated in the Table 4.1. Table 4.1 illustrates the comparison results of Modified Russian Peasant Multiplier and Proposed Vedic Multiplier.

4.6 Conclusion Single In this paper, the Coarse grained and the Fine grained reconfigurable design was designed by using Vedic multiplier. The flexible reconfigurable architecture for efficient realization is appropriate for applications. In the structural design various processing elements were designed for different applications. Power utilization is extremely critical for the coarse grained and fine grained reconfigurable design. Unnecessary power consumption by the circuit can be avoided. It offers 23.51% reduction of power utilization compared to the existing design. Delay is another important parameter communication between the processing elements. Delay of the path may be minimized by the technique of bi-directional routing. It successfully minimized occurring of delay in the design. The proposed method offers 17.18% of delay minimized when compared to the conventional technique. The synthesized report of the design is taken from Xilinx ISE.

References 1. Ramesh, G.P.: Performance Analysis of Traffic with Optical Broker for Load Balancing and Multicasting in Software Defined Data Center Networking (2017) 2. Abbas, S.M., Tsui, C.Y.: Fsnoc: a flit-level speedup scheme for network on-chips using selfreconfigurable bidirectional channels. IEEE Trans. Very Large Scale Integr. (VLSI) Syst. 23(9), 1854–1867 (2015). https://doi.org/10.1109/tvlsi.2014.2351833

38

Y. Sougoumar and Tamilselvan

3. Lan, Y.C., Lin, H.A., Lo, S.H., Hu, Y.H., Chen, S.J.: A bidirectional NoC (BiNoC) architecture with dynamic self-reconfigurable channel. IEEE Trans. Comput. Aided Des. Integr. Circuits Syst. 30(3), 427–440 (2011). https://doi.org/10.1109/TCAD.2010.2086930 4. Yao, J., Saito, M., Okada, S., Kobayashi, K., Nakashima, Y.: EReLA: a low-power reliable coarse-grained reconfigurable architecture processor and its irradiation tests. IEEE Trans. Nucl. Sci. 61(6), 3250–3257 (2014). https://doi.org/10.1109/TNS.2014.2367541 5. Kim, Y., Lee, J., Shrivastava, A., Yoon, J.W., Cho, D., Paek, Y.: High throughput data mapping for coarse-grained reconfigurable architectures. IEEE Trans. Comput. Aided Des. Integr. Circuits Syst. 30(11), 1599–1609 (2011). https://doi.org/10.1109/TCAD.2011.2161217 6. Tran, A.T., Baas, B.M.: Achieving high-performance on-chip networks with shared-buffer routers. IEEE Trans. Very Large Scale Integr. (VLSI) Syst. 22(6), 1391–1403 (2014). https:// doi.org/10.1109/tvlsi.2013.2268548 7. Killian, C., et al.: Smart reliable network-on-chip. IEEE Trans. Very Large Scale Integr. (VLSI) Syst. 22(2), 242–255 (2014). https://doi.org/10.1109/tvlsi.2013.2240324 8. Kim, Y., Mahapatra, R.N.: Dynamic context compression for low-power coarse-grained reconfigurable architecture. IEEE Trans. Very Large Scale Integr. (VLSI) Syst. 18(1), 15–28 (2010). https://doi.org/10.1109/tvlsi.2008.2006846 9. Tuwanuti, P., Thongbai, N.: Implementation of vedic multiplier technique on multicore processor. In: TENCON 2014–2014 IEEE Region 10 Conference, pp. 1–6, IEEE (2014). https:// doi.org/10.1109/tencon.2014.7022325 10. Toro, S., Patil, A., Chavan, Y.V., Patil, S.C., Bormane, D.S., Wadar, S.: Division operation based on Vedic mathematics. In: 2016 IEEE International Conference on Advances in Electronics, Communication and Computer Technology (ICAECCT), pp 450–454, IEEE (2016). https:// doi.org/10.1109/icaecct.2016.7942630 11. Oh, S., Lee, H., Lee, J.: Efficient execution of stream graphs on coarse-grained reconfigurable architectures. IEEE Trans. Comput. Aided Des. Integr. Circuits Syst. 36(12), 1978–1988 (2017). https://doi.org/10.1109/TCAD.2017.2682645 12. Cao, P., Liu, B., Yang, J., Yang, J., Zhang, M., Shi, L.: Context management scheme optimization of coarse-grained reconfigurable architecture for multimedia applications. IEEE Trans. Very Large Scale Integr. (VLSI) Syst. 25(8), 2321–2331 (2017). https://doi.org/10.1109/tvlsi. 2017.2695493

Chapter 5

Breast Cancer Classification Using Tetrolet Transform Based Energy Features and K-Nearest Neighbor Classifier A. Amjath Ali, Suman Mishra and Bhasker Dappuri Abstract The cancer that develops in the breast tissue is referred to as breast cancer. The cancer in the breast could be sometimes symptomatic and are identified by self-examination of the breast or by a physician, whereas in certain cases there could be no symptoms at all. However, signs like a lump in the breast, change of size of the breast, dimpling and fluid discharge from the nipple are some of the symptoms which are cause of grave concern. The early diagnosis of the disease is the key to combat the deadly disease paving the path for hope of life. Mammography is a very popular technique that is used for the early diagnosis of breast cancer. In this study, a technique for breast cancer classification in digitized mammogram is put-forth employing tetrolet transform based energy features and K-Nearest Neighbor (KNN) classifier. The breast mammogram images of benign and malignant category are decomposed into sub-band coefficients using tetrolet transform and the energy features are extracted. These extracted features are given as input to the KNN classifier. Results show better classification accuracy in the breast cancer images using tetrolet transform based energy features and KNN classifier. Keywords Breast cancer · Tetrolet transform · Energy features · KNN classifier

5.1 Introduction Breast cancer is a deadly disease and is responsible for loss of lives among female population in particular, world-wide. Even though researchers have not been able to attribute any particular reason for the disease, it is of paramount importance that A. Amjath Ali (B) Department of Electrical and Electronics Engineering, Ibra College of Technology, Ibra, Oman e-mail: [email protected] S. Mishra · B. Dappuri CMR Engineering College, Hyderabad, India e-mail: [email protected] B. Dappuri e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_5

39

40

A. Amjath Ali et al.

the disease is detected at its early stage for cancer prognosis leading to successful treatment and often late detection of the disease resulting casualty. Among several modalities of breast screening, mammography has attributed lot of attention because of its capability of detecting the cancer at very early stage. Over the decade several researchers have contributed phenomenally into the development of algorithms for breast cancer classification, some of those are reviewed here. Mass classification of breast image by deep texture representation is discussed in [1]. Initially the local binary pattern and gray level co-occurrence matrix is used to extract the features from the mammographic breast cancer images. The classification is made by six classifiers: classifier via clustering, bayes network, simple decision table, decision stump, decision table J48 and random tree. Comparison of machine learning classifiers for breast cancer diagnosis is discussed in [2] for feature selection. The input breast cancer image features are selected by recursive feature elimination, recursive feature elimination cross validation and principal component analysis. The classification is made by Support Vector Machine (SVM), decision tree, adaboost classifier and random forest classifier. Hybrid approach for classification of mammograms is discussed in [3], using Convolution Neural Network (CNN) and radial basis functions using SVM classifier. The input mammogram image features are extracted by CNN. The classification is made by SVM classifier based on radial basic function. Breast cancer classification with electronic medical records using hierarchical attention bidirectional networks is discussed in [4]. The breast image features are extracted by three hierarchical attentions of bidirectional networks. Bidirectional recurrent neural network is used for classification. Automatic breast cancer detection using digital thermal images are discussed in [5]. Initially from the breast mammogram image, statistical features are extracted by gray level co-occurrence matrices. Finally, classification is made by back propagation neural network. Automatic breast cancer classification is discussed in [6]. The breast image features are extracted by hybrid features like completed local binary pattern, singular value decomposition and wavelet transform. SVM is used for classification. Mammogram based cancer detection is discussed in [7] using deep CNN. The input breast image features are extracted by deep CNN. Binary classifier is used for the classification. Classification of breast cancer using contourlet transform domain is discussed in [8]. The breast image features are extracted by contourlet transform. The classification is made by artificial neural network, k-nearest neighbor and SVM. Breast cancer images classification with deep CNN is discussed in [9]. Initially deep CNN is used to extract the breast image features. The classification is made by kernel ensemble method by using tensor flow framework with back propagation training. CNN for breast cancer classification is discussed in [10]. The input breast image features are extracted by CNN and classification is made by binary classifier. In this paper, a technique for breast image classification is presented based on the tetrolet transform with energy feature and KNN classifier. The organization of the paper is as follows: Section 5.2 describes the methods and materials for breast image classification system. Section 5.3 gives the results of breast image classification

5 Breast Cancer Classification Using Tetrolet Transform …

41

system in terms of accuracy, sensitivity and specificity. The last section presents the conclusion of the breast image classification system.

5.2 Methods and Materials In this study, a technique for breast cancer classification is presented. Figure 5.1 demonstrates the classification of breast cancer classification system. The implementation of this technique includes energy based tetrolet transform decomposition and KNN classification. Tetrolet Transform Tetrolet transform is a local, scalable and effective algorithm. In tetrolet transform image matrix is separated into blocks consisting of 4 × 4 pixels. If the image size cannot be separated by four, zero padding should be applied. The tetrominoes blocks are made by four equal sized pixels that are connected to each other. The five basic tetrominoes squares are given in Fig. 5.2. Considering, tetrolet transform is applied to an image k = {(u, v) : u, v = 0, . . . , V − 1} be the set of index of an image k = f (u, v) where V = 2 P , and neighborhood index is v˜ of index (u, v) is a vertex or boundaries. The tetrolet equation can be defined as given in the Eq. 1, v˜ (u, v) = {(u − 1, v), (u + 1, v), (u, v − 1), (u, v + 1)} Fig. 5.1 Breast cancer classification system

Input breast image

Tetrolet Transform

Energy features

KNN Classifier

Benign

Malignant

(1)

42

A. Amjath Ali et al.

Fig. 5.2 Five shapes of tetrominoes with 4 × 4 pixels

The input breast cancer mammographic images are decomposed by tetrolet transform to produce sub-band coefficients. Tetrolet Based Energy Features The sub-band coefficients are produced by the tetrolet transform and the energy feature is obtained. Energy features are used in other fields like glaucoma classification [11]. In signal processing Wavelets have been employed in many applications [12]. The assumptions of energy-based methods in frequency or spatial domains are the various texturepatterns that are presented in different energy distributions. Consider ing the energy E g is the repetitions of pixel pair that measures the image uniformity, while the pixels are similar to each other, the value of pixel is more. The computation of tetrolet transform based energy feature is done using Eq. 2,   P−1 P−1   Rk,l E g (I ) = 

(2)

k=0 l=0

where E g (I ) is input image, Rk,l be the subband coefficient at location (k, l) and P be the size of the sub bands. KNN Classification KNN classification is based on data measurement storing all available instances. It discovers the nearest neighbor in particular instance of time. The unknown instances are identified by the known neighbor. The KNN classifier rule has a class that indicates the K neighbor. The K indicates any number of neighboring pixels, such as K = 1, 2, 3, 4, . . . , n, where n referred as number of cases. The output of the KNN classifier is measured by the Euclidian distance. Considering k = (m 1 , n 1 ) and l = (m 2 , n 2 ) are two points. The Euclidian distance between these two points is given in Eq. 3,

5 Breast Cancer Classification Using Tetrolet Transform …

43

 (m, n) =

(m 1 − m 2 )2 (n 1 − n 2 )2

(3)

The breast image classification for tetrolet transform based energy features are classified by KNN classifier.

5.3 Results and Discussion A set of 50 breast cancer mammogram images of benign and malignant categories are selected from skin image database for ascertaining the performance of the system. The sizes of images are 256 × 256 pixels. Figures 5.3 and 5.4 shows the benign and malignant breast cancer images from the database. The performance of the system is validated through the information provided in the breast image database. Tetrolet transform is applied and energy features are extracted and classified by KNN classifier. The performance of the system is evaluated by tetrolet transform based energy features along with KNN classifier. Table 5.1 shows the accuracy, sensitivity and specificity obtained at 1 to 4 levels of Tetrolet transform based energy features. From Table 5.1, it is observed that at 3rd level of tetrolet transform based energy features produce the higher accuracy of 92% and their sensitivity and specificity are 88 and 96%. Figure 5.5 shows the graphical representation of the breast cancer

Fig. 5.3 Benign images for breast cancer image classification

44

A. Amjath Ali et al.

Fig. 5.4 Malignant images for breast cancer image classification Table 5.1 Tetrolet transform levels with KNN classifier accuracy, sensitivity and specificity levels

Tetrolet transform Levels

Accuracy (%)

KNN classifier Sensitivity (%)

Specificity (%)

1

68

60

76

2

76

76

76

3

92

88

96

4

88

84

92

Performance of Tetrolet based Energy Features based on Breast Cancer Classification 100 80 Sensitivity

60

Specificity

40

Accuracy

20 0

Level 1

Level 2

Level 3

Level 4

Fig. 5.5 Graphical representation of the performance of tetrolet based energy features

5 Breast Cancer Classification Using Tetrolet Transform …

45

Fig. 5.6 ROC curve for tetrolet transform based energy features for breast cancer image classification

classification system. Figure 5.6 shows the ROC curve for tetrolet based energy features at four levels.

5.4 Conclusion An efficient method for breast cancer image classification using mammogram breast images is proposed. The system uses tetrolet transform based energy features along with KNN classifier. Tetrolet transform decompose the mammogram breast images into higher and lower frequency sub-bands. From the sub-band coefficients energy

46

A. Amjath Ali et al.

features are extracted. These extracted features serve as the input for the classification. KNN classifier is employed to classify the mammogram breast cancer images using the extracted energy features at four levels of tetrolet transform decomposition. Results show that the 3rd level produces better classification accuracy when compared with other levels.

References 1. Boudraa, S., Melouah, A., Merouani, H.F.: Deep texture representation for breast mass classification. In: 2018 International Conference on Signal, Image, Vision and their Applications (SIVA), pp. 1–4, IEEE (2018). https://doi.org/10.1109/siva.2018.8661052 2. Liu, B., Li, X., Li, J., Li, Y., Lang, J., Gu, R., Wang, F.: Comparison of machine learning classifiers for breast cancer diagnosis based on feature selection. In: 2018 IEEE International Conference on Systems, Man, and Cybernetics (SMC), pp. 4399–4404, IEEE (2018). https:// doi.org/10.1109/smc.2018.00743 3. Alkhaleefah, M., Wu, C.C.: A hybrid CNN and RBF-based SVM approach for breast cancer classification in mammograms. In: 2018 IEEE International Conference on Systems, Man, and Cybernetics (SMC), pp. 894–899, IEEE (2018). https://doi.org/10.1109/smc.2018.00159 4. Chen, D., Qian, G., Pan, Q.: Breast cancer classification with electronic medical records using hierarchical attention bidirectional networks. In: 2018 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), pp. 983–988, IEEE (2018). https://doi.org/10.1109/bibm. 2018.8621479 5. Soliman, O.O., Sweilam, N.H., Shawky, D.M.: Automatic breast cancer detection using digital thermal images. In: 2018 9th Cairo International Biomedical Engineering Conference (CIBEC), pp. 110–113, IEEE (2018). https://doi.org/10.1109/cibec.2018.8641807 6. Elelimy, E., Mohamed, A.A.: Towards automatic classification of breast cancer histopathological image. In: 2018 13th International Conference on Computer Engineering and Systems (ICCES), pp. 299–306, IEEE (2018). https://doi.org/10.1109/icces.2018.8639219 7. Salem, M.A.M.: Mammogram-based cancer detection using deep convolutional neural networks. In: 2018 13th International Conference on Computer Engineering and Systems (ICCES), pp. 694–699, IEEE (2018). https://doi.org/10.1109/icces.2018.8639224 8. Kabir, S.M., Bhuiyan, M.I.H.: Classification of breast tumour in Contourlet transform domain. In: 2018 10th International Conference on Electrical and Computer Engineering (ICECE), pp. 289–292, IEEE (2018). https://doi.org/10.1109/icece.2018.8636769 9. Adeshina, S.A., Adedigba, A.P., Adeniyi, A.A., Aibinu, A.M.: Breast cancer histopathology image classification with deep convolutional neural networks. In: 2018 14th International Conference on Electronics Computer and Computation (ICECCO), pp. 206–212, IEEE (2018). https://doi.org/10.1109/icecco.2018.8634690 10. Angara, S., Robinson, M., Guillén-Rondon, P.: Convolutional neural networks for breast cancer histopathological image classification. In: 2018 4th International Conference on Big Data and Information Analytics (BigDIA), pp. 1–6, IEEE (2018). https://doi.org/10.1109/bigdia.2018. 8632800 11. Dua, S., Acharya, U.R., Chowriappa, P., Sree, S.V.: Wavelet-based energy features for glaucomatous image classification. IEEE Trans. Inf. Technol. Biomed. 16(1), 80–87 (2012). https:// doi.org/10.1109/TITB.2011.2176540 12. Ramesh, G.P., Malini, M., Professor, P.G.: An efficacious method of cup to disc ratio calculation for glaucoma diagnosis using super pixel. Int. J. Comput. Sci. Eng. Commun. 2(3) (2014)

Chapter 6

Bayesian Neural Networks of Probabilistic Back Propagation for Scalable Learning on Hyper-Parameters K. Thirupal Reddy and T. Swarnalatha Abstract Extensive multilayer neural systems prepared with back proliferation have as of late accomplished best in class results in some of issues. This portrays and examines Bayesian Neural Network (BNN). The work shows a couple of various uses of them for grouping and relapse issues. BNNs are included a Probabilistic Model and a Neural Network. The plan of such a plan is to join the qualities of Neural Networks and stochastic demonstrating. Neural Networks display ceaseless capacity approximates abilities. Be that as it may, utilizing back drop for neural networks adapting still has a few disservices, e.g., tuning a substantial figure of hyper-parameters to the information, absence of aligned probabilistic forecasts, and a propensity to over fit the preparation information. The Bayesian way to deal with learning neural systems does not have these issues. Nonetheless, existing Bayesian systems need versatility to expansive dataset and system sizes. In this work we present a novel versatile strategy for learning Bayesian neural systems, got back to probabilistic engendering (PBP). Like traditional back spread, PBP works by figuring a forward engendering of probabilities through the system and afterward completing a retrogressive calculation of inclinations. A progression of analyses on ten true datasets demonstrates that PBP is essentially quicker than different methods, while offering aggressive prescient capacities. Our examination additionally demonstrates that PBP-BNN gives precise appraisals of the back change on the system weights. Keywords Bayesian Neural Networks (BNN) · Probabilistic Backpropagation (Pbp) · Probabilistic Predictions

K. Thirupal Reddy (B) Department of Computer Science and Engineering, Bharat Institute of Engineering and Technology, Hyderabad, Telangana, India e-mail: [email protected] T. Swarnalatha Department of Computer Science and Engineering, Nalanda Institute of Engineering and Technology, Guntur, AP, India e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_6

47

48

K. Thirupal Reddy and T. Swarnalatha

6.1 Introduction The Stochastic models permit coordinate particular of a model with known cooperation between parameters to produce information. Amid the forecast stage, stochastic models create an entire back dissemination and deliver probabilistic certifications on the expectations. In this way BNNs are an extraordinary mix of neural system and stochastic models with the stochastic model framing the centre of this reconciliation. BNNs would then be able to deliver probabilistic assurances on its expectations and furthermore produce the circulation of parameters that it has gained from the perceptions. Further this zone is quickly making progress as a standard machine learning approach for various issues. The mean along with difference of the yield of unit j are characterized as mz j and v z j, individually. The signify in addition to difference of the initiation or contribution in favour of component j are characterized as mama j also v a j, individually [1]. We encompass with the aim of, as a result of the ReLU enactment work,     m zj =  α j m aj + υ aj γ j  1 mz mw , |I ( j)| i∈I ( j) i j,i    2

2 1 υ aj = m iz υ wj,i + υiz m wj,i + υiz υ wj,i |I ( j)| i∈I ( j)

(6.1)

m aj = √

(6.2)

Multilayer neural systems the back propagation learning rule. This has additionally broadly been acknowledged. Afterward, different quickened adaptations of the standard have been explained that accelerate the learning procedure. This is to be accomplished by changing the system weights as indicated by a limitation change calculation, customarily carry out by a back propagation calculation that is measured as a speculation of the delta regulation [2].

6.2 Existed Methods 6.2.1 Radial Foundation Purpose Networks Occupation estimate utilizing confined premise capacities is the after effect of the exploration work done by potential capacity way to deal with example acknowledgment. This is an outcome of utilizing the sigmoid capacity as the system initiation work with its similarity to the unit step work, which is a worldwide capacity. The privately limited premise capacities ought to by and large have the shape.

6 Bayesian Neural Networks of Probabilistic Back Propagation …

F(x) =

n 

wi ϕ(x − xi )

49

(6.3)

i=1

The most critical issue here is the choice for each

−xi − ci 2 Fi = exp σi2

(6.4)

Neuron in the concealed level the middle c I, also the extend approximately the inside V I this is for the most part done utilizing the k-implies bunching calculation, which is equipped for deciding the ideal position of focuses. The system preparing procedure for the most part incorporates two preparing stages: x instatement of RBF focuses, for example utilizing unsupervised grouping techniques x yield weight preparing of the RBF utilizing a versatile calculation to evaluate its suitable qualities. An Elman arrange is a 4-layer organize made out of info layer, shrouded layer, yield layer and the setting layer, the hubs of which are the one-advance defer components implanted into the nearby input ways (Fig. 6.1). On account of a steady system this capacity must diminishing with time and at last achieve its base, or it’s esteem stays consistent. Is utilized, the adjustments in time are constantly depicted after the condition. f (x) = κ

Fig. 6.1 RBF network

1 1 + e−x

 uj du j = w ji yi − + Uj, dt D j i

(6.5) (6.6)

50

K. Thirupal Reddy and T. Swarnalatha

Fig. 6.2 Fully connected recurrent neural network

Fig. 6.3 BNN-BP

where t, is a consistent positive esteem, I y is the yield estimation of the unit I, Dj is the factor controlling the sigmoid rot opposition, and U j is the outer contribution to the unit j. The subsequent vitality work for this situation is characterized by E =−

 1  wi j u i u j − u i Ui 2 i j i

(6.7)

For the preparation of repetitive systems, proposed a general structure like that utilized for preparing feed forward systems, called back propagation through time. This, in any case, isn’t constantly acceptable. There are two fundamental learning standards for intermittent systems: x settled point learning, through which the system achieves the recommended enduring state in which a static info example ought to be put away x direction learning, through which a system Fig. 6.3 out how to pursue a direction or a grouping of tests after some time, which is important for

6 Bayesian Neural Networks of Probabilistic Back Propagation …

51

transient example acknowledgment, multistep expectation, and frameworks control. For direction learning, both the back propagation through time and the real-time repetitive learning are suitable [3]. In view of the after-effects of this work, have worked out a back engendering calculation for systems, the initiation capacity of which complies with the transformative law (Fig. 6.2), ⎛ ⎞  dvi = −vi + g ⎝ wi j v j + xi ⎠ τ (6.8) dt j        αMCEI (x;D, R) = αEI x;D ∪ xi , yit xi ∈ R p yit |xi , θ P(θ |D)dθ dy θ,y

≈

M    1   αEI x;D ∪ xi , yit xi ∈ R M k=1

(6.9)

A specific sort of intermittent systems that don’t comply with the limitations of the Hopfield systems are the dynamic repetitive systems, proposed for portrayal of frameworks whose inner state changes with time [4]. They are especially fitting for demonstrating of nonlinear unique frameworks, for the most part characterized by the state-space conditions [5]. X (k + 1) = f (x(k), u(k)) Y (k) = C x(k.)

(6.10)

6.3 Proposed Method (BNN-PB) Where {(xi, ky t I)} xi ∈ R is the arrangement of at present running capacity assessments (and their anticipated targets) and where, practically speaking, we just take M = 1 test. Note that the calculation of α EI inside the total again requires different examples from the HMC strategy [6]. With respect to EI we can separate through the calculation of Eq. (6.3) and augment it utilizing inclination rising. We characterize a variation of our calculation for settings in which we can play out numerous assessments of ft. in parallel. Using such parallel (and offbeat) work assessments in a principled way for BO is non-unimportant as we in a perfect world might want to underestimate over the results of as of now running assessments while recommending new parameters x for which we need to question to accomplish this proposed a securing capacity which we allude to as Monte Carlo EI α MCEI that we receive here for our model. Formally, we gauge α MCEI as k t yi

  ∼ p yit |xi , θ p(θ |D),

(6.11)

52

K. Thirupal Reddy and T. Swarnalatha

6.3.1 Computational Requirement For any BO technique it is vital to keep the computational necessities for preparing and assessing the model as a primary concern. To this end we need to draw a correlation between our technique and DNGO regarding computational expenses. To start with, SGHMC examining is comparably modest as standard SGD preparing of neural systems, i.e. preparing a DNGO demonstrate starting with no outside help and examining by means of SGHMC has comparative computational expenses. If we somehow managed to begin examining sans preparation with each approaching information point and would settle the quantity of MCMC ventures to be equal to K goes through the entire dataset then the computational multifaceted nature for testing would develop directly with the quantity of information point [7].

6.3.2 Obtaining Well Calibrated Uncertainty Estimates with Bayesian Neural Networks As made reference to in the principle paper, there exists a huge group of work on Bayesian strategies for neural systems. The last conduct is exemplified in Fig. 6.1 (left) where we relapsed the sink work from 20 perceptions with a two layer neural system (50 tanh units each) utilizing our usage of the Bayes by Back prop (BBB) approach from Blundell et al. Multilayer neural systems prepared with back spread have best in class results in numerous relapse issues Multilayer neural systems prepared with back propagation have cutting edge results in numerous relapse issues, yet they Require tuning of hyper-parameters. Multilayer neural systems prepared with back propagation have cutting edge results in numerous relapse issues; however they require tuning of hyper-parameters. These are Can be influenced by over fitting issues [8]. Multilayer neural systems prepared with back-spread have best in class results in numerous relapse issues, yet they require tuning of hyper-parameters. On a fundamental level, the Bayesian methodology can take care of these issues, yet most Bayesian techniques need adaptability. L layers with W = {Wl} L l = 1 as weight networks and yield Z. • • • • • • • • •

L layers with W = {Wl} L l = 1 as weight lattices and yield ZL. ReLUs actuations for the shrouded layers: a(x) = max(x, 0). L layers with W = {Wl} L l = 1 as weight frameworks and yield ZL. ReLUs actuations for the shrouded layers: a(x) = max(x, 0). The probability: p(y|W, X, γ) = QN n = 1 N (yn|zL(xn|W), γ − 1) ≡ fn. L layers with W = {Wl} L l = 1 as weight grids and yield ZL. ReLUs initiations for the concealed layers: a(x) = max(x, 0). The probability: p(y|W, X, γ) = QN n = 1 N (yn|zL(xn|W), γ − 1) ≡ fn. The priors: p(W|λ) = QL l = 1 QVl i = 1 QVl − 1 + 1 j = 1 N (wij, l |0, λ − 1) ≡ gk, • L layers with W = {Wl} L l = 1 as weight networks and yield ZL.

6 Bayesian Neural Networks of Probabilistic Back Propagation …

53

Fig. 6.4 Multi-layer network

• ReLUs enactments for the shrouded layers: a(x) = max(x, 0). • The probability: p(y|W, X, γ) = QN n = 1 N (yn|zL(xn|W), γ − 1) ≡ fn. • The priors: p(W|λ) = QL l = 1 QVl i = 1 QVl − 1 + 1 j = 1 N (wij, l |0, λ − 1) ≡ gk, p(λ) = Gamma(λ|α λ 0, βλ 0) ≡ h, p(γ) = Gamma(γ|α γ 0, βγ 0) ≡ s. • L layers with W = {Wl} L l = 1 as weight frameworks and yield ZL. • ReLUs initiations for the shrouded layers: a(x) = max(x, 0). • The probability: p(y|W, X, γ) = QN n = 1 N (yn|zL(xn|W), γ − 1) ≡ fn. • The priors: p(W|λ) = QL l = 1 QVl i = 1 QVl − 1 + 1 j = 1 N (wij, l |0, λ − 1) ≡ gk, p(λ) = Gamma(λ|α λ 0, βλ 0) ≡ h, p(γ) = Gamma(γ|α γ 0, βγ 0) ≡ s (Fig. 6.4). • L layers with W = {Wl} L l = 1 as weight lattices and yield ZL. • ReLUs actuations for the shrouded layers: a(x) = max(x, 0). • The probability: p(y|W, X, γ) = QN n = 1 N (yn|zL(xn|W), γ − 1) ≡ fn. • The priors: p(W|λ) = QL l = 1 QVl i = 1 QVl − 1 + 1 j = 1 N (wij, l |0, λ − 1) ≡ gk, p(λ) = Gamma(λ|α λ 0, βλ 0) ≡ h, p(γ) = Gamma(γ|α γ 0, βγ 0) ≡ s. • The back estimate is q(W, γ, λ) = hQL l = 1 QVl i = 1 QVl − 1 + 1 j = 1 N (wij, l |mij, l, vij, l) I Gamma(γ|α γ, βγ) Gamma(λ|α λ, βλ).

6.3.3 Neural Networks Including More Than One Hidden Layer Report on behalf of the strategies B.P moreover P.B.P, utilizing neural systems are among 2, 3 as well as 4 shrouded layers. We utilized systems amid 50 units in each one concealed sheet, aside from in the datasets sound moreover video, where we utilized 100 examples. Table 6.1 demonstrates the normal test RMSE and the relating standard blunders gotten by PBPx and BPx, where x is the quantity of concealed layers in the system. PBP has the best by and large prescient execution, with PBP2 accomplishing the best outcomes in 5 datasets. Note with the purpose of the ideal number of concealed layers in PBP is issue subordinate [9]. In informational

54

K. Thirupal Reddy and T. Swarnalatha

Table 6.1 Average test R.M.S.E in the experiment with bottomless neural networks Dataset

BP1

BP2

BP3

BP4

PBP1

PBP 2

PBP3

BNN PBP4

Boston

3.228 ± 0.1951

3.185 ± 0.2365

3.019 ± 0.1848

2.874 ± 0.1570

3.014 ± 0.1800

2.795 ± 0.1590

2.938 ± 0.1645

3.088 ± 0.1519

Concrete 5.977 ± 0.2207

5.396 ± 0.1273

5.568 ± 0.1271

5.530 ± 0.1390

5.667 ± 0.0933

5.241 ± 0.1164

5.732 ± 0.1075

5.956 ± 0.1597

0.676 ± 0.0367

0.628 ± 0.0278

0.667 ± 0.0321

1.804 ± 0.0481

0.903 ± 0.0482

1.237 ± 0.0592

1.176 ± 0.0552

Kin8 nm 0.091 ± 0.0015

0.073 ± 0.0009

0.071 ± 0.0006

0.071 ± 0.0009

0.098 ± 0.0007

0.071 ± 0.0005

0.073 ± 0.0007

0.075 ± 0.0008

Naval

0.001 ± 0.0001

0.001 ± 0.0000

0.001 ± 0.0001

0.001 ± 0.0001

0.006 ± 0.0000

0.003 ± 0.0001

0.010 ± 0.0013

0.004 ± 0.0011

Power Plant

4.182 ± 0.0402

4.220 ± 0.0744

4.112 ± 0.0378

4.184 ± 0.0591

4.124 ± 0.0345

4.028 ± 0.0347

4.065 ± 0.0382

4.075 ± 0.0366

Protein

4.539 ± 0.0288

4.188 ± 0.0313

4.014 ± 0.0326

3.960 ± 0.0110

4.688 ± 0.0115

4.251 ± 0.0153

4.094 ± 0.0285

3.970 ± 0.0376

Wine

0.645 ± 0.0098

0.651 ± 0.0108

0.652 ± 0.0101

0.650 ± 0.0158

0.635 ± 0.0079

0.643 ± 0.0077

0.641 ± 0.0086

0.637 ± 0.0079

Yacht

1.182 ± 0.1645

1.542 ± 0.1920

l.I07 ± 0.0863

1.265 ± 0.1287

1.015 ± 0.0542

0.848 ± 0.0495

0.893 ± 0.0991

1.711 ± 0.2288

Year

8.932 ± NA

8.976 ± NA

8.933 ± NA

9.045 ± NA

8.869 ± NA

8.918 ± NA

8.874 ± NA

8.934 ± NA

Energy

l.185 ± 0.1242

indexes, for example, vector and edges of one single concealed cover is ideal, while in sound track we locate with the purpose of 4 shrouded layers is improved.

6.4 Results We assess the blunder in the second estimation performed in condition (6.2) in the above discourse. This guess comprises in supplanting the vector thickness with a Gaussian thickness that has a similar mean and difference. This guess turns out to be increasingly exact as the degrees of opportunity in the sound casings thickness increment [10, 11]. This determination frequently be the situation as we repeat greater than the information furthermore we decrease our vulnerability on the estimation of

6 Bayesian Neural Networks of Probabilistic Back Propagation …

55

Fig. 6.5 RMS error window

Fig. 6.6 MSE value

the commotion limitation. We assessed the relation mistake in log Z reasons by this estimation as BNN-PBP emphasizes more the information of the Boston Housing dataset in the investigations of Segment 4 in the fundamental record. The left plot in Fig. 6.1 demonstrates the blunder amid the initial 100 emphases of PBP greater than the personality data points. We preserve see with the aim of the mistake is little in the moment container. Specifically, at this phase we are exceptionally certain on the estimation of the commotion restriction in addition to the constraint _ and_ in the back guess take moderately elevated qualities. This expands the quantity of degrees of opportunity of the sample thickness in condition (6.1), enhances the nature of the Gaussian estimation (Fig. 6.5). Graphical Analysis See Figs. 6.6, 6.7 and 6.8.

6.5 Conclusion Using neural Network based back propagation method gave more error compared to BNN_PBP method. The percentage error is 10% decreased compared to previous methods. Finally using BNN_PBP method gives 100% efficiency. Using NN-DL BP

56

K. Thirupal Reddy and T. Swarnalatha

Fig. 6.7 BNN_PBP window

Fig. 6.8 BNN_PBP versus MSE

got better results. But, complexity is high. so, decreasing this we go with BNN_PBP technique. By using this algorithm efficiency is increased by 90% compared to existed method.

References 1. Aizerman, M.A., Braverman, E.M., Rozenoer, L.I.: Theoretical foundation of potential function method in pattern recognition. Autom. Remote Control 25, 917–936 (1964). https://scinapse. io/papers/1526146785 2. Akaike, H.: Statistical predictor identification. Ann. Inst. Stat. Maths. 22, 202–217 (1970) 3. Ramesh, G.P., Kumar, N.M.: Radiometric analysis of ankle edema via RZF antenna for biomedical applications. Wirel. Pers. Commun. 102(2), 1785–1798 (2018) 4. Chakraborty, K., Mehrotra, K., Mohan, ChK, Ranka, S.: Forecasting the behaviour of multivariate time series using neural networks. Neural Netw. 5, 961–970 (1992) 5. Cichocki, A., Unbehauen, R.: Neural Networks for Optimization and Signal Processing. Wiley, Chichester, West Sussex, UK (1993). https://www.wiley.com/en-ai/Neural+Networks+for+ Optimization+and+Signal+Processing-p-9780471930105 6. Cohen, M.A., Grossberg, S.: Absolute Stability of global pattern formation and parallel memory storage by competitive neural networks. IEEE Trans. Syst. Man Cybern. 13, 815–826 (1983). https://doi.org/10.1109/TSMC.1983.6313075

6 Bayesian Neural Networks of Probabilistic Back Propagation …

57

7. Cybenko, G.: Approximation by superposition’s of a sigmoidal function. Math. Control Signals Syst. 2, 303–314 (1989). https://doi.org/10.1007/BF02551274 8. Denton, J.W.: How good are neural networks for causal forecasting? J. Bus. Forecast. 14(2), 17–20 (1995). https://www.questia.com/library/journal/1P3-6802140/how-good-areneural-networks-for-causal-forecasting 9. Elman, J.L.: Finding structure in time. Cogn. Sci. 14, 179–211 (1990). https://doi.org/10.1016/ 0364-0213(90)90002-E 10. Fogel, D.B.: An information criterion for optimal neural network selection. IEEE Trans. Neural Netw. 2, 490–497 (1991). https://doi.org/10.1109/72.134286 11. Forster, W.R., Collopy, F., Ungar, L.H.: Neural network forecasting of short, noisy time series. Comput. Chem. Eng. 16(2), 293–297 (1992). https://doi.org/10.1016/0098-1354(92)80049-F

Chapter 7

Extensive Study on Antennae for IoT Applications T. Jayanthi, N. Sai Akhila and G. Pravallika

Abstract Present world is being ruled by Internet of Things (IoT). Where, wireless communication is the first choice of communication for IoT nodes and devices. Antenna system plays a vital role in providing communication between all IoT Nodes and Devices (IND), Antenna performance may seem unimportant. The whole idea of IoT is based on having many devices close together with inexpensive transceivers. Despite the close range and low cost, though, antenna performance still matters. In this paper we present an extensive study on various antennae and their performance for IND. Keywords IoT · IND · Antenna performance · Transceivers

7.1 Introduction Over In recent years there have been dramatic changes in human social life due to the Internet of Things (IoT) which connects every Device/object, every module and every device to a network. The number of devices connected to the internet is constantly increasing. In accordance with a Cisco report [1], it is predicted that in 2020, over 50 billion objects will be linked to the internet. The result is the diffusion of an IoT that completely changes the way people interact with their environment. In this environment antenna, its size and placing on the device/Object plays a vital role in providing communication between the objects working at various frequencies ranges from AF to RF.

T. Jayanthi (B) · N. Sai Akhila · G. Pravallika Department of ECE, CMR Engineering College, Hyderabad, India e-mail: [email protected] N. Sai Akhila e-mail: [email protected] G. Pravallika e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_7

59

60

T. Jayanthi et al.

The IOT is allowed by the several advances in smart sensors, RFID, and other communication technologies using internet protocols [2]. Electronic engineers are confronting with lots of challenges as more and more devices/Objects are connecting into Internet in wireless ways. With every physical thing being connected in IoT, wireless communication is the first choice of communication for Internet of Things node devices/objects. To make radio transmitters assembled in present equipment space and size are to be conscious parameters, also how to design and manufacture devices with increasingly smaller sizes has become a major challenge. As hardware industry always relies upon absolutely new silicon process innovation, ongoing years have seen progressively littler size of silicon chips. By incorporating MCU (micro programmed control unit) and RF front end into SoC (framework on chip) structure, space issue has been effectively settled for IoT implementation [3] As the size of IoT modules reduced by including more wireless technologies, so making antennas space is an important challenge. Therefore, designers of IoT-module antenna confront the confinements of keeping up sensible execution in regularly contracting impressions and under extraordinary impedance conditions. The advancement drift towards SoC lacks issue fathomed concerning physical structure of RF transmitter, that is, antenna. We for the most part leave antenna designer to clients or encourage them to get a simple to-utilize antenna module with incorporated antenna. An antenna system selection will be a significant part in devices/objects. Selecting the right antenna for an application plays a key role in design. So, creating effective antenna will decide to get better performance. A device of IoT requires engineers to observe various factors that include size of antenna, shape of antenna, and its placement. While designing antennae for IoT applications we need to consider the following parameters [4] • • • • • • •

To check Mounting option, whether it is RF connector mount, or PCB mount etc. Omni-directional or directional operation Frequency of operation Shape and size of antenna Coverage requirement Cost of Antenna Gain of Antenna.

In this paper, a brief description about the basic concepts of IoT and how to choose antenna for the IoT applications is presented. The main aim of the paper is to provide knowledge on various types of antennas that can be used for IoT applications. Section 7.2 of this paper deals with the various types of antennas and their layouts. Section 7.3 gives the comparison of various parameters of these antennas. Finally conclusion is provided in Sect. 7.3.

7 Extensive Study on Antennae for IoT Applications

61

7.2 Antennas for IoT Applications In this section various types of IoT antennas are presented. These help IoT module designers to make right choice of antenna for specified applications.

7.2.1 SIW Antenna A SIW antenna is a compacted on-body antenna mainly employed for applications using IoT. It is built against the substrate of fabric by darning the textile sheets that are conductive via posts. Figure 7.1 illustrates the SIW antenna’s geometry [5]. SIW antenna is created out of the top area and ground base through a T-formed opening on a substrate of fabric (tangent loss = 0.02 at 5.5 GHz, εr = 1.2). The thread with conductive one is employed for associating these two conductive textiles, shaping a SIW structure with cavity-backed supply and a slot for shining. Every modes of hybrid inside the primary cavity (dotted line) shows the virtual magnetic wall by using the technique of substrate half-mode integrated waveguide (SHMIW). By employing this method, the cavity can be sliced down into half. To accomplish higher gain and better front-to-back ratio (FBR), the stretching was carried out by 15 mm at bottom ground. Therefore; the antenna total dimension of is quite compressed size compared with usual wearable slot antennas. The return loss bandwidth 10-dB of the planned antenna entirely covers Wi-Fi band of 5-GHz and include a 6.08 dBi gain through 16.8 dB FBR at 5.5 GHz. This antenna Suitability for connection through on-body was confirmed by using a 2/3rd muscle equivalent spirit as the human body.

Fig. 7.1 Geometry of SIW antenna

62

T. Jayanthi et al.

Fig. 7.2 Layout design of RFID reader antenna

7.2.2 RFID Reader Antenna Antenna of RFID reader is a circularly polarized. The antenna design comprises of substrate, ground plane, and a design for circular patch. It is structured with ground fix over which a substrate is put on. At final stage of the substrate the design of circular patch is implicated. It consists of ground plane and the design of circular patch with 23.1 mm and 15.3 mm of radius respectively. They are made of conductors. Different conducting materials are used for both ground plane and design of circular patch. Feeding system used is a micro strip line center feed that is selected at the antenna frame. Maximum efficiency and a lot less losses can be achieved by this type of feeding method. The losses are low due to the implementation of circular polarization in the antenna. The fabrication process of antenna becomes easier with Micro-strip line feed than by using the method of coaxial feeding. The design layout of antenna is shown in Fig. 7.2 [6]. The antenna resonates at 2.45 GHz frequency. It is a frequency ISM band with little return loss experimented at about −40 dB. It is also experimented that the antenna has a 6.5 dBi gain and 7.44 dBi directivity. Hence this design of patch can be basically feasible to support IoT applications based on RFID. Radio frequency identity of an object is read with the efficiency of 81%.

7.2.3 UNB Miniature Antenna IoT systems require the less bandwidth and are explained by the transmission of small amounts of information and low transmission rate that certainly direct to the

7 Extensive Study on Antennae for IoT Applications

63

Fig. 7.3 Inverted F antenna structure

enhancement of ultra-narrow band (UNB) module. The compact inverted-F antenna structure is described in Fig. 7.3 [7]. It is contains a set of folded metallic wire prior to a ground plane of 0.5λ × 0.5λ that is shorted near the feed. The result of this structure conveys a few favorable circumstances. First, being established barely by metallic wires, no contact of dielectric substrate considered. These restriction have been fluctuated from a reference structure planned to resonate at 868 MHz. The deliberate model displays great impedance matching with |S11| < −6 dB from 866 to 871 MHz. In this band the efficiency of the total antenna that measured is around −2 dB, which is reasonable for applications using IoT and verifies the simulated results. Since the antenna proposed for wearable IoT device in that the antenna ground plane is get ready by a flexible tissue of metal.

7.2.4 Dual Band UWB Antenna Its antenna size is 24 mm × 26 mm. It contains micro-strip line-fed radiator as illustrated in Fig. 7.4 [8]. The semi-circular slot which generates WLAN of 5.8 GHz, and a slot that generates 3.5 GHz WiMAX. The dual-notched attributes decrease obstruction among the effectively existing UWB working devices while antenna shows a negative gain at notched bands. The UWB antennas that are dual notched is strongly notched at 5.11–6.4 GHz and 3.28– 3.85 GHz and frequency bands for WLAN and WiMAX respectively. The finalized antenna size is 24 mm × 26 mm by a micro strip line-fed ring radiator. It contains a one slot for WiMAX and other slot for WLAN. Working data transfer capacity of 117.67% is accomplished. This fundamental structure conveys the cost for assembling without negotiating the 98.585% efficiency. Considering the

64

T. Jayanthi et al.

Fig. 7.4 Structure of micro strip ultra wide band antenna

little size and more prominent working data transmission with those quick frequency characteristics it is great for IoT applications.

7.2.5 Compact Dual Band Antenna The antenna with a patch of micro strip has been viewing the capability to operate in twice frequency. The antenna is intended by using one main patch of micro strip and two slotted patches that are linked with the main patch. Figure 7.5 represents the structure of antenna [9].

Fig. 7.5 Structure of compact dual band antenna

7 Extensive Study on Antennae for IoT Applications

65

The antenna with micro-strip patch is made reconfigurable by linking the main patch to slotted antennas on both sides, working at 2.4 GHz resonance frequency. 2.4 GHz is the antenna’s operating frequency and it can be employed in lot applications. Various spaces are added to obtain the frequencies of multiple resonance at 2.4 GHz in the proposed antenna that satisfies the WLAN standards and it having return loss fewer than -10 dB. Several works will be performing towards antenna’s gain enhancement in future. Additionally, the calculation of every single electronic gadget is being reduced widely owing to growth in the technology era. Other than that, it is additionally basic to carefully choose the substrate, as its permittivity considerably affects the antenna performance.

7.2.6 Reconfigurable Patch Antenna The antenna is made out of a circular patch with a 58.8 mm diameter placed on a substrate of RO5880 by 0.787 mm thickness and 2.2 dielectric constant. This antenna resorts to three distinct perspectives to adjust its movement between 1.58 GHz, 2.4 GHz, and 868 MHz. The main strategy executed in the plan is the incorporation of a slot with dimensions of 30 mm × 2.4 mm as shown in Fig. 7.6 [10]. The second process is the usage of Vias for the antenna tune matching and force the antenna to work at lower frequency bands without expanding its physical parts [11, 12]. The coordination of DTC on the secured opening creates this structure’s 3rd part. The location of DTC beside the antenna slot is shown in Fig. 7.6.

Fig. 7.6 DTC location with the antenna’s slot

66

T. Jayanthi et al.

7.3 Conclusion This paper is a comparative study on various types of antennas used for IoT applications. Various parameters of antennas for IoT applications like operating frequency, gain, losses, efficiency etc., are studied. These antennas can operate with LoRa, Bluetooth, and the bands of GPS. The antenna additionally shows reduction of 50% mass at 868 MHz and 20% size at 1.58 GHz. By using DTC, the antenna is digitally reconfigured and its execution is tried for approval. These antennas may provide better future for IoT Applications

References 1. Lizzi, L., Ferrero, F., Monin, P., Danchesi, C., Boudaud, S.: Design of miniature antennas for IoT applications. In: 2016 IEEE Sixth International Conference on Communications and Electronics (ICCE), IEEE, pp. 234–237 (2016) 2. Atzori, L., Iera, A., Morabito, G.: The internet of things: a survey. Comput. Netw. 54(15), 2787–2805 (2010) 3. https://www.pcbcart.com/article/content/antenna-design-in-iot-design.html 4. http://www.techplayon.com/types-iot-antennas-choose-get/ 5. Lee, H., Choi, J.: A compact all-textile on-body SIW antenna for IoT applications. In: 2017 IEEE International Symposium on Antennas and Propagation & USNC/URSI National Radio Science Meeting, IEEE, pp. 825–826 (2017) 6. Vikram, N.: Design of ISM band RFID reader antenna for IoT applications. In: Wireless Communications, Signal Processing and Networking (WiSPNET), International Conference, IEEE, pp. 1818–1821 (2016) 7. Lizzi, L., Ferrero, F.: Use of ultra-narrow band miniature antennas for internet-of-things applications. Electron. Lett. 51(24), 1964–1966 (2015) 8. Hassan, S.A., Samsuzzaman, M., Hossain, M.J., Akhtaruzzaman, M., Islam, T.: Compact planar UWB antenna with 3.5/5.8 GHz dual band-notched characteristics for IoT application. In: 2017 IEEE International Conference on Telecommunications and Photonics (ICTP), IEEE, pp. 195– 199 (2017) 9. Katoch, S., Jotwani, H., Pani, S., Rajawat, A.: A compact dual band antenna for IOT applications. In: 2015 International Conference on Green Computing and Internet of Things (ICGCIoT), IEEE, pp. 1594–1597 (2015) 10. Asadallah, F.A., Costantine, J., Tawk, Y., Lizzi, L., Ferrero, F., Christodoulou, C.G.: A digitally tuned reconfigurable patch antenna for IoT devices. In: 2017 IEEE International Symposium on Antennas and Propagation & USNC/URSI National Radio Science Meeting, IEEE, pp. 917– 918 (2017) 11. Ramesh, G.P., Kumar, N.M.: Design of RZF antenna for ECG monitoring using IoT. In: Multimedia Tools and Applications, pp. 1–6 (2019) 12. Ramesh, G.P., Rajan, A.: Microstrip antenna designs for RF energy harvesting. In: 2014 International Conference on Communications and Signal Processing (ICCSP), IEEE (2014)

Chapter 8

A Bi-spectrum Analysis of Uterine Electromyogram Signal Towards the Prediction of Preterm Birth Kamalraj Subramaniam, P. Shaniba Asmi and Nisheena V. Iqbal

Abstract Prediction of preterm birth is one of the significant perinatal hurdles for the prevention of preterm birth. The uterine Electromyogram (Uterine EMG), obtained from the abdominal surface is analyzed for the prediction or preterm labor. Many linear and non-linear features and classifiers have been analyzed in different researches. In this paper two neural network classifiers were applied to the Bi-spectrum feature obtained from the Uterine EMG signal. The Bi-spectrum analysis was done after preprocessing the signal. Three pre-processing methods were tried to improve the performance. The best classification accuracy of 99.89% was obtained with Elman neural network classifier when pre-processed with three level wavelet (db4) decomposition. The sensitivity and specificity were found to be 100% and 99.77% respectively. Keywords Electromyogram · Bi-spectrum analysis · Elman neural network classifier

8.1 Introduction In recent years there have been dramatic changes in human social life due to the Internet of Things (IoT) which connects every Device/object, every module and every device to a network. The number of devices connected to the internet is constantly increasing. According to a report by Cisco [1], it is predicted that more than 50 billion objects will be connected to the internet by 2020. The result is the diffusion of an IoT that completely changes the way people interact with their environment. Kamalraj Subramaniam (B) Karpagam Academy of Higher Education, Coimbatore, India e-mail: [email protected] P. Shaniba Asmi · N. V. Iqbal MES College of Engineering, Kuttippuram, India e-mail: [email protected] N. V. Iqbal e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_8

67

68

Kamalraj Subramaniam et al.

In this environment antenna, it’s size and placing on the device/Object plays a vital role in providing communication between the objects working at various frequencies ranges from AF to RF. The birth of an infant before the gestation of 37 weeks is the major reason for morbidity and death of infants. The illness found in such babies is large in number and hence increases the chance of hospitalization. Some of the reasons for preterm birth mother’s poor health, infectious diseases and lack of proper health care resources. If it is possible to predict the preterm birth earlier, it can be treated properly and may leads to true labour. Electromyography is one of the non invasive and accurate techniques to acquire electrical activity from the uterus. This signal is acquired from the abdominal surface of pregnant women. The analysis of power spectral measures only the distribution of power as a process of frequency, and not the information about phase. It also formulates the assumption that the signal occurs from a linear process, thus un-noticing the possible contact between components of the signal which are manifested the same as phase coupling, a general incident in signals produced from non-linear sources [1]. The analysis through bi-spectrum is a process of signal processing that measure the phase coupling degree among the signal components used in this work. Bi-spectrum estimation was applied to analyse EEG signals to test the existence of nonlinear phase coupling within the EEG signals in a certain psycho-physiological state [2]. The primary objective of this paper is prediction of preterm birth by applying bi-spectrum analysis on Uterine EMG. The signal Physionet recorded on 38 preterm labors and 38 term labor was used for this purpose. Two classifiers are employed to distinguish between preterm labour and term. Section 8.1 of this paper introduce the work, Sect. 8.2 give details about the data used and method, and Sect. 8.3 talks about the results and at last conclusion in Sect. 8.4.

8.2 Materials and Methods 8.2.1 Data Requisition In this section various types of IoT antennas are presented. These help IoT module designers to make right choice of antenna for specified applications.

8.2.2 SIW Antenna Physionet database in the Term Preterm Electro-hysterogram database (TPEHG DB) is used for downloading the EHG records used in this research. It was collected at the Department of Obstetrics and Genecology, medical centre Ljubljana, Solvenia from 1997 to 2005.

8 A Bi-spectrum Analysis of Uterine Electromyogram …

69

Fig. 8.1 Electrode placement

The records are of 30 min duration with three channels. The records were collected from the abdominal surface using four AgCl2 electrodes with sampling frequency 20 Hz [3]. The electrodes were placed above the naval spaced 7 cm apart symmetrically in two rows as in Fig. 8.1 [11]. The first channel acquires signal from electrodes E2-E1, second from E2-E3 and third from E4-E3. Randomly selected 38 term and 38 preterm records were used in this work. Each record was segmented with window size of 1200. The signal then analyzed with 75 and 50% overlap and also without overlap.

8.2.3 Pre-processing The raw signal consists of noise and may be corrupted during recording. This has to be removed using band pass filter. The EHG signal range is from 0 to 3–5 Hz. Electrical signal due to voluntary contractions of the abdominal skeletal muscles have frequency component of about 30 Hz. Respiratory artifacts mainly distributed between 0.20 and 0.34 Hz [4]. To eliminate these lower cut off frequency was kept at 0.34 Hz and upper cut off frequency selected as 3 Hz. Fourth order Butterworth and Elliptic band pass filters were considered for pre-processing the segmented signal. In addition to these filters a three level wavelet decomposition using Daubechies 4 mother wavelet also applied to evaluate the performance.

70

Kamalraj Subramaniam et al.

8.2.4 Bi-Spectrum Analysis The UEMG signal after segmentation using a window size 1200 was considered in three different scenarios: 75% overlap, 50% overlap, and without overlap. These segments were then pre-processed in three methods using the Butterworth filter, elliptic filter, and three-level db4 wavelet transform. Bi-spectrum feature was extracted from this pre-processed signal. The power spectrum can be used as a quadratic descriptor in stochastic domain. it conveys the amplitude details from the Fourier components. The phase information that may be present in non-linear processes is not feasible by this. It illustrates the Gaussian field completely. As it is not possible to have a region with density of negative value, the high amplitude field of alternating density is not Gaussian. This implies power spectrum only gives incomplete information. The higher-order correlations between Fourier components, is a method that gives the further message in Fourier space. The cubic, the three-point similarity function and its Fourier counterpart are the subsequent order descriptors of bi-spectrum. The bi-spectrum is the third order moment of the Fourier amplitudes of a random field that depends on three wave numbers. Bi-spectrum analysis was introduced in 1960s by geophysicists to study ocean wave motions. The massive computational requirements of bi-spectral analysis decreased its application in bio-signals. By the novel approach of high-speed, lowcost computing, examination utilizing bi-spectral analysis has developed. It offers useful and better information compared to the information that can be extracted from a power spectrum [5]. Power spectrum quantifies only the power distribution with respect to frequency In the case of an Uterine EMG signal, prediction is more important than time consumption for signal processing. Phase coupling is characteristic of non-linear systems. With the power spectrum methods or other quantitative methods, it is very difficult to express the quantity of phase coupling [6]. To identify the quantity of phase coupling in the signal, a signal processing method called bi-spectrum analysis must be used. It is a high-level signal processing method that expresses the quantity of quadratic non-linearity. While estimating a bi-spectrum, the signal x(k) is dissolved first into a number of epochs that approximated to a mean value of zero. This is to avoid offset in signal due to potential difference between the electrodes [7]. Subsequently, the Fourier transform of each epoch is found. A bi-spectrum, which is a function of two frequencies, is evaluated as   L     ∗ X i ( f 1 )X i ( f 2 )X i ( f 1 + f 2 ) B( f 1 , f 2 ) =   

(1)

i=1

where X i∗ ( f 1 + f 2 ) is the conjugate of X i ( f 1 + f 2 ); L is the number of epochs; X i ( f ) is the Fourier transform corresponding to the ith epoch. The phase coupling cannot be measured from a single epoch. If the phase relationship among the pairs of frequencies is random or associated, then more number of

8 A Bi-spectrum Analysis of Uterine Electromyogram …

71

epochs was considered to determine phase coupling between each pair of frequencies. The triple product X ( f 1 )X ( f 2 )X ∗ ( f 1 + f 2 ) is estimated for this purpose for each of the epoch series and then average of this product is done. From the average of this product the magnitude is evaluated and is referred as bi-spectrum [1, 8].

8.2.5 Classifier Two neural network classifiers are used for discriminating term and preterm Uterine EMG signal. They are probabilistic neural network (PNN) and the Elman neural network (ENN). PNN comes under feed-forward neural network with four layers. The layers can be named as (i) (ii) (iii) (iv)

Input layer, hidden layer, pattern/summation layer Output layer

In PNN algorithm, the probability distribution function (PDF) of any class is derived by a Parzen window and a function which is non-parametric. After applying PDF of each class, the new input is determined. For designate it to the class with the highest posterior probability, Bayes’ rule is applied [9]. ENN is a three layer network arranged horizontally. In addition to the three layers there is a context unit, which is linked to the middle hidden layer which is weighted with one. For every period of time, the measuring input is fed-forward and then a training rule is utilized. The fixed back connections store the former data of the second units (hidden layer) in the context units, as they travels through the connection prior to the application of learning rule. Thus the network can sustain a state that enables it to perform a sequence prediction [10].

8.3 Results and Discussion The results of the bi-spectrum analysis for the three cases of pre-processing with the two neural network classifiers are tabulated and discussed in this section. In each pre-processing technique the signal is applied with 75% overlap, 50% overlap and with no overlap. In the case of ENN classifier there is no significant difference when overlap is considered and the best is with 75% overlap that preprocessed with wavelet decomposition. Table 8.1 shows the accuracy of bi-spectrum feature. For PNN the best is with no overlap and is for the signal pre-processed with fourth order Butterworth filter, but its accuracy is much below ENN classifier. Comparison of accuracy of bi-spectrum feature with the two classifiers for 75% overlap is given in Fig. 8.2.

Butterworth

75

57.05

99.71

Filter

% of overlap

PNN

ENN

Window size 1200

Accuracy in %

99.78

55.47

50 99.70

58.71

0

Table 8.1 Accuracy of bi-spectrum feature with the two classifiers

99.77

57.16

75

Elliptic 50 99.78

54.50

0 99.70

55.71

99.89

54.65

75

Wt 3 level db4 50 99.78

53.45

0 99.85

54.80

72 Kamalraj Subramaniam et al.

8 A Bi-spectrum Analysis of Uterine Electromyogram … Fig. 8.2 Comparison of accuracy of bi-spectrum feature with the two classifiers

120 100

99.77

73

99.77

99.89

80 60

57.05

57.16

54.65

40

PNN ENN

20 0

Butterworth filter

Elliptic filter

Wavelet decomposition

8.4 Conclusion In this paper Bi-spectrum analysis were considered for the classification of preterm and term pregnancy. Two classifiers are used to classify bi-spectrum feature. Elman neural network shows the best classification of 99.89% with sensitivity 100% and specificity 99.77%, when the data taken with window size 1200 and 75% overlap. The signal was pre-processed using three methods and the wavelet decomposition method performed best. Thus bi-spectrum feature have good discriminating capacity as it have the phase information. More nonlinear feature has to be studied in future.

References 1. Sigl, J.C., Chamoun, N.G.: An introduction to bispectral analysis for the electroencephalogram. J. Clin. Monit. 10(6), 392–404 (1994) 2. Al-Askar, H., Radi, N., MacDermott, Á.: Recurrent neural networks in medical data analysis and classifications. In: Applied Computing in Medicine and Health, pp. 147–165. Morgan Kaufmann, Burlington (2016) 3. Goldberger, A.L., Amaral, L.A.N., Glass, L., Hausdorff, J.M., Ivanov, P.Ch., Mark, R.G., Mietus, J.E., Moody, G.B., Peng, C.K., Stanley, H.E.: PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals. Circulation 101(23):e215–e220 (2000). http://circ.ahajournals.org/content/101/23/e215.fullPMID: 10851218; https://doi.org/10.1161/01.cir.101.23.e215 4. Fele-Žorž, G., Kavšek, G., Novak-Antoliˇc, Ž., Jager, F.: A comparison of various linear and non-linear signal processing techniques to separate uterine EMG records of term and pre-term delivery groups. Med. Biol. Eng. Comput. 46(9), 911–922 (2008) 5. Ramesh, G.P., Aravind, C.V., Soysa, R.R.N.: Body area network through wireless technology. Int. J. Comput. Sci. Eng. Commun. 2(1), 129–134 (2014) 6. Ramesh, G.P., Kumar, N.M.: Design of RZF antenna for ECG monitoring using IoT. Multimedia Tools Appl. 1–6 (2019) 7. Kaplanis, P.A., Pattichis, C.S., Hadjileontiadis, L.J., Panas, S.M.: Bispectral analysis of surface EMG. In: 2000 10th Mediterranean Electrotechnical Conference on Information Technology and Electrotechnology for the Mediterranean Countries (MeleCon 2000) (Cat. No. 00CH37099), vol. 2, pp. 770–773. IEEE (2000)

74

Kamalraj Subramaniam et al.

8. Kumar, N., Khaund, K., Hazarika, S.M.: Bispectral analysis of EEG for emotion recognition. Procedia Comput. Sci. 84, 31–35 (2016) 9. Karjala, T.W., Himmelblau, D.M., Miikkulainen, R.: Data rectification using recurrent (Elman) neural networks. In [Proceedings 1992] IJCNN International Joint Conference on Neural Networks, vol. 2, pp. 901–906. IEEE (1992) 10. Raghu, S., Sriraam, N., Kumar, G.P.: Classification of epileptic seizures using wavelet packet log energy and norm entropies with recurrent Elman neural network classifier. Cogn. Neurodyn. 11(1), 51–66 (2017) 11. Goshvarpour, A., Goshvarpour, A., Rahati, S., Saadatian, V.: Bispectrum estimation of electroencephalogram signals during meditation. Iran. J. Psychiatry Behav. Sci. 6(2), 48 (2012)

Chapter 9

Application of Information Science and Technology in Academic Libraries: An Overview S. Velmurugan and G. P. Ramesh

Abstract Informational is imperative in support of person evolution since air is most needful for staying alive of all surviving things on the earth, includes humans. The slight change made by new info techniques has a key factor of people live span, working, and playing worldwide. The Information Science and Technology (IST) take part in an important function in the evolution of academic library services in proactive to the challenging task of the informational services. The paper aims to deal with the rapid evolution of IST and implementation of its services in the academic libraries. Presently all academic libraries are established to provide IST based digital library. IST includes services such as Electronic Sources to enhance the info required as per the needs of user’s effective manner. Keywords Information Science and Technology (IST) · Libraries · Library of Electronic Academics · Libraries of Digital Academics · E-Sources

9.1 Introduction In Informational is the basic ingredient of all nature of investigate and evolution. Informational is a basic source which is most needful for surviving in present viable digital world using the Internet. The informational knowledge to utilizing with modified IST based digitalized owing to the evolutions in academic libraries. It is a foremost purpose for the cultural-evolution and socio-economic of several country. In the statement of Kemp’s model, said “Information is the 5th rank of human’s need” S. Velmurugan (B) Department of Electronics and Communication Engineering, TJS Engineering College, Peruvoyal, Gummidipoondi Taluk, Thiruvalluvar District, Tamil Nadu 601206, India e-mail: [email protected] G. P. Ramesh Department of Electronics and Communication Engineering, St. Peter’s Institute of Higher Education and Research, Chennai 600054, Tamil Nadu, India e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_9

75

76

S. Velmurugan and G. P. Ramesh

as same as the immediate requirement of water, air, shelter, and food. The major requirement for information in every effort of human is not to be hard to search [1]. A speedy and simple way to use required info is the most important, particularly in libraries. IST appliance and its techniques are being used with the libraries for processing of storage, academic info, communication, distribution of informational, computerization, etc., in full info. Additionally, the base of the internet and the evolution of the World Wide Web (WWW) transform the IST. Authorize the services to be utilized using IST based libraries are vital to offer the services to their academic user’s.

9.2 Informational Informational is global, it is called to every human in every language, if or if not be an equivalent ‘word’ in a language to describe the expression ‘info-informational’ although it is accessible. Everyone receives info on a daily basis. In Shannon and Weaver stated that “Information is any stimulus that reduces uncertainty”. Similarly, Ching Chin Chen and Peter Hernon say info like “all knowledge, ideas, facts, data and imaginative works of mind which are communicated formally and or informally in any format” [2]. This informational is most important to human life, how do we know and where we get from? In the case study of info, the search is a challenging task even though it will be accomplished that research is one of the eminent areas where it takes the steps to find the required information. The contribution of specialist in the technologies, social-science, science fields, and the humanities sacrifice is helpful to the whole society. The focus of government on Research & Development (R&D) allotted the enormous funds into these fields. Therefore, more and more info-search is continuously generated. Subsequently, the world is being improved with info-search that will require to the trend the name ‘info explosion’ [3].

9.3 Informational Required The Librarian states info required as “that need which academic library services or materials are intended to satisfy”. According to Maurice B. Line statement info required as, “what an individual ought to have for his work, his research, his edification, his recreation, etc.” [4].

9 Application of Information Science and Technology …

77

9.4 Information Science and Technology • In basic components of Webster’s, “Information Science and Technology (IST) is the collective term for various technologies involved in the processing and transmission of Information they include computing telecommunications and microelectronics” [5]. • The same ‘ALA Glossary’ states, “Information Science and Technology as the application of computers and technologies to the acquisition, organization, storage, retrieval, and dissemination of information” [6]. • In the Department of British in Industry and Trade define it as, “The acquisition, processing, storage and dissemination of vocal, pictorial, textual and numerical information by a microelectronics-based combination of computing and telecommunication” [7].

9.4.1 Basic Components of IST Presently, in technical developments is becoming a mainspring in our culture. Information Science and Technology (IST) is a universal term implied in a package of technologies [8]. The basic components of IST used in modern digitalized academic library and information system are shown in Fig. 9.1.

Fig. 9.1 Basic components of IST

78

S. Velmurugan and G. P. Ramesh

9.4.2 Implementation of IST The implementation of IST involves the existence of an infrastructure where it learns, obtains, and effectively apply the latest techniques. It comprises appropriate assist accessible in human resources, R&D developments, networks of well-resourced telecommunication, and initial investments.

9.5 IST Application in Digital Library The gross improvement can be made use by the digital library. It is the main information center. IST disguised for the growth of users as an intact. The librarian needs in IST must include the techniques of IST that are estimated to be used in the access of digital academic library and further services such as processing, collection, storage, recovery and sharing of recorded information. The fast rising information technology provides almost every field of application include in the library [9]. The following environments used to enhance the academic library, • Management of Library: The management on library includes the access for the following that will be maintained by the utilization of these fast IST developments: Cataloguing, Classification, Database creation, Indexing, Database Indexing. • Computerization of Library: The automation in the library is the aim of minimizing the interference of human to the library service access; as a result any user can access the desired information at the lower cost. The automation mainly focused into two categories that are the founding of the entire library databases and the entire functions for housekeeping of the library. • Networking of Library: The networking provides to user-friendly on info search shared to all users in which common database maintained by the IST based server as called Libraries and Information Centre, used to improve the efficiency of info search. • Digital Library: The digital library provides E-Source like microfilms, photography, audio and tapes, optical disk, printing, etc. • Technological Library: It offers digital communication; it contains editing, publishing, technical writing, DTP systems, and so on.

9.6 Effect of IST in Library IST has a significant effect on the library and info search. Informational access has taken fast transformations from traditional methods, consequent upon the introduction of IST [10]. Table 9.1 summary the effect of info technology in the library.

9 Application of Information Science and Technology …

79

Table 9.1 Effect of IST in library Sl. no.

Informational access

Traditional method

IST

1

Generate, Originate

Writing, Typing

Word Processing, Text editing. Character Recognition, Voice Recognition

2

Preserve, Store

Manuscript, Paper- Print Media

Electronic Publishing, Magnetic Storage, Videotext, Tele-text. Computer disk, ROM

3

Process

Classification, Cataloguing, Indexing

Electronic data processing, Artificial intelligence/Expert systems

4

Retrieval

Catalogs, Indexes

Database management system, Information retrieval off-line, On-line

5

Disseminate/Communicate

Bibliographies, Abstracts, Hard Copies

Electronic mail, Electronic document delivery, Computer conferencing Telefacsimile, View data

6

Destroy

Physical weeding

Magnetic erasers, Optical erasers, re-use the medium

Source http://web.simmons.edu

9.7 Merits and Demerits of IST • Merits – – – – – – – – – –

The simple way to share different activities of library services Networking of library provides creation and collaboration Keep clear of repeating access within a library Improve the limitation of services offered Delay of search reduced and users time saved Efficiency increased Fast and simple access to info search QoS improved in library access Enhanced the experience to gain knowledge Association with central libraries

• De-merits – – – –

Deficiency due to lack of fund Maintenance cost increased every year Inadequately skilled staff’s Unemployment causes due to automation

80

S. Velmurugan and G. P. Ramesh

9.8 Organization of IST Based Services Access can be provided on the basis of following classes by the IST-based services They are; • Equipment and Facilities • Users Service • E-Sources

9.8.1 Types of Equipment and Facilities The types of equipment and facilities required to fulfill the availability of IST based libraries includes. a. Desktop: Desktop-based technology provides an impact of information search goal and quest behavior from various perspectives such as digital electronic environment, user study, usage of library resources, user education. b. OPAC: By using online database share the materials word wide to form a group of libraries supported by Online Public Access Catalog (OPAC) through the internet. c. Classified Catalogue: It is a catalog of shared library that contains list of a number of libraries. It offers information regarding the users search like book format, media, microform, cards and networked electronic databases. Classified catalogs are very helpful to librarians, as they support them in locating and acquiring materials from the central library by using the loan services of interlibrary [11]. d. CD-ROM: It offers a feedback presented in academic libraries, includes all aspect of libraries involved and employment implication. CD-ROM has high impact implied in academic library functions and offered the best service to their users. e. Scanner: Photo scanner is often referred as scanner; it is employed to scans printed text, images, handwriting, and converts it to a digital image. The driver of automated scanner allows the document transferring and it is more frequently used for large format documents, while flatbed design has not practical. f. RFID: It is abbreviated as Radio-Frequency Identification. It is a tag that is encoded digitally and utilized for capture of auto-read via radio waves. It means that storing a serial number recognized by a product and associated information on the microchip which is connected to an antenna. RFID is employed as a bar codes alternative. g. Tele-text: It is a rescue service of TV-information executed in the UK. It presents constraints of text-based information; frequently contains international, national, sporting news, and TV schedules. It is employed to transmit television signal in among interval of vertical blanking and frames of image [12].

9 Application of Information Science and Technology …

81

h. Facsimile: It is a replica of a previous version of article, books, art, navigation, or additional items of chronological information i.e., as true-copy to the novel source. It vary from other forms of reproduction by effort to copy the source as precisely as feasible as in the format that includes color, condition, scale, and additional material qualities. Likewise Text-books and article also requires a whole copy of entire pages; as a result a shortened copy called as “partial facsimile”. i. Photo-copier: It is a mechanical device that creates documents copies and other images of illustration in fast and cheap. Current photo-copiers employs the technology known as Xerox i.e., Photo-Copy. It is largely accessed by the client in the Digital libraries. j. Printing techniques: A printer is a secondary device that generates a printed documents and/or textbook stored in digital form, usually print media contains paper or transparencies. k. Barcode: A barcode is a scanner. It is a digital electronic device for evaluating the printed barcodes. Alike to a flatbed scanner, it comprises of a lens; light source; and a light sensor transmitting optical impulses into electrical. Even though, every barcode readers contains the circuitry for the decoder to capture the barcode’s image information offered by the sensor and transmit the content of barcode’s to the output port of scanner’s.

9.8.2 Service to Users a. DDS: It is abbreviated as Document-Delivery Service (DDS). It transfers the duplicates of text book chapter’s and journal papers from central Library at small charges. Moreover to complete the information needed by an end client through the supply of info/doc is known as a document delivery service. b. Inter-library loan: It is related to book bank, i.e., the association among single and many libraries might make use of material. It also can be stated as a lending of library materials by one library to different library. c. Index/Abstract Service: It is a procedure that is employed to obtain information from accessing the memory or the capability of compile an index. The abstracts formulation fundamentally in a narrow field; by a unit; a profitable association; the abstracts being published and send to subscribers regularly [13]. d. User chatting service: Introduce user chatting service through online using the Internet, which offer a direct transmission of text-based messages from sender to receiver; hence the delay for visual access from end to end user as reduced. e. Current Awareness Service: The objective of a Current Awareness Service (CAS) is introduced to the users to access new acquisition in their libraries. Especially in academic libraries, provides display boards and shelves to draw attention to recent additions, and many libraries produce complete or selective lists for circulation to patrons.

82

S. Velmurugan and G. P. Ramesh

f. Selective dissemination of information: Selective dissemination of information (SDI) was initially a word search related to library access employed to maintain a user learned of new resources on particular topics. g. Scan copy: A scanning service provides to make a copy of material which is not available in the digital e-source form. It contains journal papers and book chapters. Users may employ this service that the library offers within the copyright act applied to subscription limitation. h. BBS: A Bulletin Board Services is computing software which permits clients to connect and log into the system with a user ID. A client can carry out functions like reading news and bulletins, uploading and downloading software and data, and exchanging messages with other users, and so on, once they get logged into the system [14]. i. Digital library: It provides electronic materials in which collection are store in digitally and accessed by using desktop. The digital material can store in local disk, or use distantly via desktop through the internet. A digital library is a type of information retrieval system.

9.8.3 E-Sources a. Audio-Visual Contents: The Audio-Visual Content includes a broad collection of video materials to bear the research and studies require by the user. b. Intra-Internet: Development of digital growth, communication is converted into simple and instant access and update to current trends. The most recent use of internet in high speed has incised along with distance and useless it is easy to exchange info to any users at any places and at any times. c. Library Web-portal: Library Web-portal provides to access services and info E-source presented in the libraries. Almost in all library web-portal, the online database is maintained to determine a user whether the information is accessible in the library. Catalog: A catalog is a set of information for single or multipurpose, usually in digital E-Source. The info in general intended to replica significant aspect of certainty; the user’s supports to processes the required information [15].

9.9 Conclusion The Digital E-Source has the effect of rival the Information and Science Technology (IST) community. User access from academic libraries in the developed and developing countries can interact with each other on the same platform. The result is that the user community of academic library seems to move towards global regularities.

9 Application of Information Science and Technology …

83

As a prospective observation of user’s needs, it is insisted that academic libraries are performing in a form of digital, they must be conscious of the most recent technologies to persist and continue the service provided. Exploitation of IST in present libraries is optimal to acquire any information at any time in any place and at the lowest cost. It examines to increase the libraries rank and it reduce the effort of the library professions. It has expanded the service to worldwide margins, innovative techniques and methods assist to progress the best users services.

References 1. Ashok Babu, T.: Modern Information Technologies: Their Impact on Library Services. Library Information Technology in the Modern Era: Libraries and Librarians in New Millennium, Commonwealth, New Delhi, pp. 65–72 (1999) 2. Dabas, C.: IT Applications for TQM and Library Marketing, pp. 40–42. Ess Publications, New Delhi (2008) 3. Devarajan, G.: Applied Research in Library and Information Science, p. 74. Ess Publications, New Delhi (2005) 4. Kannappanawar, B.U.: Problems and Prospects of Information Technology in R&D Libraries “Dr. P.S.G Kumar festschrift Library and Information Profession in India, Delhi; B.R. Publishing Corporation 1(2), 612–617 (2004) 5. Kumar, P.A.: Impact of Information Technology on the Collection Development in University Libraries of Assam: A Study (2017) 6. Lata Suresh, Singh, S.N.: Status in ICT and health information system. Int. J. Inf. Libr. Soc. 3(1), 16–24 (2014) 7. Laloo, B.T.: Information Needs, Information Seeking Behaviour and Users. Ess Ess Publ (2002) 8. Kumar, P.S.G.: Information Technology: Basic Concepts, pp. 9–17. BR publishing Corporations, New Delhi (2003) 9. Ramesh, G.P.: Performance Analysis of Traffic with Optical Broker for Load Balancing and Multicasting in Software Defined Data Center Networking 10. Ruan, L., Qiang, Z.: The role of information technology in academic libraries’ resource sharing in Western China. Libr. Trends. 62(1), 180–204 (2013) 11. Line, M.B.: Draft definitions: information and library needs, wants, demands and uses. In: Aslib proceedings, MCB UP Ltd. 26(2), 87–87 (1974) 12. Mehrjerdi, Y.Z.: RFID: the big player in the libraries of the future. Electron. Libr. 29(1), 36–51 (2011) 13. Naggi Reddy, Y.: Information technology-based services in a University Library: a user satisfaction survey. Ann. Libr. Inform. Stud. 53, 15–17 (2006) 14. Pujar, S.M.: Information use by economists: a study. Ann. Libr. Inform. Stud. 54, 190–194 (2007) 15. Soper, M.E.: The Librarian’s Thesaurus: A Concise Guide to Library and Information Terms, p. p2. American Library Association, Chicago (1990)

Chapter 10

A Stable Routing Algorithm Based on Link Prediction Method for Clustered VANET Bhasker Dappuri, Malothu Amru and Allam Mahesh Venkatanaga

Abstract VANETs are high class, dynamic and consistent and forma main module of Intelligent Transport System (ITS) which is self-controlled, wheeled and stimulating class of MANET. We use RIVLP to improve clustering. In this paper, a new algorithm is developed and compared with existing algorithms which show improved performance in terms of stability up gradation, delay time reduction, lifetime of both CH and cluster and throughput. Keywords VANET’s · Intelligent Transport System (ITS) · Clustering

10.1 Introduction In VANET’s are of two types Vehicle to Vehicle (V2V) and Vehicle to Infrastructure (V2I) and are of great interest from industry with core units as the Road Side Unit (RSU). VANETs are IEEE which supports vehicles in 100–300 m range through dedicated Short range Communication (DSRC) or a 3G network. OBU s are fixed whereas RSUs are static put either at the road side or near to traffic lights which are similar to access points and can communication with infrastructure in 100–300 m range [1] where a vehicle will not only be a router but also a source or destination to communicate in one hop or multihop fashion (Fig. 10.1). MANET’s are widely used over VANET as they give good QoS and safety applications with fair access to use the channels [2]. Using clustering methodology, vehicles are grouped which in turn makes VANET dynamic; this decreases the terminal problem [3] and is also an efficient key for scalability issue [4]. This helps in managing vehicle, routing and accessing resources and bandwidth [2]. VANETs are MANETs B. Dappuri (B) · M. Amru · A. M. Venkatanaga Department of ECE, CMR Engineering College, Hyderabad, India e-mail: [email protected] M. Amru e-mail: [email protected] A. M. Venkatanaga e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_10

85

86

B. Dappuri et al.

Fig. 10.1 VANET system structure

which have mobile nodes as vehicles fitted with On Board Units (OBUs), Application Unit (AU), Global Positioning System (GPS) as digital maps and these vehicles move on roads following traffic instructions and communicate [5–10]. A link prediction algorithm is proposed which has real time and active vehicle monitoring and lifetime extension system having information of lanes of automobiles on road and vehicles.

10.2 Related Work This In order to increase the cluster lifespan and CH count per vehicle based on routing techniques of VANET’s using inter layer approach with information of present traffic flow and parameters as vehicle size speed density of vehicles an Intelligent Based Clustering Algorithm in VANET (IBCAV) is proposed [11]. This is a core technique is a smart technique called as an artificial neural network. An algorithm created on traffic flow [2] to get cluster strength where a route with heavy traffic flow is considered and given much importance. Based on the lanes having maximum traffic flow a CH is selected to increase network lifetime which is proposed in [12] where vehicle should hold information as lane direction, lane recognition and amp matching to road and convey the same information to neighboring vehicles which helps in selection stable and efficient CH [13]. By grouping the vehicles in lanes with high speed while the other vehicles are put into another cluster as a group, doing the grouping location and direction is considered. This is adopted in an algorithm [14] in a VANET as to increases the cluster lifespan and reduce vehicle changes per cluster. Ann algorithm [15] is proposed for connectivity up gradation and avoids the interruption of vehicles as the CH changes offer a new parameter as “Vehicle Interconnection Metric” which gives the effort in

10 A Stable Routing Algorithm Based on Link Prediction Method …

87

the neighborhood. By adjusting the alternations good communication can be established among the pairs and metric is built on beacon frames checked among vehicles which help in elimination the selection of new CH as the old CH leaves its role [16]. Considering and comparing movability of vehicles in multi hop distance and minimum value of mobility gives another algorithm [17]. Information about the lane vehicles and their transmission to adjacent vehicles will give us the ideal cluster head which improves the network lifetime which would need information as the flow of traffic depends on data related to speed, long lasting and steady CH [18].

10.3 Proposed Framework Based on traffic flow and link prediction algorithm clustering, CH selection is carried which helps in lowering the disruption between CH and cluster members and gives a best route increasing the network lifetime.

10.3.1 System Scenarios Unlike in most WSN, here CH rounds per vehicle are not considered and a node is the vehicle which has no battery problem and vehicle is assumed to have same processing power and storage [19]. As the CH leaves it either informs the CH ahead of it or 2nd winning CH is given the role and if any 2 CH’s has identical score then a random selection is made. For each intersection traffic separates and flows in 3 ways left, right and straight flow. In a way to increase lifetime CH is selected from the path having heavy traffic flow which also increase the stability. For instance consider five lanes have 3 straight lanes, as these have heavy traffic hence CH selects this lane [20]. In Fig. 10.2, 6th lane turns to right with 3 vehicles and 1st lane has 2 vehicles

Fig. 10.2 System scenario

88

B. Dappuri et al.

taking a left turn along with 2, 3, 4, 5 with 10 vehicles from which CH is selected by cluster which the network lifespan.

10.3.1.1

Clustering Process Transition Model

A unique ID is given to vehicles which convey the movement information as speed, space, size of the cluster, direction and connectivity level along with cunt of neighboring vehicles considering which CH selection is done following the below mentioned rules [21], as vehicle • • • •

It is not a present member is represented as UN—Unknown Vehicle It handles cluster responsibilities is denoted as CH—Cluster Head It acts as a part/member of cluster is represented as CM—Cluster Member A vehicle which does not get HELLO message in TI (Time Interval) from CH is TM—Temporary Member.

10.3.1.2

Clustering Creation Phase

It includes opening, connecting, CH selection leaving, merging of nodes as shown in Fig. 10.3.

Fig. 10.3 Architecture of RIVLP protocol

10 A Stable Routing Algorithm Based on Link Prediction Method …

89

10.3.2 Procedure for Crating Clusters VANET established link used to predict the network lifetime in which the edge is taken as the link in the graph calculated on distance among the neighboring nodes moving along the edge respectively [22]. Every node has got GPS which assists in know the location in order to route message. Clusters are formed inset up phase and data transmission happens in steady phase. Information about the distance will be updated by current node and is sent to next neighboring node and it continues till the information reaches destination node. Due to high mobility the network structure changes and by knowing about types of structure links can be made stable which needs traffic to be stable and not jammed as jamming makes the mobility route to degrade its performance as the message will not be sent at defined periods [23]. Using lifetime prediction link instead of using edge reliability estimation could improve RIVLP reliability as edge calculation happens based on timestamp of probe message sent to nodes in the network [24].

10.4 Experimental Results The results shown in below tabular column are values of simulation carried out in the NS-2 simulator using default 802.11 (Table 10.1). The below figures are the simulation results (Figs. 10.4, 10.5, 10.6, 10.7, 10.8, 10.9, 10.10, 10.11, 10.12, 10.13, 10.14, and 10.15). Table 10.1 System parameters

Fig. 10.4 Network deployment

Parameter

Value

Application traffic

CBR

Transmission rate

1000 bytes/0.5 ms

Radio range

250 m

Simulation time

1000 ms

Number of nodes

40

Area

1000 × 1000

Routing protocol

AODV

90 Fig. 10.5 Clustering process

Fig. 10.6 Broadcasting process in network

Fig. 10.7 Data communication between sender and receiver

Fig. 10.8 All vehicles moving while data communication

B. Dappuri et al.

10 A Stable Routing Algorithm Based on Link Prediction Method …

Fig. 10.9 Table of distance, angle and channel capacity of every vehicle

Fig. 10.10 Cluster head selection based on distance

91

92

Fig. 10.11 Routing table of nodes

Fig. 10.12 Energy table for request process

B. Dappuri et al.

10 A Stable Routing Algorithm Based on Link Prediction Method …

Fig. 10.13 Energy table for delete the request process

Fig. 10.14 Energy table for final request process

93

94

B. Dappuri et al.

Fig. 10.15 Trace file of network process

10.5 Conclusion To generate stable clusters and to select enhanced routing path a lifetime prediction link is based on traffic movement is used which involves weight of traffic, size of cluster, level of connectivity, direction of vehicle, are the basic things to select a cluster. Simulation results of this algorithm when compared with other algorithms shows improved performance in terms of stability up gradation, delay time reduction, lifetime of both CH and cluster and throughput.

References 1. Çalhan, A.: A fuzzy logic based clustering strategy for improving vehicular ad-hoc network performance. Sadhana 40(2), 351–367 (2015) 2. Gajare, S., Deore, P., Wagh, S.: Traffic management in VANET using clustering. Int. J. Eng. Tech. Res. (IJETR) 2(5) (2014). ISSN: 2321-0869 3. Mottahedi, M., Jabbehdari, S., Adabi, S.: IBCAV: intelligent based clustering algorithm in VANET. Int. J. Comput. Sci. Issues (IJCSI) 10(1), 538 (2013) 4. Bali, R.S., Kumar, N., Rodrigues, J.J.: Clustering in vehicular ad hoc networks: taxonomy, challenges and solutions. Veh. Commun. 1(3), 134–152 (2014)

10 A Stable Routing Algorithm Based on Link Prediction Method …

95

5. Yang, F., Lin, Z., Tang, Y.: A traffic flow based clustering scheme for VANETs. Sens. Transducers 180(10), 110 (2014) 6. Huang, L., Wu, J., You, F., Lv, Z., Song, H.: Cyclist social force model at unsignalized intersections with heterogeneous traffic. IEEE Trans. Industr. Inf. 13(2), 782–792 (2016). https:// doi.org/10.1109/TII.2016.2597744 7. Li, W., Song, H.: ART: an attack-resistant trust management scheme for securing vehicular ad hoc networks. IEEE Trans. Intell. Transp. Syst. 17(4), 960–969 (2015). https://doi.org/10. 1109/TITS.2015.2494017 8. Ahmed, S.H., Bouk, S.H., Yaqub, M.A., Kim, D., Song, H., Lloret, J.: CODIE: controlled data and interest evaluation in vehicular named data networks. IEEE Trans. Veh. Technol. 65(6), 3954–3963 (2016). https://doi.org/10.1109/TVT.2016.2558650 9. Nie, L., Jiang, D., Guo, L., Yu, S., Song, H.: Traffic matrix prediction and estimation based on deep learning for data center networks. In: 2016 IEEE Globecom Workshops (GC Wkshps), IEEE, pp. 1–6 (2016). https://doi.org/10.1109/glocomw.2016.7849067 10. Blanco, J.I., Song, H.: Simulation of communications and networking in vehicular Ad Hoc networks. In: Simulation Technologies in Networking and Communications: Selecting the Best Tool for the Test, pp. 547–570. CRC Press (2014) 11. Saini, H., Mahapatra, R.: Implementation and performance analysis of AODV routing protocol in VANETs. Int. J. Emerg. Sci. Eng. 2319-6378 (2014) 12. Lin, Y., Song, H.: DynaCHINA: real-time traffic estimation and prediction. IEEE Pervasive Comput. 4, 65 (2006) 13. Ramesh, G.P., Rajan, A.: Microstrip antenna designs for RF energy harvesting. In: 2014 International Conference on Communications and Signal Processing (ICCSP), IEEE (2014) 14. Mohammad, S.A., Michele, C.W.: Using traffic flow for cluster formation in vehicular ad-hoc networks. In: IEEE Local Computer Network Conference, IEEE, pp. 631–636 (2010) 15. Rawashdeh, Z.Y., Mahmud, S.M.: A novel algorithm to form stable clusters in vehicular ad hoc networks on highways. EURASIP J. Wirel. Commun. Netw. 1, 15 (2012) 16. Ramesh, G.P.: Performance Analysis of Traffic with Optical Broker for Load Balancing and Multicasting in Software Defined Data Center Networking 17. Vodopivec, S., Hajdinjak, M., Bešter, J., Kos, A.: Vehicle interconnection metric and clustering protocol for improved connectivity in vehicular ad hoc networks. EURASIP J. Wirel. Commun. Netw. 1, 170 (2014) 18. Mehmood, A., Mauri, J.L., Noman, M., Song, H.: Improvement of the wireless sensor network lifetime using LEACH with vice-cluster head. Ad Hoc Sens. Wirel. Netw. 28(1–2), 1–7 (2015) 19. Mehmood, A., Lloret, J., Sendra, S.: A secure and low-energy zone-based wireless sensor networks routing protocol for pollution monitoring. Wirel. Commun. Mobile Comput. 16(17), 2869–2883 (2016) 20. Mehmood, A., Umar, M.M., Song, H.: ICMDS: secure inter-cluster multiple-key distribution scheme for wireless sensor networks. Ad Hoc Netw. 55, 97–106 (2017) 21. Li, C., Ye, M., Chen, G., Wu, J.: An energy-efficient unequal clustering mechanism for wireless sensor networks. In: IEEE International Conference on Mobile Adhoc and Sensor Systems Conference, IEEE, p. 8 (2005) 22. Mehmood, A., Nouman, M., Umar, M.M., Song, H.: ESBL: an energy-efficient scheme by balancing load in group based WSNs. KSII Trans. Internet Inf. Syst. 10(10) (2016) 23. Zhang, Z., Boukerche, A., Pazzi, R.: A novel multi-hop clustering scheme for vehicular ad-hoc networks. In: Proceedings of the 9th ACM International Symposium on Mobility Management and Wireless Access, pp. 19–26. ACM (2011) 24. Mehmood, A., Khanan, A., Mohamed, A.H., Mahfooz, S., Song, H., Abdullah, S.: ANTSC: an intelligent Naïve Bayesian probabilistic estimation practice for traffic flow to form stable clustering in VANET. IEEE Access 6, 4452–4461 (2017)

Chapter 11

Reversible Image Watermarking for Health Informatics Systems Using Distortion Compensation in Wavelet Domain Swathi Guntupalli, M. Sreevani and M. Raja Abstract Commencing the watermarked image replacement of related inventive cover and watermark symbol is assured by Reversible image watermarking. Two fold significant consideration sin reversible watermarking are Capability and distortion of the image. Concentrating on improving the implanting capability plus decreasing the distortion in medicinal images, a reversible watermarking is explored in this paper. Intended for implanting single bit of watermark in a transform factor, we practice numeral wavelet transform. The formed distortion is recompensed in the succeeding repetition as soon as a constant is altered in single repetition and this has been formulated using a novel methodology. Condensed alteration proportion is produced using Distortion Compensation technique. Upon 4 varieties of medical images comprising MRI of the brain, cardiac MRI, MRI of breast and intestinal polyp images, the anticipated scheme is verified. Through a single-stage wavelet transform, the extreme capability of 1.5 bpp is attained. With reference to capability and alteration, the anticipated scheme is greater to the state of-the-art mechanisms which are validated using Investigational outcomes. Keywords Health informatics system · MRI · Wavelet transform · Distortion compensation technique

S. Guntupalli (B) · M. Sreevani · M. Raja Department of ECE, CMR Engineering College, Hyderabad, India e-mail: [email protected] M. Sreevani e-mail: [email protected] M. Raja e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_11

97

98

S. Guntupalli et al.

11.1 Introduction In remedial uses as well as in the enhancement of investigative abilities, Medical image analysis is broadly utilized to assist Doctors. Confidentiality complications such as, alteration and unapproved admission might be initiated by means of the Broadcast of the remedial data over a communal web. Remedial image watermarking sources unique image adaptation and alteration even though digital image watermarking has marked the aforesaid complications [1]. The exclusive image matter needs to be conserved because even a slight alteration in medical image watermarking ought to undesirable influence on physician investigative [2]. Towards restoring both the exclusive image as well as watermark data adjustable watermarking has stood presented by means of an authentic manner in latest centuries. Patient confidentiality is reserved as well as the physician investigative procedure and conducts remain unaffected using this method. By means of a significant portion of Health Information System (HIS), Reversible image watermarking can remain deliberated. In the investigation communal, reversible watermarking has appealed a percentage of responsiveness in recent times. Towards attaining improved strength, watermarking by means of transform domain which is attentive on the wavelet transform [3], is employed in numerous latest revisions. Intended for reversible watermarking, Haar discrete wavelet convert remained utilized. By means of a quantization service, fourth stage of distinct wavelet convert was employed on behalf of implanting. En route for attainment of additional safety, Watermark statistics was preset using BCH coding [4]. Towards producing a distorted image, Selvam et al. initially employed numeral wavelet convert in addition to that they practiced distinct Gould transform in distorted factors then extended a capability of 0.25 for each pixel. Numeral wavelet convert was employed, by Cohen-Daubechies Fauraue. The histogram was pre-controlled towards avoiding the excess or in flow besides at that moment further data was produced [5]. For hardiness and softness of the distorted factors, Companding was practiced. Meanwhile here is a possibility of alteration in companding progression as a result further data was produced. Watermarking was distorted near Arnold pure view and excess or inflow remained stopped using post-dispensation towards improving the safety. A section of concern and non-section of concern remained detached inevitably at leading by means of adaptive threshold in dictator procedure [6]. Formerly, by means of bin histogram individually, inserting in every extent was employed. Centered on estimate fault extension, reversible image watermarking was anticipated. When excess or inflow arose transverse neighbors were assumed by means of implanting situations. Adjustably histogram fluctuating was accomplished openly to pixels or else estimate faults. On behalf of watermarking using exclamation fault extension 2 intellectual practices comprising “genetic algorithm” and “particle swarm optimization” were employed [7, 8]. Aimed at damage recognition, intermediary substantial bit replacement was employed by means of watermark implanting procedure, as well as a insubstantial watermark was implanted. Towards increasing the safety, Watermark bits were encoded earlier to implanting. In instance of encoded images, Reversible

11 Reversible Image Watermarking for Health Informatics Systems …

99

Fig. 11.1 Block diagram of the embedding procedure

watermarking can be advantageous. Therefore on behalf of encoded images, a contingency matter of Paillier cryptosystem was employed [9]. Similarly, for the encoded image portion histogram changing was employed. A unique reversible watermarking scheme for remedial images on tele remedy uses is anticipated in this paper. In 2 repetitions, Numeral wavelet purview is employed on behalf of watermarking in addition to that watermark bits are implanted in every sub-group. The aforementioned is likely to need two-bit implanting using minor alterations in this way. Likewise, constant alteration in the leading repetition is recompensed in the succeeding repetition. The remaining paper is structured as below. Centered on numeral wavelet convert, anticipated reversible watermarking is offered in Sect. 11.2. Towards investigational outcomes, Sect. 11.3 is devoted. Final statements are offered lastly, in Sect. 11.4.

11.2 Proposed Method By means of a convert purview rounded reversible image watermarking, the distinct wavelet convert is expansively employed in latest revisions. Pixels remain transformed in the custom of numeral standards towards the variable topic 1 in distinct

100

S. Guntupalli et al.

wavelet convert. Conservation of unusual numeral significance cannot be assured owed towards altering the constant standards in implanting stage in the custom of shortening. In the anticipated reversible image watermarking, number to number wavelet convert is engaged towards reporting this difficulty. Twofold stages comprising implanting plus abstraction remain convoluted in the anticipated process [10]. Wavelet constants stay transformed into a dualistic plot in the anticipated procedure using: Q(x) = mod

 x  2

,2

 (1)

Here, aimed at computing the remains, “mod” is a value in addition . is a ground value. The unusual constant assessment remains transformed by means of a continual level affording towards the waterline bit as well as equivalent dualistic plot. In the course of the restoration of the unusual image, this method may well generate the identical level aimed at dissimilar constants as well as origins uncertainty. Follower fundamental is formed as adjacent data and marks the procedure adjustable towards avoiding this uncertainty. Implanting as well as abstraction stages remain offered in further particulars in the foregoing. A. Embedding phase In Fig. 11.1 a summary of the anticipated implanting stage is offered. Shield image and a waterline symbol are the structure contributions and watermarked image and tracing fundamental remain the structure productivities. Comprising LL, LH, HL, plus HH, the shield image is distorted by means of single stage numeral wavelet convert, as well as 4 wavelet sub-groups are anticipated mainly. Implanting in undersizedfrequency sub-group and consuming great considerate towards the human optical structure, LL sub-group can be directed towards non-cognitive alteration. Therefore, 3 sub-groups of the extreme-frequency (LH, HL, and HH) remain elected on behalf of implanting affording towards proficiency necessities. Aimed at every particular implanting sub-group, the amount of water line moments meant for implanting remains distributed keen on twofold equivalent portions. Implanting stage consists of twofold repetitions, besides every single moment of leading portion remains implanted on a factor in the leading repetition. Succeeding portion of the waterline remains implanted in the formerly implanted factors in the succeeding repetition. In Algorithm 1, Implanting procedure aimed at twofold repetitions is described. Through converse numeral wavelet convert, the implanted image in wavelet are mains distorted posterior to the spatial field. Through significance slighter than zero as well as excess aimed at individuals through a assets superior than 255 Implanting procedure might indicates to a charge external of the adequate collection of the image standards that remain in flow aimed at pixels. Pixel standards remain condensed as well as their positions and their unique standards stand deliberated by means of adjacent data for these circumstances. Assume that wavelet constant remains (u, v), as well as the dimension of the waterline by means of adjacent data is (i) i = 1… and

11 Reversible Image Watermarking for Health Informatics Systems …

101

Fig. 11.2 Block diagram of the extraction procedure

waterline remains as well as waterline constant remains (u, v). Implanting procedure (Algorithm 1) is given below.

102

S. Guntupalli et al.

Dissimilarities of repetition 1 are recompensed in such a manner that Waterline data is implanted using repetition 2. The three advantages of the Implanting procedure are specified as follows. In the leading method, amount of improved constants remains insignificant which point’s towards small non-cognitive perversion. Furthermore in the succeeding method, dissimilarities organized factors commencing repetition 1 can remain recompensed in repetition 2. Using the capability of implanting in every single sub-band (LH, HL, HH), the situation remains likely to upsurge the implanting capability in the next method. Therefore, we have attained extreme volume of 1.5 bpp. Extra sturdiness can be attained using modest approaches such as reproducing in addition to this; the aforementioned is a significance declaration. B. Extraction phase In Fig. 11.2, general ideas of the anticipated abstraction stage remain obtainable. Watermarked image and tracing fundamental are the Participations in addition to it, enhanced unusual image and detached waterline logos remains the productivities. The adjacent data stands utilized to improve the excess/inflow standards rear to their unusual positions on top, by means of a pre-succeeding period. Spatial-purview forms of the watermarked image are recovered afterwards. Numeral wavelet convert is accomplished on watermarked image in addition to it, 4 sub-groups are acquired afterwards the pre-succeeding. In twofold repetitions through recompense procedure, Watermark data is detached after implanted sub-groups. Single moment remains detached commencing every wavelet factor on every single repetition. The reversed directive of the implanted remains the directive on which watermark data is detached. Commencing leading repetition, implanted watermark in the succeeding repetition is detached. Using reverse numeral wavelet convert, the unusual image is restored as a final point. For abstraction stage, the constant rate is (u, v ),in addition to dimension

11 Reversible Image Watermarking for Health Informatics Systems …

103

of watermark as adjacent data remains t (i) i = 1, …, and detached watermark is we, and improved factor is (u, v, ). The procedure of extraction (Algorithm 2) of the anticipated watermarking technique is given below.

11.3 Experimental Results Scheduled 4 gray scales remedial image information arrays comprising brain MRI, cardiac MRI, intestinal polyp and breast MRI, presentation of the anticipated technique remains established. In Fig. 11.3, Exemplary images commencing 4 remedial information arrays stay demonstrated. Towards the dimension of 512 × 512 whole images have remained extended. Comprising an identical amount of 1’s as well as 0’s (49% ones plus 51% zeros) Contribution watermark is a dualistic image. On instance of 4 remedial images, a capability-alteration outcome of the anticipated technique is conveyed in this paper. The typical outcomes of every information-routine remain similar to the images of Brain MRI, cardiac MRI, intestinal polyp as well as breast MRI stays comprising 80, 70, 100 and 100 images [10]. The extreme capability of 1.5 bpp by small alteration remains acquired which is presented using Replication outcomes. Correspondingly, the development in the capability starting 0.1–1.5 bpp doesn’t disturb the photographic superiority. A survey is implemented on occasion of Lena image for evaluation of the anticipated scheme using additional correlated approaches. In Fig. 11.4, a capability-PSNR

104

S. Guntupalli et al.

Fig. 11.3 Input image

graphics demonstrated. Related to the additional 6 equivalent approaches, this one is detected that, aimed at the identical capability, our scheme conceives improved PSNR. Likewise, a comparatively excessive quantity of implanting the photographic superiority remains quite satisfactory [11]. Additional capability can be acquired using extra alteration stages or repetitions in the anticipated procedure are essential to consider. Lastly, the unusual as well as the watermarked images remain presented towards estimating the optical superiority of the watermarked images, in Fig. 11.5.

11 Reversible Image Watermarking for Health Informatics Systems …

105

Fig. 11.4 Watermarked image

11.4 Conclusion To Novel reversible image watermarking method was presented for medical images based on integer wavelet transform. Improving distortion and embedding capacity was considered in the proposed method. The embedding process was performed in two iterations using a compensation approach. It was possible for modified coefficients in the first iteration, to recover its original value in the second iteration. One bit was embedded on iteration. Hence the maximum capacity of 1.5 bpp was obtained. Simulation results demonstrated that the proposed reversible image watermarking provided suitable capacity-distortion in comparison with the other methods.

106

S. Guntupalli et al.

Fig. 11.5 Recovered image

References 1. Dragani´c, A., Mari´c, M., Orovi´c, I., Stankovi´c, S.: Identification of image source using serialnumber-based watermarking under compressive sensing conditions. In: 2017 40th International Convention on Information and Communication Technology, Electronics and Microelectronics (MIPRO), pp. 1227–1232. IEEE (2017) 2. Rai, A., Singh, H.V.: SVM based robust watermarking for enhanced medical image security. Multimed Tools Appl. 1–14 (2017) 3. Huang, J., Chen, G., Shu, L., Wang, S., Zhang, Y.: An experimental study of clogging fault diagnosis in heat exchangers based on vibration signals. IEEE Access 4, 1800–1809 (2016) 4. Soleymani, S.H., Taherinia, A.H.: Double expanding robust image watermarking based on spread spectrum technique and BCH coding. Multimedia Tools Appl. 12, 1–9 (2016) 5. Tan, M.H., Li, Q., Shanmugam, R., Piskol, R., Kohler, J., Young, A.N., et al.: Dynamic landscape and regulation of RNA editing in mammals. Nature 550, 249–254 (2017) 6. Tran, L.: An interactive method to select a set of sustainable urban development indicators. Ecol. Indic. 61, 418–427 (2016) 7. Ruijters, E., Schivo, S., Stoelinga, M., Rensink, A.: Uniform analysis of fault trees through model transformations. In: 2017 Annual Reliability and Maintainability Symposium (RAMS), pp. 1–7. IEEE (2017) 8. Yuan, X., Elhoseny, M., El-Minir, H.K., Riad, A.M.: A genetic algorithm-based, dynamic clustering method towards improved WSN longevity. J. Netw. Syst. Manage. 25(1), 21–46 (2017). https://doi.org/10.1007/s10922-016-9379-7

11 Reversible Image Watermarking for Health Informatics Systems …

107

9. Wu, H.T., Cheung, Y.M., Huang, J.: Reversible data hiding in Paillier cryptosystem. J. Vis. Commun. Image Represent. 40, 765–771 (2016) 10. Han, H., Hwang, J.: Cruise travelers’ environmentally responsible decision-making: an integrative framework of goal-directed behavior and norm activation process. Int. J. Hospitality Manage. 53, 94–105 (2016) 11. Ramesh, G.P., Kumar, N.M.: Design of RZF antenna for ECG monitoring using IoT. Multimedia Tools Appl. 1–6 (2019)

Chapter 12

A Digital Image Encryption Algorithm Based on Bit-Planes and an Improved Logistic Map Mohammad Jabirulah, Amgoth Srinivas and Panduga Kavitha

Abstract In digital image encryption (DIE) procedure centered on bit-planes and an improved arrangement map is offered by this paper. By the enhanced logistic map, a chaotic series is produced primarily and in the original image the pixels are scrambled. Further into a high and undersized 4-bit matrix, the scrambled image is divided correspondingly. To produce a chaotic series that is extremely connected with the image as the fundamental, the undersized 4-bit matrix is presented addicted to an enhanced logistic category and for location scrambling plus the XOR arrangement of the extreme 4-bit matrix and the basic is used. Lastly, to acquire the cipher text image, two fold matrices are joined into image matrix which is of 8-bit. One-time pad characteristic is the implication of this procedure. In the terminologies of the histogram, to evaluate the safekeeping of image encryption, plaintext sensitivity, data entropy, and pixel relationship index, MATLAB simulation investigates are added. Validating that the procedure compromises upright encryption, Investigational outcomes display that the number of pixel changes ratio (NPCR) is superior to Ninety percentages then the data entropy of the cipher text image extents 7.99. Keywords Digital image encryption · Logistic map · Bit-plane · One-time pad

12.1 Introduction Over With intense and primitive layouts, Images are the universally replaced mode of data. Digital images are extensively used in many fields such as military undertakings, medication, industrialized production and civil security by the fast improvement of hypermedia expertise besides the attractiveness of the Internet. In the target trailing in M. Jabirulah (B) · A. Srinivas · P. Kavitha Department of ECE, CMR Engineering College, Hyderabad, India e-mail: [email protected] A. Srinivas e-mail: [email protected] P. Kavitha e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_12

109

110

M. Jabirulah et al.

video analysis, digital image practices are employed [1]. For peculiar identification, electroencephalogram or face identification is employed. Pixel scrambling, pixel assessment replacement, and the widespread usage of the twofold approaches are commonly included in the existing image encryption technology. Exclusive benefits offered by image encoding procedures centered ensuing the bit-plane ought to stand planned in the current years [2]. Being openly utilized on behalf of bit-level snarling plus unused in digital image encryption procedures built on bit-levels in addition to chaotic schemes might hold a disordered group or to generate chaotic arrangements, some Eigen values of the image are being used. Preceding the dualistic image is reconstructed, in order to get the encrypted image, and towards scrambling the location of the huge dualistic image, a chaotic structure is utilized [3]. The sequences produced by changed chaotic schemes remain reprocessed to scramble every bit-plane and the plaintext image is degenerated keen on a bit-plane image. Evaluate the variation in the middle of the chaotic pseudorandom number generator (CPRNG) plus the customary accumulation customary [4]. Into the key sequence generation procedure, a number of plaintext image data is offered. This has the features of one-image-onepassword and similarly improves the relationship concerning the key series and the plaintext image. Connected to certain acknowledgements of the original image, certain current image encryption algorithms are linked, whereas others who are only linked with the initial key are unrelated to the things of the original image [5]. DIE algorithm is presented in this paper that actively associates the data of unique image through the encryption method in account to resolve these complications and progress the safety of the image encryption algorithm. To achieve XOR procedure by means of extreme 4-bit matrix, this sequence is used [6]. For instance the algorithm has an important one-time pad characteristic, it advances the condition whereby the fundamental is only connected with certain Eigen standards of the plaintext image [7, 8]. By methods for the histogram, plaintext affectability, information entropy, and neighboring pixel association, the safety of the proposed system is evaluated through MATLAB simulation tests.

12.2 Related Knowledge 12.2.1 Image Bit-Plane The information is set down as binary data, denoted by 0 and 1 in computer memory structures. Preceding the computer, the dualistic gray scales of the similar location of every pixels of an alphanumeric image which is also termed as image bit-planes can arrange an image, which consumes equal proportions towards the original, since the gray rate of a digital image pixel is denoted by means of a 0 and 1 twofold categorization. The uses of bit-level, which remain an essential quality of digital images, are digital watermarks and digital image sturdiness coding. Think through

12 A Digital Image Encryption Algorithm …

111

Table 12.1 Image bit-plane position and weight Symbol

b(8)

b(7)

b(6)

b(5)

b(4)

b(3)

b(2)

b(l)

Position

8th

7th

6th

5th

4th

3rd

2nd

1st

Weight

27

26

25

24

23

22

21

20

Table 12.2 Percentage of Image Information Contained in Each Bit-Plane Position

8th

7th

6th

5th

4th

3rd

2nd

1st

Symbol

b(8)

b(7)

b(6)

b(5)

b(4)

b(3)

b(2)

b(l)

Percentage (%)

50.196

25.098

12.549

6.275

3.137

1.568

0.784

0.393

the model of an eight-bit gray digital image P. As described in Table 12.1, a particular pixel rate of P is placed in the PC and the image can be degenerated into eight bitplanes. The quantity of image data enclosed in every single bit-plane is dissimilar by means of the weightiness of every bit-plane remain dissimilar. Of each bit-plane, the portion of data enclosed in the images can be considered by: I(i) =

2i−1 , i = 1, 2, 3, 4, 5, 6, 7, 8 255

(12.1)

In Table 12.2, the conclusions are offered. There will be a regular growth in the measure of image data enclosed in the bit-levels commencing starting small value to big value (i.e., from first to eighth). Merely 5.882% of the overall image data is enclosed in the inferior four bit-levels (starting after first to fourth), while the left over 94.118% is enclosed in the superior four bit-levels (beginning from fifth to eighth). A bit-level degeneration of the Lena image is displayed in Fig. 12.1. The image clarity will turn out to be more and more of poorer quality, but some image information can still be shown starting from the 8th to 5th bit-plane by detecting. A definite measurement of plaintext image data is enclosed in these bit-planes.

12.2.2 Logistic Map In the chaotic system, Logistic maps are one of the types. They have been extensively used for several investigation grounds and require a modest mathematical method that provides multiple dynamic performances. The typical calculation for a logistic map is: xn+1 = µxn (1 − xn )

(12.2)

where 0 < µ < 4, x ∈ (0, 1), n = 0, 1, 2, …. When 3.57 < µ ≤ 4, the plot turns out on the way to be chaotic. The logistic map calculation has remained changed as

112

M. Jabirulah et al.

Fig. 12.1 Decomposition of Lena image based on bit-planes

well as prolonged to advance its dynamic features and enlarge the logistic mapping limitation series in place of the enhanced logistic map in [14] can be engraved by means of: ⎧ ⎨ xn+1 = (L(µ, xn ) × G(k)) − f loor (L(µ, xn ) × G(k)) (12.3) L(µ, xn ) = µxn (1 − xn ) ⎩ G(k) = 2k , k ∈ Z + , k ≤ 8 The chaotic sequence is more uniformly circulated in [0, 14] and this improved calculation reveals a chaotic performance when 0 < µ ≤ 4. On the early values of µ and x0 , the iteration of Eq. (12.3) is quite reliant on. The logistic map is prearranged by means of a fragmented purpose, as well as minor four bit-levels of the image to be encoded remain presented by means of a regulated factor to additional improvement of the uncertainty of the chaotic series. This enhanced logistic map is stated using: ⎧ ⎪ xn+1 = (L(µ, xn ) × G(k)) − f loor (L(µ, xn ) × G(k)) ⎪ ⎪ ⎪ ⎪ ⎨ L(µ, xn ) = µxn (1 − xn ) G(k) = 2k  ⎪ ⎪ ⎪ j, i 200

7.3

19 Energy Conservation Strategy for DC Motor Load Applications

(a)

185

(b) Cost Vs. Firing Angle

0.5

Energy Vs. Firing Angle Energy=VI*t(KWH)

0 72 54 52 38 34 30 27 25 18 9

0

Fig. 19.6 a Cost versus firing angle graph b Energy versus firing angle graph without load on motor

(b)

(a) Energy Vs. Firing Angle

Cost Vs. Firing Angle

Fig. 19.7 a Cost versus firing angle graph b Energy versus firing angle graph with load on motor

% saving in energy When motor runs for 60 h, Energy conserved is 0.5298 kWh, % Energy savings = (E −E)/E × 100 = 29.73%. Figure 19.6a, b shows the variation of cost and energy.

19.5.2 Output with Load All the calculations are carried out with load on motor and tabulated in Table 19.1. The calculations are carried out as explained in Sect. 19.5.1 and the results obtained are E = 1.49 kWh, E = 2.14 kWh, Total amount saved during conservation = Rs. 3.3, % Energy savings = 30.47%, Fig. 19.7a, b shows the variation of cost and energy used.

19.6 Conclusion A proper experimental set up is established and the analysis is carried out with and without load for the separately DC motor. From the results it can be observed that, even a small reduction in speed can give significant savings in energy. Motor running at 80% speed consumes less energy compared to one running at full speed. From the tabular column readings noted in the hardware setup, we observe that energy consumed when motor running at full speed is four times the energy consumed when

186

U. Hari Priya et al.

motor running at 300 rpm. The speed of DC motor is directly proportional to supply voltage. By changing firing angle in SCR “phase-angle controlled” drive; variable DC output voltage can be obtained which in turn changes speed of the motor. Here in this experiment 29.73% of energy is saved if motor is without load and it is 30.47% with load. At zero firing angle, motor takes complete power. When energy is compared at different firing angles, energy consumed at lower firing angle is higher than that of at higher firing angle. It can be concluded that with minor speed variation in a particular intervals can provide a significant savings (conservation) in energy. This concept can be utilized at large scale in the industries.

References 1. “Saving Energy through Innovation and Technology”, Infineon Technologies Automotive, Industrial & Multimarket Business Group, pp. 1–20 (2017) 2. de Almeida, Anibal T., Fonseca, Paula, Bertoldi, Paolo: Energy-efficient motor systems in the industrial and in the services sectors in the European Union: characterization, potentials, barriers and policies. Energy 28(673–690), 1–18 (2013) 3. Gilbert, W.B.: Optimizing drive systems for energy savings. Siemens Energy and Automation, usa.siemens.com/motioncontrol, pp. 1–11 (2012) 4. Anil, D.: Energy conservation in ceiling fan using BLDC motor. Int. J. Adv. Eng. Res. Dev. 4(7), 1–3 (2017) 5. Liu, Z.Z., Luo, F.L., Rashid, M.H.: Speed nonlinear control of DC motor drive with field weakening. IEEE Trans. Ind. Appl. 39(2), 1–7 (2003) 6. Bernard, A.: Speed Control of Separately Excited DC Motor Using Artificial Intelligent Approach, pp. 1–10 (2013) 7. Munteanu, T., Rosu, E., Gaiceanu, M., Dumitriu, T., Dache, C.: Energy saving control for DC motor drive systems. Przeglad Elektrotechniczny 87(12), 1–8 (2011) 8. Lin, L.F.: Nonlinear field weakening controller of a separately excited DC motor. In: International Conference on Energy Management and Power Delivery, pp. 1–6 (1998) 9. Erney Fabian, C.B., Diego Alejandro, P.P., Franklin Meer, G.A.: Design and implementation of single phase fully controlled bridge rectifier using PIC microcontroller. Int. J. Sci. Res. Publ. 6(1), 1–12 (2016) 10. Awad, A.S., Said, A.I.: Speed control of DC motor drives based on efficient utilization of energy and optimal performance. In: 14th International Conference and Exhibition on Electricity Distribution. Part 1. Contributions (IEEE Conf. Publ. No. 438), 2nd–5th, pp. 1–5 (1997) 11. Ferreira, F.J.T.E., de Almeida, A.T.: Overview on energy saving opportunities in electric motor driven systems—Part 2: Regeneration and output power reduction. In: 52nd Industrial and Commercial Power Systems Technical Conference (I&CPS), pp. 1–7 (2016) 12. Deepa, K., Jeyanthi, R., Kumar, M.V.: Efficient and compact power supply for robotic application. Advances in Recent Technologies in Communication and Computing—ARTCom, Lecture Notes in Electrical Engineering, LNEE, pp. 104–109 (2012) 13. Deepa, K., Sharika, M., Kumar, M.V.: Implementation of a SISO-ZVS push-pull converter fed DC servo motor. In: 5th India International Conference on Power Electronics, pp. 1–4 (2012) 14. Deepa, K., Kumar, M.V.: Performance analysis of a DC Motor Fed from ZCS-Quasi-resonant Converters. In: International Conference on Power Electronics (IICPE 2012), pp. 1–4 (2012) 15. Deepa, K., Mahalakshmi, R., Yuvasri, Kumar, M.V.: Bi-directional push pull converter fed four quadrant DC Drive. In: IEEE International Conference on emerging trends in Communication, Control, Signal Processing and Computer Application (C2SPCA 2013), pp. 1–6 (2013)

Chapter 20

End-to-End Delay Analyses via LER in Wireless Sensor Networks K. Ramesh and V. Kannan

Abstract Virtual lives of present generation peoples are facing lack of green environmental and good health condition. Emerging developmental applications of Wireless Sensor Networks (WSNs) is the most necessary thing to deploy and monitor our physical world like disaster, agriculture and healthcare. For the most challenging characteristics of WSN facing energy drain out, unlike to easy or possible to replace the battery of WSN and also an end-to-end performance of reliability of data communication due to delay. Through this work made an analysis of IEEE 805.15.4 MAC with changing routing direction. Moreover, considering the CSMA/CA MAC with Redundant Radix Based Number size communication (RRBNs) system. Added to multi-hop networks with novel based routing schemes of Low Energy Routing (LER) direction. In particular, a various load condition of networks determine different performance regarding the delay, energy consumption and reliability of communication links. Finally, determine the network equations based on Markov chain model and communicating the data via RRBNs communication system and also finding the solution at critical condition for various load distribution of the WSNs. Keywords IEEE 802.15.4 MAC · CSMA/CA · Redundant radix based number size communication · Markov chain · Low energy routing

20.1 Introduction The main challenges faced by the WSNs are the energy drain out and the reliability of end-to-end data communication. Reference [1] described an accurate estimation of energy efficient Heterogeneous LEACH for WSNs. A comprehensive end-toend performance evaluation of priority and delay aware medium access protocol. K. Ramesh (B) St. Peter’s Institute of Higher Education and Research, Chennai, India e-mail: [email protected] V. Kannan GMR Institute of Technology, Rajam, Andhra Pradesh, India e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_20

187

188

K. Ramesh and V. Kannan

The IEEE 802.15.4 protocol has been implemented by modifying the physical layer parameters and achieved the minimum delay responsiveness of WSNs as given in the reference [2]. The mathematical equations are derived through analytical method to evaluate, analyze and compared the results of end-to-end reliability has been described in the paper [3]. In many of the applications involved in WSNs, we focus on emerging applications of disaster and healthcare monitoring. To address these emerging issues of metric value towards end-to-end delay and higher energy consumption reading. In fact, IEEE 802.15.4 MAC protocol and interdependence routing directions are taking a specified route to improve the reliability of data communication and extend the lifetime of sensor nodes. In this situation mentioned, we derive the model of a Markov chain method is presented for single-hop networks at different traffic generation rate. Extended the multi-hop topology at different traffic load condition according to the routing direction along with path route. Thus, new approaches are needed to estimate the delay value and energy consumption value. The problem of energy consumption in single data packets or multiple of data packets has to convert the binary code into Redundant Radix Based Number size ¯ has (RRBNs). The encoding/decoding data string keep silent of 1’s, in place of 1’s the number string of 0’s and 1’s. In particular, we focus on the CSMA/CA MAC with RRBNs communication of IEEE 802.15.4 standard, extended to multi-hop topology, according to the routing specifications of LER.

20.2 Related Work A mathematical model of HTNs address the better performance as compared with conventional ZigBee based WSNs [3]. Overhearing and packet coding [4] adapted in a high-fidelity analytical model and incorporating the shortest path algorithm to evaluate the communication delay. Reference [5] described the analytical model of Wireless Sensor Networks with sleep nodes and derived the equations of data delivery delay, energy consumption with respect of DTMC (Discrete–Time Markov Chain) and Fixed-Point Approximation method. Remote observations of Crop Leaf Area Index evaluated through Wireless Sensor Networks [6]. The collection of data from LAI and validated with MODIS (Moderate Resolution Imaging Spectro radiometer) LAI with low cost and low-energy consumption. Investigated, link and node failures in alternating WSN operation like active and sleep mode via the analytical model of two-dimensional Markov representation and comparative solutions presented [7]. A new protocol has been presented in [8, 9], Reconfigurable Directional Antennas (RDA) getting opportunities to minimize the loss of data packets in Wireless Sensor Networks. In [10], proposed a new protocol of Delay Guaranteed Routing and MAC (DGRAM) Protocol for Wireless Sensor Networks working towards TDMA MAC and achieved lesser delay than the analytical model delay. Some of the developed papers [11, 12] that described real-time routing in Wireless Sensor Networks. In the previous work of [13, 14] has been implemented the MAC protocol to achieve

20 End-to-End Delay Analyses via LER in Wireless Sensor Networks

189

energy efficiency, among them proposed the Sensor-MAC (S-MAC) category of synchronous duty cycle MAC protocols which all the sensor nodes under the same cluster heads are co-ordinate by changing their modes of sleep and wake-up. Reference [15], The main impact of a battery depends on the working environment of the transceiver at sensors, the metric to validate the energy cost of both communication and protocols were presented in [16], energy efficient as an overhead cost of on top of the Eb/No requirement. In contrast to EbT based communication protocols, a novel method of communication strategy is called as Communication through Silence (CtS) described in [17] that included that the silent periods of energy-based transmissions. Novel based communication scheme has been proposed in [18] that converts a binary coded data using a Redundant Radix Based Number (RRBN) representation and used in silent periods to communicate the bit value of ‘0’. RBNS that utilizes the digit from the set of (−1, 0, 1) then to represents a number with base is known as radix 2.

20.3 RRBNs Encoding and Decoding Method The RRBNs represents an input string by ‘1’ at (x + i) bit position followed by a 1¯ at bit position i exchanged in the input string, and another remaining intermediate bit of 1’s changed into zeros. Transmitter module is switched-off while the RRBNs string for 0 bit values and hence the power consumption is lesser than LER bounds of Markov chain model. Combinations of both approach of RRBNs and LER bounds, a significant level of energy reduction have been achieved during transmission period. v = 2i + 2i+1 + · · · + 2x+i=1 = 2x+i − 2i

(20.1)

Rule 1: Considering the input data string has (1111100)2 , with two ‘0’s trapped between runs of 1’s. Hence, read the substrings from LSB and spilt into every four bits. The first four-bit substring has (1100)2 , a run of x 1’s (x > 1) begins at the bit position of i, consists of a ‘1’ at bit position (x + i) as converted into 1¯ in the place of i with ‘0’ in all intermediate bits. In this case at bit position ‘i’ as ‘0’, as continue the bit position at ‘i’ as ‘0’. The input string of the first four bit from LSB, replace the bit position ‘i’ as ‘0’ and hence the position as same. And other intermediate bits replaced with (x + i) pattern, that is x = 2 (‘1’s), (x + i = 2 + 1 = 3), the equivalent RRBNs is (100)r bn . A substring of remainingfour bit  as (0111)2 , x = 3(‘1’s), (x + i = 3 + 1 = 4), the equivalent RRBNs has 1001¯ r bn . The total input data string   ¯ replaced with 1001100 , the number of 1’s reduced from the input string and r bn then achieved the low energy reading. ¯ durRule 2: After implementing the rule 1, the RRBNs data string contains 11, ¯ 11, ¯ 11, ¯ but this pattern system is not ing transmission the digit patterns with 1¯ 1, possible to encoding and decoding process transmission and receiving period. However, every occurrence of the 11¯ is converted into equivalent bit pattern of 01¯ in the

190

K. Ramesh and V. Kannan

Table 20.1 Conversion value of RRBNs RRBNs digits 1¯

Transmitting symbols

0

00

1

01

10

  ¯ RRBNs substring of 1000100 . Furthermore converting the digit patterns accordrbn ing to Table 20.1 for the RRBNs encoding symbols. The final symbol of RRBNs is (10001000)rbn , denoting the bits through hardware circuit is implementing the single-pass algorithm of x(i) and y(i). Hence, the receiver side providing equivalent hardware circuit to converting RRBNs into equivalent binary digit pattern of original data packet. Moreover, achieving low energy reading by depicts Eq. (20.2) into Eq. (20.3). After applying the reduction rule, 1 in the RRBNs coded data substring ¯ likely equal to 2S +(n + 2)·2n−3 = (3n + 2)·2n−3 . Adapting contain the 1’s and 1’s the LER bounds and obtained by implementing reduction rule 1 is given by, ηe = 1 −

3n + 2 (3n + 2)2n−3 =1− n n·2 8n

(20.2)

where, S = n · 2n−3 , for example n = 8 is equal to ηe = 59.4%, where as n = 1024, ηe = 62.5%. After applying both reduction rule 1 and 2 in the RRBNs coded data ¯ likely equal of (3n + 2) · 2n−3 − (n + 2)2n−3 + 2n−1 = substring contains 1’s and 1’s n−2 (n + 2)2 . ηe = 1 −

n+2 4n

(20.3)

Inserting, n = 8, and getting the value of ηe = 68.75% and n = 1024, ηe = 75%.

20.4 End-to-End Delay in WSN The reliability of end-to-end link of the WSN, the data packets are dropped due to two factors involved in the transmission period (i) channel busy (ii) re-transmission limits. Among these factors, the channel busy involves access failure during transmission, happened data packets fail to get a clear channel within the back- off limits of m + 1. After l + 1 time repeatedly collisions occurred and the data packets are discarded when the transmission fails to transmit the data packets. The probability that the data packets can be dropped due to channel busy or channel access failure is obtained in Eq. (20.4).

20 End-to-End Delay Analyses via LER in Wireless Sensor Networks

Pc f,dl

     m+1 m+1 l+1 kdl 1 − (Pcol,dl ) 1 − kdl   = m+1 1 − Pcol,dl 1 − kdl

191

(20.4)

The probability of a data packet is dropped, since the re-transmission limit by    m+1 l+1 Pcr,dl = Pcol,dl 1 − kdl

(20.5)

Solving Eqs. (20.4) and (20.5), the reliability of end-to-end link of WSN is Rdl = 1 − Pc f,dl − Pcr,dl

(20.6)

Multi-hop reliability is the product of each link reliability derived in Eq. (20.6).

20.4.1 Delay Analysis of Multi-hop Networks The fundamental reason for causing the delay of Ddl in a WSN for a successfully received data packet at predefined period when the data has ready to be transmitted. According to the Markov chain model considering that the WSN has source node of Ni that forwards the data packet up to neighbour availability of nodes, i.e., destination node N j that creates the traffic Fdl due to forwarding the data packet. Consider the rate of traffic generation λ at a particular sensor node and note down the forwarded traffic in the link dl as follows in the Eq. (20.8). Fdl =

gl pkt/s ad U nit Backo f f Period

(20.7)

where gl is a data packet to transmit in every unit time during the channel is in the ideal state. Where a Unit Back-off Period is the duration of basic unit time in IEEE 802.15.4. The average end-to-end (e2e) delay Ddl states that the data packet can be successfully received without collision of packets from the source node to destination node, when the data packet is ready to be the relay at the head of MAC queue. The source node has been transmitted a data packets successfully through kth attempt and the expectation delay as follows. (20.8)

192

K. Ramesh and V. Kannan

Where, k    E[Tb ] E Ddl,k = Ttr + Tc +

(20.9)

h=0

The Tb is the back-off delay, Ttr and Tc are transmission and collided data packet transmission of time periods in successful transmission. The is the event of interest denotes the occurrence of data packet transmission at the period of time (t + 1), where as the symbol mentioned at the event of interest within n attempts that the data packets successfully transmitted.

(20.10)   m+1 where 1 − kdl is the maximum channel accessing probability of m back-off stage. The expectation of back-off stage delay of E[Tb ] has been characterized as follows in the derived equations. E[Tb ] =

m 

  Pr ( Di |D)E Tb,i

(20.11)

i=0

Tb,i = (1 + i)Tse +

i 

h Tb,k

(20.12)

k=0

The sensing time Tse is the in the unslotted mechanism. Probability of event Di is well known as the successful sensing event within m attempts of D, mentioned that the node sensing an idle channel in CCA and as given in the Eq. 20.13. Pr ( Di |D) =

i kdl (1 − kdl ) m+1 1 − kdl

(20.13)

End of the event that substitute parameters from the Eqs. (20.9) to (20.13) in the Eq. (20.8) and get the average end-to-end delay in the single path. Similarly, the multi-hop is the sum of the single path delay in continuing path from the transmitter.

20 End-to-End Delay Analyses via LER in Wireless Sensor Networks

193

20.4.2 WSN Low Energy Routing Direction Routing metric decides the shortest path between the link of the source node and a destination node. According to the delay factor, the routing metric requires to be updated and considering the sensor network traffic in dynamic behavior, adapting statistical analysis in the dynamic network. The effect of the routing metric has described by the matrix form of φ ∈ R(N +1)x(N +1) . Denoting the link of dl(i, j), which represents the element of G i, j . G i, j

maxφi,h = Pr φi, j = max Nh ∈ ϑi

(20.14)

where ϑi is the candidate receiver associated with parent set ϑi ⊂ Pi . The distribution of the successful traffic flows along the sensor network have been modeled by the matrix G. The vector type of traffic generation probability G = [G 1 . . . G N ] is the solution of the system traffic flow balance equation of G = GT + , in a steady state condition and then follow this equation in the matrix form. G = Λ[I − X ]−1

(20.15)

And define a matrix X, X i, j = G i, j Q dl , where I is the identity matrix, I ∈ R(N +1)×(N +1) . The energy reading of multi-hop network can be obtained from the extension equation of single-hop network during idle listening state by idle queue stage in the Markov chain that includes the transmitting, receiving and ACK of a data packet.

20.5 Validation Results in WSN The beneficial simulations is based on Monte Carlo analysis under the specifications of the IEEE 802.15.4 with various load level of traffic generation pattern and sensing range of the sensor nodes. Considering on both single-hop and multi-hop sensor network model, simulation carried via ns-2 simulator and compared the analytical results with MATLAB tool. The first set of simulation result and validate the analytical model for considering full sensing range of |l | = N odes after that implementing the reduced sensing range of |l | = 5. The single-hop network with homogeneous traffic pattern, validate the reliability by changing the packet generation rate at different load condition respect of number of sensor nodes (N = 14, N = 20). The simulation and analytical model results can be validated as shown in the Figs. from 20.1 to 20.5, the variation between simulation value and analytical value is about only 3%. The results shown in Fig. 20.1 as the reliability of the sensor network is lesser than the full sensing capability by means of reducing the sensing capacity.

194

K. Ramesh and V. Kannan

Fig. 20.1 Reliability of single-hop network

However, in the case of higher traffic condition and number of nodes, the validation value is the negative impact on the reliability as shown in the graph. The average delay of the single-hop network that presented with homogeneous traffic pattern under varying of packet generation rate as shown in Fig. 20.2. Moreover, the positive impact curve is due to the separate busy channel indication of reducing the sensing capacity and increasing the packet generation rate. The significant effect of the delay curve is lesser value in the reducing sensing capacity and as mentioned in the same number of sensor nodes (N = 10) and comparing with full sensing capacity of a network model. The contrast of simulation and analytical results is moreover same Fig. 20.2 Average delay of the single-hop network

20 End-to-End Delay Analyses via LER in Wireless Sensor Networks

195

Fig. 20.3 Energy consumption of single-hop network

for successfully received packets. The validation results, the average delay, in turn decreasing the value, which in turn decreasing the busy channel probability. Represents Fig. 20.3 of sensor network energy reading regarding abrupt changing of traffic load. The major impact of energy reading in the sensor nodes changing their status of sleep and awake time, in this system of status belong to be in MAC protocol. Validated the energy consumption which includes RRBNs coding system, to reducing the energy loss while transmitting and receiving of data packets. Effective and minimum energy reading as shown in Fig. 20.3, which included the Low Energy Routing (LER) scheme and RRBNs coding system. Adapting lower sensing capacity and in turn the network effects of data collision and more energy reading in the sensor node. Considering multi-hop topology, assume each node generates same traffic load. Forwarded the data in multi-hop topology which selects the various paths to reach the destination. The comparison of reliability and delay analysis as shown in Figs. 20.4 and 20.5 respectively. In Fig. 20.4 presents the validated results belongs path 1 and 2, hence path 2 is lesser reliability than path 1, a reason behind is the low reliability in path 2 which constitutes more traffic than path 1. In multi-hop network, the path 1 and 2 are coupled with other sensor nodes, so dominant the traffic load. Regarding end-to-end reliability performance in path 1 is much better and as shown in Fig. 20.4. The reason is that the negative impact of path 2 is accessed more contention back-off windows than path 1. Hence, the average number of contenders in each time unit increases. Based on reliability, the end-to-end delay is minimum in path 1 as shown in Fig. 20.5, with an effect of minimum traffic load included in the path 1 multi-hop network.

196

K. Ramesh and V. Kannan

Fig. 20.4 E2E reliability of multi-hop network

Fig. 20.5 E2E delay of the multi-hop network

20.6 Conclusion In this paper, presented a model of single-hop and multi-hop sensor network. Validated the CSMA/CA MAC with RRBNs and adding with novel routing scheme of LER direction at various traffic load condition. Moreover, then, discussed the end-toend performance via reliability, end-to-end delay and energy consumption of different load network path of both single and multi-hop networks. The results shown that minimum delay and also lower level of energy consumption adapting IEEE 805.15.4 Markov chain analysis. The various effects of results have been directly depending on the carrier sensing limit of a sensor network. The simulation result is very close with analytical results, nearly 3–4% variation, which influences with sensing

20 End-to-End Delay Analyses via LER in Wireless Sensor Networks

197

range. Added advantageous view of performance validation is achieved through the combination of LER direction and RRBNs coding scheme.

References 1. Daraghma, S.M., Nuray, A.T.: A new energy efficient clustering based protocol for heterogeneous wireless sensor networks. J. Electr. Electron. Syst. 4(3), 1–7 (2015) 2. Erol-kantarci, M., Al-Anbagi, I., Mouftah Hussein, T.: Priority- and delay- aware medium access for wireless sensor networks in the smart grid. IEEE Syst. J. 1–11 (2013) 3. Pradhumna, L., et al.: Modeling latency and reliability of hybrid technology networking. IEEE Sens. J. 13(10), 3616–3623 (2013) 4. Liet, J.-W., et al.: Analytical model for coding-based reprogramming protocols in lossy wireless sensor networks. IEEE Trans. Comput. 66(1), 24–37 (2017) 5. Chiasserini, C.F., Garetto, M.: An analytical model for wireless sensor networks with sleeping nodes. IEEE Trans. Mob. Comput. 5(12), 1706–1718 (2006) 6. Qu, Y., et al.: Crop leaf area index observations with a wireless sensor network and its potential for validating remote sensing products. IEEE J. Sel. Topics Appl. Earth Observations Remote Sens. 7(2), 431–444 (2014) 7. Fredrick, A., et al.: An analytical model for bounded WSNs with unreliable cluster heads and links. In: 40th Annual IEEE Conference on Local Computer Networks, pp. 201–204. Florida, USA (2015) 8. Le, T.N., et al.: Improving energy efficiency of mobile WSN using reconfigurable directional antennas. IEEE Commun. Lett. 20(6), 1243–1246 (2016) 9. Gao, Y., et al.: COPE: improving energy efficiency with coded preambles in low—power sensor networks. IEEE Trans. Ind. Inf. 11(6), 1621–1630 (2015) 10. Shanthi, C., Shahoo, A.: DGRAM: a delay guaranteed routing and MAC protocol for wireless sensor networks. IEEE Trans. Mob. Comput. 9(10), 1407–1423 (2010) 11. Lu, C., He, T., Stankovic, J.A., Abdelzaher, T.: SPEED: a stateless protocol for real-time communication in sensor networks. In: IEEE International Conference of Distributed Computing Systems (46) (2003) 12. Ekici, E., Felemban, E., Lee, C.G.: MMSPEED: Multipath Multi-SPEED Protocol for QoS guarantee of reliability and timeliness in wireless sensor networks. IEEE Trans. Mob. Comput. 5(6), 738–754 (2006) 13. Estrinb, D., Ye, W., Heidemann, J.: Medium access control with coordinated adaptive sleeping for wireless sensor networks. IEEE/ACM Trans. Netw. 13(3), 493–506 (2004) 14. He, C., Zhang, Y., Jiang, L.: Performance analysis of S-MAC protocol under unsaturated conditions. IEEE Commun. Lett. 12(3), 210–212 (2008) 15. Martinez-Bauset, J., Lakshmikanth, G., Li Frank, Y.: Performance analysis of synchronous duty-cycled MAC protocols. IEEE Wirel. Commun. Lett. 4(5), 469–472 (2015) 16. Andrew Wang, Y., Charles Sodini, G.: On the energy efficiency of wireless transceivers. IEEE ICC Proceedings, pp. 3783–3788 (2006) 17. Zhao, Q., et al.: Resurce constrained signal processing communications and networking. IEEE Sign. Process. Mag. 24(3), 12–14 (2007) 18. Sinha, K.: An energy efficient communication scheme for applications based on low power wireless networks. In: IEEE Consumer Communications and Networking Conference (2009)

198

K. Ramesh and V. Kannan Dr. K. Ramesh was born in Odugathur, Tamilnadu, India on 10th June 1966. He received his Bachelor Degree of Electrical and Electronics Engineering from Anna University in the year 2007. Masters Degree in VLSI Design from Sathyabama University in the year 2011 and Ph.D., from St. Peter’s University, Chennai, in the year of 2017. He has presently worked as the Professor. His current research focused on Wireless Sensor Networks and his interested areas of Low-Power VLSI design, WSN based Electrical Apparatus and Nano Electronics. He has eight research publications in National/International Journals and conferences to his credit.

Dr. V. Kannan was born in Ariyalore, Tamil Nadu, India in 1970. He received his Bachelor Degree in Electronics and Communication Engineering from Madurai Kamarajar University in the year 1991, Masters Degree in Electronics and control from BITS, Pilani in the year 1996 and Ph.D., from Sathyabama University, Chennai, in the year 2006. His interested areas of Research are Optoelectronic Devices, VLSI Design, Nano Electronics, Digital Signal Processing and Image Processing. He has more than 200 Research publications in National/International Journals/conferences to his credit. He has 25 years of experience in teaching and presently working as Professor in GMR Institute of Technology, Rajam, Andhra Pradesh, India. He is a life member of ISTE.

Chapter 21

Multi Band Antenna System for Quality Evaluation Application of Apple Fruit Angeline M. Flashy and G. P. Ramesh

Abstract This paper presents an antenna sensing technique for quality assessment of Apple samples. The Compact rectangular 2 × 2 antenna array is designed with Bandpass filter for the triple band microwave sensing at 2.4/3.5/5.8 GHz and the parameters are analyzed using Vector Network Analyser. The Microstrip Antenna senses and reads the spectra produced from the apple sample for quality evaluation. In Existing, Granulation of Different Apple fruits can be analyzed by using Atomic Absorption Spectrophotometry (AAS) technology and Signal Processing. To overcome the cost-effectiveness, high-profile, the grading and quality evaluation is done by microstrip antenna and Wavelet Decomposition Technique in the proposed work. The parameters for Apple sample quality evaluation such as Standard deviation, Variance, Skewness, Crest factor, RMS and ZCR at different positions was investigated from 2.2 to 7 GHz using MATLAB. It was found that the reflected signal from the sample provides the obvious variation and can be used as indicators for justification. The analysis of device stability with Stability factor and the input and output impedance matching with Smith Chart and Tuning are analysed and validated with the simulation of S parameters using Advanced Design System (ADS) Simulation tools. The designed antenna array evaluates the grade of apple fruit and report the data to the consumer through IOT. Keywords Band pass filter · Standard deviation · Variance · Skewness · Crest factor · RMS · ZCR · ADS · Stability factor

A. M. Flashy (B) · G. P. Ramesh St. Peter’s Institute of Higher Education and Research, Chennai, India e-mail: [email protected] G. P. Ramesh e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_21

199

200

A. M. Flashy and G. P. Ramesh

21.1 Introduction Fruit nutrition is an important part of the human diet; therefore it is necessary to evaluate and determine its quality. The analysis and grading of fruit quality is also a critical task in the commercial market. With the existing growing need for low production costs with high efficiency, the agricultural industry has to overcome the number of challenges, including maintenance of high-quality standards and assurance of food safety while avoiding liability issues [1–5]. Grading food products for different markets has become crucial to meet the challenges. Microstrip patch sensors have become useful and widely used in agriculture, industrials, food products, telemedicine fields, etc. [6, 7]. Diseases in fruits cause a difficult and unsolved problem in the agricultural industry [8]. Another method is Atomic Absorption Spectrophotometry (AAS) [9]. Each sample was connected to the AAS and the grading of apple was analysed. The Antenna sensing technique can improve accuracy since it penetrates deep into fruits compared to Near Infrared (NIR). Microwave sensor operates based on the determination of dielectric properties of unknown objects [10, 11]. It is designed with inverted C Shape for the triple band microwave sensing at 2.4/3.5/5.8 GHz, the LPF and HPF are cascaded to develop a Bandpass filter (BPF) [12–15], for passing frequencies 2.4/3.5/5.8 GHz and analyzed. The proposed antenna parameters are fabricated and analysed using Vector Network Analyser results in cost-effectiveness.

21.2 Dielectric Properties for Quality Assessment To assess the qualities of living tissues of fruits, there is a need to develop informative sensing techniques. The dielectric properties of biological and food products have become important parameters in food engineering and technology. Dielectric spectroscopy widely covers the extraordinary frequency spectral range from 10−6 to 1012 Hz. There are number of different dielectric polarization mechanisms operating at the molecular or microscopic levels is shown in Fig. 21.1. The equation for relative complex permittivity ε* is represented as in Eq. (21.1): ε * = ε − jε where ε denotes the dielectric constant and ε denotes the loss factor and j =

(21.1) √

− 1.

21.2.1 Grading of Apple Fruit A better electrical characterization of the dielectric properties of fruits is required for this purpose. Different Apple samples were graded such as Fresh Apple, Apple

21 Multi Band Antenna System for Quality …

201

Fig. 21.1 Dielectric constant and loss factor variations

Table 21.1 Grading of apple samples S.No

Samples

Description

1.

Fresh apple

A fresh, high-quality apple should have a pleasant aroma

2.

Apple scab

Gray or brown corky spots are present on the surface of apple

3.

Apple blotch

The surface of fruit having fungal disease

4.

Apple rot

spore-bearing structures appear in concentric circles on the diseased apple surface

5.

Wax coated apple

Apples are often coated with a layer of wax to look more appealing

Scab, Apple blotch and Apple rot, Wax Coated Apple at different positions was investigated from 2.2 to 7 GHz using MATLAB is shown in Table 21.1.

21.3 Antenna Design and Geometry The Single rectangular antenna and 2 × 2 antenna array is designed with FR-4 substrate of dielectric constant 4.4. The dimensions for the proposed single antenna and antenna array are given in Table 21.2. The Geometry of the Single rectangular antenna and 2 × 2 antenna array is shown in Fig. 21.2.

21.3.1 Prototype Design of 2 × 2 Antenna Array The Prototype design and fabrication model of the antenna array is shown as Fig. 21.3. Return loss (S11 ) is the important parameter to analyse the effective power delivery of the designed antenna performance in wireless environment.

202

A. M. Flashy and G. P. Ramesh

Table 21.2 Dimensions of the rectangular antenna Parameters

Dimensions (mm)

Rectangular patch

Length (L1 )

8.9

Width (W1 )

13

Inverted c-shaped patch

Length (L2 )

6

Width (W2 )

10

Length (L3 )

1.4

Width (W3 )

1

Length (Lf )

1.8

Slot Feed line

Width (Wf ) Substrate thickness (Fr4)

Fig. 21.2 Geometry for the antenna

Fig. 21.3 Layout design of antenna array

1 1.6

21 Multi Band Antenna System for Quality …

203

Fig. 21.4 Simulated return loss S11 and radiation pattern of antenna array

The simulation plot of return loss S11 and Radiation Pattern of antenna array is − 11.850 dB at 2.4 GHz, −12.186 dB at 3.39 GHz and −20.050 at 5.8 GHz as shown in Fig. 21.4. The narrow bandwidth required by IoT applications are obtained.

21.4 Antenna Sensing Technique The Antenna sensing technique can improve the accuracy since it receives the reflected signal from the fruit and the coupled signal to the antenna, oriented perpendicular to the fruits, which provides the obvious variation and can be used as indicators for justification from 2.2 to 7 GHz using MATLAB. The Reflected signal from the Apple is recorded using Sigview tool and analysed is shown in Fig. 21.5.

Fig. 21.5 Recorded signal from apple fruit

204

A. M. Flashy and G. P. Ramesh

Fig. 21.6 Stability factor (K) versus frequency

Stability Factor, K

20

Stab(S)

15 10 5 0 -5 7.0

6.5

6.0

5.5

5.0

4.5

4.0

3.5

3.0

2.5

2.0

freq, GHz

21.5 Stability Analysis of Antenna System Proper Bias network design is essential for any nonlinear circuit design as it is essential to ensure that right amount of bias reaches the device and also it doesn’t load/leak the desired RF energy. Choice of bias network topology is pretty much dependent on the frequency of operation. For lower frequencies designers can use Inductors/Choke in the DC bias path and for higher frequencies high impedance quarter wavelength line is the preferred choice. Necessary and sufficient conditions for device to be stable are Stability Factor (K) >1, Stability Measure >0. The Stability Factor versus Frequency and RF Power is shown in Fig. 21.6. The Stability Factor of PA is achieved as below 1.

21.6 Data Transmission Using IOT The measured Apple sample data is transferred using IoT technique. The data absorbed by the designed antenna are transferred from the source to the destination point using a wireless technique. The Launchpad is a type of Wi-Fi device, which would be programmed using the Proteus software and designed using antenna design structure. The antenna will receive the entire information from the source and then the collected data transferred using the IoT device. The entire system minimized in a small chip which is more useful for the farmer peoples. The designed antenna is connected to the Launchpad device to transmit the data to the operator through wireless IOT system. Figure 21.7 describes the ADC coding for the programmer to operate the system in the wireless medium.

21 Multi Band Antenna System for Quality …

205

Fig. 21.7 Circuitry of the launchpad circuit

21.7 Evaluation of Apple Sample To analyse the grading of Apple sample, One Dimensional Wavelet Transform is done by loading the electrical signal as Wavelet 1-D tool parameters such as, (a) Standard deviation, the measure of the signal fluctuations from the mean. (b) Variance, the measure of the power of the signal fluctuations, (c) Skewness, a measure of the asymmetry of the data around the sample mean. (d) Crest factor, specifies the properties of an electrical system such as the purity of a signal or waveform. (e) RMS, root-mean-square value of a signal. (f) ZCR, Zero Crossing rate is the threshold for a given sample and the samples are analysed and obtained the results with high performance and reported the data to the consumer through IOT is shown in Table 21.3. Table 21.3 Evaluatio analysis of apple sample Samples

Natural apple

Apple scab

Apple rot

Apple blotch

Wax coated apple

SD

3.72

3.16

3.04

3.44

2.58

Variance

12.5

10.03

9.28

11.20

8.92

Skewness

0.03

0.09

0.02

0.008

0.01

Crest factor

3.06

4.42

4.26

4.48

4.48

RMS

3.72

3.16

3.047

3.348

2.98

ZCR

3339

2928

3252

2934

2934

206

A. M. Flashy and G. P. Ramesh

21.8 Conclusion The microwave sensing technique for quality assessment of fruits operates at resonant frequency of 2.4/3.5/5.8 GHz is analysed. The Compact rectangular 2 × 2 antenna array senses and reads the spectra produced when the samples excited by radiation. The parameters such as Standard deviation, Variance, Skewness, Crest factor, RMS and ZCR of the sample fruit such as Natural Apple, Apple scab, Apple blotch and Apple rot, Wax Coated Apple at different positions was investigated from 2.2 to 7 GHz using MATLAB. It was found that the antenna senses and monitors the Quality assessment of Apple fruit. This cost effective antenna sensor is a good candidate for fruit classification.

References 1. Rahman, M.M., Islam, M.F.: Double U-slot microstrip patch antenna for WLAN and WiMAX applications. Int. J. Comput. Appl. (0975–8887) 162(6) (2017) 2. Manouare, A.Z., Ibnyaich, S., Idrissi, A.E., Ghammaz, A.: A compact dual-band CPW-Fed planar monopole antenna for 2.62–2.73 GHz frequency band, WiMAX and WLAN applications. J. Microwaves, Optoelectron. Electromagn. Appl. 16(2) (2017) 3. Kumar, R.D., Ramesh, G.P.: Dipole Micro-strip patch antenna design for tri-band frequencies. J. Eng. Appl. Sci. 12(11), 2845–2852 (2017) 4. Alam, M.N., Bhuiyan, R.H., Dougal, R.A., Ali, M.: Concrete moisture content measurement using interdigitated near-field sensors. IEEE Sens. J. 10(7), 1243–1248 (2010) 5. Kumar, R.D., Ramesh, G.P.: Reconfigurable antenna design for soil testing to improve soil quality. Int. J. Eng. Technol. 7, 70–73 6. Balanis, C.A.: “Antenna Theory” Analysis and Design, 3rd edn. Wiley, Hoboken (2005) 7. You, K.Y., Salleh, J., Abbas, Z., You, L.L.: A rectangular patch antenna technique for the determination of moisture content soil. In: PIERS Proceedings, Cambridge, USA, pp. 850–854, 5–8 July 2015 8. Limpiti, T., Krairiksh, M.: In site moisture content monitoring sensor detecting mutual coupling magnitude between parallel and perpendicular dipole antennas. IEEE Trans. Instrum. Meas. 61(8), 61 (2012) 9. Kumar, R.D, Ramesh, G.P.: Survey of feeding techniques to improve bandwidth of microstrip patch antenna. Int. J. Appl. Eng. Res. 10(17), 13109–13111 10. Papadopoulos K.A., Papagianni C.A., Gkonis, P.K., Venieris, I.S., Kaklamani, D.I.: Particle swarm optimization of antenna arrays with efficiency constrants. Prog. Electromagn. Res. M 17, 237–251 (2011) 11. Tantisopharak, T., Moon, H., Youryon, P., Bunya-athichart, K.: Non-destructive determination of the maturity of Durian fruit in the frequency domain using its natural frequency. IEEE Trans. Antennas Propag 64(5), 1779–1787 (2016) 12. Oda, M., Mase, A., Uchino, K.: Non-destructive measurement of sugar content in apples using millimeter wave reflectometry and artificial neural networks for calibration. In: Proceedings Asia-Pacific Microwave Conference APMC, pp. 1386–1389 (2011) 13. Yu, Y., Yi, L., Liu, X., Gu, Z., Rizka, NM.: Dual-frequency two-element antenna array with suppressed mutual coupling. Int. J. Antennas Propag. 2015 (2015) 14. Kumar, R.D., Ramesh, G.P.: Multiband reconfigurable antenna design for medical applications. J. Adv. Res. Dyn. Control. 18(4), 9–13 (2017) 15. Sun, X.L., Cheung, S.W., Yuk, T.I.: Dual band monopole antenna with frequency tunable feature for WiMAX applications. IEEE Antennas Wirel. Propag. Lett. 12, 100–103 (2013)

Chapter 22

Effective Utilization of Image Information Using Data Mining Technique D. Saravanan, Dennis Joseph and S. Vaithyasubramanian

Abstract In recent, video databases data mining is widely used for various applications such as crime prevention, web searching, cultural heritage, advertising, news broadcasting, video, education and training and military. The advancement of databases specially the multimedia dates are in need to efficiently handle due to the growing amount of multimedia data include audio video, sound, animation, image etc. Revolution in the extensive database of computerized medias gives rise to the study of useful information from database. The study such as multimedia information retrieval, productive storage and organization of available information are in focus. This paper discuss how effectively handle the image data’s. Keywords Data mining · Image mining · Image data base · Information retrieval · Querying · Image histogram · Image color cue · Hierarchical clustering

22.1 Introduction Information extraction or information recovery is the process of identify the relevant information in the document or searching the relevant content in the web repository or find the exact information in the whole document [1]. There is overlap in the usage of the terms data retrieval, document retrieval, information retrieval, and text retrieval. Even if the terminology may differ the actual function never changes. Any D. Saravanan (B) · D. Joseph Faculty of Operations and IT, ICFAI Business School (IBS), Hyderabad. The ICFAI Foundation for Higher Education (IFHE) (Deemed to Be University U/S 3 of the UGC Act 1956), Hyderabad, India e-mail: [email protected] D. Joseph e-mail: [email protected] S. Vaithyasubramanian Faculty of Mathematics, Department of Mathematics, Sathyabama Institute of Science and Technology, Chennai, India e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_22

207

208

D. Saravanan et al.

information retrieval brings the relevant structure or relevant information [2]. This term gives the associative terminology for different field of science and engineering, medicine, management, economics, image searching and more. Because of its associative nature instantaneous procedures are required to extract the relevant document or content. This process gives the special attention in the field of image processing specially in video image retrieval process [3, 4]. For retrieve specific information based on video content effectively, and to provide better ways for entertainment and multimediaapplications.

22.1.1 Steps of Image Mining Image mining requires that images be retrieved according to some requirement specifications. The requirement specifications done into the various steps of increasing complexity [5]: Step 1: For extracting image data user need to extract image attributes such as image shape, image pixel position, pixel color value, and pixel location in the image frame. Step 2: After extracting image features it is necessary in video data mining video files are divided into individual frames. Step 3: After the frames each frame pixel values are calculated. It helps to differentiate the frames.

22.2 Preprocessing Steps of Data Mining Extracting the needed information from the data repository is not an easy step. Any mining process started with preprocessing function [6]. It helps to improve the processing speed and also reduces the storage space. Mining process starts when we store our information in any computer repository. Extracting the relevant information from the stored huge content involves preprocessing steps [7]. For any domain especially in business operations it helps in the following ways [8]: (i) Massive data collection (ii) Powerful multiprocessor computers and (iii) Data mining algorithms (Fig. 22.1).

22.2.1 Image Extraction Image extraction it differ from normal extraction because of the nature of the image data’s. Extracting the specific images based on the image attributes, image pixel values, image frame values and more. Image mining is differing from content based mining, content based mining based on any content specification or content attribute

22 Effective Utilization of Image Information Using Data …

209

Fig. 22.1 Data mining preprocessing steps

values [9, 10]. But image processing based on image attributes and frame values of the particular image. Main functions of any image mining finding the relevant pattern from the collections of the image data base.

22.2.2 Relational Database Versus Image Database Relation data values are differ from image attribute values. Example in relation data base user specifies any values give the actual value of the domain [11]. In image data base we cannot specify the values like that. Unique versus multiple interpretations. A third important difference deals with image characteristics of having multiple interpretations for the same visual patterns Any traditional data mining algorithms are not work well for image data functions because of the image nature [12]. For that a special procedures are required for extracting information from the image data base.

22.3 Information Retrieval System There is various methods helps to extract the relevant information from the stored database. All this procedure works with higher order image property or lower order

210

D. Saravanan et al.

image property. Certain procedure follows the combination of this both [13]. The higher level functions are explain the operational functions of the image. Lower level image functions such as image pixel values, image pixel position, frame RGB values [14, 15]. Based on this image extraction are done in two types such as target search and category search. Target search bring the specified image i.e. whatever the users specified on his image query corresponding image extracted from the image database ex user try to extract some shape or some symbol or specify structure and more. In target search user may not know the output image [16]. Based on the users query information are extract and retrieved, such as scenery images or skyscrapers (Fig. 22.2). Increasing the demand in information technology growth of image based data’s also increased drastically. Extracting of such information is not an easy process, user need certain domain knowledge for extracting the needed information. In earlier periods image data extraction used in climate prediction after this applications are further extended into: Map generation and manipulation; Visual analysis of experiment data - in chemistry, physics and statistics; Visualization of mathematical functions and CAD/CAM (computer aided design and manufacture) for ship building, quickly followed by architecture, landscaping, interior decorating and fashion. As computer capacity has grown, image-based application areas have expanded to include: (i) Satellite image analysis for monitoring weather, climate, agricultural quality, human activity (ii) Analysis of medical images (iii) Navigation—for submarines, ships and aircraft and (iv) Security—photo identity and finger prints. Because of the growing internet technology today it applied to various fields such as painting, traditional and more. This attracts the new sets of user community in this domain.

Fig. 22.2 Image retrieval using image content

22 Effective Utilization of Image Information Using Data …

211

22.3.1 Image Mining Algorithm Steps Due to the complexity of the input data sets video data mining need preprocessing steps, the steps as follows: Step 1 Characteristic Withdrawal: Before processing the actual mining, video data sets are divided into section called shots, each shot or frames used for actual mining process. Each frame values of RGB are extracted and stored separately for further process. This value helps for extraction and removal of duplication shots. Step 2 Object identification and record creation: Compare objects in one image to objects in every other image. Label each object with an id; it helps the user to differentiate the objects. Step 3 Assign identifiers with objects: After step 2, objects identifiers are assigned with pruned frames. Step 4: Repeat step 1–3 un till all the frames are assigned unique identifiers. Apply data mining algorithm to produce object.

22.3.2 Creation of Index on Image Data Base Image databases can be and normally are indexed by at least the first 3 of the 4 main index types listed above, listed below in falling frequency: 1. Atomicindexes used for standard RDB attributes containing context (photographer/artist, owner) or structural (encoding type, size) metadata. 2. Term indexes, defined on terms selected from semantic metadata attributes such as title, caption, subject and description. 3. Image attributes such as image pixel values, histogram values are extracted and stored for further process. 4. Frames observation such as time line or objects in the frames, texture are identified and stored.

22.4 Application of Data Mining 22.4.1 Video Data Mining Shot Detection Video shot detection is the first step for video parsing, and the detected shot boundaries are the basic units for video feature extraction. Use this values user query or user matching process are done.

212

D. Saravanan et al.

22.4.2 Creation of Histogram on Images Image extraction or image matching are done with help of image histogram values. Histogram values are calculated with help of image color pixel values i.e. values of basic building block of any color images Red, Green and Blue values are used for this process. For this the image pixels are divided into three basic color mechanisms.

22.4.3 Experimental Results See Figs. 22.3, 22.4, 22.5 and 22.6 and Table 22.1.

22.5 Conclusion Data mining describes a class of brings the unknown knowledge or patterns in large amounts of data. Most of data mining research has been dedicated to alpha-numeric databases, and text data sets only. Very few mechanisms are available for the multimedia data mining. The current status and the challenges of video data mining which Table 22.1 Different frame splitting time

Number of frames spitted 100

2

1000

5

10,000

Fig. 22.3 To view the image

Search time (s)

14

22 Effective Utilization of Image Information Using Data …

213

Fig. 22.4 Image retrieval

is a very premature field of multimedia data mining are discussed in this paper. The issues discussed should be dealt with in order to obtain valuable information from vast amounts of video data.

214

Fig. 22.5 Relevant to image query Fig. 22.6 Sampling query comparison

D. Saravanan et al.

22 Effective Utilization of Image Information Using Data …

215

References 1. Regunathan, R., Xiong, Z., Divakaran, A., Ishikawa, Y.: Generation of sports highlights using a combination of supervised and unsupervised learning in the audio domain. In: ICICS-PCM Conference, Singapore (2003) 2. Divakaran, A., Peker, K.A., Radhakrishnan, R., Xiong, Z., Cabasson, R.: Video sumarization using MPEG-7 motion activity and audio features. In: Rosenfeld, A., DoDoermann, D., DeMenthon, D. (eds.) Video Mining. Kluwer Academic Publishers (2003) 3. Saravanan, D.: Video data image retrieval using—BRICH. World J. Eng. 14(4), 318–323 (2017) 4. Saravanan, D.: Image frame mining using indexing technique. In: Data Engineering and Intelligent Computing, Chapter 12, pp. 127–137. Springer Book series. ISBN:978-981-10-3223-3, July 2017 5. Xie, L., Chang, S-F., Divakaran, A., Sun, H.: Unsupervised mining of statistical temporal structures in video. In: Rosenfeld, A., Doermann, D., DeMenthon, D. (eds.) Video Mining. Kluwer Academic Publishers (2003) 6. Alemu, Y., Koh, J.B., Ikram, M., Kim, D-K.: Image retrieval in multimedia databases: a survey. In: Fifth International Conference on Intelligent Information Hiding and Multimedia Signal Processing (2009) 7. Hilbert, D.: Uber die stetige Abbildung einer Linie auf ein Flachenstuck. Math. Annalen, 38–40. [10] Bartolini, I., Ciacci, P., Waas, F.: Feedback bypass: a new approach to interactive similarity query processing. In: Proceeding of 27th Int’l Conference Very Large Data Base (VLDB’01), pp. 201–210 (2001) 8. Brunelli, R., Mich, O.: Image retrieval by examples. IEEE Trans. Multimed. 2(3), 164–171 (2000) 9. Saravanan, D.: Effective video data retrieval using image key frame selection. In: Advances in Intelligent Systems and computing, pp. 145–155 (2017) 10. Saravanan, D.: Clustering the irregularity events in intelligence surrounding systems. J. Pure Appl. Math. 119(12), 15025–15035 (2018) (Special Issues), ISSN:1311-8080 11. Fan, J., Luo, H.: Emantic video classification by integrating flexible mixture model with adaptive em algorithm. In: ACMSIGMM, pp. 9–16 (2003) 12. Wang. J.Z.: A text book on. In: Integrated Region-Based Image Retrieval. Kluwer Academic Publishers (2001) 13. Zhang, J., Hsu, W., Lee, M.L.: An information driven framework for image mining. In: Proceedings of 12th International Conference on Database and Expert Systems Applications (DEXA). Munich, Germany (2001) 14. Saravanan, D.: Effective video content retrieval using image attributes. EAI Endorsed Trans. Energy Web Inf. Technol. 5(18), e8, 1–5 (2018) 15. Saravanan, D.: Efficient video indexing and retrieval using hierarchical clustering techniques. Adv. Intell. Syst. Comput. 712, 1–8 (2018). ISBN:978-981-10-8227 16. Vailaya, A., Figueiredo, M., Jain, A.K., Zhang, H.J.: Image classification for content-based indexing. IEEE Trans. Image Process. 10(1), 117–130 (2001)

Chapter 23

Particle Swarm Optimization Algorithm Based PID Controller for the Control of the Automatic Generation Control Ali Abdyasser Kadhum, Thaeer Mueen Sahib and Mohsın Mousa Mohammed Ali Abstract When the automatic generation control (AGC) is used in power system safety, the aim is to ensure that the power systems’ expected frequency is maintained at a perceived stable value. The role of the AGC lies in the adjustment of a given system, especially with the intention of meeting or achieving the required load. Also, SGC aids in ensuring that systems adjust to changes in frequency, besides enhancing the ACE adjustment to zero. However, one of the challenges that face frequency control processes in interconnected zones is the role of single areas. Therefore, this study applies the PSO (Particle Swarm Optimization) algorithm towards the realization of fine PID controllers, especially in contexts involving two area load frequency controls. From the findings, the investigation demonstrates that the selected controller improve the performance of targeted systems and also enhances operations in AGC supplies; a trend confirmed by the resultant sensible dynamic response. Also, SIMULINK and MATLAB are used to investigate the two areas’ performance control. Similarly, the study employed K–800–23.5–0.0034 or the AL-Dura power plant form. Keywords Two are and single area control system · Generation frequency control · Particle swarm optimization (PSO) · Automatic generation control (AGC)

23.1 Introduction To improve system controls, power systems play an important role. In power system operations and control, one of the most important aspects involves the continuous supply of power to the target system; especially due to the need for equal distribution of power to various customers relying on the system at hand. Should the latter be achieved, system stability tends to be realized. With an equilibrium realized between the power generated and the power demanded, the interconnected power systems require the total generation to be matched with the load demand [1]. Should changes A. A. Kadhum (B) · T. M. Sahib · M. M. M. Ali Al-Furat al-Awsat Technical University, 31003 Al-Kufa, Iraq e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_23

217

218

A. A. Kadhum et al.

occur to a given power system, it becomes important to gain knowledge about nominal system frequency changes, as well as some of the expected power connections in supplementary regions. Applying the intelligent control approach ensures that decisions are made accurately and quickly, especially due to the ability of the technique to offer precise information. The eventuality is that in single areas with thermal generation components, AGC plays the role of maintaining the expected system frequency values, which is expected to lie at 50 Hz [2]. It is also worth noting that control systems aid in maintaining frequency reliability, besides ensuring that the zero steady state error is realized. To discern AGC problems, different controllers are applied. The role of these controls is to assure good response, ensure that the oscillation is reduced or controlled and maintained, and promote the achievement of the steady state error. Indeed, the intelligent controller has been documented to yield superior results compared to the case of the conventional controller [3]. From the literature, insights highlight further that AGC shapes the operations of thermal systems—in the context of different types of controller. In the study by Kaur [4], the PSO algorithm and PI were applied to conventional controllers. The target setting entailed the two-area interconnected power system. Form the findings, it was noted that when different controllers are compared, responses vary from one system to another. In another investigation, Ibraheem and Singh [5] applied PI to determine the performance of the AGC in three-area interconnected power systems—using the PSO algorithm. Indeed, the proposed algorithm’s results were compared to those reported when the conventional Ziegler Nichols (ZN) was applied, with the findings indicating that the proposed algorithm was promising. In the investigation by Yarlagadda et al. [6], the impact of fast-acting ALFC and AVR loops was investigated, especially in relation to their application on dynamic stability improvement when applied to PID and PD controllers. The study compared how the various techniques would promote improvements in dynamic stability. Similarly, Krishna et al. [7] focused on load frequency control in power system generators via the use of PID. The study established that when the PID controller is turned suitably, there is likely to be a reduction in single-area frequency variations, as well as a decrease in the area control area; with similar observations also made in relation to the case of the two-area systems—in which the PSO algorithm was applied. Overall, findings indicated that the proposed system yielded improvements in the steady state. It I also worth highlighting that in the investigation by Panda et al. [8] the Hybrid Neuro-Fuzzy (NHF) technique, an artificial intelligence (AI) technique, was implemented in relation to the context of the AGC. Due to the proposed system’s capability to handle non-linearity, it was affirmed to improve the performance of the controllers. Also, the system was observed to be quicker compared to situations involving conventional controllers, especially in two-area systems. As such, it was concluded that through intelligence controllers, the dynamic responses is likely to improve, also proving faster than the case in which conventional controllers are employed.

23 Particle Swarm Optimization Algorithm Based PID …

219

23.2 Materials and Methods The role of AGC lies in the description of how controllable generators’ power production could be regulated. The description occurs in certain accepted areas experiencing variations in system frequencies, as well as tie-line leading. Similarly, it is used during schedule system frequency maintenance. These arrangements are shown in Figs. 23.1 and 23.2. In single-area power systems, the roe of the AGC is seen to lie in the maintenance of the frequency of the system, ensuring that it stands at a constant value; even in the wake of increased system perturbation. To ensure that the electrical power system is maintained at a given operating rate and steady state, the resultant power that is generated is expected to be equal to that which is demanded. However, it is worth indicating that when sensible power systems are presented, there is a constant variation in the load (relative to time); a trend suggesting that the irregular state is unlikely to be satisfied through power equilibrium. To counter the problem of the control action, controllers are used in such a way that they extract operating states and also balance out minor alterations in the system’s conditions, especially regarding load demands. As they achieve these aspects, the controllers do not compromise the expected frequency limits [9, 10]. The following figures illustrate the AGC controller model (Figs. 23.3 and 23.4). Several advantages have been associated with AGC. Some of these merits include better efficiency in the detection and solution to power faults, improvements in load variation recovery capability, increased generation ability via two- or one-area connection, and the ability to maintain the generation of various units economically.

Fig. 23.1 The power system’s block diagram

220

Fig. 23.2 Block diagram of LFC and the power system

Fig. 23.3 An illustration of a single-area block diagram

A. A. Kadhum et al.

23 Particle Swarm Optimization Algorithm Based PID …

221

Fig. 23.4 The two-area block diagram

Others include the ability to achieve accurate values regarding control area interchange power, and the ability to achieve a consistent value (50 Hz) for the given system frequency. An example of an area that could be consider is a case of problems experienced during power production control, especially in knit electric area generators. The problem may involve the capacity to maintain certain, set frequencies. In such an area, all generators have coherent groups to ensure that they can slow down and speed up simultaneously—while maintaining the power angles. The setting defines a control area. To gain insights into frequency control problems facing the AGC system, a single turbo-generator system supply can be considered, especially on an isolated load [10, 11]. Given PL as the total load change, frequency variations in the steady state (in the context of two-area systems or settings) become: −PL     f SS =   1 R1 + 1 R2 + · · · + 1 Rn + D −PL =  1 Req + D 1   Req =  1 Req + 1 R2 + · · · + 1 Rn β=

1 −PL = +D  f SS Req

(23.1) (23.2) (23.3)

ACE of each area is linear combination of biased frequency and tie-line error. For area1 : ACE1 = P12 + β1  f

(23.4)

222

A. A. Kadhum et al.

For area 2 : ACE2 = P21 + β2  f

(23.5)

23.3 Particle Swarm Optimization (PSO) PSO reflects an evolutionary computation technique guided by swarm intelligence, especially those that aim at locating a fertile feeding zone. Russell Eberhart and James Kennedy coined this algorithm in 1995. Indeed, PSO is easy to implement and also requires the operator to adjust fewer parameters, especially regarding system position and velocity. In this case, a swarm reflects a group of disorganized organisms seeking to cluster in a given place jointly; with each member having assumed a different, random path. Hence, the PSO algorithm adopts various particles that are similar to swarms moving and in search of the best outcome. Hence, the respective particles are treated as those existing in 3-D spaces. With each particle maintaining its coordinates in a given problem space, the best solution is achieved [12].

23.4 Simulink Model of AGC with PSO Algorithm To ensure that a simulink model of AGC is established, the governor’s transfer function is applied, besides the power system and turbine for gussa ACE, ∇ f . Also, the transfer functions are applied to ensure that the frequency-domain analysis is enhanced. The simulink model’s functionality is summarized and highlighted in Fig. 23.5.

Fig. 23.5 Simulink model of the AGC with PSO algorithm

23 Particle Swarm Optimization Algorithm Based PID …

223

23.5 Simulation Results Following the application of the Simulink toolbox or MATLAB aided in understanding the two-area system’s performance. With 0.1 p.u set as the load disturbance, frequency deviations were investigated, as well as the intelligent controller’s role in shaping the performance of the AGC. Under load disturbance, variations in system frequency were observed. Figures 23.6, 23.7 and 23.8 illustrate the study’s results.

Fig. 23.6 Change in ACE1, ACE2

Fig. 23.7 Change in  f 1 and  f 2

224

A. A. Kadhum et al.

Fig. 23.8 Change in  f 1 and  f 2 with PSO algorithm

23.6 Conclusion In this investigation, the main aim was to investigate the AGC of two areas. To unearth the degree to which the proposed scheme was effective, conventional PID techniques and the PO algorithm were applied. In turn, the respective controller’s nature of performance was analyzed. From the findings, the planned controller exhibited superior dynamic performance. Particularly, the proposed scheme was found to be capable of monitoring power systems in terms of the timely maintenance of the voltage and system frequency. Hence, it was noted that most of the optimization techniques rely on controller designs, a trend that has seen two-area systems utilize the controls to monitor and regulate their operations. In summary, the study established that contexts such as two-area systems experience frequency variations, with parameter differences in the given systems playing a moderating role in shaping the overall performance.

References 1. Sadat, H.: Power system analysis. Tata MCGraw Hill (1999) 2. Sivanagaraju, S., Sreenivasan, G.: Power system operation and control. PEARSON (2011) 3. Prakash, S., Sinha, S.K.: Four area load frequency control of interconnected hydro—thermal power system by intelligent PID control technique. 978-1-4673-0455-9/12. IEEE (2012) 4. Garg, K., Kaur, J.: Particle swarm optimization based automatic generation control of two area interconnected power system. Int. J. Sci. Res. Publ. 4(1) (2014) 5. Singh, O.: Design of particle swarm optimization (PSO) based automatic generation control (AGC) regulator with different cost functions. J. Electr. Electron. Eng. Res. 4(2), 33–45 (2012)

23 Particle Swarm Optimization Algorithm Based PID …

225

6. Yarlagadda, V., Saroja, P.S., Avinash, P., Nikhilesh P.: Comparative analysis of fast acting AVR and ALFC loops with classical controllers on dynamic stability improvement. Int. J. Adv. Res. Electr., Electron. Instrum. Eng. 4(6) (2015) 7. Krishna, S.R., Singh, P., Das, M.S.: Control of load frequency of power system by PID controller using PSO. Int. J. Recent. Dev. Eng. Technol. Website: www.ijrdet.com (ISSN 23476435(Online), 5(6), 2016) 8. Panda, G., Panda, S., Ardil, C.: Hybrid neuro fuzzy approach for automatic generation control of two—area interconnected power system. Int. J. Electr., Electron. Sci. Eng. 3(3) (2009) 9. Challa, K.K., Nagendra Rao, P.S.: Analysis and design of controller for two area thermalhydro-gas AGC system. 978-1-4244-7781-4/10. IEEE (2010) 10. Kothari, D.P., Nagrath, J.: Modern power system analysis, 3rd edn. Tata McGraw Hill (2003) 11. Magla, A., Nanda, J.: Automatic generation control of an interconnected hydro—thermal system using conventional integral and fuzzy logic control. In: Proceedings IEEE Electric Utility Deregulation, Restructuring and Power Technologies (2004) 12. Kawata, K., Fukuyama, Y;, Takayama, S., Nakanishi, Y.: A particle swarm optimization for reactive power and voltage control. Technical Research Center, Power Technology Lab., The Kansai Electric Power Co., Inc., Power Technology Lab, Fuji Electric Corporate R &D, Ltd., IEEE Trans. on Power Systems, Vol. 15, No. 4, pp. 1232–1239, November 2001

Chapter 24

Proposed Improving Self-management Support System for Chronic Care Model (Heart Diseases) Jammel Mona, Mohammad Dosh and Wafaa Kamel Al-Jibory

Abstract Software engineering can be characterized as the definite examination of the engineering to the structure, the improvement and the support of the software. Software engineering models and methods are imposed on software architecture in order to make this movement efficient, repeatable and at last oriented towards success. Utilizing the models can be used in order to provide the way to deal with critical thinking, documentation, and strategies for the model development and the investigation. The Chronic Care Model (CCM) can be defined as the organizing framework which is used to improve the incessant ailment care and an astounding apparatus for improving consideration at both the individual and populace level. In this paper, one of the important CCM model parts, which is Self-management Support, will be improved in order to make the model more efficiency in the processing. Keywords CCM model · Heart Diseases · Software engineering models · Software architecture

24.1 Introduction Software engineering is the utilization of an orderly, restrained, quantifiable way in order to deal with the progression, movement, and backing of programming, and the examination of these approaches; that is, the utilization of structure to programming [1]. Ed Wagner developed the Chronic Care Model (CCM), and is often known as the ‘Wagner Model’. Figure 24.1 illustrates this model [2]. J. Mona (B) College of Medicine, University of Kufa, Kufa, Iraq e-mail: [email protected] M. Dosh Department of Computer Science, College of Education for Girls, University of Kufa, Kufa, Iraq W. K. Al-Jibory Department of Software Engineering, Imam Al-Kadum (a) University College for Islamic Sciences, Baghdad, Iraq © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_24

227

228

J. Mona et al.

Fig. 24.1 An illustration of the chronic care model

The Chronic Care Model (CCM) was advanced with the aim of ensuring that human service framework execution is enhanced. Also, CCM was advanced to ensure that populace and individual well-being mediations are supported accordingly. Therefore, the use of incessant consideration framework is informed by the need to address the prevalence of perpetual ailment, especially with recent changes in the application of models, which have seen growing demands for patient-centered care. In the latter trend, the emphasis is to ensure that a more personal care process is extended to individuals who are handicap or have unending ailments [3, 4]. In the CCM model, six major components are documented. These component include healthcare organizations, community resources, practice-level clinical information systems, decision support, the self-management support (SMS), and the delivery system design (DSD) [5, 6]. Currently, the Internet of Things (IoT) got a staggering thought from investigators, since it transforms into a noteworthy advancement that ensures a wise individual life, by allowing correspondences between things, machines and everything together with society. IoT addresses a system which contains things in all actuality, and sensors annexed to or solidified to these things, related with the Internet by methods for wired and remote framework structure. Figure 24.2 illustrates the Internet of things Concept [7]. The Web of Things is a specialization of the Internet of Things that utilizes what made the web so effective and applies it to implanted gadgets so as to make the most recent improvements in the Internet of Things available to however many designers as could be allowed. On the Web of Things—simply like on the web-anybody with

24 Proposed Improving Self-management Support System …

229

Fig. 24.2 Internet of things concept

a content manager and the fundamental comprehension of web norms (HTML and HTTP) can rapidly begin interfacing gadgets and items to the web. In any case, it additionally empowers setting off to the following dimension and serves to successfully manufacture intelligent and inventive genuine applications that mix the physical and computerized universes [8].

24.2 Related Work In [9], the main aim was to apply the CCM framework in a socio-environmental context. The specific objective was to discern the applicability of the model in handling diabetes in community-based organization and also through self-administration among executives. In the findings, the study established that the CCM model is informative whereby it yields improvements in self-organization among diabetic patients, especially regarding medical preparation and self-management of the condition. As such, the study recommended the future use of this model in enhancing self-management among patients. Additional insights suggested that the parts that most a great part of the time influenced on physiological extents of disorder, prosperity and limit status, and individual fulfillment were self-organization sponsorship and movement system structure particularly when in blend. In [10] to get to more readily coordinate parts of aversion and wellbeing advancement into the criticality of the CCM framework; an upgraded adaptation was presented. This adoption was presented in the form of the Expanded Chronic Care Model. This new model incorporates components of the populace wellbeing advancement field so extensively based aversion endeavors, acknowledgment of the social determinants of wellbeing, and improved network investment can likewise be a piece of crafted by wellbeing framework groups as they work with unending sickness issues.

230

J. Mona et al.

24.3 The Proposed Method In the wake of healthcare trends such as general prosperity and informatics, one of the trends that have played a crucial role entails the concept of e-health. Through this provision, data is shared via the Internet and other platforms that have evolved due to technological advancement. Therefore, the decision by e-health providers to embrace the CCM framework is seen as a promising step towards improved service provision in the healthcare industry. Some of the areas in which the CCM framework seeks to streamline and enhance healthcare outcomes include community and health systemrelated issues, decision support, clinical information systems, the self-management support system, and the delivery system design. In the proposed modification to the model the self-management support domain will be improved. The modification to the self-management support will be through the using of the Web of things and Internet of things. WoT and IoT will be used to provide alert feature to the model based on the data which are collected through the sensors. These alerts will be used in many cases such as care system, informing doctors and other cases. Figure 24.3 illustrates the proposed modification to CCM.

24.4 Conclusions In summary, software development models refer to different methodologies or strategies employed to enhance the operations of different systems, with the healthcare sector unexceptional. One of these models entails the CCM framework. Specifically,

Delivery system design

Self-Management Support

The Chronic Care Model

Clinical Information Systems

Decision Support

Community

Health System-Related Issues

Fig. 24.3 The proposed modification to CCM

IoT

Alert Feature

WoT

24 Proposed Improving Self-management Support System …

231

this framework has played a crucial role in ensuring that there are system improvements and organization in a manner that promotes the realization of the intended goals. Overall, this paper has examined the CCM concept and its role in supporting the aspect of self-management. Coming in the wake of WoT and IoT, the model is affirmed to play a crucial role towards system improvement, as well as enabling users such as those in the e-health system to make informed decisions.

References 1. Standard glossary of software engineering terminology. IEEE (1990) 2. http://www.improvingchroniccare.org/index.php 3. Australian government department of health and ageing (AGDHA). National Chronic Disease Strategy (2006) 4. Anderson, N.A., Bridges-Webb, C., Chancellor, A.H.B.: General practice in australia, pp. 3–4. Sydney University Press, Sydney (1986) 5. Stellefson, M., Dipnarine, K., Stopka, C.: The chronic care model and diabetes management in US primary care settings: a systematic review. Prev Chronic Dis. 10, E26 (2013) 6. Wagner, E., Austin, B., Von Korff, M.: Organizing care for patients with chronic illness. Millbank Q. 74(4), 511–544 (1996) 7. Mohammed, Z.K., et al.: Internet of things applications, challenges and related future technologies. WSN 67(2), 126–148 (2017) 8. Zeng, D.: The web of things: a survey. J. Commun. 6(6) (2011) 9. Ansari, R.M., et al.: Application of chronic care model for self management of type 2 diabetes: focus on the middle aged population of pakistan 10. Barr, V., et al.: The expanded chronic care model: an integration of concepts and strategies from population health promotion and the chronic care model. Hosp. Q. 7(1) (2003)

Chapter 25

DWINE Your Fear—Defensive Device for Women in Need A. B. Sarada Pyngas, B. Ruchitha Chowdary and R. Kavitha

Abstract The crime rate against women and girls in India is rapidly increasing every day. Women are not able to lead their lives independently even in this modernized world. A lot number of physical abuse attacks are being reported on a daily basis. According to the recent government reports, every fifteen minutes, one child is being sexually abused in India. But the criminal is not getting punished due to lack of proper evidence. 99% of such cases go unreported. The presence of CCTV’s is helping in some way, but when the attack happens in a remote area, there is no way to find the accused. The lack of education about the good touch and bad touch among school kids is one of the important factors. In case of girls who have a little knowledge about the bad touch, are either hesitant to inform their parents, or informing them only after the crime has been committed, which is of no use. The chances of punishing the criminal are quite low in such cases. Therefore, a new solution is expected as a modern system which prevents the child abusement crime from happening or at least to find the criminal immediately to prevent another child from the same crime. A wearable device which records the Adrenaline and Oxytocin levels of the body using sensors is suggested. The heart beat rate is also taken into consideration to detect if the girl is in any danger. Pictures of the surroundings are also clicked using an in-built camera to have a proper evidence. Keywords Women security · Oxytocin · Adrenaline · QCM chips · GPS tracker · In-built camera · Sensors · Accurate evidence

A. B. Sarada Pyngas (B) · B. Ruchitha Chowdary · R. Kavitha School of Computing, SASTRA Deemed University, Thanjavur, India e-mail: [email protected] B. Ruchitha Chowdary e-mail: [email protected] R. Kavitha e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_25

233

234

A. B. Sarada Pyngas et al.

25.1 Introduction Recently many cases are being reported against women security, especially in India. The latest Pollachi incident where girls and women were molested and sexually abused by a gang of men, brings tears to our eyes. A recent study says that 10% of all the crimes reported in India are against women [1]. As responsible citizens of India, it is our duty to eradicate this serious issue from the society. The proposed device “DWINE Your Fear” contributes its part in protecting the women and girls. DWINE is an acronym for Defensive device for Women In Need, while ‘DWINE’ literally means ‘to decline’. So, the device basically helps the women who are in need of protection against any sexual assault, which ultimately chases away their fear of getting molested or abused.

25.2 Literature Review In paper [2], a device has been designed with a pressure switch. When the girl is in danger, the switch is to be pressed by her and then a message followed by a call is sent to her relatives and the police. Paper [3] talks about a foot device which helps the woman if her foot is tapped ten or more times when she is in danger. An audio recording facility is available which records the voices in her surroundings that later act as an evidence. In paper [4], a smart ring called SMARISA is talked about. Raspberry pi zero, Raspberry pi camera, buzzer and a button are used for implementation. The attacker’s image is also taken. When the button is pressed, the device is activated. In paper [5], an amperometric biosensor has been developed to detect adrenaline to support tumor diagnostic technology. Another colorimetric sensor has been developed in paper [6] to detect adrenaline based on functionalized gold nanoparticles.

25.3 Drawbacks in the Current Systems The current systems always need involvement of the person wearing the device. When it comes to the case of children or women who don’t know what is happening around them, such devices will be of no use. Even if the victim is somehow able to inform her parents/guardians about her danger, the existing devices cannot help in preventing the crime. No immediate action is done as an attempt for self-defence. Systems like alarm buzz can help to some extent, but cannot be used to collect proper evidences.

25 DWINE Your Fear—Defensive Device for Women in Need

235

25.4 Proposed Idea Our idea is to build a wearable device that gives a solution for the above-mentioned problem by detecting two hormones of human body. ‘Oxytocin’ is the hormone that is generated during any physical intercourse, especially in women, while ‘Adrenaline’ is the hormone that is responsible for fear. When a molester misbehaves with a young girl who doesn’t even know what the other person is doing to her, she gets afraid and the fear in her increases. This means that there is an adrenaline rush in her body at that time. Proposed system will have such a device which uses Oxytocin and Adrenaline sensors to record the levels of oxytocin and Adrenaline. When the levels exceed a minimum limit, as an attempt for self-defence, an alarm is buzzed to alert the people nearby and an electric shock is generated against the criminal. It could stop the attacker from continuing harassment. Meanwhile a message is sent to the victim’s guardians and the nearby police station. The GPS tracker helps them to find the exact location of victim. A micro camera is also fixed in the device. This clicks pictures of the attacker and constantly keeps uploading them to the cloud under proper network connection. When there is a poor network, the data is stored in the internal storage memory in unreadable format and once the network is available, the data gets updated to the cloud. Therefore, the attacker will not have a chance to contaminate the evidence.

25.5 Architecture Diagram The proposed idea is depicted as a pictorial flow. See Fig. 25.1

25.6 Scenario Referring to the recent Pollachi incident or the famous Nirbhaya case or the Kathua rape case, the most targeted females are of adolescent age. They do not even have a minute idea of what is happening with their body. Let there be a girl named Sia. Sia is just nine years old. She comes from a lower middle-class family and attends the local government school. She uses a bicycle to go to school. She is sometimes accompanied by her friends, but mostly she goes alone. Her parents are daily wage workers and they have no enough time to sit and teach good and bad things to Sia. There is a mechanic shop on her way to school maintained by Raavan, where Sia gets her cycle repaired often. Taking this as the background, let us consider a day where Sia is going to her school all alone. Raavan notices this and wants to make a move. He approaches Sia and offers a chocolate. Being a kid, she gets attracted to the chocolate and takes it from Raavan, since she knew him already. While Sia is busy tasting the chocolate, Raavan

236

A. B. Sarada Pyngas et al.

Fig. 25.1 The process flow (Image source Google)

makes her believe that if she comes with him, he would give her more chocolates. The innocent Sia immediately follows him. That day, Raavan had planned everything such that there is no one in his mechanic shop to disturb him from performing his pleasure act. Sia enters the shop and comes to know that Raavan had not brought her there to give chocolates, but for something else. She does not understand what his intention is. As Raavan starts approaching her, she starts panicking and is incapable of doing anything. Sia shouts at the top of her voice, but no one could hear her. She pleads him; but her innocent looks and words could not melt Raavan. Out of no choice, she surrenders herself to Raavan. After the act, Raavan threatens her not to inform about it anywhere, else he would kill her parents. With the fear of losing her parents, she does not reveal anything to anybody. So Raavan is now safe and is ready to play with some other Sia’s life. This is the pathetic reality that we need to believe. This is how such incidents are happening and getting unnoticed. In order to change this, we have come up with a solution. This device prevents any of such incidents from happening and also protects many such ‘Sia’s. Assume the same girl, Sia again, going to her school with the ‘DWINE’. Raavan calls her into his shop. This time, the moment Sia gets a panic attack, she has an Adrenaline (the fear hormone) rush in her body. The Adrenaline sensor detects the high levels and immediately buzzes an alarm. When Raavan approaches her and gets in contact with her, a mild electric shock is produced by the device, which stops him for some time. Meanwhile, Sia’s parents, school guardian and the nearest police station would have received information that Sia is in trouble. The GPS tracker of the device helps them to know Sia’s exact location. In case Raavan still tries to

25 DWINE Your Fear—Defensive Device for Women in Need

237

get intimate with Sia, her Oxytocin (vital hormone in sexual reproduction) levels will be under constant check. If they exceed a minimum level, then another high alert is sent to her well-wishers. Due to the detection of abnormal hormone level, DWINE immediately takes random pictures of the surroundings in all directions. These pictures are immediately uploaded into the cloud. Now, by the time Raavan has done something serious with Sia, ‘DWINE’ came to her rescue. The electric shock has stopped Raavan for some time. The loud alarm sound caught the pedestrians’ attention. The location tracer informed the police about Sia’s situation. Her school guardian and the police were present at the location to protect Sia from Raavan. Knowing the arrival of police, Raavan tries to escape. But with the help of pictures that were uploaded already, police came to know that it was none other than Raavan who had tried to molest Sia. Since they have a proper digital evidence, the police easily find him and produce in the court. He is given life imprisonment and many such ‘Sia’s are hence protected from the clutches of this ‘Raavan’.

25.7 Implementation The main motive behind designing the device is to protect the girl or woman from danger before anything wrong happens. This is also to collect proper evidence which later is used to find the accused.

25.7.1 Hardware Components Used • Raspberry Pi Camera • Adrenaline Sensor • Oxytocin Sensor.

25.7.2 Main Distinct Modules • • • • • • • •

Sensing the level of Adrenaline Sensing the Oxytocin level Keeping track of heartbeat Recording these values into database Checking the two levels with minimum threshold limits Buzzing the alarm if limit exceeds Generation of mild shock Clicking pictures using the camera and uploading to database

238

A. B. Sarada Pyngas et al.

• Tracking the location of device • Sending a message to the previously saved contacts.

25.7.3 Inputs Given • The hormone count in the girl’s body is given to the sensors • The device location is traced by GPS tracker and made ready to be sent as a message • Pictures taken by the camera • Emergency phone numbers saved in the device’s memory.

25.7.4 Outputs Obtained • • • • •

A buzzy alarm A mild shock Message sent to the contacts Recorded levels of Oxytocin and Adrenaline Clicked pictures stored in memory.

25.7.5 Framework Challenges In the process of implementation, there are a few difficulties to overcome. Since the adrenaline and oxytocin are the hormones to be detected, there needs to be a sensor that can sense them without any direct contact with the blood. Developing a wearable sensor is a challenging task which still requires more time to be deployed. There is no available sensor that detects any of these hormones directly. Taking Adrenaline alone into consideration might lead to false alarms. So relating it along with Oxytocin and then coming to a conclusion is time-consuming.

25.8 Result The recorded levels of both hormones and the pictures of accused uploaded to the cloud act as a proper evidence. The location shared is helpful to know the exact place to extend immediate help.

25 DWINE Your Fear—Defensive Device for Women in Need

239

25.9 Conclusion and Future Work This is the functioning of ‘DWINE’, the device which helps women to protect themselves from the evil eyes of this generation’s Raavans. The full implementation is to be done and the work needs to be extended with some more features. A proper sensor to detect Adrenaline and Oxytocin is to be designed. This sensor is to be integrated with other hardware components and developed as a wearable device.

References 1. https://economictimes.indiatimes.com/news/politics-and-nation/what-crime-stats-dont-say/ articleshow/66787792.cms 2. Punjabi, S.K., et al.: Smart intelligent system for women and child security. In: 2018 IEEE 9th Annual Information Technology, Electronics and Mobile Communication Conference (IEMCON), IEEE (2018) 3. Ruiz, R., Richardson, M.T.: Functional balance training using a domed device. Army War College Carlisle Barracks Pa Physical Fitness Research Institute (2005) 4. Sogi, N.R., et al.: SMARISA: a raspberry pi based smart ring for women safety using IoT. In: 2018 International Conference on Inventive Research in Computing Applications (ICIRCA). IEEE (2018) 5. Molinnus, D., et al.: Detection of adrenaline based on bioelectrocatalytical system to support tumor diagnostic technology. In: Multidisciplinary Digital Publishing Institute Proceedings, 1(4) (2017) 6. Chen, Z., et al.: A highly sensitive colorimetric sensor for adrenaline detection based on organic molecules-functionalized gold nanoparticles. Sens. Actuators B: Chem. 207, 277–280 (2015)

Chapter 26

Microstrip Patch Antenna for Peripheral Arterary Disease Diagnosis G. P. Ramesh

Abstract A microstrip Patch antenna is designed for Peripheral Artery Disease with the resonant frequency ranges from 0.7 to 4.8 GHz. An introduced design covers the requisite bandwidth for GSM/DCS/PCS/UMTS cellular phone system and IEEE Wireless Local Area Network (WLAN) standards: 802.11n, b/g (Wi-Fi), Bluetooth. Light weight, low-cost, plain configuration and multi-band functionality are the advantages. All results of the resonant frequency, return loss, radiation patterns and fields distributions are analysed and verified. The simulation analysis was performed using the ADS software. In this project, a diagnosed peripheral artery disease based on frequency from 0.7 to 4.8 GHz analysis is proposed. The patch antenna keeps in close proximity over the nerve skin surface. The signal from the antenna acquires and process with signal processing toolbox to determine blood fluid dynamics for peripheral artery disease diagnosis. Keywords Peripheral artery disease · GSM · DCS · PCS · UMTS · WLAN · Blood fluid · ADS

26.1 Introduction Interest in multi-band antennas is increasing, especially in order to reduce the number of antennas embedded in combining multiple applications on a single antenna. To answer this, several techniques are used. Various techniques like Frequency Selective Surface using thicker profile for folded shorted patch antennas, the use of slots with thicker substrate E-shaped patch antenna E-shaped is compatible feeding techniques like L-probe feed are used to enhance bandwidth of microstrip patch antenna. A triband E-shaped printed monopole antenna loaded with narrow slots, suitable for MIMO application [1]. The size of feeding patch and thickness of dielectric should be taken care. This combination of techniques, with the use of the method of the microstrip line, which is easier compared to other methods and gives a good G. P. Ramesh (B) St. Peter’s Institute of Higher Education and Research, Chennai, India e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_26

241

242

G. P. Ramesh

overview of the physical operation antenna [2, 3], allowed us to obtain an antenna capable of covering the GSM standard, so the standard DCS, PCS, UMTS and WLAN applications namely in ISM band used by systems Bluetooth (2.4–2.485 GHz) and Wi-Fi (2.4 GHz for 802.11b/g/n). The MIMO antenna utilizes two size reduction techniques to downsize the antenna element; and proposes a self-decoupling analogy that greatly eases the 28 complexity in the decoupling structure design [4]. In this paper, an internal, low profile multi-band antenna will be designed with the help of several techniques, such as the ones mentioned above [5, 6]. The original design and its effects is presented in this paper. Antennas can be designed to transmit and receive radio waves in all horizontal directions equally (omnidirectional antennas) [7, 8], or preferentially in a particular direction (directional or high gain antennas). An antenna may include parasitic elements [9], parabolic reflectors or horns, which serve to direct the radio waves into a beam or other desired radiation pattern.

26.2 Proposed Antenna System The 3D perspective of top and side views of the original antenna is shown in Fig. 26.1. The antenna is simulated on an FR4_epoxy substrate of 45 × 70 mm2 with a dielectric constant εr = 4.4. The thickness of the substrate is H = 0.8 mm. A rectangular patch

Fig. 26.1 Geometry of the proposed antenna

26 Microstrip Patch Antenna for Peripheral … Table 26.1 Specifications of the proposed antenna

Width

243 Length

W = 30 mm

L = 41.5

W1 = W2 = 2 mm

L1 = L2 = 17 mm

W3 = 24 mm

L3 = 2.45 mm

M4 = M5 = 20 mm

L4 = L5 = 2.45 mm

W6 = 0.925 mm

L6 = 8 mm

W7 = 3 mm

L7 = 10 mm

antenna including technical slots with different dimensions is shown in Table 26.1. The εr is chosen such that it gives better efficiency and larger bandwidth. The antenna is fed by a microstrip line in order to increase the bandwidth and gain. The proposed antenna system is placed near blood flow in level tube. In initial condition, the signal from normal and Anomalous required to differentiate for comparative analysis. The electromagnetic signals from and was recorded using data acquisition tool for the more the required signal was processed with program for frequency component analysis. Peripheral artery disease simulates by making the blood flow in level tube at low speed and made to accumulate in tube. And an Electromagnetic signal required for both conditions acid above and the same signal process with spectrogram in matlab for analysis as shown in Fig. 26.2.

Fig. 26.2 Electromagnetic signal acquisition for different blood dynamics

244

G. P. Ramesh

26.3 Antenna System Analysis Without Blood Fluid Sample The electromagnetic signal acquired from proposed antenna process with spectrogram and the corresponding output is shown below. The signals were recorded under three conditions such as antenna only condition, electromagnetic signal acquired from antenna during normal blood flow and electromagnetic signal acquired from antenna during blood accumulation condition. The signal acquired from the antenna system without blood fluid sample is recorded is shown in Fig. 26.3 and the electromagnetic signal acquired from the antenna system without blood fluid sample ranges from 20 to 140 dB is recorded is shown in Fig. 26.4. The magnitude signal acquired without blood fluid sample is 15.82 at the normalized frequency is 0.54 GHz is shown in Fig. 26.5. The probability distribution comparison of electromagnetic signal acquired without blood fluid sample, the power distribution is upto 1.5 is shown in Fig. 26.6.

26.4 Antenna System Analysis During Blood Flow The signal acquired from the antenna system with normal blood flow is recorded is shown in Fig. 26.7 and the electromagnetic signal acquired from the antenna system normal blood flow ranges from 50 dB to 150 dB is recorded is shown in Fig. 26.8. The magnitude signal acquired normal blood flow is 15.82 at the normalized frequency is 0.54 GHz which is shown in Fig. 26.9. The probability distribution comparison of The signal in the time domain

Normalized amplitude

1

0.5

0

-0.5

-1 0

2

4

6

Time, s

Fig. 26.3 Signal acquired from the antenna system without blood fluid

8

10

26 Microstrip Patch Antenna for Peripheral …

245

Fig. 26.4 Electromagnetic signal acquired without blood fluid

Magnitude Response (dB) 70

Magnitude (dB)

60 50 40 30 Normalized Frequency: 0.5428467 Magnitude: 15.32767

20 10 0

0.1

0.2

0.3

0.4

0.5

Normalized Frequency (

0.6

0.7

0.8

0.9

rad/sample)

Fig. 26.5 Magnitude signal without blood fluid

electromagnetic signal acquired normal blood flow, the probability of signal increases upto 7000 is shown in Fig. 26.10.

246

G. P. Ramesh

2

10

Probability distribution of the signal

4

probability distribution of the signal standard normal distribution

Number of samples

1.5

1

0.5

0

-1

-0.5

0

0.5

1

Signal amplitude

Fig. 26.6 Probability distribution signal acquired without blood fluid

The signal in the time domain

Normalized amplitude

1

0.5

0

-0.5

-1 0

2

4

6

8

10

Time, s

Fig. 26.7 Signal acquired during normal blood flow

26.5 Antenna System Analysis with Blood Accumulation The signal acquired from the antenna system with blood accumulation for Peripheral artery disease diagnosis is recorded and shown in Fig. 26.11. The antenna system with blood accumulation ranges from 50 dB to 130 dB is recorded is shown in Fig. 26.12. The magnitude signal acquired in blood accumulation is 15.82 at the normalized frequency on 0.54 GHz is shown in Fig. 26.13. The probability distribution comparison of electromagnetic signal acquired normal blood flow, increases upto 7000 upto 1.5 is shown in Fig. 26.14.

26 Microstrip Patch Antenna for Peripheral …

247

Fig. 26.8 Electromagnetic signal acquired during normal blood flow

Magnitude Response (dB) 60

Magnitude (dB)

50 40 30 20 10 0 0

0.1

0.2

0.3

0.4

0.5

Normalized Frequency (

0.6

0.7

rad/sample)

Fig. 26.9 Magnitude signal response during normal blood flow

0.8

0.9

248

G. P. Ramesh

Probability distribution of the signal

8000

probability distribution of the signal standard normal distribution

7000

Number of samples

6000 5000 4000 3000 2000 1000 0

-0.5

-1

0.5

0

1

Signal amplitude

Fig. 26.10 Probability distribution of signal acquired during normal blood flow

The signal in the time domain Normalized amplitude

1 0.5 0 -0.5 -1 0

2

4

6

Time, s

Fig. 26.11 The signal acquired during accumulation blood flow

8

10

26 Microstrip Patch Antenna for Peripheral …

249

Fig. 26.12 Electromagnetic signal acquired during blood accumulation Magnitude Response (dB) 60

Magnitude (dB)

50 40 30 20 10 0 0

0.1

0.2

0.3

0.4

0.5

Normalized Frequency (

0.6

0.7

rad/sample)

Fig. 26.13 Magnitude signal response during blood accumulation

0.8

0.9

250

G. P. Ramesh

Probability distribution of the signal

7000

probability distribution of the signal standard normal distribution

Number of samples

6000 5000 4000 3000 2000 1000 0

-1

-0.5

0

0.5

1

Signal amplitude

Fig. 26.14 Probability signal acquired during blood accumulation

26.6 Conclusion The Analysis and design of a novel multi-band miniature microstrip Patch antenna for wireless communication is presented. The designed microstrip Patch antenna is optimized to cover the GSM, DCS, PCS, UMTS applications in the ISM band with Bluetooth and Wi-fi. The proposed antenna is very compact, very easy to fabricate, and is fed by a 50 microstrip line which makes it very attractive for current and future cellular phones applications. The simulated bandwidth of the antenna was 89% for the first band, 19.45% for the second band and 36.25% for the third band. Thus the effectiveness of antenna was validated with testing of antenna on blood electromagnetism and signal processing techniques. The variation in proportion to blood dynamics is used for Peripheral artery disease diagnosis.

References 1. Nezhad and Hassani: A novel triband E-shaped printed monopole antenna for MIMO application. IEEE Antennas Wirel. Propag. Lett. 9, 576–579 (2010) 2. Fang, H.S., Lin, P.Y. Chuang, C.S.: Triple-band MIMO antenna for mobile wireless applications. IEEE Antennas Wirel. Propag. Lett., vol. 15, pp. 500–503, 2016 3. Liou, C.Y., Mao, S.G.: Miniaturized triple-band antennas with high isolation for MIMO antenna system applications. Microw. Opt.Technol. Lett. 57(11), 2555–2558 (2015) 4. Ojaroudi, N., Halili, K.: Design of triple-band monopole antenna with meander line structure for MIMO application. Microw. Opt. Technol. Lett. 54(9), 2168–2172 (2012) 5. Ojaraoudi, N., Ojaraoudi, M.: A novel design of triple-band monopole antenna for multi-input multi-output communication. Microw. Opt.Technol. Lett. 55(6), 1258–1262 (2013) 6. Mehranpour, M., Ojaroudi, Y.: A new design of triple-band WLAN/WiMAX monopole antenna for multiple-input/multiple-output applications. Microw. Opt. Technol. Lett. 56(11), 2667–2671 (2014)

26 Microstrip Patch Antenna for Peripheral …

251

7. Chu, Q.X., Huang, T.G.: A compact wideband MIMO antenna with two novel bent slits. IEEE Trans. Antennas Propag 60(2), 482–489 (2012) 8. AMcEwan and Excell: Wideband printed MIMO/Diversity monopole antenna for WiFi/WiMAX applications. IEEE Trans. Antennas Propag 60(4), 2028–2035 (2012) 9. Caloz, C., Itoh, T.: Electromagnetic Metamaterials: Transmission Line Theory and Microwave Applications. Wiley

Chapter 27

Wireless EAR EEG Signal Analysis with Stationary Wavelet Transform for Co Channel Interference in Schizophrenia Diagnosis V. Nithya and G. P. Ramesh Abstract Schizophrenia is a mental disorder where the patient experience changes in thought process, behavior and emotion. The changes listed above occur due to chemical imbalance in brain. Due to the nature of the disorder, the patients confront the family members about the things they hear and hallucinate. Initially, the family members deny to queries made by patient and later the responses evolve into anger and quarrel. The family members often lack awareness about schizophrenia disorder. Hence, there is a need to diagnose auditory hallucination at early stage. The auditory hallucination alters the EEG signal in ear. EEG sensor is designed and the same place behind the ear lobe to acquire the change in EEG pattern. For study the EEG pattern, acquire for normal and schizophrenia person while watching different videos namely funny video and horror video. The EEG signal acquire during movie watching task and transmit EEG to the base station through wireless sensor network for the wavelet analysis and classification to evaluate the efficiency of data transmission in various routing algorithms such as AODV and DSR and co channel interference of spread spectrum modulation address. Keywords Schizophrenia · Ear EEG · Auditory hallucination · WSN · AODV · DSR

27.1 Introduction Schizophrenia is a cognitive psychological disorder which causes delusions and hallucinations in patients. Schizophrenia occurs in people at age between 16 and 40. The males are prone to affect by schizophrenia at younger age than females. The schizophrenia in some cases develops at a slow rate and the patients are unaware V. Nithya (B) · G. P. Ramesh Department of E.C.E, St. Peter’s Institute of Higher Education and Research, Avadi, Chennai, India e-mail: [email protected] G. P. Ramesh e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_27

253

254

V. Nithya and G. P. Ramesh

about the disease for many years. In marginally low cases, the schizophrenia develops at a faster phase. Schizophrenia symptoms are categorized as negative, positive, emotional and cognitive symptoms. The negative symptoms includes facial expression loss, no interest in anything and liveliness. The positive symptoms include psychotic symptoms such as hallucinations and delusions. Unlike positive and negative symptoms the cognitive symptoms affects the thought process of the patients. The thought process results in either positive or negative symptoms. The emotional symptoms are negative symptoms such as lack of expression and no liveliness. In addition, Schizophrenia patients exhibit major symptoms such as delusions, thought disorder, hallucination, poor expression of emotion, lack of motivation, unawares to surrounding, social withdrawal and cognitive difficulties. The multi-layer RBMs profound system can separate profound various leveled portrayals of the genomic information, and afterwards guarantees a more exhaustive approach for the psychological illness conclusion [1]. The authors demonstrate that the name expectation strategy given system combination from numerous information composes indicates guarantees for the more definite conclusion of schizophrenia, which can likewise be stretched out to other ailment models [2]. These outcomes have consistency with Classical network examination works and exhibit the MST potential as a ground-breaking strategy to be utilized as a part of looks into, considering schizophrenic brain connective [3]. The starter comes about underscore the utility of such a VR-based framework, to the point that empowers accurate and quantitative evaluation of social aptitude deceits in patients with schizophrenia [4]. The examination researched functional availability contrasts in some of resting state networks (first consideration network, worldly network, sound network, and the default mode network) in early-onset schizophrenia patients and offspring of control gathering (ordinary improvement) utilizing resting state functional MRI technique. This first examination planned to add to the portrayal of schizophrenia infection by distinguishing the particular changes that happen in the resting state networks of schizophrenia brain [5]. The authors demonstrate that our outcomes are steady with past findings in writing, yet the authors likewise show that the fuse of nonlinear relationship in the information empowers the location of spatial examples that are not identified by straight ICA. Specifically, The authors indicate networks including the basal ganglia, cerebellum and thalamus that show significant contrasts in patients versus controls, some of which demonstrate clear nonlinear examples [6]. The authors outcomes demonstrate that the blend of these two strategies gives significant data that catches the basic qualities of brain network availability in schizophrenia. Such data is helpful for the expectation of schizophrenia patients. Classification precision execution was enhanced significantly (up to ≈7%) concerning just the fMRI strategy and (up to ≈21%) in respect to just the MEG technique.

27 Wireless EAR EEG Signal Analysis with Stationary Wavelet …

255

27.2 Related Work Functional network components as an element of task contain valuable data for the singular expectation of schizophrenia patients. Such data is valuable for preparing and replicates in testing. An execution was enhanced significantly (up to ~20%) concerning a solitary FNC (resting-state) measure [7]. The authors can foresee the gathering of the unlabelled SCZ subjects. The gathering forecast technique was connected to test the intensity of network-based highlights, and the execution was assessed by a 10-overlay cross-validation. The outcomes demonstrate that the forecast exactness is the most astounding while applying the authors expectation technique to the combined network got from three data composes among seven tried networks. As a conclusion, integrative methodologies that can exhaustively use different sorts of data are more helpful for finding and expectation [8]. The gathering expectation strategy was connected to test the intensity of network-based highlights, and the execution was assessed by 10-overlap cross-validation. The outcomes demonstrate that the expectation exactness is the most noteworthy while applying the authors forecast strategy to the combined network got from three data writes among seven tried networks. As a conclusion, integrative methodologies that can exhaustively use numerous kinds of data are more helpful for finding and forecast [9]. The RMCL process that is manufactured utilizing R programming dialect is connected to PPI networks of schizophrenias chance elements applicant qualities data got from the Bio-GRID database. RMCL calculation reproduction performed with the various parameter of swelling. At that point, the consequences of RMCL calculation recreation are contrasted with MCL calculation reproduction with similar parameters. RMCL calculation gives brings about the covering bunches, which mean there is the connection between groups. In this manner, in light of the after effects of RMCL calculation reenactment, there is the connection between protein bunches of a few applicant qualities, one of which is the connection of NRG1 and CACNG2 quality item [10]. The latest discoveries propose a vital change in the brain design towards a more adjusted network, for the most part in the associations identified with longrange cooperation’s. Since these progressions are significantly more articulated in controls, a shortfall in the neural network revamping can be related with SCH. Also, an exactness of 72.5% was gotten utilizing a recipient working trademark bend with a forget one cross-validation method. The freedom of network topology has been shown by the different intricacy measure proposed in this examination, along these lines, it supplements conventional diagram measures as a way to describe brain networks [11]. The exactness of the proposed two-layer strategy can achieve 91.63% through leave-one-out cross-validation. The test comes about to show the achievability of scaling musical observation by investigating brain AEPs evoked by musical boosts with various multifaceted nature and consonance. These outcomes can be connected to the improvement of musical preparing assessment and mental issue finding frameworks later on [12]. The motivation behind this investigation is to portray the intellectual elements of schizophrenic patients utilizing unique sound-related and visual event-related possibilities (ERPs) given the Wisconsin Card Sorting Test

256

V. Nithya and G. P. Ramesh

(WCST). From the test comes about, it is indicated that there is gradualness of programmed psychological handling and controlled subjective preparing amid WCST in correlation with ERPs for schizophrenic patients [13]. The research indicates mitochondrial complex I complex, particularly NADH: ubiquinone oxidoreductase (EC 1.6.5.3) gamma subunit could be the potential biomarker for schizophrenia. These outcomes proposed that the methylation profile of the genomic arrangement in the promoter locale of mitochondrial qualities may be a critical factor of pathogenesis for schizophrenia.

27.3 Methodology The ear EEG signal acquire via two electrodes shown in Fig. 27.1. The EEG signal acquired from sensor process with stationary wavelet transform for different routing algorithms in wireless sensor network as shown in Fig. 27.2. The Stationary wavelet transform (DWT), provides enough information for synthesis and analysis of the original signal. A digital signal’s time scale representation is by using digital filtering techniques. The continuous wavelet transform was computed by varying the window’s scale, window shifting in time, multiplying the signal and integrating the signal. In the Stationary wavelet transform, filters with various cut-off frequencies and analyze at various scales. The signal is sent through high pass filters and low pass filters to analyze the high and low frequencies. Through the process of filtering operations, the resolution of the signal can be changed and through up sampling and down sampling, the frequency scale can be exchanged. The Stationary wavelet transform’s coefficients are sampled from the continuous wavelet transform on a dyadic grid i.e., s0 = 1 and t0 = 1 giving s = 2j and t = k * 2j. The Stationary wavelet transform analyze the signal at various frequency bands with various resolutions by decomposing the signal into a coarse approximation and detail information. The Stationary wavelet transform has two sets of functions called scaling and wavelet functions. The standard Stationary wavelet transform is based on filters H and G and a binary decimation operator D0 . The filter H is a low pass filter and represented by a sequence {h n }. The effect of low pass filter on a infinite sequence that is double is represented as

Fig. 27.1 Two electrode ear EEG sensor

27 Wireless EAR EEG Signal Analysis with Stationary Wavelet …

Fig. 27.2 workflow for EEG signal analysis

(Hx)k =



h n−k xn

n

The filter is expected to satisfy the internal orthogonality relation that is 

h n h n+2 j = 0

n

for all integers j = 0 and having the sum of squares



h 2n = 1.

257

258

V. Nithya and G. P. Ramesh

27.4 Co-channel Interference in WSN for Dynamic Signal Transmission In this paper, direct spread spectrum modulation (DSSM) in WSN has evaluated for co channel interference during dynamic signal transmission over different routing algorithm. DSSM communication attract more substantial interest because of the many promising inherent advantages such as high spectral efficiency with low average error probability, high energy efficiency, and very simple transmitter and receiver architectures. Hence, WSNs use DSSM because of promising to ameliorate. The performance of a WSN with DSSM and co-channel interference is analyzed through the transmission of dynamic Schizophrenia signal and evaluate for the loss and high spectral efficiency. The received signal analyze with stationary wavelet transform.

27.4.1 Dynamic Signal Transmission in WSN Figure 27.3 shows the deployment of WSN network for the dynamic signal transmission with four nodes. Each node consists of MAC and Physical of IEEE 802.15.4/g with +19 dBm of transmits power output. The four nodes are communicated with AODV and DSR algorithms. The transmitted signal is evaluate for the residual noise for to identify the co channel interference and to evaluate the pattern of the signal of schizophrenia patients. The simulation results for the patients with abnormality when seeing funny videos are performed. Here they obtained signal is de-noised by using the stationary wavelet transform de-noising. The stationary wavelet transform de-noising is performed by importing a signal and performing a stationary wavelet decomposition of a signal. Fig. 27.4 show the original signal recorded with patients with abnormality when the patient is seeing funny videos. From the Figure the original signal has peak noise variation at amplitudes 0.22, 0.2, 0.15 and −0.15 in the range 0.12 to 0.14 s, 1.14 s

Fig. 27.3 WSN for dynamic signal transmission

27 Wireless EAR EEG Signal Analysis with Stationary Wavelet …

259

Fig. 27.4 Original signal recorded from patients with abnormality when seeing funny videos

and 2.1 s. The original signal shown in Fig. 27.4 shows that there is significant amount of noise components present in the signal. First the stationary wavelet decomposition for the original signal is performed without de-noising up to 5 levels. This generates the approximation and detail coefficients up to level 5. Then the coefficients of approximation and detail up to level 5 are displayed Fig. 27.5. The figure shows the non decimated approximation coefficients and non decimated detail coefficients before de-noising the original signalD6T Thermal Sensor (Fig. 27.6).

Fig. 27.5 Non decimated approximation and detail coefficients before de-noising

260

V. Nithya and G. P. Ramesh

Fig. 27.6 Residuals of the original signal before de-noising

Residuals are differences between the one-step-predicted output from the model and the measured output from the validation data set. Residual analysis show various information’s depending on time domain or frequency domain input output validation data. The residuals of the signal show that more noise components present in the signal. Then the stationary wavelet decomposition is performed with de-noising up to 5 levels. Figure 27.7 shows the de-noised signal. The residual of the signal shown implies that the residual of the signal after de-noising is the compressed form of the residual signal before de-noising. The histogram and cumulative histogram representation of the de-noised signal is shown in Fig. 27.8. The histogram of a signal shows the number of samples in the signal. Figure 27.9 shows the original signal recorded with patients with abnormality when the person is seeing horror videos. From the Figure, the signals recorded when the patient is seeing horror videos has maximum amplitude variations as compared to the signals recorded when the patient the patient is seeing funny videos (Table 27.1). From the Figure, the signal has maximum to maximum variation of noise in amplitudes ranging from −0.075 to 0.15 in 0.18 to 2.2 s First the stationary wavelet

Fig. 27.7 De-noised signal from SWT

27 Wireless EAR EEG Signal Analysis with Stationary Wavelet …

261

Fig. 27.8 Histogram and cumulative histogram of the de-noised signal

Fig. 27.9 Original Signal recorded from patients with abnormality when seeing horror videos Table 27.1 EEG statistical analysis for schizophrenia patient watching funny video Parameter

Value

Median

−0.0007481

Mean

−0.00213

Minimum

0.0692

Maximum

−0.05982

Range

0.1227

Standard dev

0.01454

Median Abs Dev

0.007694

Mean Abs Dev

0.01077

L1 norm

2382

L1 norm

6.836

Max norm

0.06292

262

V. Nithya and G. P. Ramesh

decomposition for the original signal is performed without de-noising up to 5 levels. This generates the approximation and detail coefficients up to level 5. Then the coefficients of approximation and detail up to level 5 are displayed Fig. 27.10. Shows the non decimated approximation coefficients and non decimated detail coefficients before de-noising the original signal. Residuals are differences between the one-step-predicted output from the model and the measured output from the validation data set. Fig. 27.11 shows the residual of the original signal before de-noising.

Fig. 27.10 Non decimated detail and approximation coefficients before de-noising

Fig. 27.11 Residual of the signal before de-noising

27 Wireless EAR EEG Signal Analysis with Stationary Wavelet …

263

Fig. 27.12 De-noised Signal

Fig. 27.13 Histogram and cumulative histogram of the de-noised signal

From Fig. 27.11, the residual of the signal has more noise components as compared to the residual of the signal taken for patients with abnormality while seeing funny videos. Figure 27.12 shows the de-noised signal. The histogram and cumulative histogram representation of the de-noised signal is shown in Fig. 27.13. The histogram of a signal shows the number of samples in the signal (Table 27.2). Table 27.3 shows the Residual value of the signal obtain from the same person through the WSN network with different routing algorithm and the residual measure for the signal shows the value of co channel interference in the network.

27.5 Conclusion Ear EEG electrode is designed and place behind the ear to analyze the brain signals to plot the changes in emotion due to influence of watching funny and horror videos and transmit through WSN for various routing algorithms and the signal residual estimate the co channel interference during the dynamic signal transmission. Patients were selected to analyze the effects of schizophrenia disease in their emotional activities. The results of the schizophrenia patients were compared with the results of normal patients to identify the severity in recorded signal due to the cause of disease. The

264

V. Nithya and G. P. Ramesh

Table 27.2 EEG statistical analysis for schizophrenia patient watching horror video Parameter

Value

Median

−0.0002397

Mean

−0.0002568

Minimum

0.04028

Maximum

−0.03621

Range

0.007641

Standard dev

0.008386

Median Abs Dev

0.005306

Mean Abs Dev

0.006554

L1 norm

1450

L1 norm

3.944

Max norm

0.04028

Table 27.3 Residual value of the signal S. No.

Parameter

AODV

DSR

1

Median

−0.0001945

−0.0001715

2

Mean

−0.0001845

−0.0001645

3

Standard deviation

0.007515

0.0064258

experiments were conducted on 10 persons of normal group and schizophrenia group. The people were shown same movie watching task. The results shows, schizophrenia person show different ear EEG response to watching movie compared to normal person. Similarly the signal show the residual for the both horror and funny video the EEG signal shows the AODV algorithm has less co channel interference.

References 1. Qiao, C., Lin, D., Wang, Y.-P.: The effective diagnosis of schizophrenia by using multi-layer RBMs deep networks. In: IEEE International Conference on Bioinformatics and Biomedicine (BIBM), pp. 603–606 (2015) 2. Deng, S.P., Lin, D., Calhoun, V.D., Wang, Y.P.: Diagnosing schizophrenia by integrating genomic and imaging data through network fusion, In: Proceedings of the 2016 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), pp. 1307–1313 (2017) 3. Anjomshoa, M., Dolatshahi, F., Amirkhani, F., Rahmani, M.M., Mirbagheri, M.H., Aarabi, Structural brain network analysis in schizophrenia using minimum spanning tree. In: 2016 38th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, pp. 4075–4078 (2016) 4. Bekele, E., Bian, D., Peterman, J., Park, S., Sarkar, N.: Design of a virtual reality system for affect analysis in facial expressions (VR-SAAFE); application to schizophrenia. IEEE Trans. Neural Syst. Rehabil. Eng. 25(6), 739–749 (2017)

27 Wireless EAR EEG Signal Analysis with Stationary Wavelet …

265

˙ 5. Içer, S., Esra, Ö., Görüntüleme, T., Uygulama, T., Üniversitesi, E.: Obtaining resting state networks in early onset schizophrenia disease by independent component analysis. 2016 Med. Technol. Natl. Congr. 1(2), 176–180 (2016) 6. Castro, E., Hjelm, R.D., Plis, S.M., Dinh, L., Turner, J.A., Calhoun, V.D.: Deep independence network analysis of structural brain imaging: application to schizophrenia. IEEE Trans. Med. Imaging 35(7), 1729–1740 (2016) 7. Cetin, M.S., Stephen, J.M., Calhoun, V.: Sensory load hierarchy-based classification of schizophrenia patients. In: Proceedings of International Conference Image Process. ICIP, vol. 2015–Dec, pp. 467–471 (2015) 8. Deng, S.P., Lin, D., Calhoun, V.D., Wang, Y.P.: Predicting schizophrenia by fusing networks from SNPs, DNA methylation and fMRI data. In: Proceedings of Annual International Conference of the IEEE Engineering in Medicine and Biology Society EMBS, vol. 2016–Oct, pp. 1447–1450 (2016) 9. Fajnerova, S.I., Rodriguez, M., Spaniel, F., Horacek, J., Vleck, K., Levcik, D.: Spatial navigation in virtual reality—from animal models towards schizophrenia. 2015 Int. Conf. Virtual Rehabil. 1(2), 44–50 (2015) 10. Ginanjar, R., Bustamam, A., Tasman, H.: Implementation of regularized markov clustering algorithm on protein interaction networks of schizophrenia’s risk factor candidate genes. 2016 Int. Conf. Adv. Comput. Sci. Inf. Syst. 1(2), 297–302 (2016) 11. Gomez-Pilar, J. et al.: Novel measure of the weigh distribution balance on the brain network: graph complexity applied to schizophrenia. In: Annual International Conference of the IEEE Engineering in Medicine and Biology Society EMBS, vol. 2016–Oct, pp. 700–703 (2016) 12. Hsieh, T.H., Sun, M.J., Liang, S.F.: Musical perception scaling of AEPs from musicians, schizophrenia and normal people. TAAI 2015—2015 Conf. Technol. Appl. Artif. Intell. 1(3), 358–362 (2016) 13. Huang, M., Lo, P., Chen, C., Chen, C., Cheng, K.: The application of computerized WCST and long-term evoked potentials for schizophrenia analysis. 2015 Int. Conf. Virtual Rehabil. 2(1), 5165–5168 (2006)

Chapter 28

Advance Approach for Effective EEG Artefacts Removal Rudra Bhanu Satpathy and G. P. Ramesh

Abstract Electrical impulses generated by nerve firings in the brain diffuse through the head and can be measured by electrodes placed on the scalp and is termed as electroencephalogram (EEG). The artefacts, such as eye blinks etc., in EEG recordings obscures the underlying processes and makes analysis difficult. Large amounts of data must often be discarded because of contamination by artefacts. To overcome this difficulty, signal separation techniques are used to separate artefacts from the EEG data of interest. Some artefacts, such as eye blinks, produce voltage changes of much higher amplitude than the endogenous brain activity. EEG data may be contaminated at many points during the recording and transmission process. Most of the artefacts are biologically generated by sources external to the brain. Improving technology can decrease externally generated artefacts, such as line noise, but biological artefacts signals must be removed after the recoding process. This paper proposes a new technique for removing the artefacts from the EEG signal which uses kurtosis based on difference of Gaussian and Super-Gaussian signal and Spatially-Constrained ICA (SCICA) and Daubechies wavelet techniques. Threshold plays an important role in separating the artefacts from the non-artefact EEG. Otsu’s Threshold is been adopted as the thresholding method in this paper. Keywords EEG · Artefacts · SCICA

28.1 Introduction The EEG signal works as a good tool to explore brain activity and is able to detect changes within a milliseconds time-span. Depending upon the type of neuron, an action potential takes 0.5–130 ms approximately to propagate across a single neuron, R. B. Satpathy (B) · G. P. Ramesh Department of Electronics and Communication Engineering, St. Peter’s Institute of Higher Education and Research, Chennai, Tamil Nadu, India e-mail: [email protected] G. P. Ramesh e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_28

267

268

R. B. Satpathy and G. P. Ramesh

Fig. 28.1 a Clean EEG, b Eye blink, c Eye movement, d 50 Hz, e Muscle activity, f Pulse

whereas, other methods likewise functional Magnetic Resonance Imaging (fMRI) and Positron Emission Tomography (PET) have time resolution in terms of seconds and minutes and makes these methods less efficient. Moreover, EEG directly measures the brain’s electrical activity, whilst other methods such as Single Photon Emission Computed Tomography (SPECT), fMRI record changes in blood flow or PET record changes in metabolic activity, which are indirect markers of electrical activity belonging to the brain. EEG data may be contaminated at many points during the recording and transmission process. Most of the artefacts are biologically generated by sources external to the brain. Improving technology can decrease externally generated artefacts, such as line noise, but biological artefacts signals must be removed after the recoding process. Figure 28.1 shows waveforms of some common EEG artefacts. Eye Blink artefacts: It is very common in EEG data, produces a high amplitude signal that can be many times greater than EEG signals into consideration. Because of its high amplitude an eye blink can corrupt data on all electrodes, even those which are at the back of the head. Eye artefacts are often measured more directly in the Electro-Oculogram (EOG), pairs of electrodes placed above and around the eyes. But unfortunately, these measurements are contaminated with EEG signals of interest and so simple subtraction is not a removal option even if an exact model of EOG diffusion across the scalp is available. Eye Movement: These artefacts are caused by the reorientation of the retinocorneal dipole. The effect of this artefacts is sturdier than that of the eye blink artefacts. Eye blinks and movements often occur at close intervals. Line Noise: Strong signals from A/C power supplies can corrupt EEG data during transfer from the scalp electrodes to the recording device. Notch filters are often used to filter this artefacts containing lower frequency line noise and harmonics. Notch

28 Advance Approach for Effective EEG Artefacts Removal

269

filtering at these frequencies can remove useful information. Line noise can disturb the data from some or all of the electrodes depending on the source of the problem. Muscle Activity: These artefacts are caused by activity in different muscle groups including neck and facial muscles. These signals have a wide frequency range and can be distributed across different sets of electrodes depending on the location of the source muscles. Pulse: When an electrode is placed on or near blood vessel, it causes pulse, or heartbeat, artefacts. The expansion and contraction of the vessel introduce voltage changes into the recordings. The artefacts signal has a frequency around 1.2 Hz, but can vary with the state of the object. This artefacts can appear as a sharp spike or smooth wave. This paper proposes a new technique for removing the artefacts from the EEG signal which uses kurtosis based on difference of Gaussian and Super-Gaussian signal and Spatially-Constrained ICA (SCICA) and daubechies wavelet techniques. Threshold plays an important role in separating the artefacts from the non-artifact EEG. Otsu’s Threshold is been adopted as the thresholding method in this paper. This method pre-assumes that EEG contains two classes namely, artifact and non artifact signal and further it calculates the optimum threshold separating those two classes.

28.2 Related Work Artifacts noise in EEG are usually handled by dumping the affected segments of EEG. The humblest methodology is to discard a fixed length segment, perhaps one second, from the time an artifacts is detected. Discarding segments of EEG data with artifacts can greatly decrease the amount of data available for analysis. EEG data collected from children is especially problematic in this respect. The first attempts at removing artifacts focused on eye blinks. Regression using the EOG channel was attempted in the time and frequency domain. These methods all rely on a clean measure of the artifacts signal to be subtracted out. Since the EOG is contaminated with EEG signals, the regression of ocular artifacts has the undesired effect of removing EEG signals from the observations. A good review can be found in. Kenemans et al. gave a general lagged regression model. Jung et al. used this regression model for a baseline artifacts removal method. Multivariate statistical analysis techniques, such as principal component analysis, have been used to separate and remove noise signals from the brain activity of interest. Comparison of four methods for artifacts removal by artificially mixing an artifacts signal from one subject with a set of EEG signals from another subject is given in. The artificial mixing matrices were chosen to approximate mixing in the scalp. Two independent component analysis methods studied in, were significantly better than principal component analysis and simple EOG subtraction. Performance was measured using the mean squared error between the true artifacts signal and the extracted artifacts signal. Significance was measured using an F-statistic and Tukey’s studentized range test.

270

R. B. Satpathy and G. P. Ramesh

The common spatial patterns (CSP) technique, which requires the use of two data sets was used by Kolesto remove abnormal Components, It uses data from 80 patients. No quantitative evaluation was done on the removal but it was visually observed that the artifacts were extracted into a small number of components that would allow their removal. In online filtering systems, artifacts recognition is important for achieving their automatic removal. One approach to recognition of noise components is based on measuring structure in the signal. The fractal dimension and a metric based on auto-regressive (AR) coefficients have been used for this purpose. Eye blinks and heart beats were found to have consistent fractal dimensions on the data studied. Jung suggests that the spectral structure might be distinct for certain artifacts components (e.g., line noise) and that this would allow for automatic removal of these artifacts. Kalman filters and extended Kalman filters have also been used for artifacts detection with success. Depending heavily on the artifact type. This approach was most successful at recognizing one second windows containing muscle and movement artifacts. The common signal separation approaches to artifact removal are: principal components analysis, maximum signal fraction analysis, canonical correlation analysis, and independent component analysis. Shao et al., proposed an automatic EEG Artifact removal which uses a Weighted Support Vector Machine approach with error correction. An automatic electroencephalogram (EEG) artifact removal method is presented in this paper. Compared to past methods, it has two unique features: a. A weighted version of support vector machine formulation that handles the inherent unbalanced nature of component classification and b. The ability to accommodate structural information typically found in component classification. The advantages of the proposed method are demonstrated on real-life EEG recordings with comparisons made to several benchmark methods. Results show that the proposed method is preferable than the other methods in the context of artifact removal by achieving a better tradeoff between removing artifacts and preserving inherent brain activities. Qualitative evaluation of the reconstructed EEG epochs also demonstrates that after artifact removal inherent brain activities are largely preserved. Kavitha et al., suggested a modified ocular artifact removal technique from EEG by adaptive filtering. Electroencephalogram (EEG) is the reflection of brain activity and is widely used in clinical diagnoses and biomedical researches. EEG signals recorded from the scalp contain many artifacts that make its interpretation and analysis very difficult. One major source of artifacts is the eye movements that generate the Electrooculogram (EOG). Many applications of EEG such as Brain Computer Interface (BCI) need real time processing of EEG. Adaptive filtering is one of the most efficient methods for removal of ocular artifacts which can be applied in real time. In the conventional adaptive filtering, primary input is the measured EEG signal and the reference inputs are vertical EOG (VEOG) and horizontal EOG (HEOG) signals.

28 Advance Approach for Effective EEG Artefacts Removal

271

In this paper, an adaptive filtering approach is proposed which includes radial EOG (REOG) signal as a third reference input. By the analysis based on the performance of adaptive algorithms using two reference inputs i.e. HEOG and VEOG and that with three reference inputs i.e. VEOG, HEOG and REOG, it is found that the reference method gives more accurate results than the existing reference method. Furthermore [1] have proposed EEMD-CCA and EEMD-IVA (Independent Vector Analysis) method for EMG artifact removal from single channel EEG data and concluded that EEMD-CCA is outperformed than IVA method. The EEG motion artifacts removal approach is extended by using cascaded method of EEMD and multi-set CCA (EEMD-MCCA). The method presented good results [2] by increasing PSNR and reducing the RMSE values in comparison to the existing muscle artifact removal methodologies available. The Multivariate-EMD approach has proposed by Teng and Wang [3] for EMG artifact removal and compared with ICA algorithm based on SNR and MSE. However, Zhao [4] has proposed a method for automatic ocular artifacts (OA) removal from EEG recordings based on Wavelet-enhanced Canonical Correlation Analysis (WCCA) approach. The WCCA algorithm outperforms over popular ocular artifacts removal methods as CCA, ICA and WICA. This WCCA method removed the most ocular artifacts with minimal cerebral information loss [5].

28.3 Proposed System A system model is used to show the schematic representation of the proposed algorithm to understand the concept as depicted in Fig. 28.2. Work zones such as bridges and tunnel construction or maintenance can give traffic incidents, a boost. One of the ways or methods by which traffic incidents can get prevalent, is due to existing work zones.

28.3.1 Implementation Algorithm The Proposed EEG Motion Artifact Removal Algorithm is as Follows Define the reference and artifactual EEG data from standard multi-channel EEG data set for correlation match. The artifactual EEG data is passed through EEMD to convert single channel EEG into multi-channel EEG data (IMFs). Generated IMFs are passed to CCA algorithm for source separation. The signal and artifact sources are separated by CCA algorithm. Further, Wavelet Transform is applied on correlated components. The Pearson’s correlation coefficient is used for artifacts recognition and their elimination and finally, remaining signal sources are reconstructed to find the artifact-free signal.

272

R. B. Satpathy and G. P. Ramesh

Load the Acquired EEG data

Baseline Wandering

Artifactual EEG Data Generation

EEMD Algorithm

IMFn

IMF1 IMF2

CC1

Blind Source

CC2 Application of Wavelet Transform Algorithm of 3

Separation (BSS)

level CCn WT

WT

WT W

Pearson’s Correlation Coefficient based motion artifact removal

Reconstruction of all the remaining wavelet modes

Reconstructed Motion Artifact Free signal

Fig. 28.2 Block diagram of the proposed cascaded EEMD-ICA Technique

Parametric evolution is carried out based on SNR and correlation improvement. The correlation is carried out with respect to reference original EEG data initially defined. The parametric evaluation of the proposed method is done with existing motion artifact removal methods. The two-stage artifact removal techniques have been employed in the literature for single channel EEG signal artifact removal. The techniques are termed as two-stage as they need an additional algorithm for artifact identification and effective removal. The Blind Source Separation algorithms (CCA and ICA) are applicable for multichannel EEG signal to improve the artifact removal capacity. However, if the input signal is single channel then this measurement is required to convert into the multichannel signal. Further, this multi-channel signal is applied for source separation. The algorithms which is applied for conversion from single channel to multi-channel is Ensemble Empirical Mode decomposition (EEMD). This algorithms is applied in cascade with Blind Source Separation algorithms (ICA or CCA) to improve the artifact removal efficiency. The state of the art algorithms are implemented to evaluate the efficiency of the proposed algorithm. The two-stage cascaded algorithms EEMD-ICA and EEMDCCA are implemented on data set [6] and results are evaluated qualitatively and quantitatively as discussed below.

28 Advance Approach for Effective EEG Artefacts Removal

273

EEMD- ICA Algorithm Implementation The cascaded two-stage approach of EEMD and ICA was first documented by Mijovic in 2010 for the removal of ECG artifact from EMG signals and the removal of seizure events from EEG signals [7]. This cascaded approach is applied on single channel EEG signal. The EEMD approach converts the single channel signal to multi-channel (IMFs). This EEMD is a data-adaptive approach for the input signal which provides robustness to the input EEG signal. Further, these generated IMFs are applied as an input to the second stage ICA approach for source separation. The ICA approach separates the input components into the underlying source signals which are generated due to their independent nature [8]. The sources identified as artifacts are removed and remaining components are reconstructed to acquire artifact-free signal. Figure presents the comparison plot of EEG signal before artifact removal (Synthetically generated motion artifactual EEG signal) and after artifacts removal with EEMD-ICA technique. It is concluded from the plot that some of the epochs are suppressed by EEMD-ICA algorithm. These epochs are generated due to motion artifacts. Even after EEMD-ICA cascaded approach filtering, the artifacts are present in the EEG signal in the form of randomness. This ICA algorithm is based on higher order statistics [7] and is found to be more complex in terms of computation. However, the CCA algorithm (based on second order statistics) reduces the computational complexity to a great extent. The low computational complexity results in a fast execution time [8] (Fig. 28.3). EEMD-CCA Algorithm Implementation The CCA algorithm is found to be highly efficient for artifact removal if the multichannel signal is applied as an input to it. The EEMD approach is used to convert 0.3

signal before artifact removal signal after artifact removal

0.2 0.1 0 -0.1 -0.2

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

Fig. 28.3 Plot of ground truth signal and EEG cleaned signal following artifact removal with EEMD-ICA

274

R. B. Satpathy and G. P. Ramesh

the single channel (x) signal to multi-channel signal X (consist of IMFs). The output of EEMD (X) is applied to CCA for source separation and identified artifact sources are removed. The signal sources are reconstructed to find artifact free signal [9]. This two-stage cascaded approach EEMD-CCA [10] can be applied as one level or two level. In one level the role of EEMD is to just convert the single channel input EEG signal into multi-channel signal. The CCA approach is further applied to remove the artifacts based on the correlation. However, in the two-level approach, most of the dominating artifact are suppressed by EEMD approach and then forwarded to CCA technique for further filtering. This CCA algorithm removes remaining lower amplitude artifact components. Thus, this two level artifact removal approach EEMDCCA is preferred [11]. The two-stage EEMD-CCA algorithm is implemented on synthetic artifact EEG data set and the efficiency of algorithm is evaluated. The Proposed Three Stage Algorithm This proposed algorithm is particularly dedicated to the removal of motion artifact from EEG signal. These motion artifacts rigorously affect the visual analysis of EEG signal, complicate the EEG signal features interpretation, thus, removal is quite difficult. The motion artifact removal efficiency is improved by proposing a threestage cascading approach of EEMD, CCA and Wavelet Transform. It is concluded from Fig. 28.4, that EEMD-CCA approach suppresses the motion artifact effectively, although the randomness due to artifacts and external noise are still present in the filtered EEG signal. The Wavelet Transform algorithm is found to be more effective for removing the randomness and unpredictability of motion artifacts [12].

0.3

signal before artifact removal signal after artifact removal

Amplitude (mV)

0.2 0.1 0 -0.1 -0.2

0

0.5

1

1.5

2

2.5

3

Time (Seconds) Fig. 28.4 EEG motion artifact removal using EEMD-CCA

3.5

4

4.5

5

28 Advance Approach for Effective EEG Artefacts Removal

Amplitude (mV)

0.3

275 signal before artifact removal signal after artifact removal

0.2 0.1 0 -0.1 -0.2 0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

Time (Seconds) Fig. 28.5 Output EEG signal after EEMD-CCA-DWT artifact removal approach

This three-stage approach is based on a cascaded combination of EEMD-CCADWT. This algorithm provides more fast and accurate separation of motion artifact from EEG signal [6]. EEMD is applied to convert input single channel EEG signal into the multi-channel signal by creating Intrinsic Mode Function (IMFs). These IMFs are applied to BSS-CCA approach for improved correlation-based signal and artifact source separation. Supplementary, the Discrete Wavelet Transform (DWT) is applied to minimize the randomness available after this two-stage filtering approach (EEMD-CCA). The smoothened output obtained from EEMD-CCA-DWT are free from motion artifact. This is observed from Fig. 28.5.

28.4 Results and Discussion In order to evaluate more efficiently, quantitative evaluation is performed based on some evaluation parameters DSNR, Lambda (λ), Power Spectral Density (PSD), Correlation Coefficient and RMSE [9]. The numerical value comparison is performed among two stage cascaded approach for EEG motion artifact removal and summarized in Table 28.1. In order to evaluate the efficiency of the algorithm this two-stage EEMD-ICA algorithm is applied on synthetic artifact EEG data set. The quantitative parametric evaluation of EEMD-ICA technique generated parametric value of λ as 52%, DSNR value of 15.23 dB. The improvement in PSD and Correlation are found to be 0.36 and 0.014 respectively. Moreover, Spectral Density improvement is evaluated as 0.36 and RMSE value as 0.16. The quantitative parametric evaluation of EEMDCCA technique generated a λ of 60%, DSNR value of 20.1 dB. The improvement in PSD and correlation are found to be 0.49 and 0.07 respectively.

276

R. B. Satpathy and G. P. Ramesh

Table 28.1 Comparison between the evaluation parameters DSNR(dB), Lambda, Correlation Coefficient, RMSE and Artifact Removal algorithm using EEMD-ICA, EEMD-CCA, EEMD-CCADWT techniques Evaluation parameters Artifact removal algorithm

DSNR in (dB)

Lambda (λ) in percentage

Power Spectral Density (PSD)

Correlation coefficient

RMSE

EEMD-ICA

15.23

52

0.36

0.014

0.16

EEMD-CCA

20.1

60

0.49

0.07

0.14

EEMD-CCADWT

31.20

75

0.65

0.110

0.075

To compare the performance of the existing artifact removal method (EEMDCCA), the parametric evaluations also have performed and encapsulated. It is seen that DSNR value is observed to be 31.20. The proposed algorithm provides 53% improvement than EEMD-CCA artifact removal approach. The parameter Lambda (λ) generated is 75%, offers improvement by 25% than existing approach. The attained PSD improvement is 0.65 with 31% progress than EEMD-CCA approach and the correlation coefficient is 0.075. The parameter RMSE is reduced to 0.110. Thus, it is observed that the parameter RMSE is diminished by 18.6% than existing method (EEMD-CCA). It is concluded from the quantitative parametric evaluation that, the EEG motion artifacts removal performance is improved by (EEMD-CCA-DWT) approach with improvement in lambda (λ). The DSNR value increases than existing EEMD-CCA [9] method and thus, presents the accomplishment of this three-stage approach (EEMD-CCA-DWT). The BSS-CCA approach supports the best correlation based artifact source separation. Moreover, the DWT removes the present low amplitude artifact signal randomness which is still present after two-stage artifact removal (EEMD-CCA) approach. The performance comparison in terms of bar chart is shown in Fig. 28.6. It is concluded that the proposed method (EEMD-CCA-DWT) outperforms in each parameter with improvement in DSNR, Lambda (λ) and Power Spectral Density and with a reduction in the value of RMSE in comparison of existing EEMD-CCA motion artifact removal approach.

28 Advance Approach for Effective EEG Artefacts Removal

277

Fig. 28.6 Evaluation parameter matrices comparison for EEMD-CCA-DWT with EEMD-CCA

28.5 Conclusion The Physiological EEG Signal is preferred as a good tool for analysis of neurological diseases and its measurement is highly sensitive to the human activity, external noises and electronic device interferences. These artifacts corrupt the quality and information of the EEG signal. Thus, the accurate separation of artifacts from the desired signal is important to support the medical doctor’s evaluation for better analysis and diagnosis of human neurological diseases. This is found after comprehensive literature study that BSS approaches are not sufficient enough for removal of the EEG artifacts. The Wavelet Transform algorithm is found to be more effective for eradicating the randomness and unpredictability of artifact if it is applied after BSS techniques. Thus, three-stage cascaded approach EEMD-CCA-DWT is proposed. The results are analysed qualitatively and quantitatively. The quantitative analysis is done based on some evaluation parameters and compared with existing EEMD-CCA [9]. The evaluation parameters like Difference in Signal to Noise Ratio (DSNR), Correlation, Spectral Distortion, Root Mean Square Error (RMSE), Lambda (percentage artifact reduction) show the success of proposed algorithm in comparison to existing EEMD-CCA algorithm for EEG motion artifacts removal. Similar conclusion is drawn by visual observations in qualitative analysis.

278

R. B. Satpathy and G. P. Ramesh

References 1. Ghandeharion, H., Erfanian, A.: A fully automatic method for ocular artifact suppression from EEG data using wavelet transform and independent component analysis. Eng. Med. Biol. Soc. 5265–5268 (2006) Hyvarinen, A., Erkki, O.: A fast fixed-point algorithm for independent component analysis. Neural Comput. 9(7), 1483–1492 (1997) 2. Chen, X.H., Peng, H.: Removal of muscle artifacts from single-channel EEG based on ensemble empirical mode decomposition and multi-set canonical correlation analysis. J. Appl. Math. 1–11 (2014) 3. Teng, C.Z., Wang, G.: The removal of EMG artifact from EEG signals by the multivariate empirical mode decomposition. Signal Processing, Communications and Computing, pp. 873–876 (2014) 4. Zhao, C., Qiu, T.: An automatic ocular artifacts removal method based on wavelet-enhanced canonical correlation analysis. Eng. Med. Biol. Soc. (EMBC), 4191–4194 (2015) 5. Islam, M.K., Rastegarnia, A., Yang, Z.: A wavelet-based artifact reduction from scalp EEG for epileptic seizer detection. IEEE J. Biomed. Health Inf. (2015) 6. PhysioNet-motion artifact contaminated EEG and EEG data (motion artifact). [Online]. http:// physionet.org/cgi-bin/atm/ATM 7. Mijovic, B.V., Huffel, S.V.: Source separation from single-channel recordings by combining empirical-mode decomposition and independent component analysis. Biomed. Eng. 57(9), 2188–2196 (2010) 8. Soomro, M.H., Yusoff, M.: Comparison of blind source separation methods for removal of eye blink artifacts from EEG. Intelligent and Advanced Systems (ICIAS), pp. 1–6 (2014) 9. Sweeney, K.T., Ward, T.E.: The use of ensemble empirical mode decomposition with canonical correlation analysis as a novel artifact removal technique. Biomed. Eng. 60(1), 97–105 (2013) 10. Sweeney, K.T., Onaral, B.: A methodology for validating artifact removal techniques for physiological signals. Inf. Technol. Biomed. 16(5), 918–926 (2012) 11. Soomro, M.H., Jatoi, M.A.: Automatic eye-blink artifact removal method based on EMD-CCA. Complex Medical Engineering (CME), pp. 186–190 (2013) 12. Ng, S.C., Raveendran, P.: Enhanced μ rhythm extraction using blind source separation and wavelet transform. IEEE Trans. Biomed. Eng. 56(8), 2024–2034 (2009)

Chapter 29

Security in Internet of Things Shivam Kolhe, Sonia Nagpal and Jesal Desai

Abstract IoT simply means that all the entities—devices to be more specific are connected with each other and they talk to each other via communication channel without any human help or interaction. Since there is no Human to Machine interaction in IoT, security is necessary. Nowadays security and privacy are one of the main issues IoT is facing. Paper explains the architecture of IoT and along with general introduction of what IoT is with some examples and continue on to security problems IoT is facing, challenges and its solutions. Keywords Internet of Things · IoT · IoT security challenges · IoT security issues · IoT security solutions

29.1 Introduction In Internet of Things, sophisticated sensors and chips are embedded into any probable physical entity surrounding us. Each sensor transmits the valuable data. This data lets us understand how things works and that data is then analyzed so that it can be of relevant use to us. The question here is the interconnection of these devices and our benefit from the data they generate. The sensors use a common language to talk with each other and they securely communicate with Internet of Things platform. S. Kolhe (B) Department of Computer Engineering, Chandubhai S. Patel Institute of Technology, Charusat University, Changa District, Kheda, Gujarat 388421, India e-mail: [email protected] S. Nagpal Department of Electronics and Communication Engineering, V. T. Patel Institute of Technology, Charusat University, Changa District, Kheda, Gujarat 388421, India e-mail: [email protected] J. Desai Department of Computer Engineering, Devang Patel Institute of Advance Technology and Research, Charusat University, Changa District, Kheda, Gujarat 388421, India e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_29

279

280

S. Kolhe et al.

The data from the devices is collected at the common Internet of Things platform and then here the data is integrated and applied with analytics to infer valuable and interesting data to the applications that address industry’s or users’ specific needs. Imagine your life after 15–20 years with your home will be connected, your body is connected and your vehicles, etc. everything is connected. And with this connectivity feature all these will be able to communicate with each other and with the other devices around them to better predict and fulfill our needs. Our bodies will be connected with embedded medical devices and other devices that will keep track of us. In future we will have self-driving cars that are able to navigate safely than they can today. So, threat against bad illness, bad drivers or even a burglar targeting our homes will be less than a software flaw that may be in one of these devices. If these devices are isolated or not connected, then there is no problem. But this is not the condition in reality. All our devices are and will be interconnected, they are able to talk to each other and they are connected up to the global internet and when that takes place, they open themselves to attacks and are vulnerable. The IoT devices we are using are developed faster and released faster. But they lack the ability of security. Manufacturers and higher parties or bosses force the developers top develop and deploy the IoT product faster. For keeping up in competitions against other companies they develop and deploy the products faster. They don’t care about security. Writing 100 million lines of code is useless if the personal data of user is not secure. Consider an example of a politician. If a politician has heart disease and that is made in control by embedding a smart pacemaker. Now as it is a smart pacemaker, politician gets an android app through which he can see the status of the pacemaker and his heart. He can manipulate the pacemaker and also doctors have the data of his heart and pacemaker status. Now if the politician has an enemy from the other party, and if they get to know that a smart pacemaker is embedded in politicians’ heart, and if there is no security provided in software of pacemaker then as the pacemaker is connected to internet, it can be easily hacked and the hacker can manipulate the pacemaker without even the politician knowing anything. So the life of the user can be at risk (here politician). That is why security in IoT appliances is a serious issue. Consider a car is fully connected with the internet. And if this car is vulnerable to attacks, then a hacker can hack the system of the car and misuse the technology for causing accidents in the traffic and harm to the people. Let’s take a car for an example. A car is embedded with many sensors that continuously monitor each compartment and specifically each important part of the car. Now if there is any problem in the car or in its engine, the sensors will send the data to the computing core of the car which will integrate and analyze all the data received from the sensors. The core then converts the data into knowledge and sends it to the manufacturer or a nearby garage. So that when user reaches the garage, the mechanic is already familiar with the problem and the car can be quickly repaired. Now if any part is damaged, then the information of the damaged part has already been sent to its manufacturer. And till user reaches the garage, the new part is already present there sent by manufacturer to the garage. If any major issue comes, then the platform will alert the user about the issue and will provide location of the nearest

29 Security in Internet of Things

281

garage. The manufacturing platform not only provides alerts but also uses the data such that they can know more about the problem and so that they can update the existing models of the car as well as they can take precautions during manufacturing new cars. Keep in mind, this is not just one car. It is assumed that till 2020, each and every device or thing will be connected with internet. Thus, from this single example we can say that every sensor and device is communicating with each other and the core knows everything about you. A machine knows everything about you starting from you wake up to till you go to bed at night. The sensors are continuously tracking you and we don’t want this sensitive data to be hacked. For this there is a need for a secure communication between the device and the Internet of Things platform. The platform maintains thousands of information from each sensor and maintains historical records in secured database.

29.2 IoT Layered Architecture Figure 29.1 represents the layered architecture of IoT. All the layers are briefly discussed below. This architecture is a combination of four layers [1, 2].

29.2.1 Sensor Connectivity Layer The Sensor Connectivity Layer consists of large number of smart devices that are integrated with sensors, which are able to capture information from the physical world and translate it to the digital world. Thus, we can say that sensor devices are acting as the intermediary between physical world and the rest of the digital world. This information is then transmitted to the next layers so that the data can be analyzed and processed. We have sensors that are able to capture number of measurements like monitoring activity, temperature, air quality, health data and many more. High capability sensors are embedded into these kinds of smart devices. These sensors are categorized as Industrial, automation, body sensors, etc. Figure 29.2 depicts the Sensor Connectivity Layer.

29.2.2 Gateway Network Layer Figure 29.3 depicts the IoT Gateway implementation. Tons of millions of data is generalized by sensors. So, we need a strong transmission system to transmit the data. That’s why we have gateways. So, gateways are used to mitigate the information of the data from one end to the other end. Some of the gateways also send information from internet to sensor nodes. Gateway performs two-way transmission. This layer integrates different types of networks into one IoT platform.

282

Fig. 29.1 IoT layered architecture

S. Kolhe et al.

29 Security in Internet of Things Fig. 29.2 Sensor connectivity layer

Fig. 29.3 Simple gateway

283

284

S. Kolhe et al.

There are two types of gateway implementation for IoT [3, 4]: (a) Simple Gateway: This is responsible for just transmission of data or forwarding of data generated by sensors and actuators. (b) Embedded Control Gateway: This gateway extends the functionalities of the simple gateways but adds responsive processing and intelligent handling of different nodes. The gateway is able to filter various kinds of data. This reduces cost and complexity of the nodes. It provides interoperability.

29.2.3 Management Layer Management layer takes care of (a) Analysis of Information (b) Security Control (c) Management of Devices (d) Modelling of Process. Figure 29.4 shows the components of Management Service Layer. Data Management is actually divided into two categories:

Fig. 29.4 Management service layer

29 Security in Internet of Things

285

Fig. 29.5 IoT gateway implementation

(a) Periodic: Here the sensor data from IoT devices require filtering. That means we can apply analytics, perform actions, and control actuators using this filtered data. Everything is done on the filtered data. (b) Aperiodic: Here the data is directly operated, used or sent to the user without filtering. Sensor data from some IoT devices need instantaneous delivery and reaction. For example, Data from the sensor connected to the patient. Suppose a heart of a patient is sending abnormal readings then that data should be sent first considering it as a priority (Fig. 29.5). Data Management governs the flow of information of data, access of information, amalgamation and control. Figure 29.4 portraits the Management Service Layer. Data Abstraction (Information Extraction Processing): This is used as a common business model. Too much information is available there. So, there is a need of providing an abstract view of the data that you own.

29.2.4 Application Layer The Application Layer is the last layer in the IoT layered architecture. The processed data from previous layer is transferred to this layer. The application layer provides services to the clients or the end users. IoT can be implemented practically in all sections. Applications are divided into two main sections.

286

S. Kolhe et al.

Table 29.1 Service domain of application layer Service domain Smart home

Internet access, Entertainment

Smart city

City management, resource management, transportation management, disaster management

Smart agriculture

Soil sensing, area sensing, trespassing, water requirement alert, fire sensing

Smart transportation

Intelligent systems, traffic light control, navigation support (GPS), smart car technology, traffic status

Smart energy and fuel

Fuel level indication, pipeline monitoring

29.2.4.1

Applications Based on Sector

IoT can contribute in various sectors like Transportation, Healthcare, Retail, Energy and Military. IoT devices can help monitor the effect of environmental changes on different areas of cities. These devices can be used to monitor various natural resources of the earth like water bodies, soil, sunlight, trees, etc. Various sensors are used to accomplish this task. Vehicles can be embedded with IoT devices so that every vehicle can communicate with other vehicles on road. Owner can have complete details about his/her vehicle. IoT enabled healthcare devices can be used to keep people healthy and safe. This can also improve the diagnosis by the physicians. Smart store can be an example of IoT in Retail sector. Smart store includes applications like connected customer, product tracking and prediction, keeping track on equipment for predicting its failure, energy consumption and other issues.

29.2.4.2

Applications Based on Horizontal Market

In this section IoT contributes in Fleet Management, Asset Management, Supply Chain, People Tracking and Surveillance (Table 29.1).

29.3 Security Issues Here we are not talking about web services, but these devices are present in your office, your car, your home, and simply everywhere and they can record your data and also control things. So, security here cannot be compromised. Because if any of the device or sensor node is hacked then that can be a jumping point to attack other things in that area. Let’s take an example of webcam that monitors your home. The audio and video data is transmitted to the cloud as well as to your mobile phones, to computer devices.

29 Security in Internet of Things

287

Thus, the data could be stolen from any of these places. Not just data on my webcam, but if the hackers somehow manage to hack the credentials then they can also control the webcam remotely and main thing is that we will not be knowing that they are doing it. So, we have to take care of every minute point in the system so as to build secure products. The owner needs to be able to grant and revoke specific permissions to the people, apps, and other devices. And there needs to be a way to authenticate the owner and those other people, apps, and devices. Any user data needs to be stored on the devices and on the servers and importantly the information needs to be securely transmitted among these things. The devices need to be resistant to attacks, should be updatable and recoverable if there is any problem [5]. Let’s talk about home automation system. When you are registered to a cloud server, you get an authorization token. And whoever is connected to your LAN gets the token. So, the cloud server presents that token to your house and you can now control your home appliances and devices. But if any outsider or hacker hacks into your home network or if he somehow gets your token, then he can access and control your home too. But if the token is given to any trusted person, it means that you trust this device too. Also, it grants trust that device is not running any malware or hasn’t been compromised. Now if this device is running any malicious application, that device can request this token, get this token and can control your home remotely. So, it’s better that you don’t trust any other device in local network and use any kind of an end to end authentication and encryption model with all your communication [5]. Think of how malware gets onto a laptop or a phone. We know and are advised by everyone that not to click on any link that you get from someone or any untrusted link. But if a right kind of exploit is used then a malicious website can install malware right on your machine just by your loading a webpage. Example is java Runtime Exploit that can load up on the page and install some code on and you are unaware of what’s happening. So, to stop this we can do end to end authentication encryption on devices. So, you can mitigate the impact of this kind of attack also. We can install Ad-Blockers because these software programs block the malicious ads from opening in the browser. A secure browser helps guard against threats like malwares and phishing. These can steal the passwords and can damage your system. Browsers like Google Chrome built a feature known as “Sandboxing” that will enhance the browser’s security. Every website we open in the browser is an independent process. Each tab opened in the browser corresponds to a separate process. If there is any problem in one of the tab or if it crashes, then the other tabs remain unaffected. Thus, sandboxing provides a layer of protection around each of these processes. If accidently we open a harmful and malicious web page or a hacked website, the mechanism of sandbox works in such a way that it prevents the malicious code from causing damage to the computer. The malicious code runs within the sandbox, thus preventing damage to other tabs that are opened in the browser. Thus, whenever the infected tab is closed, the malicious code is gone. Thus, sandboxing also helps prevent threat to our network. Many antivirus programs like Quick Heal also provide the facility of sandboxing. Advantage should be taken from these techniques so as

288

S. Kolhe et al.

Fig. 29.6 Structure of Cloud Service

to protect our network. Figure 29.6 shows the internal structure of the cloud service and its working. The laptops and PC’s we use today have a lot of security features like firewall and Antivirus but Internet of Things don’t have these. Popular websites like Twitter and others went down for about a day because they were suffering from an attack called Distributed Denial of Service (DDOS) attack [6]. These websites provide services to the users for registering and then using the services. Now if the website is bombarded with too many requests or too many fake users, then the website gets a lot of stress and the website server goes down. This is DOS attack. Interesting thing is that, this attack was not launched from a PC but it was launched using IoT devices like a security camera and some network attached storage. Let’s talk about one of such malwares, ‘Mirai’. Mirai is a malicious program that attacks the smart devices that are connected with your network. This malware has been used in some of the largest cyber attacks ever recorded. In 2016, hackers hacked thousands of households connected devices like printers, baby monitors, cameras and smart refrigerators. They flooded the servers by taking control over the smart devices. This is an example of DDOS attack. Once main system or computer gets infected by this malware, it infects the connected smart devices and the user is unaware of this. Many sites like Twitter, Netflix, Spotify, The New York Times were affected by a cascading string of DoS (DDoS) attack [6]. Because of the 10/21 attacks a large amount of useless traffic was generated at servers that were targeted. These servers mainly belonged to a company called ‘Dyn’. This company provides DNS services to the other companies. Because of large number of devices that were connected to the internet and were unsecure made this attack happen. Because of default passwords, the DDOS-enabling infections were made possible. Cyber attacks and its evolution are shown in Table 29.2.

29 Security in Internet of Things

289

Table 29.2 Cyber attacks and its evolution Years

Attacks

1997–2004

Malicious Code, Trojan, Advanced Worms

2004–2007

Identity Theft, Phishing

2007–2010

DNS attacks, Rise of BOTNETS, SQL attacks, Anti Spam sites, Competitive Sabotage escalation

2010–2013

Social Engineering, DoS attack, BotNets, Malicious Emails, Ransomware attack, PoS compromised

2013-Present

Banking Malware, Keylogger, BitCoin Wallet Stealer, Identity Theft, Phone Hijacking, Ransomware, PoS attack, Cyber Warfare, Android Hack, loT nodes and connected devices hacking, MIRAI attack, etc. And many more new attacks are arising

The smart devices are connected to microcontrollers which are embedded with Bluetooth or Wi-Fi capabilities. These microcontrollers run on either Linux or some simpler microcontroller operating systems like RTOS and they have got image processing capabilities and they have got lots of computing power and hackers are able to take control over them and they can launch DDOS attack over them. So security in IoT devices is a major issue nowadays. Developers should take care about security In these devices.

29.4 Solutions IoT security, previously ignored, has now become an issue of high concern [1, 2]. Authentication: Using cryptographic certificates devices can prove that they are the correct devices and they can prove that they are who they say they are. We should not assign a default password to a device that can be same across devices all over the world. A strong and complex password including lower-case and upper-case alphabets, symbols and numbers must be assigned to the devices or products. Means only an owner can log in. Data Confidentiality: Cryptographic Encryption techniques must be used to protect privacy of data in devices. Make sure that no personal data is readily accessible. Availability: It means ensuring that the data are available when needed and that IT systems are operating reliably. Encryption: Best Encryption techniques like SSL/TLS and many other techniques should be used to protect any web interface against any hackers. Web Interface: Websites can be hacked using standard hacking techniques like SQL Injection and cross-site scripting. Thus, web interfaces should be protected from these attacks. Firmware Updates: Bugs are just irritating. Security bugs cause danger. Thus, Over-The-Air (OTA) updates must be supported by all IoT devices. Before applying

290

S. Kolhe et al.

these updates, they should be verified. So, if any security issue is found, the devices can be upgraded and security vulnerabilities can be closed. Data Integrity: The meaning of data integrity means that the data should not be changed in-flight. Thus, cryptographic hashing should be used to ensure this. It means ensuring that the data within IT systems are accurate. A study by Hewlett Packard revealed that “70% of Internet of Things devices are vulnerable to attacks.”

29.5 Conclusion IoT has power to change this world but if developers don’t focus on IoT security, then trust will not be developed in users that there IoT devices are secure. Thus, steps should be taken such that IoT devices should be made immune to cyber attacks. Data privacy of users should be given priority and that data must be secured.

References 1. Farooq, M.U., Waseem, M., Khairi, A., Mazhar, S.: A Critical Analysis on the Security Concerns of Internet of Things (IoT) (2015) 2. Singh, D., Tripathi, G., Jara, A.J.: A Survey of Internet-of Things: Future Vision, Architecture, Challenges and Services. In: Internet of Things (WF-IoT) (2014) 3. Chen, H., Jia, X., Li, H.: A brief introduction to IoT gateway. Communication Technology and Application (ICCTA 2011) IET International Conference, pp. 610–613, 14–16 Oct (2011) 4. Emara, K.A., Abdeen, M., Hashem, M.: A gateway-based framework for transparent interconnection between WSN and IP network. EUROCON ’09, pp. 1775–1780 5. Bhabad, M.A., Bagade, S.T.: Internet of Things: Architecture, Security Issues and Countermeasures (2015) 6. Zeinab, M., Ahmed, (E.S.A).: Internet of Things Applications, Challenges and Related Future Technologies (2017)

Chapter 30

A Hybrid TLBO Algorithm by Quadratic Approximation for Function Optimization and Its Application Sukanta Nama, Apu Kumar Saha and Sushmita Sharma

Abstract Recently hybrid optimization algorithms enjoy growing attention in the optimization community. However, over the last two decades, many new hybrid meta-heuristics optimization techniques are developed and are still developing. On the hybrid optimization algorithm, the most common criticism is that they are not well balanced in respect of the local search and global search of the algorithm. Viewing this, in the present work a modified adaptive based teaching factor is suggested for the basic TLBO algorithm. Also, a novel hybrid approach is proposed that combines the Teaching Learning Base Optimization (TLBO) Algorithm and Quadratic approximation (QA). The QA is applied to improve the global as well as local search capability of the method that also represents the characters of “Teacher Refresh”. For the performance investigation, the suggested algorithm is involved to solve twenty classical optimization functions and one real life optimization problem and the performances are differentiated with different state-of-the-arts methods in terms of numerical results of the solution. Keywords Hybrid optimization method · Teaching learning based optimization (TLBO) · Quadratic approximation (QA) · Unconstrained optimization problem

S. Nama (B) Department of Mathematics, Ramthakur College, A.D. Nagar, Agartala, West Tripura 799003, Tripura, India e-mail: [email protected] A. K. Saha · S. Sharma Department of Mathematics, National Institute of Technology Agartala, Barjala, Jirania, Agartala 799046, Tripura, India e-mail: [email protected] S. Sharma e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_30

291

292

S. Nama et al.

30.1 Introduction The field of optimization has been taken a revolutionary change during past five decades for searching answers of some of the difficulties based on the characteristic of the problem, like multi-modality, dimensionality, and differentiability. For removing these difficulties, metaheuristic optimization techniques are introduced and the development of such research is going on tremendously and researchers are already obtained and still trying to get new techniques so that these techniques can deal with every type of optimization problem. Some of the well-known population-based metaheuristic methods are: PSO (particle swarm optimization) [1, 2], GA (genetic algorithm) [3], ACO (ant colony optimization) [4], DE (differential evolution) [5], GSA (gravitational search algorithm) [6], ABC (artificial bee colony) [7, 8], TLBO [9] etc. Also, many authors have attempted to improve an existing method by modifying the different aspects of the method. In this direction, some of the improvement on PSO can be found in literature. These are known as PSO-variants. Some of the PSO-variants are PSO-cf [10], PSO-cf-local [11], FI-PSO [12], UPSO [13], FDR-PSO [14], CPSO-H [15] and CLPSO [16]. In a similar way, there are many research article have been published on improving or modifying DE and known as DE-variants. Some of the DE-variants available in the literature are DE-rand-2 [17], DE-current-to-rand-1 [18] DE-best-2 [19], and EPSDE [20]. A good number of applications of these algorithms have already been done in different branches of Science, Humanities, Medical and Engineering field. During past two decades, many authors suggested that the combination of one meta-heuristics algorithm (or the component of one algorithm) with other metaheuristics algorithm can perform better compared to a single optimization method. These kind of combined algorithms are called a hybrid metaheuristic algorithm. Pant and Thangaraj suggested a hybrid method called DE-PSO [21] to solve the global optimization problems. DE-PSO is designed with the association of the component of both DE and PSO algorithm. By integrating the quadratic approximation (QA) tools in the GAs, Deep and Das [22] proposed four hybrids GAs, called HGA1, HGA2, HGA3 and HGA4. In [23], an integrated algorithm has been proposed by integrating two heuristic optimization techniques namely PSO and GA. In that procedure, a set of particle is randomly generated inside the optimum space. During the optimization process, every particle is governed by the combined rule of PSO and GA. Also, to regulate the particle velocity, a new modified namely ‘constriction factor’ has been introduced. DESQI [24] is a combined method of DE algorithm and simple quadratic interpolation (SQI) [17, 18]. The SQI is merged into the DE algorithm to increase the rate of search capability of the basic DE and it is seen that addition of SQI with DE enhances the efficiency and robustness of DE. Also, the employment of SQI in DE seems to help improving the exploitability of the algorithm and also helps in acquiring the accurate optimum functional value. Another hybrid metaheuristic algorithm PSOGSA [25] is proposed by the combination of PSO and GSA. The primary knowledge of this method is to incorporate the exploitative ability of basic PSO and the explorative proficiency of GSA. This can improve the robustness of

30 A Hybrid TLBO Algorithm by Quadratic Approximation …

293

the algorithm. In PSO algorithm, the most popular variant is PSO-W, i.e. PSO with inertia weight. By incorporating the QA into the PSO-W, a new algorithm called qPSO-W [26] is proposed. The QPSO [27] is another ensemble algorithm, which is introduced using PSO and QA. In QPSO, QA is incorporated into basic PSO to growth the capability of the algorithm. E-BSADE [28] is an ensemble algorithm by the combination of BSA and DE algorithms. Also, in E-BSADE the control parameter of DE and BSA are considered based on self-adaption scheme. Recently, incorporating the QA with the SOS algorithm, a new HSOS [29] has been proposed. In the present study, based on the combination of QA and the TLBO algorithm, the HTLBO algorithm has been proposed. In HTLBO, first, improvement of TLBO is suggested by modification of teaching factor. Then QA is combined with this improved TLBO and performance of this hybrid TLBO algorithm is investigated. At the beginning of the optimization process, improved TLBO is executed to increase the exploration proficiency and also to growth the convergence speed and then the Quadratic Approximation (QA) operator is used to enhance the exploitative behaviour of original TLBO. The key contributions of the present study are shortened below: (i)

Based on the adaptation, a modification of teaching factor has been done in this paper due to heuristically change of teaching factor in the original TLBO. (ii) An improved TLBO is proposed initially to progress the explorative ability as well as the search speed of the method. (iii) A novel hybrid TLBO algorithm (called HTLBO) that combines the improved TLBO algorithm and QA is proposed which can find the higher accuracy of the global optimum value with faster convergent. (iv) Finally, the proposed HTLBO is applied on one real-life optimization problem, namely, spread spectrum radar polyphase code design problem.

30.2 Related Work A detailed discussion of the reviews paper related to this work has been presented in this section. These review papers are on the basis of hybridization of TLBO and others algorithms. These algorithms are compared with others algorithms which are swarm intelligence based algorithms, evolutionary algorithms, different improve TLBO algorithms and so on. At first, some modified version of TLBO algorithm has been discussed. The tabulated form of these algorithm are presented below:

Author

Satapathy and Naik [30]

Satapathy et al. [31]

Rao and Patel [32]

S. No.

1

2

3

elitist TLBO for solving single objective functions

A orthogonal design factor is incorporate into basic TLBO algorithm to proposed OTLBO

mTLBO is based on the modification of learner phase for global search

Contribution

(continued)

In the elitist TLBO algorithm, the wickedest student is changed by elite student after the execution of learner phase. Before execution of the iteration, identify the similar student; if the similar student (solution) exists then modify this student (solution) to reach faster in global optimum. This improve method is experienced on seventy six single objective test functions. Variation of the parameter such as class size i.e. number of student and number of iteration (fitness evaluation) are executed on the performance results of test functions. The execution results with others algorithms present the acceptable of this algorithm

Orthogonal design factor is robust factor. It is used to generate new offspring by an optimum statistical process and to reduce the function evaluation, increase the faster convergence rate. Using this factor a new improve version of algorithm called OTLBO is proposed. It is used to evaluate the optimum value of some benchmark functions. An obtained result indicates the good performance of OTLBO

A modified TLBO called mTLBO is suggested for improving the faster global search ability i.e. the convergence rate and computational time during the execution of optimization. It is used to obtain the intra cluster, quantization errors and inter cluster distances on clustering problems which are the existing data

Description

294 S. Nama et al.

Author

Satapathy et al. [33]

Nayak et al. [34]

Xia et al. [35]

S. No.

4

5

6

(continued)

A modified TLBO called simplified TLBO (STLBO) is proposed

A MO-TLBO is applied to minimize the multi-objective optimal Power Flow (OPF)

A weight factor is incorporated to the basic TLBO and proposed Weighted TLBO

Contribution

(continued)

STLBO is the composition of three phases one is Feasible Solution Generator phase, Learning Phase, Teaching Phase. The Feasible Solution Generator phase is incorporated to obtain the feasible disassembly sequence during the execution of algorithm, whereas the other two phases are used to improve and execute the solution space towards optimum. This method is applied to solve complex non-linear combinatorial optimization problem which is Disassembly Sequence Planning (DSP)

OPF is a popular multi-objective static, nonlinear, non-convex complex hard global optimization problem. In this problem, the three functions are there and these are to be minimized which are fuel cost, Transmission losses and L-index. In this problem, the constraints are load flow equations and the system operating limits. MOTLBO with pareto-optimal front is applied to solve this problem

By defining a new weight factor to the basic TLBO algorithm, another improved TLBO which is called WTLBO is proposed. In the WTLBO, the weight factor is conjugated to improve faster overall search ability of TLBO. This algorithm is applied on twenty well known standard test function and founded solutions are compared on some popular algorithms which shows the acceptable performance of WTLBO

Description

30 A Hybrid TLBO Algorithm by Quadratic Approximation … 295

Author

Roy et al. [36]

Roy et al. [37]

S. No.

7

8

(continued)

Quasi-oppositional TLBO is suggested to obtain the optimal solution of short-term hydro-thermal scheduling optimization problem

Heat and power dispatch optimization problem solved by Oppositional TLBO

Contribution

Short-term hydrothermal scheduling optimization problem is also a non-linear hard complex optimization problem. Quasi-oppositional strategy is incorporate into the original TLBO to develop a new algorithm called the quasi-oppositional TLBO (QOTLBO). This algorithm is examined to calculate the optimum value of the problem optimization problem. The obtained optimum results are found to be satisfactory presentation associated to the other existing methods

The combination of heat and power dispatch optimization problem is a non-linear hard complex problem. An opposition based learning (OBL) strategy is combined to the basic TLBO algorithm to form a new algorithm, called OTLBO. The OBL is added to original to improve the acceleration of convergence rate and optimum result. The result obtained by OTLBO, are compared with other existing algorithms and found to be satisfactory performance

Description

296 S. Nama et al.

30 A Hybrid TLBO Algorithm by Quadratic Approximation …

297

As already mentioned that due to the lack of equilibrium of explorative and exploitative behaviour of a single algorithm, sometimes can’t efficiently search the entire domain of the problem and also lack of desired convergence speed. As a solution of it, a lot of researchers are inclined to the hybrid methods where the strength of each component algorithms can be put together to get the preferred search and convergence speed. It has been seen that many a times, the hybrid methods give better performance than a single method. For this benefit of hybrid methods, a lot of works on hybrid metaheuristics can be found in the literature and thus various works are also available where TLBO is used as a component method. In the following table, some of the hybrid methods where TLBO is an component algorithm are given. For more details, one can consult [38, 39].

S. No.

Author

Component algorithm

Description

1.

Xie et al. [40]

TLBO and simulated annealing (SA)

This paper introduced a novel algorithm called hybrid TLBO is presented which combines the VNS (variable neighbourhood search) and SA. VNS is used to increase the solution faster and SA is to the algorithm as a local search process. It determines the job sequence with minimization of maximum lateness criterion and minimization of make span criterion for PFSP. A largest order value (LOV) rule is utilized to convert the individual to the job permutation

2.

Azad-Farsani et al. [41]

TLBO and Chaotic PSO(CPSO)

To find the global optima in more efficient way, proposed algorithm combines the CPSO and TLBO. PSO has some parameters in order to achieve a proper performance i.e. inertia weight factor, which is an adjustable parameter. But TLBO is efficient for finding the optimum without tuning any parameters because TLBO is free from adjusting parameters. For tuning of inertia weight factor, algorithm used chaotic framework and to increase the quality of results, CPSO is mixed to TLBO

3.

Dokeroglu [42]

TLBO and Tabu Search

To obtain the optimum solution of complex combinatorial optimization problem and Quadratic Assignment, A TLBO-based hybrid algorithm is proposed and recombination operators is used to trained individuals and after that the Robust Tabu Search is executed to them (continued)

298

S. Nama et al.

(continued) S. No.

Author

Component algorithm

Description

4.

Gnanambal et al. [43]

TLBO and GA

With the help of cross over property of GA,a new hybrid algorithm called Hybrid-TLBO is suggested. To verify the performance, it is applied to find total cost of active optimal power generation. Performance results indicate the useful and high quality result of this method

5.

Khare and Kumar [44]

Many optimizing liaisons (MOL) and TLBO

In this project, a novel algorithm MOL-TLBO is introduced with the association of many optimizing liaisons (MOL) and TLBO. Also, for techno-economic-socio analysis, in this work a new integrated hybrid renewable energy sources (IHRES) is introduced. Finally the new MOL-TLBO is applied to solve the problem

6.

Sahu et al. [45]

Local Unimodal Sampling (LUS) and TLBO

In this project, a novel algorithm called hybrid LUS–TLBO is introduced based on fuzzy-PID controller to the original TLBO. A Local Unimodal Sampling (LUS) is also incorporate to the algorithm. After that the LUS–TLBO is applied to solve optimization problem which is load frequency control (LFC) problem of a two-area interconnected multi-source power system with and without HVDC link

7.

Babazadeh and TavakkoliMoghaddam [46]

GA and TLBO

A novel hybrid algorithm called GA-TLBO is introduced with the association of GA and TLBO. A new optimization problem on the field of operation research based on the capacitated three-stage supply chain network design (SCND) problem is introduced. The GA-TLBO is applied to solve the problem. During the execution of the algorithm, an encoding scheme is applied based on random key and priority for removing the infeasible solutions

8.

Deb et al. [47]

Chicken Swarm Algorithm (CSO) and TLBO

In this project, a hybridization of CSO and TLBO is proposed to improve the search capability and accurateness. Due to involvement of no tuning parameters, the variability of solutions is less in TLBO and throughout focus of the proposed algorithm to remove the variability of CSO without compromising with the accuracy. The algorithm performs TLBO in all the generations and CSO is performed only in some particular generation (continued)

30 A Hybrid TLBO Algorithm by Quadratic Approximation …

299

(continued) S. No.

Author

Component algorithm

Description

9.

Patsariya et al. [48]

TLBO and maximum power point tracking (MPPT)

In this project, a novel converter topology algorithm is proposed to the TLBO to produce TLBO-MPPT for maximizing the output voltage which is based on Heuristic method of simulation to design a DC converter and can be easily integrated with the grid. A new TLBO method has been implemented for fast convergence of the optimal parameters to design a suitable MPPT technique and to extract the maximum output from Solar Photovoltaic system

10.

Shahbeig et al. [49]

TLBO and mutated fuzzy adaptive PSO algorithm

To identify the most relevant genes involved in breast cancer development, a novel hybrid algorithm (hybrid TLBO–PSO)is proposed which is the combination of the mutated fuzzy adaptive PSO and TLBO algorithm. This algorithm is used to calculate the smallest subset of genes involved in breast cancer with the highest amount of classification accuracy, sensitivity and specificity

11.

Tuo et al. [50]

HS (Harmony Search) and TLBO

To solve the complex optimization problems, with the association of TLBO and HS, a hybrid algorithm is proposed, namely, HSTLBO is introduced. With the purpose of balancing the global investigation and manipulation abilities i.e. exploration of unknown regions and the exploitation of high accuracy optimum results in the recognized areas is done by HS and TLBO to the HSTLBO. Also, a self-adaptive selection strategy is used to HSTLBO

12.

Ding et al. [51]

Opposing-based initialization with TLBO

In this exertion, a novel hybrid TLBO (called HTLBO) is proposed which has mainly two strategies. Firstly, an opposing-based initializing strategy is used to enhance the structure of the initial distribution. In order to increase the local search capability of HTLBO, with the suggested algorithm combines the local search with the teaching phase of TLBO and also a special teaching process is used to some specific students to improve the convergence speed (continued)

300

S. Nama et al.

(continued) S. No.

Author

Component algorithm

Description

13.

Singh et al. [52]

TLBO and PSO

The proposed algorithm is developed by combining TLBO and PSO which is known as hybrid teaching–learning particle swarm optimization (HTLPSO). The foremost concept of the HTLPSO algorithm is that it merges the best half of population obtained after the teacher phase in TLBO with the best half of the population obtained after PSO and the obtained population is used in learner phase of TLBO

14.

Chenet al. [53]

TLBO with learning enthusiasm mechanism

In this paper, a new algorithm is suggested to enhance the efficiency of basic TLBO by adding new feature learning enthusiasm mechanism and the proposed algorithm is known as a learning enthusiasm based TLBO (LebTLBO). High learning enthusiasm is considered a student from a set of student with best objective function value. Low learning enthusiasm is considered a student with worst objective function value. Furthermore, to increase the superiority of the poor knowledgeable student, a poor student training phase is applied

15.

Nenavath and Jatoth [54]

TLBO and SCA (sine cosine algorithm)

Here, a novel hybrid SCA–TLBO algorithm is suggested which is efficient to solve optimization problems and visual tracking. The hybrid SCA–TLBO algorithm is the association of sine-cosine algorithm and TLBO. The hybrid SCA–TLBO-based tracing structure is employed to through an experiment live object pursuit error, root mean sq. error, tracking detection rate, absolute error, and time cost as parameters

16.

Zhanget al. [55]

ABC and TLBO

A framework of a novel hybrid method, so-called ABC-TLBO is suggested which uses the framework of basic ABC and an individualized based on TLBO is employed in onlooker bee phase to improve the exploitation capability of the proposed algorithm (continued)

30 A Hybrid TLBO Algorithm by Quadratic Approximation …

301

(continued) S. No.

Author

Component algorithm

Description

17.

Sevinç and Dökero˘glu [56]

TLBO algorithm and extreme learning machines (ELM)

A framework of a novel hybrid TLBO algorithm is suggested with the association of extreme learning machines (ELM) for the solution of data classification problems. It has continued to be riveting to revisit this classical problem and investigate the efficiency of new techniques. The performance of TLBO-ELM is observed to be competitive for both binary and multiclass data classification problems compared with state-of-the-art other methods from the literature

18.

Guo et al. [57]

TLBO and DE

To estimate the PEM cell model parameters, a novel and efficient algorithm is proposed in this paper which is the hybridization of TLBO and DE. The abbreviation of algorithm is TLBO-DE method, which can be validated by several validation functions

19.

Zhang et al. [58]

TLBO and BMO (bird mating optimizer)

A new hybrid method is suggested to defeat the drawback i.e. speed of the convergence and quality of solution and it is inspired by intelligent mating behaviour of birds. The presented algorithm combines the benefits of TLBO) and bird mating optimizer (BMO)

20.

Tang et al. [59]

TLBO and VNS with seven neighbourhood operators

A framework of a novel hybrid TLBO so called HTLBO approach is developed which is the associations TLBO and VNS with local search approach neighborhood operators. It is applied to obtain the two-sided assembly line balancing stochastic optimization problem. To handle the multiple constraints of this problem, a new priority-based decoding methodology is suggested. Execution of comparison with other algorithms suggests the acceptability of the hybrid HTLBO (continued)

302

S. Nama et al.

(continued) S. No.

Author

Component algorithm

Description

21.

Shao et al. [60]

TLBO with discrete teaching and learning (probabilistic) phase, reconstruction of population and also neighbourhood search

To minimize the no wait flow shop scheduling (NWFSSP) problem, a hybrid discrete method is suggested. This method is the composition of TLBO with the association of probabilistic learning mechanism (HDTPL). It is executed by four processes which are population reconstruction, discrete teaching phase, neighbourhood search and probabilistic learning phase. The calculation begins with discrete teaching stage for example Forward-embed and Backward-embed are embraced to copy the instructing procedure. In the second stage which is the discrete probabilistic learning stage and it set up a successful probabilistic model with thought of both employments arranges in the grouping and comparable activity squares of chosen predominant students. After that by utilizing the crossover operator to gain information, every student interfaces through the probabilistic process. In light of accelerate the technique, there are three kinds of neighborhood search structures for example Insert-search, Referenced-embed search, and Swap-search, which is intended to increase the nature of the worldwide best student and momentum student and for the guess, Taguchi technique is utilized to research the principle parameters of HDPTL

22.

Shao et al. [61]

TLBO with discrete teaching phase based on probabilistic model, discrete learning phase based on hierarchical structure together with reinforcement learning

A novel framework of hybrid method called HDTLM is suggested with the association of discrete teaching, learning and reinforcement learning technique in the original TLBO algorithm. It is applied to find the optimum output of no-idle flow shop scheduling problem (NIFSP). To generate a In the discrete teaching phase, To generate a series of position sequences, a probabilistic approach based on the elite learners and the best learner is used and to replace the mean individual in the teaching-learning based optimization, consensus permutation’s concept is applied (continued)

30 A Hybrid TLBO Algorithm by Quadratic Approximation …

303

(continued) S. No.

Author

Component algorithm

Description

23.

Das and Padhy [62]

Support vector machine (SVM) and TLBO

In this work, a new hybrid algorithm called SVM–TLBO is recommended which is the association of TLBO and a support vector machine (SVM). The SVM-TLBO algorithm is efficient to avoid parameters of algorithms, which are essential during the evaluation of process of optimization methods. It is used to predict the daily closing costs of the COMDEX commodity futures index, listed within the Multi commodities market of India Limited

24.

GonzálezÁlvarez et al. [63]

TLBO with different leaning strategy

In this paper, a novel algorithm called H-MOTLBO is recommend to predict common patterns in sets of protein sequences which is developed by combining basic TLBO with a process of local search. To increase the information of individuals a group of solution executed in different process of learning. It is used to predict common patterns which can gives relevant biological evidence as regards to those protein functions in sets of protein sequences. The execution results gives the acceptable solution quality that others well-known biological tools

25.

Chen et al. [64]

TLBO with multi-classes cooperation and simulated annealing operator

In this framework, a novel SAMCCTLBO is suggested based on the composition of TLBO with multi-classes cooperation and simulated annealing operator. A microteaching approach is incorporated into SAMCCTLBO to take the full benefits from teaching process in which the population is divided into several sub-classes. The modification would possibly make the mean solutions increased rapidly for the result of microteaching is commonly higher than teaching in massive categories. With thinking about the dilemma of studying ability of learner, the inexperienced persons in specific sub-classes solely examine new understanding from others in their sub-classes in learner section of SAMCCTLBO, and all learners are rearranged in to a small group randomly after some process to enhance the diversity of the sub-classes (continued)

304

S. Nama et al.

(continued) S. No.

Author

Component algorithm

Description

26.

Zou et al. [65]

TLBO and Gaussian sampling learning

In this study, a new hybrid algorithm is which is known as bare-bones teaching-learning-based optimization (BBTLBO). In the initial stage of the BBTLBO, every student executed by teaching process which employs process of learning. It is the combination of process of learning and teaching process in the original TLBO. After that, apply a learning process based on Gaussian sampling with neighbourhood search is used to execution the optimization process. Finally the suggested method is applied to solve the non-linear global optimization problems

27.

Ghasemi et al. [66]

Gaussian bare-bones and TLBO

In this paper, An improved form of Gaussian bare-bones TLBO (MGBTLBO) algorithm is suggested, which is very efficient method for solving the problem with discrete and continuous control variables. The MGBTLBO is applied on optimal reactive power dispatch (ORPD) problem which has both discrete and continuous parameters with IEEE standard norms

28.

Wang et al. [67]

TLBO and DE

A TLBO–DE is established based on the composition of TLBO and DE to obtain the optimum value of chaotic time series prediction. In TLBO–DE, DE is integrated into update the preceding first-class positions of individuals to pressure TLBO soar out of stagnation, due to the fact of its robust looking ability and suggested method accelerates the combination rate and increases the calculation’s exhibition

29.

Ghasemi et al. [68]

modified teaching learning algorithm (MTLA) and double DE (DDE) algorithm

In this project, a novel algorithm is proposed which is the combination of MTLA i.e. modified teaching learning algorithm and double differential evolution (DDE) algorithm. The hybrid MTLA-DDE algorithm is applied to solve optimal reactive power dispatch (ORPD) optimization problem which is the non-linear, multi variable and non-convex (continued)

30 A Hybrid TLBO Algorithm by Quadratic Approximation …

305

(continued) S. No.

Author

Component algorithm

Description

30.

Zou et al. [69]

TLBO with differential learning

In this study, a novel algorithm is suggested based on differential learning approach to the basic TLBO called DLTLBO. In this algorithm, to generate a new mutation vector, based on neighbourhood search approach learning strategy incorporated into the teacher phase in the standard TLBO, while to produce another mutation vector, a differential learning is used and to increase the diversity of the solution space and to generate new solutions, a crossover operation is performed in the two newly obtained mutant vector. For achieving the stability of exploitation and exploration, global search and local search are combined

31.

Dib and Boumhidi [70]

DE and TLBO

To modify the basic framework of optimally, the unknown parameters of the (PID-PSS) controller and the exponential form of the tracking error modified, A novel and robust hybrid algorithm is proposed, namely DE–TLBO. The aim of the proposed algorithm is to ensure a good tracking accuracy and to enhance the level of the oscillations damping in the multi-machine power system with an optimal choice of the parameters of all proposed controllers

32.

Turgut and Coban [71]

DE and TLBO

In this framework, a novel algorithm is suggested which is the composition of TLBO and DE. It is applied to solve the proton exchange membrane fuel cell (PEMFC) model parameters optimization problem. The purpose of the algorithm is to simulate the accuracy level of the optimization problem which is proton exchange membrane fuel cell (PEMFC) problem

33.

Lim and Isa [72]

TLBO and PSO

A hybrid algorithm called TPLPSO is suggested with the combination of cutting edge TLBO and PSO. In the TPLPSO, first the particle increase the position based on previous knowledge and after that, particle arrives into the peer-learning phase, which fails to improve its fitness and also an exemplar is selected in this phase as the supervision particle and to avoid the premature convergence situation, a stagnation prevention strategy (SPS) is engaged (continued)

306

S. Nama et al.

(continued) S. No.

Author

Component algorithm

Description

34.

Lim and Isa [73]

Bidirectional teaching and peer-learning particle swarm optimization

A new algorithm called BTPLPSO is proposed. It is executed by two phase one is bidirectional teaching phase in which the particle update their position by teaching phase and another is peer-learning phase in which the particle increase the global best position using their previous knowledge. Also, an elitist learning strategy is used with the basis of efficient orthogonal investigational to increase the position of global best particle

35.

Cheng et al. [74]

PSO, TLBO and circular crowded sorting (CCS)

To solve multiobjective optimization problem, a HTL-MOPSO algorithm is proposed which is the composition of TLBO and PSO with circular crowded sorting (CCS). To increase the diversity of solution space, HTL-MOPSO algorithm combines the TLBO with canonical PSO search and also, to improve the diversity and spread of solutions, CCS technique is developed

36.

AzizipanahAbarghooee et al. [75]

Gradient TLBO and black hole (BH) algorithm

In this project, a new optimization approach is proposed which is the composition of gradient-based modified TLBO and BH algorithm so called MTLBO–B algorithm to obtain the optimum operational cost. To overcome the premature convergence of both MTLBO and BH algorithms, a wavelet mutation operator is incorporated to MTLBO–B which is proposed based on self-adaptation. After that, those problems which have been remaining unsolved i.e. one of them is to produce robust solution can be overcome by classical gradient-based technique

37.

Güçyetmez and Çam [76]

conventional genetic and TLBO

In this project, a new hybrid algorithm, called G-TLBO is introduced based on conventional genetic GA and TLBO for solving wind-thermal power systems. To examine the comparison, the fuel cost and algorithm run time are considered as performance criteria

38.

Chen et al. [77]

TLBO and ABC

To solving the solar PV parameter estimation problems, the TLABC algorithm is introduced which is the composition of TLBO and ABC. The process of TLABC is executed by three search phases i.e. learning-based on looker bee phase, teaching-based employed bee phase, and generalized oppositional scout bee phase to proficiently examine solution space (continued)

30 A Hybrid TLBO Algorithm by Quadratic Approximation …

307

(continued) S. No.

Author

Component algorithm

Description

39.

Tefek et al. [78]

GSA and TLBO

A framework of a novel hybrid gravitational search–teaching–learning-based optimization method is suggested to develop linear, quadratic and exponential models and with help of algorithm, energy demand estimation (EDE) was implemented. For the analysis of the problem, five indicators are used such as population, installed power, the socio-economic indicators (gross domestic product) and the electrical indicators (net electric consumption and gross electric generation)

40.

Huang et al. [79]

TLBO and CS (cuckoo search)

With the association of TLBO and CS, a new version of hybrid algorithm, so-called teaching-learning-based cuckoo search (TLCS) is suggested to solve structure designing optimization problem. To evolve a co-evolutionary tool for population space, it combines the teaching-learning process with levy flight and to generate new solutions from the abandoned solutions in cuckoo search, levy flight is operated on those solutions and for other best results, the process of teaching-learning is applied to increase the diversity of the population space during the execution

41.

Huang et al. [80]

TLBO and CS (cuckoo search)

To calculate the constrained design optimization problems, a new hybrid algorithm called TLCS is introduced. It is the composition of TLBO and CS algorithm

42.

Tuo et al. [81]

HS and TLBO

In this project, an improved version of global HS algorithm with the association of TLBO is introduced for high dimension complex global optimization problems, known as harmony search based on teaching-learning (HSTL). To maintain the proper balance between diversity of population and execution rate convergence and population diversity, there are four strategies i.e. teaching-learning strategy, harmony memory consideration, random mutation and local pitch adjusting, and change in the parameters is done by adopting dynamic strategy (continued)

308

S. Nama et al.

(continued) S. No.

Author

Component algorithm

Description

43.

Mahdad and Srairi [82]

TLBO and Pattern search (PS) algorithm

A novel flexible planning strategy is proposed to improve the security optimal power flow (SOPF) by minimizing the total fuel cost, total power loss and total voltage deviation considering critical load growth which is based on the TLBO and pattern search algorithm (PS). The primary feature of the proposed hybrid algorithm is that TLBO algorithm is adapted and coordinated dynamically with a local search algorithm (PS)

44.

Jiang and Zhou [39]

DE and TLBO

hDE-TLBO is a new method and it is produce by combination of TLBO and DE. In this method DE is used to refresh the teacher during the execution of the algorithm. This method is applied to optimize the total fuel cost and emission effects of short-term optimal hydro-thermal scheduling model

30.3 Details of Basic TLBO and QA For the understanding of the reader, the descriptions of component algorithms of present study i.e. basic TLBO and QA have been presented in this section.

30.3.1 Teaching Learning Based Optimization TLBO is an optimization technique, proposed by Rao et al. [9]. This algorithm is inspired by the natural phenomenon of a classroom learning process between the teacher and student. It is similar to other population based algorithm like PSO, GSA DE etc. In a class, a set of student are study and they offered some subject. From the class room lecture by the teacher, teacher tries to improve the knowledge of the student in a whole class. In TLBO, a set if student is similar to the population set. Subject offered to each student in the classroom is similar to the different decision variables of the optimization problem. The results of the students in the execution of the teacher are similar to the objection function value corresponding to the decision variable in an optimization problem. As a hypothesis, teacher is most knowledgeable person in a society. In TLBO, a student with most optimum objective function value from a set of student is considered as a teacher. In a classroom, the student learn by teacher and self-study or by interaction between the learners, so the execution of TLBO is based on two phase one is ‘Teacher Phase’ and another is ‘Learner Phase’.

30 A Hybrid TLBO Algorithm by Quadratic Approximation …

309

In the population based algorithm, the common parameters are NP (population size) and Iteration. But some algorithm has other parameters such as mutation rate and crossover rate of the DE and GA [3, 5], inertia weight, social and cognitive parameters of the PSO; employed, scout and onlookers parameters, limits of ABC [7, 8]; harmony memory consideration rate, the number of improvisations, and pitch adjusting rate of HS. Based on the proper choice of the parameters, these algorithms perform better. Sometimes, it is difficult to select the appropriate values of this parameter. The TLBO has only common controlling parameters like class size (NP), iteration and teaching factor for its working and it has no other controlling parameters, which makes the algorithm robust and powerful.

30.3.1.1

Teacher Phase

In a class a teacher spread knowledge during the class room lecture in a classroom among the set of student on a subject. Teacher tries to increase the average knowledge of the student in a whole class. The mean parameter M g of the ith student is denoted g g g g g by Mi = (Mi1 , Mi2 , Mi3 , . . . , Mid ) at generation ‘g’ of each subject in the class. The learner which has the minimum fitness value (for minimization problem) is g representing the teacher X i for the respective iteration. The ith learner is updated in the teacher phase according to the following expression:  g g g g X newi = X i + rand ∗ X i − TF ∗ Mi

(30.1)

Here the teaching factor,TF is calculated by Eq. (30.2), TF = r ound[1 + rand(0, 1){2 − 1}] 30.3.1.2

(30.2)

Learner Phase

After learn from the teacher, students learn from the communication of others which known as learner phase. From the interaction, a learner tries to increase his/her knowledge by other learners. The mathematical representation of learning phenomenon is g g expressed below. In this phase, two learner X i and X j (i = j) select randomly from the entire set of learner. Then the ith learner will interact with jth learner and attempt to learn some new information. The new ith learner is calculated by Eq. (30.3).  g X newi

=

 g  g g g X i + rand ∗  X i − X r , f  X i  > g g g g X i + rand ∗ X r − X i , f X i
a −a < xi < a ⎪ ⎪ ⎩ k(−x − a)m x < a i i

0

⎧ m ⎪ ⎪ ⎨ k(xi − a)

⎧ ⎫  D−1 (yi − 1)2 [1 + 10 sin2 (3π yi+1 )⎬ π ⎨ 10 sin2 (π yi ) + i=1 f (x) = ⎭ D ⎩ + (y − 1)2 ]

f (x) =

f (x) = 20 + e − 20e



f (x) = max{|xi |, 1 ≤ i ≤ D} D f (x) = i=1 [100(xi+1 − xi2 )2 + (xi − 1)2 ] D f (x) = i=1 (xi + 0.5)2 D f (x) = i=1 i xi4 + random(0, 1) √  f (x) = 418 : 9829n − xi sin |xi | D f (x) = 10D + i=1 [xi2 − 10 cos(2π xi )]

Formulation of benchmark functions D 2 f (x) = i=1 xi D

D |xi | f (x) = i=1 |xi | + i=1 D i 2 f (x) = i=1 x j=1 i

(continued)

[−50, 50]

[−600, 600]

[−32, 32]

[−5.12, 5.12]

[−500, 500]

[−1.28, 1.28]

[−100, 100]

[−30, 30]

[−100,100]

[−100, 100]

[−10, 10]

[−100, 100]

Search space

Table 30.1 Standard test functions used to verify the efficiency of the suggested algorithm. Fmin = 0;(All the standard test functions are minimization functions)

316 S. Nama et al.

f (x) =

F17. Ellipsoidal

F20. Cosine mixture

F19. Exponential

F18. Cigar

2 i=1 i x i

− i)2

D i xi i=1 2



2 +

D i xi i=1 2



2

i=1 (x i D 2 f (x) = x12 + 100,000 i=2 xi  D 2 f (x) = 1 − exp −0.5 i=1 xi D 2 D f (x) = 00.1D + i=1 xi − 0.1 i=1 cos(5π xi )

D

D

f (x) =

F16. Axis parallel hyper ellipsoid

+

2 i=1 x i

D

Formulation of benchmark functions ⎧ ⎫  D−1 ⎨ 10 sin2 (π x ) + (xi − 1)2 [1 + 10 sin2 (3π xi+1 )⎬ i i=1 f (x) = 0.1 ⎩ ⎭ + (x D − 1)2 [1 + sin2 (2π x D )]] D u(xi , 5, 100, 4) + i=1

 D 2 D 2 + 0.1 f (x) = 1 − cos 2π x i=1 i i=1 x i

f (x) =

F15. Zakharov

F14. Salomon

F13. Penalized2

Functions

Table 30.1 (continued)

[−1, 1]

[−1, 1]

[−10, 10]

[−100, 100]

[−5.12, 5.12]

[−5.12, 5.12]

[−100, 100]

[−50, 50]

Search space

30 A Hybrid TLBO Algorithm by Quadratic Approximation … 317

1.14e+001 ± 1.97e+000

6.48e+003 ± 1.93e+003

2.45e+001 ± 5.43e+000

3.19e+005 ± 1.67e+005

1.30e+003 ± 4.41e+002

2.41e−001 ± 1.24e−001

2.09e+003 ± 2.06e+002

6.15e+001 ± 6.78e+000

1.21e+001 ± 1.49e+000

1.47e+001 ± 4.12e+000

5.51e+003 ± 9.91e+003

3.47e+005 ± 3.88e+005

4.38e+000 ± 6.90e−001

1.53e+001 ± 6.46e+000

1.53e+001 ± 4.86e+000

1.36e+003 ± 3.82e+002

9.61e+005 ± 3.71e+005

6.26e−002 ± 1.68e−002

8.30e−001 ± 1.49e−001

1.88e+001 ± 3.54e+000

2.53e+004 ± 4.67e+003

4.09e+001 ± 3.59e+000

3.93e+006 ± 2.00e+006

5.45e+003 ± 1.06e+003

1.18e+000 ± 2.85e−001

2.22e+003 ± 2.49e+002

6.26e+001 ± 8.54e+000

1.70e+001 ± 6.68e−001

4.98e+001 ± 1.21e+001

2.76e+006 ± 1.54e+006

9.92e+006 ± 6.16e+006

7.85e+000 ± 7.62e−001

2.70e+001 ± 7.97e+000

6.69e+001 ± 1.82e+001

5.55e+003 ± 8.64e+002

4.46e+006 ± 1.08e+006

2.34e−001 ± 3.95e−002

1.21e+000 ± 1.56e−001

F2

F3

F4

F5

F6

F7

F8

F9

F10

F11

F12

F13

F14

F15

F16

F17

F18

F19

F20

6.91e−001 ± 1.75e−001

1.07e−001 ± 4.47e−002

7.19e+005 ± 7.30e+005

1.07e+003 ± 9.46e+002

1.01e+001 ± 8.37e+000

3.40e+001 ± 9.81e+000

6.77e+000 ± 1.39e+000

2.75e+005 ± 1.27e+006

1.01e+001 ± 2.83e+001

9.41e+000 ± 6.15e+000

1.39e+001 ± 1.78e+000

3.74e+001 ± 7.03e+000

1.38e+003 ± 2.08e+002

4.77e−001 ± 2.92e−001

9.78e+002 ± 7.00e+002

8.01e+004 ± 1.05e+005

4.56e+001 ± 9.07e+000

3.91e+003 ± 3.70e+003

3.77e+000 ± 1.70e+000

9.11e+002 ± 6.87e+002

Mean ± SD

Mean ± SD

1.42e+003 ± 4.49e+002

Mean ± SD

ABC

DE/rand/1/bin/

5.38e+003 ± 9.69e+002

PSO

F1

Function

1.28e+000 ± 2.87e−001

2.37e−001 ± 8.62e−002

3.32e+006 ± 1.30e+006

5.27e+003 ± 2.17e+003

5.86e+001 ± 2.72e+001

3.85e+001 ± 1.48e+001

7.96e+000 ± 1.19e+000

1.90e+007 ± 2.34e+007

3.61e+006 ± 4.12e+006

4.31e+001 ± 1.44e+001

1.76e+001 ± 1.31e+000

7.30e+001 ± 1.09e+001

1.89e+003 ± 2.06e+002

1.42e+000 ± 1.07e+000

4.83e+003 ± 2.28e+003

5.23e+006 ± 4.57e+006

4.13e+001 ± 9.55e+000

2.41e+004 ± 8.66e+003

1.80e+001 ± 5.86e+000

5.59e+003 ± 2.65e+003

Mean ± SD

BSA

2.14e−001 ± 6.89e−002

4.80e−003 ± 1.97e−003

5.74e+004 ± 2.91e+004

1.10e+002 ± 4.02e+001

1.11e+000 ± 5.12e−001

2.45e+000 ± 9.96e−001

1.56e+000 ± 2.96e−001

3.75e+000 ± 1.05e+000

2.66e+000 ± 8.96e−001

1.88e+000 ± 3.57e−001

5.62e+000 ± 5.48e−001

4.73e+001 ± 7.00e+000

2.34e+003 ± 1.57e+002

3.19e−002 ± 1.57e−002

9.46e+001 ± 3.29e+001

1.56e+003 ± 1.14e+003

6.53e+000 ± 1.15e+000

4.58e+002 ± 1.59e+002

2.73e+000 ± 5.97e−001

8.62e+001 ± 3.91e+001

Mean ± SD

HTLBO

Table 30.2 Investigational outcomes of DE/rand/1/bin, PSO, ABC, BSA and HTLBO at dimension (D) 10 once attained 1000 FEs of 20 benchmark functions given in Table 30.1 over 25 runs and 50 PS

318 S. Nama et al.

30 A Hybrid TLBO Algorithm by Quadratic Approximation …

319

Table 30.3 Results of Friedman rank test which are calculated by utilizing the mean result of all the standard functions with D = 10 (PSO, DE, ABC, BSA, HTLBO) Algorithms

Mean rank

PSO

4.25

DE

2.70

ABC

2.35

BSA

4.45

HTLBO

1.25

given. On observing Table 30.6, it has been found that the execution of suggested HTLBO technique is higher to all the compared algorithms except PSO-cf for the test function F8. By observing Table 30.7, it is found the HTLBO to have the minimum mean rank than all other compared algorithms. This means that the proposed HTLBO statistically superior to the compared algorithms. In Table 30.8 comparison performance of the proposed HTLBO algorithm with DNLPSO [89] and some hybrid variants (DE-PSO [21], PSOGSA [25] and qPSO-W [26]) has been given. Table 30.9 gives the statistical rank test by evaluating the mean outcomes of every standard test functions of all the algorithms with Friedman rank test. The bold face indicates the superior results for each function in Table 30.8. By the observation of Table 30.9, the proposed HTLBO algorithm is found to be the minimum rank holder, which signifies HTLBO to be statistically superior to the compared methodologies. So, from the above observation, it is evident that the suggested HTLBO algorithm is significantly higher to the former differentiated methods.

30.5.2 Comparison Results for 30 Dimensional Test Functions Table 30.10 has given the comparative performance outcome of the suggested HTLBO algorithm with some basic algorithms (PSO, DE, ABC and BSA) at dimension thirty for all the test functions. All the algorithms were run twenty five times with population size fifty and 3000FEs. Table 30.11 gives statistical mean ranks acquired by Friedman rank test by evaluating the mean results of every standard test function. In Table 30.10, it is seen that for function F8 and F9, the performance of HTLBO is inferior to ABC but superior to PSO, DE and BSA. For all other test functions, HTLBO is superior to all differentiated methods. From Table 30.11 it has seen that the mean rank is lesser than to all other compared algorithms. This means HTLBO statistically superior to the remaining compared methodologies. Table 30.12 describes the comparison result of HTLBO with some DE variants (DE/best/2/bin [19], DE/current-to-rand/1/bin [17], DE/rand/2/bin [18] and EPSDE [20]) taking thirty dimensional test problems. Table 30.13 gives, the rank acquired by Friedman rank test by evaluating all the mean results of every functions for all

DE/current-to-rand/1/bin (F = 0.5, CR = 0.9)

Mean ± SD

3.40e+002 ± 1.26e+002

7.52e+000 ± 1.37e+000

1.80e+003 ± 5.13e+002

1.55e+001 ± 2.91e+000

2.51e+004 ± 1.07e+004

3.60e+002 ± 1.04e+002

7.84e−002 ± 3.14e−002

2.39e+003 ± 2.18e+002

Mean ± SD

4.47e+002 ± 2.19e+002

7.71e+000 ± 1.82e+000

1.80e+003 ± 7.32e+002

1.90e+001 ± 3.75e+000

3.14e+004 ± 2.45e+004

4.48e+002 ± 1.30e+002

1.28e−001 ± 6.62e−002

2.15e+003 ± 3.09e+002

F1

F2

F3

F4

F5

F6

F7

F8

5.69e+001 ± 9.07e+000

8.47e+000 ± 6.55e−001

4.15e+000 ± 1.14e+000

1.23e+001 ± 3.66e+000

2.50e+001 ± 1.19e+001

2.56e+000 ± 4.06e−001

6.08e+001 ± 7.10e+000

8.18e+000 ± 1.01e+000

4.80e+000 ± 1.86e+000

2.15e+001 ± 1.48e+001

1.53e+004 ± 7.61e+004

2.90e+000 ± 4.73e−001

F9

F10

F11

F12

F13

F14

Function

DE/best/2/bin (F = 0.5, CR = 0.9)

6.03e+000 ± 9.44e−001

2.84e+006 ± 2.57e+006

4.71e+005 ± 4.65e+005

2.52e+001 ± 7.04e+000

1.49e+001 ± 8.15e−001

6.76e+001 ± 7.03e+000

2.23e+003 ± 1.47e+002

4.94e−001 ± 2.10e−001

2.83e+003 ± 8.52e+002

9.96e+005 ± 4.65e+005

3.60e+001 ± 3.85e+000

1.30e+004 ± 3.33e+003

1.63e+001 ± 3.27e+000

3.01e+003 ± 1.06e+003

Mean ± SD

DE/rand//2/bin (F = 0.5, CR = 0.9)

4.64e+000 ± 9.25e−001

9.91e+005 ± 2.30e+006

5.17e+004 ± 1.54e+005

1.30e+001 ± 5.87e+000

1.37e+001 ± 1.53e+000

6.47e+001 ± 9.11e+000

8.16e+000 ± 5.51e+000

2.83e−001 ± 1.22e−001

1.27e+003 ± 3.95e+002

2.93e+005 ± 2.02e+005

2.72e+001 ± 4.97e+000

7.70e+003 ± 4.45e+003

1.25e+001 ± 2.37e+000

1.25e+003 ± 5.28e+002

Mean ± SD

EPSDE

(continued)

1.56e+000 ± 2.96e−001

3.75e+000 ± 1.05e+000

2.66e+000 ± 8.96e−001

1.88e+000 ± 3.57e−001

5.62e+000 ± 5.48e−001

4.73e+001 ± 7.00e+000

2.34e+003 ± 1.57e+002

3.19e−002 ± 1.57e−002

9.46e+001 ± 3.29e+001

1.56e+003 ± 1.14e+003

6.53e+000 ± 1.15e+000

4.58e+002 ± 1.59e+002

2.73e+000 ± 5.97e−001

8.62e+001 ± 3.91e+001

Mean ± SD

HTLBO

Table 30.4 Investigational outcomes of DE/best/2/bin, DE/current-to-rand/1/bin, DE/rand/2/bin, EPSDE and HTLBO at dimension (D) 10 on the attainment of 1000 FEs of 20 benchmark functions which is given in Table 30.1 over 25 runs and 50 PS

320 S. Nama et al.

4.48e+000 ± 1.46e+000

3.55e+002 ± 1.03e+002

2.29e+005 ± 7.88e+004

1.69e−002 ± 4.55e−003

5.56e−001 ± 1.19e−001

5.42e+000 ± 1.91e+000

4.00e+002 ± 1.34e+002

2.38e+005 ± 1.30e+005

2.15e−002 ± 8.65e−003

7.08e−001 ± 1.34e−001

F16

F17

F18

F19

F20

1.06e+000 ± 1.48e−001

1.31e−001 ± 4.48e−002

1.89e+006 ± 6.48e+005

2.85e+003 ± 8.93e+002

3.39e+001 ± 7.47e+000

2.36e+001 ± 7.51e+000

Mean ± SD

Mean ± SD

1.07e+001 ± 2.73e+000

Mean ± SD

DE/rand//2/bin (F = 0.5, CR = 0.9)

DE/current-to-rand/1/bin (F = 0.5, CR = 0.9)

7.33e+000 ± 3.18e+000

DE/best/2/bin (F = 0.5, CR = 0.9)

F15

Function

Table 30.4 (continued)

7.82e−001 ± 1.43e−001

6.51e−002 ± 2.48e−002

9.87e+005 ± 4.96e+005

1.63e+003 ± 6.93e+002

1.84e+001 ± 6.93e+000

3.45e+001 ± 1.24e+001

Mean ± SD

EPSDE

2.14e−001 ± 6.89e−002

4.80e−003 ± 1.97e−003

5.74e+004 ± 2.91e+004

1.10e+002 ± 4.02e+001

1.11e+000 ± 5.12e−001

2.45e+000 ± 9.96e−001

Mean ± SD

HTLBO

30 A Hybrid TLBO Algorithm by Quadratic Approximation … 321

322

S. Nama et al.

Table 30.5 Results of Friedman rank test which are calculated by utilizing the mean result of all the standard functions with D = 10 (DE/best/2/bin, DE/current-to-Rand/bin, DE/rand/2/bin, EPSDE, HTLBO) Algorithms

Mean rank

DE/best/2/bin

2.82

DE/current-to-rand/1/bin

2.28

DE/rand/2/bin

4.85

EPSDE

3.90

HTLBO

1.15

algorithms. In Table 30.12 it has been found that the HTLBO performs inferior to EPSDEfor test function F8 and F17. But for other functions, HTLBO is superior to all differentiated metaheuristics. From the founding of Table 30.13, the mean rank of HTLBO is lesser than the other compared algorithms. This means HTLBO statistically superior to the rest compared methodologies. Table 30.14 presents the performance results obtained by some PSO variants (CLPSO [16], UPSO [13], FI-PSO [12], PSO-cf [10] and PSO-cf-local [11]) and proposed HTLBO algorithm at dimension thirty for all benchmark problems. Table 30.15 represents ranks gathered by evaluating all the mean rank of every standard test functions for all algorithms w.r.t Friedman rank test. In Table 30.14, it cab be perceived that the presentation result of proposed HTLBO algorithm is inferior to PSO-cf for test function F17, but superior to all other compared algorithms. For test function F8, the proposed HTLBO algorithm is inferior to UPSO, but superior to the all other differentiated methodologies. Also for all other benchmark problems, HTLBO is superior to the rest differentiated methods. By observing Table 30.15, it can be realized that the mean rank of HTLBO is lesser than the former associated method. This indicates the developed HTLBO algorithm is statistically superior to other said PSO variants. Table 30.16 presents the statistical results of the twenty five runs of the DNLPSO [89] and hybrid variants (DE-PSO [21], PSOGSA [25] and qPSO-W [26]) on twenty benchmark problems taking dimension as thirty. The best results are bolded. Table 30.17 shows the statistical rank test acquired by evaluating mean results of all the test problems of each algorithm with respect to Friedman rank test. In Table 30.16, it can be seen that the presentation of proposed HTLBO algorithm is lesser to DEPSO for test function F8 and F9 and qPSO-W for test function F17, but for other test function superior to other compared algorithms. In Table 30.17, it is realized that Friedman rank of the suggested HTLBO algorithm is minimum to rest of the compared methodologies. That signifies the efficiency of HTLBO is statistically superior to other differentiated methodologies. Hance, the above observation concludes that the proposed HTLBO algorithm is efficient as well as satisfactory.

Mean ± SD

1.20e+003 ± 4.49e+002

1.12e+001 ± 2.62e+000

6.90e+003 ± 3.08e+003

2.69e+001 ± 6.09e+000

3.30e+005 ± 2.18e+005

1.43e+003 ± 3.72e+002

3.02e−001 ± 1.58e−001

5.65e+001 ± 7.20e+001

5.69e+001 ± 1.09e+001

1.21e+001 ± 1.24e+000

Mean ± SD

5.22e+003 ± 1.40e+003

1.96e+001 ± 3.83e+000

2.65e+004 ± 9.89e+003

4.63e+001 ± 7.62e+000

6.99e+006 ± 2.97e+006

5.50e+003 ± 1.87e+003

1.62e+000 ± 6.82e−001

1.55e+003 ± 3.20e+002

6.92e+001 ± 1.04e+001

1.74e+001 ± 1.03e+000

F2

F3

F4

F5

F6

F7

F8

F9

F10

UPSO

CLPSO

F1

Function

1.43e+001 ± 1.20e+000

6.91e+001 ± 6.98e+000

8.62e+002 ± 3.28e+002

5.30e−001 ± 2.36e−001

2.48e+003 ± 7.83e+002

8.09e+005 ± 4.68e+005

2.98e+001 ± 3.93e+000

1.18e+004 ± 3.78e+003

1.44e+001 ± 1.98e+000

2.68e+003 ± 6.48e+002

Mean ± SD

FI-PSO

7.93e+000 ± 1.35e+000

5.56e+001 ± 7.52e+000

1.30e+002 ± 1.94e+002

1.07e−001 ± 4.35e−002

4.00e+002 ± 1.95e+002

4.31e+004 ± 4.19e+004

1.33e+001 ± 3.04e+000

1.80e+003 ± 7.10e+002

5.00e+000 ± 1.42e+000

3.37e+002 ± 1.49e+002

Mean ± SD

PSO-cf

1.15e+001 ± 1.29e+000

5.84e+001 ± 8.03e+000

2.23e+002 ± 2.59e+002

2.22e−001 ± 9.88e−002

1.25e+003 ± 4.97e+002

1.43e+005 ± 9.47e+004

2.06e+001 ± 3.93e+000

5.60e+003 ± 2.08e+003

9.80e+000 ± 2.49e+000

1.20e+003 ± 4.23e+002

Mean ± SD

PSO-cf-local

(continued)

5.62e+000 ± 5.48e−001

4.73e+001 ± 7.00e+000

2.34e+003 ± 1.57e+002

3.19e−002 ± 1.57e−002

9.46e+001 ± 3.29e+001

1.56e+003 ± 1.14e+003

6.53e+000 ± 1.15e+000

4.58e+002 ± 1.59e+002

2.73e+000 ± 5.97e−001

8.62e+001 ± 3.91e+001

Mean ± SD

HTLBO

Table 30.6 Investigational outcomes of CLPSO, UPSO, FI-PSO, PSO-cf, PSO-cf-local and HTLBO at D = 10 after attainment of 1000 FEs of 20 benchmark functions which is given in Table 30.1 over 25 runs and 50 PS

30 A Hybrid TLBO Algorithm by Quadratic Approximation … 323

UPSO

Mean ± SD

1.30e+001 ± 3.54e+000

7.67e+003 ± 1.52e+004

4.82e+005 ± 6.72e+005

4.62e+000 ± 5.26e−001

3.03e+001 ± 1.03e+001

1.55e+001 ± 4.73e+000

1.19e+003 ± 3.22e+002

8.66e+005 ± 4.01e+005

6.78e−002 ± 2.37e−002

7.73e−001 ± 1.46e−001

CLPSO

Mean ± SD

5.25e+001 ± 1.32e+001

5.60e+006 ± 8.02e+006

1.77e+007 ± 9.61e+006

8.30e+000 ± 1.30e+000

3.30e+001 ± 7.47e+000

6.74e+001 ± 2.11e+001

5.66e+003 ± 1.64e+003

4.37e+006 ± 1.50e+006

2.46e−001 ± 7.15e−002

1.30e+000 ± 3.00e−001

F11

F12

F13

F14

F15

F16

F17

F18

F19

F20

Function

Table 30.6 (continued)

9.79e−001 ± 1.20e−001

1.18e−001 ± 2.64e−002

1.97e+006 ± 5.51e+005

2.42e+003 ± 5.36e+002

3.18e+001 ± 8.95e+000

2.84e+001 ± 8.59e+000

5.70e+000 ± 1.08e+000

2.14e+006 ± 1.55e+006

5.23e+004 ± 6.64e+004

2.53e+001 ± 7.37e+000

Mean ± SD

FI-PSO

PSO-cf

5.06e−001 ± 1.00e−001

1.69e−002 ± 7.02e−003

2.37e+005 ± 1.13e+005

3.20e+002 ± 1.38e+002

5.37e+000 ± 2.48e+000

9.61e+000 ± 6.45e+000

2.89e+000 ± 4.28e−001

2.37e+001 ± 1.34e+001

7.37e+001 ± 3.08e+002

3.61e+000 ± 1.59e+000

Mean ± SD

PSO-cf-local

7.21e−001 ± 1.14e−001

4.85e−002 ± 1.59e−002

7.51e+005 ± 3.14e+005

1.23e+003 ± 3.68e+002

1.26e+001 ± 3.57e+000

1.86e+001 ± 7.02e+000

4.37e+000 ± 5.36e−001

1.60e+005 ± 2.08e+005

2.85e+003 ± 6.66e+003

1.01e+001 ± 3.06e+000

Mean ± SD

HTLBO

2.14e−001 ± 6.89e−002

4.80e−003 ± 1.97e−003

5.74e+004 ± 2.91e+004

1.10e+002 ± 4.02e+001

1.11e+000 ± 5.12e−001

2.45e+000 ± 9.96e−001

1.56e+000 ± 2.96e−001

3.75e+000 ± 1.05e+000

2.66e+000 ± 8.96e−001

1.88e+000 ± 3.57e−001

Mean ± SD

324 S. Nama et al.

30 A Hybrid TLBO Algorithm by Quadratic Approximation …

325

Table 30.7 Results of Friedman rank test which are calculated by utilizing the mean result of all the standard functions with D = 10 (CLPSO, UPSO, FI-PSO, PSO-cf, PSO-cf-local, HTLBO) Algorithms

Mean rank

CLPSO

5.95

UPSO

3.78

FIPSO

4.90

PSO-cf

2.00

PSO-cf-local

3.12

HTLBO

1.25

30.5.3 Comparison Results for 50 Dimensional Test Functions Table 30.18 is the comparison table of the statistical results of the twenty five runs of the proposed HTLBO algorithm and hybrid variants DESQI [24], DE-PSO [21], PSOGSA [25] and QPSO [27] on twenty benchmark problems taking dimension as fifty. The superior outcomes among the compared methods are bolded. Table 30.19 shows the statistical rank test obtained by evaluating the mean results of all the functions for each algorithm w.r.t Friedman Rank Test. In Table 30.18, it can be realized that the presentation of proposed HTLBO algorithm is inferior to PSOGSA for problem F9, DESQI for problem F17. Also for test function F15, the performance of QPSO and HTLBO is similar. But for all other functions, the performance of HTLBO is superior to rest differentiated methodologies. From Table 30.19, it has seen that rank of the suggested HTLBO is the minimum than said hybrid algorithms; hance we may say that the efficiency of HTLBO is superior to the other compared hybrid algorithms. Also, from the convergence graph (Fig. 30.4a–h), it can be noticed that the suggested HTLBO is faster convergence speed than other methods. From the above observations, it can be said that the overall performance of the proposed HTLBO algorithm is statistically significant as well as satisfactory.

30.6 Application to Real Life Problems In this division, we have considered one real world optimization problem form [90]. Mathematical formulation of this real world optimization problem has been discussed elaborately in this section.

Mean ± SD

4.63e+001 ± 2.18e+001

4.47e+000 ± 5.40e+000

1.54e+003 ± 3.47e+003

8.41e+000 ± 3.18e+000

2.59e+004 ± 3.82e+004

5.66e+001 ± 3.40e+001

6.70e−002 ± 3.74e−002

1.34e+003 ± 3.70e+002

4.00e+001 ± 1.50e+001

5.02e+000 ± 1.06e+000

1.65e+000 ± 4.37e−001

6.98e+000 ± 5.01e+000

5.65e+000 ± 6.52e+000

1.86e+000 ± 6.87e−001

4.54e+000 ± 7.20e+000

2.89e+000 ± 7.15e+000

9.06e+001 ± 8.15e+001

2.98e+004 ± 2.14e+004

4.73e−003 ± 4.12e−003

3.49e−001 ± 2.79e−001

Mean ± SD

1.11e+002 ± 1.98e+002

6.69e+000 ± 1.21e+001

1.81e+003 ± 3.77e+003

8.59e+000 ± 6.49e+000

2.07e+004 ± 7.53e+004

1.56e+002 ± 1.92e+002

6.81e−001 ± 4.56e−001

2.10e+002 ± 3.36e+002

4.61e+001 ± 1.02e+001

5.55e+000 ± 3.72e+000

1.74e+000 ± 2.05e+000

4.38e+001 ± 1.87e+002

8.70e+000 ± 7.76e+000

2.27e+000 ± 2.25e+000

4.08e+000 ± 5.16e+000

4.66e+000 ± 1.23e+001

6.97e+002 ± 2.81e+003

1.96e+005 ± 5.28e+005

1.27e−002 ± 2.55e−002

3.14e−001 ± 2.49e−001

F2

F3

F4

F5

F6

F7

F8

F9

F10

F11

F12

F13

F14

F15

F16

F17

F18

F19

F20

PSOGSA

DNLPSO

F1

Function

9.52e−001 ± 9.13e−002

1.13e−001 ± 2.94e−002

1.70e+006 ± 6.04e+005

2.20e+003 ± 6.70e+002

2.59e+001 ± 9.44e+000

1.90e+001 ± 5.93e+000

5.54e+000 ± 6.91e−001

1.79e+006 ± 1.36e+006

1.07e+005 ± 1.16e+005

2.19e+001 ± 5.83e+000

1.47e+001 ± 1.23e+000

6.21e+001 ± 6.22e+000

2.04e+003 ± 2.34e+002

4.07e−001 ± 1.60e−001

2.48e+003 ± 6.75e+002

8.20e+005 ± 4.04e+005

2.93e+001 ± 4.03e+000

9.97e+003 ± 3.83e+003

1.43e+001 ± 2.00e+000

2.21e+003 ± 7.68e+002

Mean ± SD

DE-PSO

2.84e−001 ± 8.77e−002

4.91e−003 ± 1.94e−003

8.14e+004 ± 5.83e+004

1.25e+002 ± 6.13e+001

1.31e+000 ± 6.33e−001

1.29e+000 ± 7.84e−001

1.54e+000 ± 3.02e−001

4.23e+000 ± 1.80e+000

3.08e+000 ± 1.41e+000

1.80e+000 ± 4.10e−001

5.75e+000 ± 8.24e−001

4.70e+001 ± 8.80e+000

2.36e+003 ± 3.66e+002

2.58e−002 ± 1.46e−002

1.04e+002 ± 4.99e+001

1.72e+003 ± 1.49e+003

5.82e+000 ± 1.38e+000

5.12e+002 ± 2.73e+002

2.95e+000 ± 7.52e−001

1.21e+002 ± 6.64e+001

Mean ± SD

qPSO-W

2.14e−001 ± 6.89e−002

4.80e−003 ± 1.97e−003

5.74e+004 ± 2.91e+004

1.10e+002 ± 4.02e+001

1.11e+000 ± 5.12e−001

2.45e+000 ± 9.96e−001

1.56e+000 ± 2.96e−001

3.75e+000 ± 1.05e+000

2.66e+000 ± 8.96e−001

1.88e+000 ± 3.57e−001

5.62e+000 ± 5.48e−001

4.73e+001 ± 7.00e+000

2.34e+003 ± 1.57e+002

3.19e−002 ± 1.57e−002

9.46e+001 ± 3.29e+001

1.56e+003 ± 1.14e+003

6.53e+000 ± 1.15e+000

4.58e+002 ± 1.59e+002

2.73e+000 ± 5.97e−001

8.62e+001 ± 3.91e+001

Mean ± SD

HTLBO

Table 30.8 Investigational outcomes of DNLPSO, PSOGSA, DE-PSO, qPSO-W and HTLBO at D = 10 once attained 1000 FEs of 20 benchmark functions given in Table 30.1 over 25 runs and 50 PS

326 S. Nama et al.

30 A Hybrid TLBO Algorithm by Quadratic Approximation …

327

Table 30.9 Results of Friedman rank test which are calculated by utilizing the mean result of all the standard functions with D = 10 (DNLPSO, PSOGSA, DE-PSO, qPSO-W, HTLBO) Algorithms

Mean rank

DNLPSO

3.40

PSOGSA

2.30

DEPSO

4.85

qPSO-w

2.45

HTLBO

2.00

30.6.1 Spread Spectrum Radar Polyphase Code Design Problem This optimization issue is constructed on the concepts of the aperiodic autocorrelation function and the premises of coherent radar pulse processing in the receiver. This optimization problem is extensively applied in the radar system design and it has no polynomial time result. The issue underneath thought is demonstrated as a min–max nonlinear non-convex advancement issue through persistent factors and by various nearby optimal. It tends to be communicated as follows: Globalmin f (X ) = max{φ1 (X ), φ2 (X ), . . . φ2m (X )}

(30.7)

  where X = (x1 , x2 , x3 , . . . x D ) ∈ R D |0 ≤ x j ≤ 2π, j = 1, 2, 3, . . . , D and m = 2D − 1, with ⎛ ⎞ j D   φ2i−1 (X ) = cos⎝ X k ⎠, i = 1, 2, 3, . . . D (30.8) j=i

φ2i (X ) = 0.5 +

D  j=i+1

⎛

cos⎝

k=|2i− j−1|+1 j 

⎞

X k ⎠, i = 1, 2, 3, . . . D − 1

(30.9)

k=|2i− j|+1

φm+i (X ) = −φi (X ), i = 1, 2, 3, . . . , m

(30.10)

Here the object is to optimize the minimum module of the biggest among the samples of the auto-correlation task which are associated with the complex envelope of the compacted radar pulse at the optimal receiver output. And the parameters of this optimization problem are denoted as symmetrized phase differences. Details can be found in [90]. For this real life problem, the algorithm runs twenty five times with class size fifty and 5000Fes. The best, median, worst, mean and standard deviation of the fitness value achieved by HTLBO is 1.62e+000, 2.09e+000, 2.35e+000, 2.07e+000 and 1.74e−001 respectively. Table 30.20 presents the comparison of the performance

4.16e+001 ± 6.30e+000

5.47e+004 ± 8.31e+003

4.46e+001 ± 5.04e+000

2.62e+006 ± 1.25e+006

5.01e+003 ± 1.34e+003

1.46e+000 ± 6.21e−001

8.77e+003 ± 3.27e+002

2.61e+002 ± 1.37e+001

1.27e+001 ± 8.64e−001

4.42e+001 ± 9.20e+000

4.35e+005 ± 4.30e+005

4.35e+006 ± 2.31e+006

8.12e+000 ± 9.17e−001

1.14e+002 ± 1.70e+001

1.59e+002 ± 3.08e+001

4.71e+003 ± 1.09e+003

4.29e+006 ± 1.07e+006

1.88e−001 ± 3.45e−002

2.99e+000 ± 2.08e−001

1.30e+001 ± 3.18e+000

6.52e+003 ± 3.70e+003

8.89e+000 ± 9.61e−001

1.14e+004 ± 9.35e+003

4.71e+002 ± 1.84e+002

1.04e−001 ± 5.01e−002

8.92e+003 ± 3.62e+002

1.68e+002 ± 2.08e+001

5.97e+000 ± 8.43e−001

4.88e+000 ± 1.83e+000

4.83e+000 ± 1.69e+000

3.00e+001 ± 1.29e+001

3.12e+000 ± 4.70e−001

1.28e+001 ± 4.01e+000

1.50e+001 ± 6.42e+000

3.07e+003 ± 8.68e+002

3.60e+005 ± 1.60e+005

2.00e−002 ± 9.90e−003

1.11e+000 ± 3.60e−001

F2

F3

F4

F5

F6

F7

F8

F9

F10

F11

F12

F13

F14

F15

F16

F17

F18

F19

F20

2.93e+000 ± 4.84e−001

4.83e−001 ± 8.81e−002

8.99e+006 ± 2.99e+006

1.11e+004 ± 3.71e+003

3.55e+002 ± 1.37e+002

1.80e+002 ± 1.96e+001

1.82e+001 ± 1.72e+000

2.34e+007 ± 2.96e+007

8.55e+006 ± 1.31e+007

9.04e+001 ± 4.37e+001

1.72e+001 ± 8.46e−001

1.55e+002 ± 1.73e+001

5.21e+003 ± 3.99e+002

4.94e+000 ± 4.20e+000

1.10e+004 ± 3.35e+003

6.34e+006 ± 8.82e+006

7.82e+001 ± 4.06e+000

1.25e+005 ± 6.25e+004

8.69e+000 ± 3.56e+000

8.32e+003 ± 3.45e+003

Mean ± SD

Mean ± SD

4.81e+003 ± 1.12e+003

Mean ± SD

ABC

DE/rand/1/bin

3.90e+002 ± 1.63e+002

PSO

F1

Function

4.74e+000 ± 8.90e−001

5.30e−001 ± 1.94e−001

1.85e+007 ± 9.07e+006

1.62e+004 ± 8.92e+003

6.16e+002 ± 2.79e+002

1.36e+002 ± 3.94e+001

1.45e+001 ± 3.80e+000

1.23e+008 ± 1.71e+008

2.75e+007 ± 2.98e+007

1.33e+002 ± 5.29e+001

1.76e+001 ± 1.08e+000

2.73e+002 ± 3.39e+001

7.08e+003 ± 4.09e+002

1.05e+001 ± 1.09e+001

1.79e+004 ± 6.38e+003

2.58e+007 ± 2.67e+007

5.49e+001 ± 1.00e+001

2.39e+005 ± 1.30e+005

3.59e+002 ± 1.07e+003

1.58e+004 ± 7.02e+003

Mean ± SD

BSA

8.02e−003 ± 3.07e−003

1.20e−004 ± 3.22e−005

2.13e+003 ± 6.48e+002

1.67e+003 ± 3.37e+002

8.25e−002 ± 2.95e−002

8.09e+000 ± 2.08e+000

9.17e−001 ± 1.13e−001

1.80e+000 ± 3.48e−001

2.29e−001 ± 2.67e−002

9.91e−001 ± 3.87e−002

9.10e−001 ± 2.05e−001

1.81e+002 ± 1.29e+001

9.11e+003 ± 2.91e+002

1.91e−002 ± 8.86e−003

2.16e+000 ± 1.57e+000

5.21e+001 ± 8.46e+000

1.35e+000 ± 1.60e−001

3.58e+001 ± 1.18e+001

7.25e−001 ± 9.96e−002

2.34e+000 ± 5.75e−001

Mean ± SD

HTLBO

Table 30.10 Investigational outcomes of DE/rand/1/bin, PSO, SOS and QOSOS at D = 30 when attained 3000 FEs of 20 benchmark functions given in Table 30.1 over 25 runs and 50 PS

328 S. Nama et al.

30 A Hybrid TLBO Algorithm by Quadratic Approximation …

329

Table 30.11 Results of Friedman rank test which are calculated by utilizing the mean result of all the standard functions with D = 30 (PSO, DE, ABC, BSA, HTLBO) Algorithms

Mean rank

PSO

2.15

DE

3.15

ABC

3.70

BSA

4.70

HTLBO

1.30

Table 30.12 Investigational outcomes of DE/best/2/bin, DE/current-to-rand/1/bin, DE/rand/1/bin, DE/rand/2/bin, EPSDE and HTLBO at D + 30 on attainment of 3000 FEs of each 20 benchmark functions given in Table 30.1 over 25 runs and 50 PS Function

DE/best/2/bin (F = 0.5, CR = 0.9)

DE/currentto-rand/1/bin (F = 0.5, CR = 0.9)

DE/rand/2/bin (F = 0.5, CR = 0.9)

EPSDE

HTLBO

Mean ± SD

Mean ± SD

Mean ± SD

Mean ± SD

Mean ± SD

F1

1.52e+003 ± 4.52e+002

3.43e+002 ± 8.94e+001

2.15e+004 ± 2.52e+003

1.63e+003 ± 1.26e+003

2.34e+000 ± 5.75e−001

F2

2.96e+001 ± 8.07e+000

1.44e+001 ± 2.64e+000

1.08e+003 ± 2.59e+003

4.29e+001 ± 9.36e+000

7.25e−001 ± 9.96e−002

F3

1.84e+004 ± 6.37e+003

4.47e+003 ± 1.25e+003

2.70e+005 ± 3.95e+004

2.12e+004 ± 1.54e+004

3.58e+001 ± 1.18e+001

F4

3.84e+001 ± 6.97e+000

1.95e+001 ± 2.80e+000

6.30e+001 ± 4.10e+000

4.19e+001 ± 2.45e+001

1.35e+000 ± 1.60e−001

F5

5.72e+005 ± 3.16e+005

2.72e+004 ± 9.50e+003

3.29e+007 ± 9.35e+006

6.54e+005 ± 7.61e+005

5.21e+001 ± 8.46e+000

F6

1.48e+003 ± 4.70e+002

3.81e+002 ± 1.14e+002

2.17e+004 ± 2.90e+003

1.35e+003 ± 6.98e+002

2.16e+000 ± 1.57e+000

F7

6.10e−001 ± 1.84e−001

1.16e−001 ± 4.40e−002

1.53e+001 ± 3.96e+000

8.01e−001 ± 9.74e−001

1.91e−002 ± 8.86e−003

F8

8.73e+003 ± 3.31e+002

9.24e+003 ± 3.55e+002

8.80e+003 ± 2.66e+002

5.32e+000 ± 3.72e+000

9.11e+003 ± 2.91e+002

F9

2.60e+002 ± 1.69e+001

2.29e+002 ± 1.67e+001

3.02e+002 ± 1.71e+001

2.14e+002 ± 1.72e+001

1.81e+002 ± 1.29e+001

F10

9.73e+000 ± 1.15e+000

5.82e+000 ± 5.60e−001

1.83e+001 ± 3.79e−001

9.07e+000 ± 1.76e+000

9.10e−001 ± 2.05e−001

F11

1.58e+001 ± 5.30e+000

4.21e+000 ± 8.91e−001

2.04e+002 ± 2.46e+001

1.19e+001 ± 7.97e+000

9.91e−001 ± 3.87e−002

F12

1.46e+004 ± 3.76e+004

1.20e+001 ± 2.37e+000

4.75e+007 ± 1.57e+007

1.69e+006 ± 2.83e+006

2.29e−001 ± 2.67e−002

F13

3.09e+005 ± 3.81e+005

4.01e+001 ± 7.60e+000

1.17e+008 ± 2.88e+007

3.45e+006 ± 5.77e+006

1.80e+000 ± 3.48e−001 (continued)

330

S. Nama et al.

Table 30.12 (continued) Function

DE/best/2/bin (F = 0.5, CR = 0.9)

DE/currentto-rand/1/bin (F = 0.5, CR = 0.9)

DE/rand/2/bin (F = 0.5, CR = 0.9)

EPSDE

HTLBO

Mean ± SD

Mean ± SD

Mean ± SD

Mean ± SD

Mean ± SD

F14

5.96e+000 ± 8.40e−001

3.32e+000 ± 3.84e−001

1.57e+001 ± 9.17e−001

4.67e+000 ± 1.32e+000

9.17e−001 ± 1.13e−001

F15

1.01e+002 ± 1.71e+001

7.08e+001 ± 1.47e+001

1.47e+002 ± 1.38e+001

2.37e+002 ± 5.05e+001

8.09e+000 ± 2.08e+000

F16

5.17e+001 ± 1.96e+001

1.03e+001 ± 3.16e+000

6.68e+002 ± 1.35e+002

4.51e+001 ± 3.25e+001

8.25e−002 ± 2.95e−002

F17

1.52e+003 ± 6.08e+002

5.04e+002 ± 1.50e+002

2.12e+004 ± 3.63e+003

1.17e+003 ± 5.52e+002

1.67e+003 ± 3.37e+002

F18

1.54e+006 ± 6.39e+005

3.00e+005 ± 7.97e+004

1.93e+007 ± 2.94e+006

1.25e+006 ± 8.19e+005

2.13e+003 ± 6.48e+002

F19

7.77e−002 ± 2.32e−002

1.66e−002 ± 3.94e−003

6.44e−001 ± 4.72e−002

7.02e−002 ± 4.12e−002

1.20e−004 ± 3.22e−005

F20

2.64e+000 ± 3.12e−001

1.30e+000 ± 2.60e−001

4.90e+000 ± 3.67e−001

1.92e+000 ± 4.44e−001

8.02e−003 ± 3.07e−003

Table 30.13 Results of Friedman rank test which are calculated by utilizing the mean result of all the standard functions with D = 30 (DE/best/2/bin, DE/current-to-Rand/bin, DE/rand/2/bin, EPSDE, HTLBO) Algorithms

Mean rank

DE/best/2/bin

3.40

DE/current-to-rand/bin

2.15

DE/rand/2/bin

4.85

EPSDE

3.30

HTLBO

1.30

result of HTLBO with QPSO [27], DESQI [24]. With the content of Table 30.20, it is seen that the mean output and standard deviation acquired by HTLBO algorithm is minimum than QPSO and DESQI. So, it can be said that the HTLBO algorithm efficient than other associated algorithms, in regard of quality of solution and robustness.

30 A Hybrid TLBO Algorithm by Quadratic Approximation …

331

Table 30.14 Investigational outcomes of CLPSO, UPSO, FI-PSO, PSO-cf, PSO-cf-local and HTLBO at D = 30 when attained 3000 FEs on 20 benchmark functions given in Table 30.1 over 25 runs and 50 PS Function

CLPSO

UPSO

FI-PSO

PSO-cf

PSO-cflocal

HTLBO

Mean ± SD

Mean ± SD

Mean ± SD

Mean ± SD

Mean ± SD

Mean ± SD

F1

2.05e+004 ± 3.59e+003

4.84e+003 ± 1.10e+003

1.43e+004 ± 1.93e+003

1.07e+003 ± 4.48e+002

5.14e+003 ± 1.00e+003

2.34e+000 ± 5.75e−001

F2

7.18e+001 ± 8.30e+000

4.36e+001 ± 7.93e+000

6.69e+001 ± 1.11e+001

2.34e+001 ± 8.38e+000

3.78e+001 ± 6.13e+000

7.25e−001 ± 9.96e−002

F3

2.76e+005 ± 6.22e+004

6.42e+004 ± 1.33e+004

1.86e+005 ± 2.73e+004

1.30e+004 ± 5.86e+003

6.32e+004 ± 1.19e+004

3.58e+001 ± 1.18e+001

F4

6.77e+001 ± 6.00e+000

4.64e+001 ± 4.80e+000

5.03e+001 ± 5.25e+000

3.35e+001 ± 6.04e+000

4.20e+001 ± 4.13e+000

1.35e+000 ± 1.60e−001

F5

3.10e+007 ± 1.08e+007

3.04e+006 ± 1.23e+006

1.59e+007 ± 3.61e+006

3.44e+005 ± 2.68e+005

2.48e+006 ± 1.34e+006

5.21e+001 ± 8.46e+000

F6

2.07e+004 ± 3.82e+003

4.65e+003 ± 1.09e+003

1.43e+004 ± 1.82e+003

1.20e+003 ± 4.13e+002

4.80e+003 ± 1.09e+003

2.16e+000 ± 1.57e+000

F7

1.46e+001 ± 4.43e+000

1.74e+000 ± 5.31e−001

7.81e+000 ± 1.93e+000

4.88e−001 ± 1.99e−001

1.50e+000 ± 5.11e−001

1.91e−002 ± 8.86e−003

F8

6.24e+003 ± 6.73e+002

5.77e+001 ± 2.35e+002

6.13e+003 ± 8.16e+002

6.42e+002 ± 1.03e+003

2.14e+003 ± 1.63e+003

9.11e+003 ± 2.91e+002

F9

2.70e+002 ± 1.71e+001

2.12e+002 ± 2.53e+001

2.77e+002 ± 1.57e+001

1.99e+002 ± 3.15e+001

2.27e+002 ± 1.41e+001

1.81e+002 ± 1.29e+001

F10

1.82e+001 ± 7.92e−001

1.30e+001 ± 8.32e−001

1.66e+001 ± 7.27e−001

8.11e+000 ± 1.13e+000

1.32e+001 ± 8.94e−001

9.10e−001 ± 2.05e−001

F11

1.81e+002 ± 3.27e+001

4.19e+001 ± 7.24e+000

1.28e+002 ± 1.72e+001

1.03e+001 ± 3.74e+000

4.82e+001 ± 8.61e+000

9.91e−001 ± 3.87e−002

F12

3.03e+007 ± 1.95e+007

6.85e+005 ± 6.72e+005

1.13e+007 ± 5.59e+006

9.19e+003 ± 2.34e+004

7.75e+005 ± 8.94e+005

2.29e−001 ± 2.67e−002 (continued)

332

S. Nama et al.

Table 30.14 (continued) Function

CLPSO

UPSO

FI-PSO

PSO-cf

PSO-cflocal

HTLBO

Mean ± SD

Mean ± SD

Mean ± SD

Mean ± SD

Mean ± SD

Mean ± SD

F13

9.82e+007 ± 3.98e+007

4.63e+006 ± 2.14e+006

4.88e+007 ± 1.44e+007

2.41e+005 ± 3.59e+005

4.91e+006 ± 2.97e+006

1.80e+000 ± 3.48e−001

F14

1.48e+001 ± 1.02e+000

8.59e+000 ± 7.26e−001

1.29e+001 ± 6.26e−001

5.13e+000 ± 8.09e−001

8.18e+000 ± 9.57e−001

9.17e−001 ± 1.13e−001

F15

1.79e+002 ± 2.79e+001

1.84e+002 ± 3.00e+001

2.64e+002 ± 7.49e+001

1.11e+002 ± 2.36e+001

1.56e+002 ± 2.07e+001

8.09e+000 ± 2.08e+000

F16

7.00e+002 ± 1.13e+002

1.58e+002 ± 4.44e+001

4.73e+002 ± 6.53e+001

3.74e+001 ± 1.50e+001

1.74e+002 ± 3.55e+001

8.25e−002 ± 2.95e−002

F17

2.36e+004 ± 5.20e+003

5.21e+003 ± 1.34e+003

1.28e+004 ± 1.85e+003

1.20e+003 ± 4.23e+002

5.02e+003 ± 8.29e+002

1.67e+003 ± 3.37e+002

F18

1.79e+007 ± 3.81e+006

3.93e+006 ± 1.08e+006

1.31e+007 ± 2.18e+006

9.01e+005 ± 3.43e+005

4.77e+006 ± 9.57e+005

2.13e+003 ± 6.48e+002

F19

6.38e−001 ± 4.64e−002

2.02e−001 ± 3.01e−002

5.16e−001 ± 4.07e−002

4.73e−002 ± 1.95e−002

2.14e−001 ± 3.18e−002

1.20e−004 ± 3.22e−005

F20

4.61e+000 ± 3.57e−001

2.64e+000 ± 2.41e−001

4.13e+000 ± 2.62e−001

1.61e+000 ± 3.84e−001

2.59e+000 ± 2.72e−001

8.02e−003 ± 3.07e−003

Table 30.15 Results of Friedman rank test which are calculated by utilizing the mean result of all the standard functions with D = 30 (CLPSO, UPSO, FI-PSO, PSO-cf, PSO-cf-local, HTLBO) Algorithms

Mean rank

CLPSO

5.80

UPSO

3.40

FI-PSO

5.05

PSO-cf

1.95

PSO-cf-local

3.50

HTLBO

1.30

30 A Hybrid TLBO Algorithm by Quadratic Approximation …

333

Table 30.16 Investigational outcomes of DNLPSO, PSOGSA, DE-PSO, qPSO-W and HTLBO at dimension (D) 30 on attainment of 3000 FEs of 20 benchmark functions given in Table 30.1 over 25 runs and 50 PS Function

DNLPSO

PSOGSA

DE-PSO

qPSO-W

HTLBO

Mean ± SD

Mean ± SD

Mean ± SD

Mean ± SD

Mean ± SD

F1

1.31e+003 ± 2.23e+003

3.00e+003 ± 4.66e+003

2.89e+003 ± 6.04e+002

3.75e+002 ± 1.05e+002

2.34e+000 ± 5.75e−001

F2

2.80e+001 ± 2.32e+001

6.04e+001 ± 2.28e+001

3.97e+001 ± 6.08e+000

9.42e+000 ± 1.16e+000

7.25e−001 ± 9.96e−002

F3

1.43e+004 ± 1.52e+004

7.47e+004 ± 7.07e+004

3.49e+004 ± 8.94e+003

5.32e+003 ± 2.52e+003

3.58e+001 ± 1.18e+001

F4

1.75e+001 ± 6.95e+000

5.81e+001 ± 1.94e+001

2.77e+001 ± 2.41e+000

8.80e+000 ± 1.08e+000

1.35e+000 ± 1.60e−001

F5

2.26e+005 ± 5.58e+005

6.51e+006 ± 2.21e+007

7.26e+005 ± 3.77e+005

1.40e+004 ± 8.54e+003

5.21e+001 ± 8.46e+000

F6

2.43e+003 ± 6.00e+003

2.86e+003 ± 3.80e+003

2.88e+003 ± 6.56e+002

4.17e+002 ± 1.11e+002

2.16e+000 ± 1.57e+000

F7

1.10e+000 ± 4.42e−001

1.41e+000 ± 2.03e+000

6.23e−001 ± 1.89e−001

6.10e−002 ± 2.58e−002

1.91e−002 ± 8.86e−003

F8

8.96e+002 ± 1.40e+003

5.59e+003 ± 6.93e+002

8.46e+003 ± 3.69e+002

9.04e+003 ± 5.10e+002

9.11e+003 ± 2.91e+002

F9

1.18e+002 ± 4.50e+001

1.65e+002 ± 3.39e+001

2.50e+002 ± 1.30e+001

2.00e+002 ± 1.40e+001

1.81e+002 ± 1.29e+001

F10

8.14e+000 ± 3.69e+000

1.54e+001 ± 3.38e+000

1.25e+001 ± 7.08e−001

5.98e+000 ± 7.73e−001

9.10e−001 ± 2.05e−001

F11

1.31e+001 ± 1.77e+001

2.63e+001 ± 3.28e+001

2.43e+001 ± 3.29e+000

4.94e+000 ± 1.44e+000

9.91e−001 ± 3.87e−002

F12

1.44e+006 ± 7.19e+006

5.12e+003 ± 2.45e+004

1.65e+003 ± 2.45e+003

6.82e+000 ± 1.97e+000

2.29e−001 ± 2.67e−002

F13

8.31e+005 ± 2.82e+006

1.65e+007 ± 8.20e+007

3.35e+005 ± 3.26e+005

2.28e+001 ± 7.37e+000

1.80e+000 ± 3.48e−001

F14

4.03e+000 ± 1.08e+000

9.47e+000 ± 2.99e+000

8.13e+000 ± 8.74e−001

2.85e+000 ± 5.68e−001

9.17e−001 ± 1.13e−001

F15

8.57e+001 ± 9.05e+001

1.08e+002 ± 4.91e+001

7.31e+001 ± 8.81e+000

9.70e+000 ± 3.05e+000

8.09e+000 ± 2.08e+000

F16

8.41e+001 ± 1.52e+002

8.47e+001 ± 9.37e+001

8.94e+001 ± 1.68e+001

1.26e+001 ± 4.11e+000

8.25e−002 ± 2.95e−002

F17

1.55e+003 ± 2.01e+003

6.67e+003 ± 5.89e+003

3.18e+003 ± 6.33e+002

1.43e+003 ± 4.12e+002

1.67e+003 ± 3.37e+002

F18

2.31e+006 ± 4.70e+006

1.56e+006 ± 2.76e+006

2.27e+006 ± 5.65e+005

3.70e+005 ± 1.31e+005

2.13e+003 ± 6.48e+002

F19

4.96e−002 ± 6.50e−002

3.96e−001 ± 2.20e−001

1.32e−001 ± 2.45e−002

2.20e−002 ± 6.68e−003

1.20e−004 ± 3.22e−005

F20

1.86e+000 ± 1.66e+000

2.43e+000 ± 8.37e−001

2.13e+000 ± 2.70e−001

8.15e−001 ± 1.77e−001

8.02e−003 ± 3.07e−003

334

S. Nama et al.

Table 30.17 Results of Friedman rank test which are calculated by utilizing the mean result of all the standard functions with D = 30 (DNLPSO, PSOGSA, DE-PSO, qPSO-W, HTLBO) Algorithms

Mean rank

DNLPSO

3.10

PSOGSA

4.45

DEPSO

3.90

qPSO-W

2.15

HTLBO

1.40

Table 30.18 Investigational outcomes of DE/rand/1/bin, PSO, SOS and QOSOS at dimension (D) 50 once attainment 5000 FEs of 20 benchmark functions which is given in Table 30.1 over 25 runs and 50 PS Function

DESQI

PSOGSA

DE-PSO

QPSO

HTLBO

Mean ± SD

Mean ± SD

Mean ± SD

Mean ± SD

Mean ± SD

F1

2.45e+002 ± 5.04e+001

1.82e+004 ± 9.05e+003

1.85e+003 ± 4.95e+002

2.58e+002 ± 7.12e+001

1.16e−001 ± 4.34e−002

F2

7.26e+000 ± 9.22e−001

1.42e+002 ± 3.42e+001

4.85e+001 ± 8.62e+000

8.92e+000 ± 1.54e+000

1.38e−001 ± 2.12e−002

F3

5.09e+003 ± 1.08e+003

3.32e+005 ± 1.74e+005

3.83e+004 ± 1.27e+004

5.69e+003 ± 1.35e+003

1.79e+000 ± 4.51e−001

F4

9.26e+000 ± 1.90e+000

7.74e+001 ± 1.35e+001

1.34e+001 ± 1.42e+000

6.33e+000 ± 7.56e−001

3.54e−001 ± 6.50e−002

F5

9.05e+003 ± 4.47e+003

2.36e+007 ± 4.15e+007

1.62e+005 ± 8.12e+004

5.99e+003 ± 2.26e+003

4.97e+001 ± 2.79e−001

F6

2.84e+002 ± 7.68e+001

1.77e+004 ± 9.05e+003

2.00e+003 ± 5.38e+002

2.73e+002 ± 6.22e+001

0.00e+000 ± 0.00e+000

F7

1.28e−001 ± 4.61e−002

4.31e+000 ± 2.29e+000

4.94e−001 ± 1.69e−001

8.35e−002 ± 2.45e−002

1.53e−002 ± 7.72e−003

F8

1.62e+004 ± 3.59e+002

1.08e+004 ± 1.05e+003

1.53e+004 ± 5.21e+002

1.62e+004 ± 7.95e+002

4.48e+001 ± 6.03e+001

F9

3.73e+002 ± 1.82e+001

3.14e+002 ± 4.97e+001

4.21e+002 ± 2.73e+001

3.16e+002 ± 2.89e+001

3.17e+002 ± 2.15e+001

F10

4.49e+000 ± 3.47e−001

1.83e+001 ± 8.50e−001

9.55e+000 ± 8.15e−001

4.71e+000 ± 3.48e−001

7.93e−002 ± 1.76e−002

F11

3.26e+000 ± 6.99e−001

1.70e+002 ± 7.27e+001

1.87e+001 ± 4.11e+000

3.55e+000 ± 5.55e−001

1.52e−001 ± 6.61e−002

F12

4.29e+000 ± 1.54e+000

4.15e+007 ± 9.59e+007

1.62e+001 ± 4.05e+000

4.21e+000 ± 1.11e+000

1.92e−001 ± 1.53e−002

F13

2.25e+001 ± 8.71e+000

6.88e+007 ± 1.46e+008

1.14e+002 ± 3.10e+001

1.69e+001 ± 6.09e+000

3.16e+000 ± 4.37e−001 (continued)

30 A Hybrid TLBO Algorithm by Quadratic Approximation …

335

Table 30.18 (continued) Function

DESQI

PSOGSA

DE-PSO

QPSO

HTLBO

Mean ± SD

Mean ± SD

Mean ± SD

Mean ± SD

Mean ± SD

F14

3.15e+000 ± 3.67e−001

1.89e+001 ± 3.54e+000

7.45e+000 ± 5.51e−001

2.96e+000 ± 3.42e−001

7.15e−001 ± 8.66e−002

F15

1.91e+001 ± 3.98e+000

3.06e+002 ± 1.06e+002

1.16e+002 ± 1.61e+001

1.35e+001 ± 3.33e+000

1.35e+001 ± 3.75e+000

F16

1.29e+001 ± 4.20e+000

3.62e+002 ± 2.36e+002

1.00e+002 ± 2.51e+001

1.45e+001 ± 3.43e+000

5.72e−003 ± 2.17e−003

F17

1.27e+003 ± 2.67e+002

2.45e+004 ± 1.22e+004

5.66e+003 ± 1.48e+003

5.07e+003 ± 8.44e+002

1.06e+004 ± 1.65e+003

F18

2.23e+005 ± 4.89e+004

1.05e+007 ± 5.25e+006

1.73e+006 ± 4.67e+005

2.65e+005 ± 6.59e+004

8.80e+001 ± 3.20e+001

F19

1.18e−002 ± 3.93e−003

6.59e−001 ± 1.44e−001

9.29e−002 ± 2.00e−002

1.28e−002 ± 3.40e−003

4.83e−006 ± 1.80e−006

F20

2.92e−001 ± 7.27e−002

5.27e+000 ± 1.01e+000

2.23e+000 ± 3.56e−001

4.61e−001 ± 1.39e−001

2.77e−004 ± 1.19e−004

Table 30.19 Results of Friedman rank test which are calculated by utilizing the mean result of all the standard functions with D = 50 (DESQI, PSOGSA, DE-PSO, qPSO-W, HTLBO) Algorithms

Mean rank

DESQI

2.58

PSOGSA

4.65

DEPSO

3.95

QPSO

2.55

HTLBO

1.28

30.7 Conclusion TLBO is a global optimization method which is motivated from the natural phenomena of classroom teaching and learning process. During the execution it is utilized teaching and learning phase. However, the learners can also learn through selfmotivation, which has not taken in account in the unique TLBO algorithm. Also, the assessment of the teaching factor into the original TLBO algorithm is either 2 or 1, which can lessen the convergence rate of the algorithm. In this work, an adaptive principal based modified teaching factor is added into the innovative TLBO algorithm. Also, the QA operator is suggested into the original TLBO to upgrade the efficiency as well as the convergence speed to calculate the optimal value. The suggested algorithm experimented on twenty well known standard functions shown in Table 30.1. And the obtained outcomes are differentiated with some standard algorithm, some DE variants, some PSO variants and some hybrid algorithms. The non-parametric Friedman rank test is applied to find the mean rank among the compared algorithm.

336

S. Nama et al. 25

Function error value

Function error value

15 10 5

DE-PSO DESQI HTLBO PSOGSA QPSO

0 -5

0

20 15

5 0

1000 2000 3000 4000 5000

QPSO PSOGSA HTLBO DESQI DE-PSO

10

0

1000 2000 3000 4000 5000

Fitness evaluation (FEs)

Fitness evaluation (FEs)

(b) F5

(a) F1 4 3

9

DE-PSO DESQI HTLBO PSOGSA QPSO

8 7 6 5 4 3

Function error value

Function error value

10

0

QPSO PSOGSA HTLBO DESQI DE-PSO

-2

1000 2000 3000 4000 5000

0

Fitness evaluation (FEs)

(c) F8

(d) F10 25

6 4 DE-PSO DESQI HTLBO PSOGSA QPSO

2 0

0

15 10 5 0 -5

1000 2000 3000 4000 5000

QPSO PSOGSA HTLBO DESQI DE-PSO

20

0

(f) F12

(e) F11 4

5

3 2 1

DE-PSO DESQI HTLBO

0

PSOGSA QPSO

0

1000 2000 3000 4000 5000

Fitness evaluation (FEs) (g) F14

Function error value

Function error value

1000 2000 3000 4000 5000

Fitness evaluation (FEs)

Fitness evaluation (FEs)

-1

1000 2000 3000 4000 5000

Fitness evaluation (FEs)

Function error value

Function error value

0 -1

-3

8

-2

2 1

0 QPSO PSOGSA

-5

HTLBO DESQI DE-PSO

-10

0

1000 2000 3000 4000 5000

Fitness evaluation (FEs) (h) F20

Fig. 30.4 Convergence graphs of eight functions, a F1, b F5, c F8, d F10, e F11, f F12, g F12, h F12

30 A Hybrid TLBO Algorithm by Quadratic Approximation …

337

Table 30.20 Comparison of statistical result of real world optimization problem with PS = 50 over 25 independent run after 5000 FEs. Results in Boldface indicate that a better result obtained among compared algorithms Algorithms

Best

Median

Worst

Mean

SD

DESQI

1.58e+000

2.26e+000

2.42e+000

2.20e+000

1.80e−001

QPSO

1.63e+000

2.29e+000

2.57e+000

2.23e+000

2.45e−001

HTLBO

1.62e+000

2.09e+000

2.35e+000

2.07e+000

1.74e−001

Also, this proposed HTLBO algorithm is involved to resolve a real-life optimization problems and the obtained presentation is associated with hybrid algorithms. From the outcomes of the obtained results and the statistical rank, it may declare that the general execution of the suggested HTLBO algorithm is satisfactory. Additionally, the suggested HTLBO algorithm may be pertain to solve constraint and multi objective including large scale optimization problem and engineering problem. Also, this method can be applied to the industrial environment to optimize any structure including a huge number of parameters and objectives and can be incorporate to power system problem, safety-critical system performance, modelling, simulation and optimization of complex systems etc. Acknowledgements The authors would like to thank Dr. P. N. Suganthan, School of Electrical and Electronic Engineering, NTU, Singapore for shearing the source codes of PSO variants. Also thanks to the editors, anonymous referees for their valuable suggestion towards improving the book chapter.

References 1. Kennedy, J., Eberhart, R.: Particle swarm optimization. In: Proceeding of the IEEE International Conference on Neural Networks, 1995, 1942–1948 2. Shi, Y., Eberhart, R.: A modified particle swarm optimizer. In: IEEE International Conference on Computational Intelligence, pp. 69–73 (1998) 3. Holland, J.: Adaptation in Natural and Artificial Systems. University of Michigan Press (1975) 4. Dorigo, M.: Optimization, Learning and Natural Algorithms. Thesis (Ph.D.), Politecnico di Milano (1992) 5. Storn, R., Price, K.: Differential evolution—a simple and efficient heuristic for global optimization over continuous spaces. J. Glob. Optim. 11(4), 341–359 (1997) 6. Rashedi, E., Nezamabadi-pour, H., Saryazdi, S.: GSA: a gravitational search algorithm. Inf. Sci. 179, 2232–2248 (2009) 7. Akay, B., Karaboga, D.: Artificial bee colony algorithm for large-scale problems and engineering design optimization. J. Intell. Manuf. (2010). https://doi.org/10.1007/s10845-010-0393-4 8. Akay, B., Karaboga, D.: A modified artificial bee colony (ABC) algorithm for constrained optimization problems. Appl. Soft Comput. (2010). https://doi.org/10.1016/j.asoc.2010.12.001 9. Rao, R.V., Savsani, V.J., Vakharia, D.P.: Teaching–learning-based optimization: a novel method for constrained mechanical design optimization problems. Comput. Aided Des. 43, 303–315 (2011)

338

S. Nama et al.

10. Clerc, M., Kennedy, J.: The particle swarm-explosion, stability, and convergence in a multidimensional complex space. IEEE Trans. Evol. Comput. 6(1), 58–73 (2002) 11. Kennedy, J., Mendes, R.: Population structure and particle swarm performance. In: Proceedings of IEEE Congress Evolutionary Computation, Honolulu, HI, 2002, pp. 1671–1676 12. Mendes, R., Kennedy, J., Neves, J.: The fully informed particle swarm: simpler, maybe better. IEEE Trans. Evol. Comput. 8, 204–210 (2004) 13. Parsopoulos, K.E., Vrahatis, M.N.: UPSO—a unified particle swarm optimization scheme. Lect. Ser. Comput. Sci. 1, 868–873 (2004) 14. Peram, T., Veeramachaneni, K., Mohan, C.K.: Fitness-distance-ratio based particle swarm optimization. In: Proceedings of Swarm Intelligence Symposium, pp. 174–181 (2003) 15. van den Bergh, F., Engelbrecht, A.P.: A cooperative approach to particle swarm optimization. IEEE Trans. Evol. Comput. 8, 225–239 (2004) 16. Liang, J.J., Qin, A.K., Suganthan, P.N., Baskar, S.: Comprehensive learning particle swarm optimizer for global optimization of multimodal functions. IEEE Trans. Evol. Comput. 10(3) (2006) 17. Qin, A.K., Huang, V.L., Suganthan, P.N.: Differential evolution algorithm with strategy adaptation for global numerical optimization. IEEE Trans. Evol. Comput. 13(April), 398–417 (2009) 18. Iorio, A., Li, X.: Solving rotated multi-objective optimization problems using differential evolution. In: Australian Conference on Artificial Intelligence, Cairns, Australia, 2004, pp. 861–872 19. Storn, R.: On the usage of differential evolution for function optimization. In: Biennial Conference of the North American Fuzzy Information Processing Society (NAFIPS), pp. 519–523. IEEE, Berkeley (1996) 20. Mallipeddi, R., Suganthan, P.N., Pan, Q.K., Tasgetiren, M.F.: Differential evolution algorithm with ensemble of parameters and mutation strategies. Appl. Soft Comput. 11, 1679–1696 (2011) 21. Pant, M., Thangaraj, R.: DE-PSO: a new hybrid meta-heuristic for solving global optimization problems. New Math. Nat. Comput. 7(3), 363–381 (2011) 22. Deep, K., Das, K.N.: Quadratic approximation based hybrid genetic algorithm for function optimization. Appl. Math. Comput. 203, 86–98 (2008) 23. Abd-El-Wahed, W.F., Mousa, A.A., El-Shorbagy, M.A.: Integrating particle swarm optimization with genetic algorithms for solving nonlinear optimization problems. J. Comput. Appl. Math. 235, 1446–1453 (2011) 24. Zhang, L., Li, H., Jiao, Y.-C., Zhang, F.-S.: Hybrid differential evolution and the simplified quadratic interpolation for global optimization. Copyright is held by the author/owner(s). GEC’09, 12–14 June 2009, Shanghai, China. ACM 978-1-60558-326-6/09/06 25. Mirjalili, S., Mohd Hashim, S.Z.: A new hybrid PSOGSA algorithm for function optimization. In: International Conference on Computer and Information Application, ICCIA 2010 26. Deep, K., Bansal, J.C.: Hybridization of particle swarm optimization with quadratic approximation. OPSEARCH 46(1), 3–24 27. Pant, M., Thangaraj, R., Abraham, A.: A new PSO algorithm with crossover operator for global optimization problems. Innov. Hybrid Intell. Syst., ASC 44, 215–222 (2007) 28. Nama, S., Saha, A.K., Ghosh, S.: A new ensemble algorithm of differential evolution and backtracking search optimization algorithm with adaptive control parameter for function optimization. Int. J. Ind. Eng. Comput. 7, 323–338 (2016) 29. Nama, S., Saha, A.K., Ghosh, S.: A hybrid symbiosis organisms search algorithm and its application to real world problems. Memetic Comput. (2016). https://doi.org/10.1007/s12293016-0194-1 30. Satapathy, S.C., Naik, A.: A modified teaching-learning-based optimization (mTLBO) for global search. Recent Pat. Comput. Sci. 6, 60–72 (2013) 31. Satapathy, S.C., Naik, A., Parvathi, K.: A teaching learning based optimization based on orthogonal design for solving global optimization problems 32. Rao, R.V., Patel, V.: Comparative performance of an elitist teaching-learning-based optimization algorithm for solving unconstrained optimization problems. Int. J. Ind. Eng. Comput. 4, 29–50 (2013)

30 A Hybrid TLBO Algorithm by Quadratic Approximation …

339

33. Satapathy, S.C., Naik, A., Parvathi, K.: Weighted teaching-learning-based optimization for global function optimization. Appl. Math. 4, 429–439 (2013) 34. Nayak, M.R., Nayak, C.K., Rout, P.K.: Application of multi-objective teaching learning based optimization algorithm to optimal power flow problem. In: 2nd International Conference on Communication, Computing & Security [ICCCS-2012], Procedia Technology, vol. 6, pp. 255–264 (2012) 35. Xia, K., et al.: Disassembly sequence planning using a simplified teaching–learning-based optimization algorithm. Adv. Eng. Inform. (2014). http://dx.doi.org/10.1016/j.aei.2014.07.00 36. Roy, P.K., Paul, C., Sultana, S.: Oppositional teaching learning based optimization approach for combined heat and power dispatch. Electr. Power Energy Syst. 57, 392–403 (2014) 37. Roy, P.K., Sur, A., Pradhan, D.K.: Optimal short-term hydro-thermal scheduling using quasioppositional teaching learning based optimization. Eng. Appl. Artif. Intell. 26, 2516–2524 (2013) 38. Venkata Rao, R.: Teaching Learning Based Optimization Algorithm: And Its Engineering Applications, 1st edn. Springer Publishing Company, Incorporated (2015) 39. Jiang, X., Zhou, J.: Hybrid DE-TLBO algorithm for solving short term hydro-thermal optimal scheduling with incommensurable objectives. In: Proceedings of the 32nd Chinese Control Conference, 26–28 July 2013, Xian, China 40. Xie, Z., Zhang, C., Shao, X., Lin, W., Zhu, H.: An effective hybrid teaching–learning-based optimization algorithm for permutation flow shop scheduling problem. Adv. Eng. Softw. 77, 35–47 (2014) 41. Azad-Farsani, E., Zare, M., Azizipanah-Abarghooee, R., Askarian-Abyaneh, H.: A new hybrid CPSO-TLBO optimization algorithm for distribution network reconfiguration. J. Intell. Fuzzy Syst. 26(5), 2175–2184 (2014). https://doi.org/10.3233/IFS-130892 42. Dokeroglu, T.: Hybrid teaching–learning-based optimization algorithms for the quadratic assignment problem. Comput. Ind. Eng. 85, 86–101 (2015) 43. Gnanambal, K., Jeyavelumani, K.R., Juriya Banu, H.: Optimal, power flow using hybrid teaching learning based optimization algorithm. GRD Journals. Global Research and Development Journal for Engineering. International Conference on Innovations in Engineering and Technology, (ICIET)—2016, July 2016. e-ISSN: 2455-5703 44. Khare, R., Kumar, Y.: A novel hybrid MOL–TLBO optimized techno-economic-socio analysis of renewable energy mix in island mode. Appl. Soft Comput. 43, 187–198 (2016) 45. Sahu, B.K., Pati, T.K., Nayak, J.R., Panda, S., Kar, S.K.: A novel hybrid LUS–TLBO optimized fuzzy-PID controller for load frequency control of multi-source power system. Int. J. Electr. Power Energy Syst. 74, 58–69 (2016) 46. Babazadeh, R., Tavakkoli-Moghaddam, R.: A hybrid GA-TLBO algorithm for optimizing a capacitated three-stage supply chain network. Int. J. Ind. Eng. Prod. Res. 28, 151–161 (2017) 47. Deb, S., Kalita, K., Gao, X., Tammi, K., Mahanta, P.: Optimal placement of charging stations using CSO-TLBO algorithm. In: 2017 Third International Conference on Research in Computational Intelligence and Communication Networks (ICRCICN), Kolkata, pp. 84–89 (2017) 48. Patsariya, A., et al.: Implementation of noble TLBO-MPPT technique for SPV in hybrid DC-DC boost converter. In: 2017 International Conference on Energy, Communication, Data Analytics and Soft Computing (ICECDS), pp. 1622–1627 (2017) 49. Shahbeig, S., Helfroush, M.S., Rahideh, A.: A fuzzy multi-objective hybrid TLBO–PSO approach to select the associated genes with breast cancer. Signal Process. 131, 58–65 (2017) 50. Tuo, S., Yong, L., Deng, F., Li, Y., Lin, Y., Lu, Q.: HSTLBO: a hybrid algorithm based on harmony search and teaching-learning-based optimization for complex high-dimensional optimization problems. PLoS ONE 12(4), e0175114 (2017). https://doi.org/10.1371/journal.pone. 0175114 51. Ding, Y., et al.: A novel hybrid teaching learning based optimization algorithm for function optimization. In: 2017 Chinese Automation Congress (CAC), pp. 4383–4388 (2017) 52. Singh, R., Chaudhary, H., Singh, A.K.: A new hybrid teaching–learning particle swarm optimization algorithm for synthesis of linkages to generate path. Sadhana 42(11), 1851–1870 (2017)

340

S. Nama et al.

53. Chen, X., Xu, B., Yu, K., Du, W.: Teaching-learning-based optimization with learning enthusiasm mechanism and its application in chemical engineering. J. Appl. Math. (2018). https:// doi.org/10.1155/2018/1806947 54. Nenavath, H., Jatoth, R.K.: Hybrid SCA–TLBO: a novel optimization algorithm for global optimization and visual tracking. Neural Comput. Appl. (2018). https://doi.org/10.1007/s00521018-3376-6 55. Zhang, M., Pan, Y., Zhu, J., Chen, G.: BC-TLBO: a hybrid algorithm based on artificial bee colony and teaching-learning-based optimization, pp. 2410–2417 (2018). https://doi.org/10. 23919/chicc.2018.8483829 56. Sevinç, E., Dökero˘glu, T.: A novel hybrid teaching-learning-based optimization algorithm for the classification of data by using extreme learning machines. Turk. J. Electr. Eng. Comput. Sci. 27, 1523–1533 (2019). https://doi.org/10.3906/elk-1802-40 57. Guo, C., Lu, J., Tian, Z., Guo, W., Darvishan, A.: Optimization of critical parameters of PEM fuel cell using TLBO-DE based on Elman neural network. Energy Convers. Manag. 183, 149–158 (2019) 58. Zhang, Q., Yu, G., Song, H.: A hybrid bird mating optimizer algorithm with teaching-learningbased optimization for global numerical optimization. Stat. Optim. Inf. Comput. 3 (2015). https://doi.org/10.19139/soic.v3i1.86 59. Tang, Q., Li, Z., Zhang, L.P., Zhang, C.: Balancing stochastic two-sided assembly line with multiple constraints using hybrid teaching-learning-based optimization algorithm. Comput. Oper. Res. 82, 102–113 (2017) 60. Shao, W., Pi, D., Shao, Z.: A hybrid discrete optimization algorithm based on teaching–probabilistic learning mechanism for no-wait flow shop scheduling. Knowl.-Based Syst. 107, 219–234 (2016) 61. Shao, W., Pi, D., Shao, Z.: A hybrid discrete teaching-learning based meta-heuristic for solving no-idle flow shop scheduling problem with total tardiness criterion. Comput. Oper. Res. 94, 89–105 (2018) 62. Das, S.P., Padhy, S.: A novel hybrid model using teaching–learning-based optimization and a support vector machine for commodity futures index forecasting. Int. J. Mach. Learn. Cyber. 9, 97 (2018). https://doi.org/10.1007/s13042-015-0359-0 63. González-Álvarez, D.L., Vega-Rodríguez, M.A., Rubio-Largo, Á.: Finding patterns in protein sequences by using a hybrid multiobjective teaching learning based optimization algorithm. IEEE/ACM Trans. Comput. Biol. Bioinform. 12(3), 656–666 (2015) 64. Chen, D., Zou, F., Wang, J., et al.: A multi-class cooperative teaching–learning-based optimization algorithm with simulated annealing. Soft Comput. 20, 1921 (2016). https://doi.org/ 10.1007/s00500-015-1613-9 65. Zou, F., Wang, L., Hei, X., Chen, D., Jiang, Q., Li, H.: Bare-bones teaching-learning-based optimization. Sci. World J. 2014, 17p (2014). Article ID 136920. https://doi.org/10.1155/2014/ 136920 66. Ghasemi, M., Taghizadeh, M., Ghavidel, S., Aghaei, J., Abbasian, A.: Solving optimal reactive power dispatch problem using a novel teaching–learning-based optimization algorithm. Eng. Appl. Artif. Intell. 39, 100–108 (2015) 67. Wang, L., Zou, F., Hei, X., et al.: A hybridization of teaching–learning-based optimization and differential evolution for chaotic time series prediction. Neural Comput. Appl. 25, 1407 (2014). https://doi.org/10.1007/s00521-014-1627-8 68. Ghasemi, M., Ghanbarian, M.M., Ghavidel, S., Rahmani, S., Moghaddam, E.M.: Modified teaching learning algorithm and double differential evolution algorithm for optimal reactive power dispatch problem: a comparative study. Inf. Sci. 278, 231–249 (2014) 69. Zou, F., Wang, L., Chen, D., Hei, X.: An improved teaching-learning-based optimization with differential learning and its application. Math. Probl. Eng. 2015, 19p (2015). Article ID 754562. http://dx.doi.org/10.1155/2015/754562 70. Dib, F., Boumhidi, I.: Hybrid algorithm DE–TLBO for optimal H∞ and PID control for multimachine power system. Int. J. Syst. Assur. Eng. Manag. (2017). https://doi.org/10.1007/s13198016-0550-z

30 A Hybrid TLBO Algorithm by Quadratic Approximation …

341

71. Turgut, O.E., Coban, M.T.: Optimal proton exchange membrane fuel cell modelling based on hybrid teaching learning based optimization–differential evolution algorithm. Ain Shams Eng. J. 7(1), 347–360 (2016) 72. Lim, W.H., Isa, N.A.M.: Teaching and peer-learning particle swarm optimization. Appl. Soft Comput. 18, 39–58 (2014) 73. Lim, W.H., Isa, N.A.M.: Bidirectional teaching and peer-learning particle swarm optimization. Inf. Sci. 280, 111–134 (2014) 74. Cheng, T., Chen, M., Fleming, P.J., et al.: A novel hybrid teaching learning based multi-objective particle swarm optimization. Neuro Comput. 222, 11–25 (2017) 75. Azizipanah-Abarghooee, R., Niknam, T., Bavafa, F., Zare, M.: Short-term scheduling of thermal power systems using hybrid gradient based modified teaching–learning optimizer with black hole algorithm. Electr. Power Syst. Res. 108, 16–34 (2014) 76. Güçyetmez, M., Çam, E.: A new hybrid algorithm with genetic-teaching learning optimization (G-TLBO) technique for optimizing of power flow in wind-thermal power systems. Electr. Eng. 98, 145 (2016). https://doi.org/10.1007/s00202-015-0357-y 77. Chen, X., Bin, X., Mei, C., Ding, Y., Li, K.: Teaching–learning–based artificial bee colony for solar photovoltaic parameter estimation. Appl. Energy 212, 1578–1588 (2018) 78. Tefek, M.F., U˘guz, H., Güçyetmez, M.: A new hybrid gravitational search–teaching–learningbased optimization method for energy demand estimation of Turkey. Neural Comput. Appl. (2017). https://doi.org/10.1007/s00521-017-3244-9 79. Huang, J., Gao, L., Li, X.: An effective teaching-learning-based cuckoo search algorithm for parameter optimization problems in structure designing and machining processes. Appl. Soft Comput. 36, 349–356 (2015) 80. Huang, J., Gao, L., Li, X.: A teaching–learning-based cuckoo search for constrained engineering design problems. Adv. Glob. Optim. (2015). https://doi.org/10.1007/978-3-319-083773_37 81. Tuo, S., Yong, L., Zhou, T.: An improved harmony search based on teaching-learning strategy for unconstrained optimization problems. Math. Probl. Eng. (2013). https://doi.org/10.1155/ 2013/413565 82. Mahdad, B., Srairi, K.: Optimal power flow improvement using a hybrid teaching-learningbased optimization and pattern search. Int. J. Mod. Educ. Comput. Sci. 10, 55–70 (2018). https://doi.org/10.5815/ijmecs.2018.03.07 83. Mohan, C., Shanker, K.: A random search technique for global optimization based on quadratic approximation. Asia Pac. J. Oper. Res. 11, 93–101 (1994) 84. Ali, M.M., Torn, A., Viitanen, S.: A numerical comparison of some modified controlled random search algorithms. J. Glob. Optim. 11, 377–385 (1997) 85. Venkata Rao, R., Patel, V.: Multi-objective optimization of heat exchangers using a modified teaching-learning-based optimization algorithm. Appl. Math. Model. 37, 1147–1162 (2013) 86. Venkata Rao, R., Patel, V.: Multi-objective optimization of two stage thermoelectric cooler using a modified teaching–learning-based optimization algorithm. Eng. Appl. Artif. Intell. 26, 430–445 (2013) 87. Crepinsek, M., Liu, S.-H., Mernik, M.: Exploration and exploitation in evolutionary algorithms: a survey. ACM Comput. Surv. (CSUR) 45(3), 35 (2013) 88. Civicioglu, P.: Backtracking search optimization algorithm for numerical optimization problems. Appl. Math. Comput. 219, 8121–8144 (2013) 89. Nasir, M., Das, S., Maity, D., Sengupta, S., Halder, U., Suganthan, P.N.: A dynamic neighborhood learning based particle swarm optimizer for global numerical optimization. Inf. Sci. 209, 16–36 (2012) 90. Das, S., Suganthan, P.N.: Problem definitions and evaluation criteria for CEC 2011 competition on testing evolutionary algorithms on real world optimization problems. Jadavpur University, Nanyang Technol. University, Kolkata, India, 2010

Chapter 31

Home Automation Using IoT Shahzadi Tayyaba, Salman Ayub Khan, Muhammad Waseem Ashraf and Valentina E. Balas

Abstract Application based communication and data transfer through the internet is one of the most powerful and advanced methods. Internet facilitates the people to stay connected with each other within the organization and out of the organization. Physical devices and objects used for connection through the internet are called the Internet of Things (IoT). These devices are integrated with wireless routers that permit communication by using cloud services to store, retrieve and analyze the information. Advancements in various technologies have been achieved day-by-day using IoT for online analysis, use and control of sensors, embedded system, and automation. Automated home through IoT technology is called shrewd household. IoT is utilized to monitor and switch the devices and applications. IoT is manufactured and programmed in such a way that it can control vehicle, home appliances, health care devices, wearable and electronics devices remotely. Home automation is an integrated system comprising of illumination, warming, environmental, broadcasting and safety structures. The advantage of a smart home is to save energy automatically by turning off lights and other electronic devices. Security of home is one of the major aspects that can be enhanced by using IoT. Another important application of IoT based home automation is to assist disable and elderly persons. These home systems are equipped with a wireless system (end to end connectivity) to control and monitor the appliances by using packet PC, window PC, and smart phones. Web applications and software are developed to be installed in Smartphone or tablet PC to control and monitor the home appliances remotely. In this book chapter, Some S. Tayyaba · S. A. Khan The University of Lahore, Lahore, Pakistan e-mail: [email protected] S. A. Khan e-mail: [email protected] M. W. Ashraf (B) GC University Lahore, Lahore, Pakistan e-mail: [email protected] V. E. Balas AurelVlaicu University of Arad, Arad, Romania e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_31

343

344

S. Tayyaba et al.

basic ideas related to internet and working of internet, IoT and it’s architecture, embedded system, and automation have been discussed. Further IoT devices and application, home automation and embedded system’s requirements for designing the home automation have been discussed in detail. At the end home automation using IoT and advantages of IoT for the home automation system have been reported.

31.1 Introduction Internet is 5th generation technology to connect the people worldwide. Internet is also called internetworking of networks. Information can be shared, using the internet, inside an organization or outside the organization following the set of rules. Internetworking is the combination of Local Area Network (LAN) and Wide Area Network (WAN) in which switches and routers are required to communicate with each other. The Internet has some host and end system for sharing the information wirelessly with physical topologies. It plans to give some new points of view on IPv6 tending to the potential for real and future needs [1]. Layer architecture is used to design the whole network in the organization that is robust, flexible and interoperable. The set of protocols are required to implement the whole network. Internet architecture is also called the TCP/IP suite. TCP/IP suite is a set of rules used for the internet today. TCP/IP suite is also organized for layer architecture. There are five generalized layers used in TCP/IP suite as shown in Fig. 31.1 [2] and each layer is working at its protocol on the internet. The first layer is a physical layer also called hardware layer in which only host to host communication is required. The responsibility of the physical layer is to transmit information in the form of bits. Physical media is used to carry information Fig. 31.1 Generalized layers in TCP/IP suite

31 Home Automation Using IoT

345

from sender to receiver. A medium can be wired or wireless. The physical layer is actually the lowest layer in architecture and there is no requirement of any special protocol to transmit data. Links connected with routers is called as the link layer. Datalink layer has no specific protocol. Switches are used in the data link layer that are further connected with routers to communicate with other networks either wireless or wired. Data is carried in the form of frame in the data link layer. Error detections and corrections are done in the data link layer. Machines can communicate with each other using media assess control(MAC) and link layer control (LLC). Physical and data link layers use only ethernet protocol underlying the LAN and WAN technology. Internet layer is also called as network layer. Routers are used in the network layer for the host to host communication. A router is an intelligent device used to find the shortest path in the network towards the destination. Routers are required to connect the other network to communicate globally outside the organization. The internet protocol (IP) is defined for the internet layer that carries information and delivers to the destination. IP has header and information in which header contains source and destination address. IP is a connectionless protocol that does not provide flow and error control. Some other protocol are used to provide support in the transmission of data such as Internet Control Message Protocol (ICMP), Internet Group Management Protocol (IGMP), Dynamic Host Configuration Protocol (DHCP)and Address Resolution Protocol (ARP). Fragmentation is done on the transport layer. Message received from application layer is broken into bytes that are combined in a group. This group is transmitted to destination in the form of the segments. Transport layer is used to transmit segments to the destination. End to end communication is performed using port numbers. The transport layer is responsible to carry the information towards destinations safely. Transport layer has two protocols named as Transmission Control Protocol (TCP) and User Datagram Protocol (UDP). The TCP is connection-oriented protocol and provides error control, flow control and acknowledgment. While UDP is connectionless and faster than TCP. End to end communication is used in the application layer. Messages are exchanged by the application layer. It supports the process to process communication. The application layer is the user end layer in which the developer design the rule to drive the information as per system specification. There are some predefined protocols for application layers on the internet to transmit information like world wide web, hypertext transfer protocol, hypertext transfer protocol secure, secure shell, tele network, domain name system, simple mail transfer protocol, message query telemetry transport and file transfer protocol. The complete basic network architecture is depicted in Fig. 31.2 [2]. Open system for interconnection (OSI) model is a layered architecture used to design and develop the structure of the internet. There are seven layers in the OSI model. Two models have been conceived to characterize PC organized tasks that are the TCP/IP convention suite and the OSI model [2]. Some pleasing advancements of sustainable power source situating to vitality web are likewise examined, for example, ideal dispatch, coordinated vitality framework, showcase utilization, etc., to give a reference to the improvement of future power network activity and settlement of

346

S. Tayyaba et al.

Fig. 31.2 Complete network architecture

sustainable power source [3]. A consolidated power gas-heat energy internet planning technique has been developed and study demonstrate by catching inexhaustible power vulnerability, the framework cost can be extraordinarily decreased. The vitality and gas stockpiling can incredibly provide support for lessening the expense of sustainable power vulnerability [4]. The points of interest and weaknesses of microgrid and virtual power plants, the similitudes and contrasts of microgrid and virtual power plant have been discussed [5]. The coordination of the thermo-electric framework has been investigated, and thermo-electric joint task structure and system dependent have developed [6]. Five layers are the same as in TCP/IP suite but the application layer is separated into three further layers as shown in Fig. 31.3 [2].

31.2 Internet of Things Physical electronic devices that can communicate with each other for exchanging useful information by using the internet without the interception of a human being is called the Internet of Things. Meaningful information exists on the internet and human beings can get and share information. Now a day’s human has less time and attention to get information about things in real-time. Technology is increasing day by day. Because of the Internet of Things,

31 Home Automation Using IoT

347

Fig. 31.3 OSI model layer architecture

cost and time have been reduced to track and get the information in the real world. The strategy for gathering target information from the Internet, for example, sustainable power sources information, atmosphere information, and monetary information, and the use of the information for vitality Internet arranging. The strategy for gathering and preparing fluffy data from the Internet is discussed also. The occasions identified with the vitality Internet arranging can be seen and be connected to the multi-situation arranging of vitality Internet. In addition, the challenges and viewpoints of the users are given [7]. A Cloud-centric vision for the overall usage of the IoT has been discussed. The key empowering advances and various application are going to convert into IoT have studied [8]. An all-encompassing system that joins various parts from IoT models/structures has been discussed, so as to productively incorporate savvy home objects in a cloud-driven IoT based arrangement [9]. Internet of Things is not only electronic devices that are connected on the internet. Internet of Things is the latest technology to form a system that is able to sense and answer the movement of information without interruption of human. The preferences and drawbacks of IoT have been broken down and furthermore how current methodologies are guaranteeing principal and essential security imperatives and verifying intercommunication of IoT, alongside the moving changes and extension for work in this field in the coming future [10]. Break down the qualities and difficulties of DNS (domain name system) when it is utilized to help pervasive IoT. The advantages and disadvantages of the DNS in fulfilling with regards to the future development of the IoT condition [11]. Thorough audits and prospects were made dependent on

348

S. Tayyaba et al.

the foundation and ebb and flow inquire about the status of Energy Internet. Additionally, the central idea, fundamental structure, and task method of Energy Internet have been discussed and examined [12]. A shrewdness home structure dependent on the Internet of Things (IoT) has been planned and executed. Investigated the present state of IoT. SOA (service-oriented architecture) and part innovation have been developed and applied. Besides, programming design and fundamental modules have been discussed in detail [13]. Because of the absence of programmed key administration support, IoT gadgets either wind up utilizing preshared keys or no security by any means. Utilizing TinyIKE, take care of the key foundation issue for various IoT conventions utilizing a solitary IKEv2based arrangement and actualize TinyIKE for asset obliged IoT gadgets that run the Contiki OS. The TinyIKE usage supports full declaration based IKEv2 that utilizations elliptic bend cryptography. So as to guarantee the practicality of TinyIKE in the IoT, TinyIKE utilizing an arrangement comprising of genuine IoT equipment has been assessed in detail [14]. Web of Things based design covering terminal, passage, arrange, stage, application, security and different abilities have developed and implemented. Furthermore, structure a full-administration pervasive power Internet of Things standard framework, has been defined as related key specialized norms, manage the arranging, plan, and development of the power Internet of Things [15]. Intuitive key administration convention and a non-intelligent key administration convention to limit the correspondence cost of the things have been discussed. The security investigation demonstrates that the developed design is flexible to different kinds of assaults [16]. The new calculation has been intended for putting away geographic spatial data in IoT gadgets and the Cloud safely while protecting the current organization [17]. PUF (physical unclonable function) based Authentication Scheme (PAS) alongside session entering so as to guarantee the safe connection among keen gadgets in IoT and have been additionally proposed the enlistment and verification component alongside session keying dependent on test reaction pair [18]. The architecture of the internet of thing mainly consists of three parts as shown in Fig. 31.4 [2].

Fig. 31.4 Architecture of IoT

31 Home Automation Using IoT

349

31.2.1 Sensors/Electronic Devices Sensors are used to sense the signals and transform these signals in the required format. There is various type of sensors used to gather information like a sound sensor, humidity sensor, fog sensor, and smoke sensor. Electronic devices that are connected on the internet like a laptop, computer, tablet pc, mobile phone and security system other appliances using a wireless connection.

31.2.2 Data Processing A system received data from sensors and processed this information is called data processing system. Embedded systems are required to process the data and extract the useful information and send this extracted information to the database that is a cloud-based system or server.

31.2.3 Cloud-Based System Applications are developed to monitor and control the devices. The information gathers in a server that is called cloud server. Displayed a sharp Cloud Computing framework that keeps running on underutilized PCs inside an association/network. This framework showed that the “no server farm” arrangement, in fact, works [19]. The unpredictability of satellite information preparing with the arrangements has been examined that distributed computing can offer. Different cases utilized that featured the requirement for brought together access to information between handling assets with confinement imperatives, for adaptable capacity and conveyed figuring abilities, and for equipment streamlining to limit the two expenses and preparing time [20]. Application is linked with the cloud server to share and retrieve the information real-time anywhere in the world. A lot of a smart research facility the board framework dependent on the Internet of things has been structured. The equipment stage of this framework is STM32 small scale controller, receives the WIFI savvy control module, RFID-RC522 card peruser, utilizing Android/Java language to create raspberry 3. As indicated by the data, the framework would sign the information in the cloud database. Accomplished a proficient research insightful data the board framework dependent on the Internet of things [21]. A safe and productive kNN characterization calculation has been built up that masked the subsequent class name and information access designs [22]. Distributed computing is a mutual pool of processing assets prepared to give an assortment of figuring administrations at various levels, from essential foundation to most refined application administrations, effectively assigned and discharged with insignificant endeavors or specialist co-op

350

S. Tayyaba et al.

cooperation. Practically speaking, it figures out how to process, capacity, and correspondence assets that are shared by different clients in a virtualized, what’s more, disengaged condition. IoT and savvy home can profit by the wide assets and functionalities of cloud to make up for its impediment away, handling, correspondence, support in pick request, reinforcement and recuperation. For instance, the cloud can bolster IoT supervision the board and satisfaction and execute corresponding applications utilizing the information created by it. A shrewd home can be dense and center just around the fundamental and basic capacities thus limit the neighborhood home assets and depend on the cloud abilities and assets. Savvy home and IoT will concentrate on information gathering, fundamental preparing, and transmission to the cloud for further handling. To adapt to security challenges, the cloud might be secluded for exceptionally verified information and the general population for the rest. Processing assignments can be either executed on the IoT and savvy home gadgets or redistributed to the cloud. An impressive framework has been developed that are brought together, a property based support module that used Attribute-Based Encryption and grants for designated safecontact to patient records. This framework moves the bearing the board overhead from the patient to the helpful affiliation and grants straightforward task of cloud-based EHR’s passage pro to the restorative providers. Where to process relies upon the overhead tradeoffs, information accessibility, information reliance, a measure of information transportation, interchanges reliance, and security contemplations [23]. From one perspective, the triple processing model including the cloud, IoT and savvy home, ought to limit the whole framework cost, as a rule with more spotlight on decreasing asset utilizations at home. Then again, an IoT and savvy home processing administration model ought to improve IoT clients to satisfy their interest when utilizing cloud applications and address complex issues emerging from the new IoT, savvy home what’s more, cloud administration model. Internet of Things is emerging technology used in all aspect of life. Internet of thing made life easy because of a device to device communication. A working framework for distributed computing has been designed which is called vStarCloud and executed a model to demonstrate that approach is promising. Normal issues that have been investigated about for the cloud condition can be successfully fathomed by POSIV (Portable Operating System Interface of vStarCloud) interface [24]. The inside and out examination of Load Balancing Algorithms have been developed. The Load of Cloud Balancing is a procedure of reassigning the all-out burden to the individual hubs in a given system. At that point, the Comparative investigation of burden offsetting calculations with its quality measurements has condensed [25]. The RECAP (reality capture) test system and related models offered help for comprehension and foreseeing the effect on assets, remaining tasks at hand, and nature of administration (QoS) measurements just as exchange offs for energy effectiveness and cost inside the cloud, edge and haze figuring situations, while keeping up the administration level understandings (SLAs) of clients [26]. The mCloud model structure has been planned and actualized. Genuine trials directed on the actualized framework to assess the presence of the calculation. Results showed the framework and implanted choice calculation can give choices on choosing remote medium and cloud assets dependent on the diverse set of the cell phones and accomplish a noteworthy decrease on make

31 Home Automation Using IoT

351

span and vitality, with the improved administration accessibility when contrasted and existing offloading plans [27]. A three-pronged grouping and examination system discussed for merchant stages and applications. A scope of explicit agent’s improvement concerns like design, programming, and quality has been discussed [28]. The effect of the all-inclusive cloud studied on present communication and system supervision architecture of the mist. Advancements have inspected that executed these models and designs, and dissect them as for safety and power prerequisites. Besides, ways to deal with safety and flexibility related systems have examined in the mist (explicitly, abnormality location and strategy based strength the board) [29]. A system has been developed to help the request. A sheltered posting once-over is also complete to rank the planned outcomes while taking the assurance and characterization of the buyer information and redeemable the compensations of the client phones. Sweeping examinations have been done and the outcomes exhibit that the projected arrangement is beneficial and fitting for a safe available disseminated storing structure [30].

31.3 Embedded System A system in which all components work together with a fixed plan. A digital clock is a system in which all units follow the same rule to display time and date. Digital Clock does not work if one of its components fails to perform. All units in a system depend on each other. An inserted framework programming model has been created to depict programming process and inside and out the component, so as to give the direction of security controlling and programming wellbeing advancing [31]. The service-oriented architecture has portrayed for train hostile to impact framework utilizing installed multiprocessor system-on-chip. The improved design of the recently created framework has been implemented, and essential conditions for fruitful work has characterized, and the benefits of utilizing the multi-processor framework structure have been exhibited [32]. Embedded means to attach other things. Software is embedded in computer hardware system is called an embedded system. It can be a microcontroller or microprocessor-based system to achieve a precise duty. It comprises of three basic components. • Hardware • Software • The real-time operating system that is used to control the application software and applies a set of rules while execution of the program. RTOS is not required for small scale embedded system. Embedded system preforms only single task and not repeat the same function. A smoke detection system is the example of an embedded system that performs a single function to detect the smoke. Microcontroller based system or control system cannot be multitasked. These systems are pre-preprogrammed systems. There is no compromise in cost, size, and power while designing an embedded system. All components

352

S. Tayyaba et al.

must be assembled in single-chip and processes each task very fast with less power. Embedded system can compute result in online without any delay. Embedded system has built-in memory. There is no need to attach any external memory. Input and output devices can be connected with the embedded system. Installed frameworks have been structured and actualized for client recognizable proof and access to various items and administrations utilizing a cell phone are considered, proposed and depicted. This frameworks are applying diverse correspondence advances and inserted gadgets and are important for various distinguishing proof applications. The developed, structured and executed implanted framework arrangement, utilizing shrewd cell phones and inserted gadgets with virtualized server stages, for client recognizable proof in access to vehicle open parking structures reasons for existing has exhibited. Information trade advances between shrewd cell phones, unified framework, just as equipment and programming stages, used to acknowledge stopping booking and access in such a distinguishing proof framework have considered. Gotten results alongside created Web and portable applications, just as installed applications for interfacing shrewd cell phones have inserted gadgets and virtualized server stage has portrayed [33]. Software system provides the flexibility and hardware provides performance as well as security in the embedded system. Embedded system can be time consumed in term of project development. An ongoing framework has structured which is fit for perceiving four signals that associate to human feelings dependent on the arm developments. The 3D skeleton utilizing Kinect v2 sensor has characterized utilizing an SVM strategy. The framework has tried progressively on a Kinect database with the implanted framework utilizing an upgraded calculation for skeleton extraction continuously [34]. The basic architecture of the embedded system can be shown as in Fig. 31.5 [2].

Fig. 31.5 Architecture of embedded system

31 Home Automation Using IoT

353

31.3.1 Microcontroller The microcontroller is a single chip which performs a specific task. It has limited capabilities and increases the input-output process with a number of functional units. The microcontroller is used in an embedded system to compute the result in a realtime environment with memory and devices. The microcontroller is used for special purpose systems that perform a single task. There is no need to add external memory. Input and output peripherals can be attached with a microcontroller in the embedded system. Less power required to operate the microcontroller. The microcontroller takes input and process the data as per given instruction. The microcontroller contains input and output ports, central processing unit and memory unit. The microcontroller processes the data in digital format.

31.3.2 Sensor Sensors are used in embedded system to detect the change in environment and transform into an electrical signal. Sensors convert one form of energy into another form. Sensor data can be stored in the memory unit. Various IoT sensors have been analyzed and discussed their utilization as a quantum chip used in smart cities management systems [35]. Marker plan and a calculation have developed to recognize the markers under various encompassing conditions, with a long-range to be executed on installed frameworks with low computational prerequisites. The strategy decreased the current issues in the best in the class identified with the utilization of various situations and conditions, for example, various separations or distinctive light. Additionally, the requirements of the technique are being utilized negligibly to decrease the expense of arrangement [36]. The implanted framework has developed and actualized on GaneshBlue, including usage of the control system in coasting development. Installed framework intended to suit a few subsystems in the GaneshBlue gadget, and coordinated it into a greater complete framework. Frameworks have been worked with multithread capacities as certain subsystems like correspondence and control ought to be worked in a tight time limitation [37]. The joint advancement issue with vitality proficiency and viable asset usage explored for heterogeneous and circulated multi-center installed frameworks [38]. An effective arrangement has been designed to augment secrecy, while likewise ensuring the planning necessities of constant applications on shared stages [39]. A tale approach has been actualized that representing the effect of on-chip memory (or reserve) reuse on the presentation of control applications persuades new procedures for the structured control calculations. This prompts noteworthy improvement in the nature of control for given asset accessibility, or increasingly effective executions of implanted control applications [40]. The architecture has designed that contrived a strategy called PPA-LTF to oversee pinnacle control utilization. This strategy averted task execution that devoured higher power as indicated by the undertakings’ capacity follows. The exploratory

354

S. Tayyaba et al.

outcomes show that this architecture gives up to 50% (by and large by 39%) top power decrease contrasted with cutting edge plans [41]. The overhead has investigated in utilizing the generally utilized document allotment table record framework and developed an altogether quicker, littler impression, and henceforth a lower power document framework named SlimFS. The designed framework has an application to low power installed shopper applications, explicitly battery-driven wearable gadgets for human services and “green” electronic frameworks [42].

31.4 Automation Automation is a technology to control and monitor the process of production and distribution of goods in the industry. Automation is used in every aspect of life. Automation has minimized the interference of human during the manufacturing process of goods. The automatic system is used nowadays to minimize the time and less possibility of errors. Automation has three common types as given below: • Simple Automation • Variable Automation • Extended Variable Automation. Simple automation is processing things with a given set of instruction. Low level of automation in which no need of any change required in processing. Equipment configured once and processing the goods. Simple automation is also called fixed automation and some key points of simple automation are vary in term of costly, the higher rate of production and do not provide flexibility while changing equipment. Variable Automation is also called programmable automation. The program controls the processes of production. The system can be interpreted by a set of instruction while configuring the equipment. If a new product is required then the new set of instruction can be used in equipment for the new process. The characteristics of variable automation are the high cost of equipment, low production rate, most appropriate for reprogrammable while new process, fit for batch production like robotic technology used in the industry. The extended form of variable automation is flexible automation. This system is more flexible than programmable automation. There is no loss of time if one process of production change to another product. When reconfiguring the system for a new product then there is no loss in the production time. This system can produce different products in the same batch. Electrical, electronic and mechanical devices combined with a computer to achieve the automation. All techniques employed in a complex system like industries and aircraft. The advantage of automation is to save the cost of labor, electricity, materials and increases the value of goods and precision. Information technology involves in two forms as automation and user end to visualize the process also a communication tool with digital media. Automation falls in various categories like good processed by the central controller this is called orchestration, independently processes the events called

31 Home Automation Using IoT

355

choreography, the system that depends on human not fully automated and help to improve the workflow of the manufacturing, Actionable events can be examined by the user interface, equipment automation is to process the goods, another advancement of automation is sensor technology used to monitor and control the process and artificial intelligence is used for decision making. The job of constancy appraisal for remote systems in mechanical mechanization frameworks specifically concerning future adaptable creation procedures have been discussed [43]. A variable advance size LMS (VSSLMS) is connected to the procedure controlling of gas parts discovery. Hypothetical reproduction and tests have completed demonstrating that VSS-LMS is of preferred execution over custom calculations [44]. The dispersion utilities to settle on insightful choices, empowering quick and proficient reactions to support intrusion. The robotization gave on Fault Detection, detachment and power rebuilding (FDIR) over the displayed manual task of feeder and substation at IISC (Indian Institute of Science) grounds. A commonplace ring fundamental unit at the dispersion framework used to give continuous power supply to ideal conditions [45]. Roaming is a characteristic that permits such remote gadget portability between base station cells, however cyclic procedure information correspondence between PLC on the one side and the meandering remote gadget on the opposite side gets normally interfered with during the handover procedure, not permitting ongoing activity during this handover stage [46]. RFID based prepaid vitality meter and home computerization framework with an android application running at the supporter’s portable station has been designed and implemented. The vitality meter can be credited through RFID innovation and the robotization of various apparatuses can be constrained by the supporter by means of an android versatile application. The endorsers can check their present credit, devoured units, current burden and they can oversee machines remotely. A heap the administrator’s framework has additionally discussed, when the heap surpassed a specific characterized level then machines consequently closed down. A programmed robbery identification instrument has been set up for the framework [47]. A home security framework is a key concern that will be remote. Security over a system has accomplished utilizing AES (advance encryption standards) encryption. The security of the house is overseen by sending warnings to the client utilizing the Internet if there should arise an occurrence of any trespasser and it can likewise ring a caution whenever required. Home computerization is used by utilizing proper sensors introduced around the house. Raspberry pi utilized as a server and controller. Raspberry pi has the errand of controlling electrical machines and giving confirmation and security to the client [48]. Codesys IEC 61,131-3 programming has been utilized to change over a Raspberry pi as a delicate PLC. Simple and advanced parameters from the field have been observed and controlled utilizing GPIOs pins, interfacing transfers and correspondence ports accessible in raspberry pi. The computerization groupings for control observing and assurance of the framework are characterized utilizing stepping stool outline programming in the IEC 61131-3 Codesys advancement stage. UI (user interface) and HMI (hardware machine interface) screens have been designed utilizing the Codesys web representation fragment which can be gotten to

356

S. Tayyaba et al.

both locally just as over a wired or remote system [49]. An essential value for accumulation and recovery of use information includes executed inside the current home robotization framework. Information accumulation modules have actualized, which keep running on the home computerization portal and inside the home mechanization cloud, and enabled us to associate with the officially existing enormous information middleware stage [50]. Different possibilities of implemented technology have been discussed in automation to reduce the cost, time, and labor [51].

31.5 IoT Devices and Applications Devices that are connected wirelessly with internet and transmitting data called the Internet of Things. It encompasses the connectivity of the internet for devices like laptops, mobile phones, PC, tablet and other non-internet devices as shown in Fig. 31.6. These devices can communicate and control by using the internet. Various technologies embedded in devices for remotely monitored and controlled.

Fig. 31.6 IoT devices

31 Home Automation Using IoT

357

All devices of IoT that can communicate with each other to complete the process intelligently at home and industry. Devices combined with internet can be classified into three categories such that consumer, commercial and industrial. Consumer devices that can be connected with internet. These devices include smart tv, speakers, appliances and wearables devices. Security systems, energy meters, smart buildings, and cities all other devices that are used to monitor and control the traffic and weather conditions are the example of commercial and industrial internet of thing devices. Smart cooling system, smart home are also an example of IoT devices. A smart home is the example of consumer IoT devices in which a user reaches home and garage can be communicated by vehicle to open the gate. Temperature is also adjusted by the thermostat, the intensity of light is very low and data indicated by a pacemaker that it is a tense day. Sensors are used in meeting areas that can help the workers to find the location of meeting and time for the meeting, other features like type and size of meeting place. When employees enter the place, the room temperature automatically adjusted, lights of the room can dim according to the load on the PowerPoint screen and speaker starts the presentation. This is the example of a commercial sensory system for IoT devices. Machinery equipped with sensors in the industry that deliver the sensor data to the machine operator. Sensor data is used to inform the operator about disability and when should replace the machine for a better process. Some alerts generated to notify the user about the issue and where need to solve the issue, avoid to send workers to find out the problems. Various difficulties can thwart the effective arrangement of an IoT framework and its associated gadgets, comprising safety, interoperability, control/preparing capacities, adaptability, and accessibility. Huge numbers of these can be tended to with IoT gadget the panel either by receiving standard conventions or utilizing administrations offered by a seller. Contraption the officials empowers associations to fuse, orchestrate, screen and remotely supervise web-enabled devices at scale, offering the fundamental to keeping up the prosperity, system, and security of the IoT devices along their entire lifecycles. Such key points comprise gadget enlistment, gadget confirmation/approval, gadget design, gadget provisioning, gadget checking, and diagnostics, and gadget investigating. Accessible institutionalized gadget the executive’s conventions incorporate the Open Mobile Association’s Machine Organization and Lightweight D2D. Internet of Things gadget the panel administrations and programming are likewise accessible from merchants. The systems administration, correspondence and network conventions utilized with web-engaged devices, all things considered, depending upon the specific IoT application guided. Similarly, as there is a wide range of IoT applications, there are various network and correspondences choices. Protocols are used for communication consist of the constrained application protocol, datagram transport layer security, and message query telnet transport protocol, among others. Wireless protocol IPv6, low power WAN, Zigbee, LBE, Z-Wave, RFID, and network identification card. Cell, satellite, Wi-Fi, and Ethernet can likewise be utilized. Every alternative has its tradeoffs regarding power utilization, range and data transfer capacity, which must all be viewed as when picking associated gadgets and conventions for a specific IoT application. To exchange the detector’s data they collect, IoT gadgets interface with an IoT door or another edge gadget where data can either be investigated locally or

358

S. Tayyaba et al.

sent to the cloud for examination. The interconnection of customarily unwise gadgets carries up to several problems in connection to safety and shield. As if much of the time the case, IoT advancement has moved more quickly than the instruments open to shield the devices and their customers.

31.5.1 Application of IoT Devices The Fig. 31.7 depicts the application of IoT devices that can be classified into three types. IoT devices are a bit of the greater thought of home computerization, which can fuse illumination, warming, and refrigeration, broadcasting, and safety structures. Whole deal points of interest could fuse imperativeness hold reserves through normally ensuring lights and equipment is diminished. The security-related difficulties and wellsprings of danger in the IoT applications have been discussed. Blockchain, haze figuring, edge registering, and AI (artificial intelligence), to expand the degree of security in IoT have investigated [52]. Various application of smart home based on IoT has been discussed as well as discussed the challenges and utilization of smart

Fig. 31.7 Application of IoT devices

31 Home Automation Using IoT

359

home applications [53]. A wearable Internet of Things (IoT) sensor hub for checking unsafe ecological conditions for security applications by means of LoRa remote innovation. The designed sensor hub is low-power and supports numerous natural sensors. A LoRa based portal can be utilized to interface sensors to the Internet. Examined various parameters such as carbon monoxide, carbon dioxide, bright, and some broad natural parameters. The information would then be able to be shown to approved clients through an electronic application situated in the cloud server and the gadget will offer alarm to the client by means of a portable application when a crisis condition happens [54]. The usage, challenges and cutting edge utilization of lightweight cryptography (LWC) methods for keen IoT gadgets have been discussed, particularly the Long-Range Wide Area Network which is an open standard that characterizes the correspondence convention for LPWAN innovation [55].

31.5.1.1

Consumer Application

A savvy home or electronic home can be created on a step or focus points that switch shrewd appliances and machines. One key application of a savvy home is to give support to those handicaps and old persons. These home architecture can be utilized as supportive modernization to suit the owner’s specific handicaps. A system that is used voice to control and help consumers with a view and versatility constraints while prepared architecture can be linked legitimately. They can likewise be furnished with extra wellbeing things. These things can be incorporated sensors that screen for therapeutic problems, for example, falls or seizures. Three methods of the single controlled footer voltage (SRFV), single directed header voltage (SRHV) and twofold can be managed lower and higher voltage (DRFHV) contrasted with make end in term of intensity sparing. Demonstrate that, twofold directed voltage system can be more productivity power sparing in dynamic mode than three different procedures in maintenance mode [56]. Savvy home invention associated along these lines can provide consumers with more chance and advanced personal satisfaction. Various types of consumer application are depicted as shown in Fig. 31.8.

31.5.1.2

Commercial Application

The Internet of Medical Things is a use of the Internet of Things for restorative and security-related determinations, data accumulation and investigation for research, and observing. it is likewise called, prompted the assembly of a digital structure, linking reachable therapeutic assets and medicinal services administrations. IoT gadgets can be used to permit wireless checking and notice architecture. Nobel NFV empowered IoT design focused for a best in working room condition. Web administrations dependent on the authentic state move web engineering as the IoT application’s southbound interface and represent its relevance by means of two distinct situations [57]. These gadgets can encompass from cardiac and rhythm shows to gadgets prepared for

360

S. Tayyaba et al.

Fig. 31.8 Consumer application

examination exact inserts. A few medical clinics have implemented “keen beds” that can identify when the patient wants to get up. It can likewise adjust itself to assurance appropriate weight and provide support to the patient in backing without the manual collaboration of therapeutic caretakers. Particular sensors can likewise be designed inside the home to show the security and general wealth of older residents, while furthermore assuring that suitable treatment is being controlled and serving persons to recover by treatment. Imagine utilizing exercises gained from setting mindful registering, explicitly setting sharing among related vertical IoT applications to address this defer necessity of such bound together IoT applications by sanctioning setting sharing among Fog hubs for limiting framework delay [58]. Shrewd sensors are used to make a smart system that can collect, process, transfer, and breakdown important data in several conditions, such as interfacing in-house seeing appliances to emergency medicsstructure. Extra customer gadgets to help the sound existing, such as wearable displaying device for heart, are additionally acceptability with the IoT. Start to finish security observing IoT levels are additionally handy for reproductive and never-ending patience, serving one manage wellbeing strength and reiterating drug prerequisites. The utilization of the Internet of Things in people facilities a major job in control constant illnesses and in illness aversion and control. The wireless examination is conceivable over the association of incredible wireless arrangements. The network empowers wellbeing professionals to catch patient’s information and applying complex methods in wellbeing information investigation. The Internet of Things can support in the reconciliation of interchanges, switch, and information formulating using various conveyance architecture. The utilization of the Internet

31 Home Automation Using IoT

361

of Things gives out to all parts of conveyance architecture. The active connection among these parts of an automobile structure allows among and within the vehicular correspondence, astute road traffic switch, keen departure, automatic toll collecting architecture, coordination, and fleet administration, automobile control, security, and road assistance. In Coordinations and Armada Management, for example, an IoT phase can perpetually show the region and conditions of cargo and assets by methods for remote sensors and send unequivocal cautions when any special cases happen. This must be plausible with the Internet of Things and its accessibility between appliances. Sensors such as a global position system, Moisture, and Temperature transmit data to the internet of thing phase and so that, the data is surveyed and sent to the customers. Clients can follow the ongoing status of vehicles and can settle on proper choices. FPGA based novel IoT get to design and system on chip (SoC), which gave bound together access to the IoT for a wide assortment of low-speed and fast gadgets with related extendibility and configurability. The IEEE1451.2 standard has adopted for this structure and connected this plan to the vehicle observing framework [59]. Whenever combined with the advanced technology such as machine learning, at that point it supports in lessening car accident by informing sluggishness serine with chauffeurs and giving identity pushed automobiles. Different types of commercial application of IoT devices figure out in Fig. 31.9.

Fig. 31.9 Commercial application

362

31.5.1.3

S. Tayyaba et al.

Industrial Application

IoT gadgets can be used to show and switch the machine-driven, electrical structures applied in different kinds of the framework as well as in-house and building control architecture. In this explicit condition, three primary sections are being covered: The combination of the Internet with structure vivacity the panel architecture so as to make vivacity operative and Internet of Things ambitious “shrewd buildings”. The potential methods for examination for reducing vivacity consumption and detecting inhabitant performs. An ultra-wideband impulse radio-based indoor situating framework and its engineering has been designed and actualized. The exactness of the best situating of around 30 cm has been accomplished from the actualized framework [60]. The reconciliation of smart gadgets in the accumulated way and realize how generally can be utilized in future applications. A cost-effective, supple and wireless home computerization system has designed and implemented. This system can be utilized to switch the home applications from far away using a mobile application [61]. The IoT can recognize the constant mix of various accumulating gadgets outfitted with sensing, distinguishing proof, management, communication, initiation, and systems administration measurements. In light of such an exceptionally incorporated keen digital-physical space, it opens the entryway to make an entirely different business and market open doors for assembling. Framework switch and the management of accumulating kit, source and condition the panel, or accumulating method control carry the Internet of Things in the area of mechanical things and smart manufacturing also. The Internet of Things architecture empowers speedy manufacturing of new things, taking action to element desires, and ongoing improvement of creation and store network architecture, by server management device, detectors, and switch structures. Progressive switch architecture to robotize process controls, management devices and network data architecture to elevate plant comfort and security includes the area of the Internet of Things. A recent system architecture can be synchronized with brilliant lattices, allowing continuous vivacity streamlining. Approximations, electronic panels, plant streamlining, comfort and security the executive, and various measurements are specified by countless arranged sensors. IoT application design to get to different IoT administrations utilizing mist registering has been designed. Fog computing idea embraced by the application level, the implemented design has demonstrated as an administration portal to numerous IoT administrations for IoT applications [62]. There are different forms of application in cultivating such as collecting data about temperature, snow, mugginess, the speed of air, virus incursion, and mudas well. This data can be used to computerize cultivating measures, to improve value and quantity, bound risk and waste, and reduce the action vital to manage yields. Ranchers would now be able to see mud temperature and mugginess from a far distance and even apply Internet of Things attained data to exactness grounding plans. The Internet of Things architecture can be used for detecting any events or variations in secondary situations that can reduce security and increase the hazard. The Internet of Things can yield the business by reducing cost, time reduction, improved value, and reduce the use of paper at the workplace. It can support in taking faster selections and used for analysis with Real-Time Data Analytics. It can be used for

31 Home Automation Using IoT

363

planning fix and support in a productive way, by establishing actions among several professional coops and customers of these offices. Internet of Things gadgets can be used to switch the elementary framework like platforms to give entree to boats. Use of IoT gadgets for observing and occupied framework is maybe working to improve existence and emergency response management, and nature of the organization, less time and lessen expenses of activity in all foundation related zones. Indeed, even areas, for example, squander the panel can profit by robotization and improvement that could be gotten by the IoT. An application planned mindful remaining task at hand distribution plot for edge processing based Internet of Things to decrease the answer time of internet of thing devices request by selecting the part of the cloud for every internet of thing the user’s different kinds of applications and the number of registering resources apportioned for every device in each cloud. Both the system deferral and registering postponement have measured, i.e., the Internet of Things clients’ application is almost certain appointed to nearer and daintily stacked cloud [63]. Ecological observing the usage of the Internet of Things ordinarily use detectors to support natural assurance, by checking the level of air and water, weather and mud situations, and can uniform include areas like checking the growths of the wildlife cycle. Enhancement of quality obliged appliances linked with the Internet furthermore infers that various things like seismic tremor or tsunami early-alerted systems can in like manner be used by emergency organizations to give an undeniably ground-breaking guide. Internet of thing devices in this area ordinarily range a huge geographic zone and can be versatile. Various type of industrial application of IoT devices is shown in Fig. 31.10.

Fig. 31.10 Industrial application

364

S. Tayyaba et al.

31.6 Home Automation Shrewd home is the private expansion of structure computerization and includes the control and mechanization of all its inserted innovation. It characterizes a home that has machines, illumination, heating, refrigeration, television, personal computer, excitement frameworks, enormous home machines, for example, washers/dryers and coolers/coolers, security, what’s more, camera frameworks equipped for speaking with one another and being controlled, remotely by a period plan, telephone, portable or web. These frameworks encompass of controllers and detectors linked with the main focus point constrained by the home inhabitant utilizing divider mounted terminal or versatile unit associated with web cloud services. Internet of things (IoT) based structure has created for checking the action of customers inside the home system. The Elgar structure handles the administration of movement by means of IoT benefits in an IoT domain with different gadgets. The exhibition assessment pointed out that the structured framework can vigorously distinguish the exercises utilizing IoT in a savvy home condition with high precision [64]. A savvy home gives security, vitality effectiveness, low working expenses, and comfort. Establishment of shrewd items gives comfort and investment of time, cash and vitality. Such frameworks are versatile and movable to meet the progressing changing needs of home inhabitants. As a rule, its foundation is adaptable enough to incorporate with a wide scope of gadgets from various suppliers. The essential engineering empowers estimating home conditions, process instrumented information, using microcontroller empowered sensors for estimating home conditions, what’s more, actuators for observing home implanted gadgets as shown in Fig. 31.11. The prevalence and entrance of the savvy home idea are developing at a decent pace, as it turned out to be a piece of the modernization and decrease in cost patterns. This is accomplished by inserting the capacity to keep up a brought together occasion log, execute AI procedures to give principle cost components, sparing proposals, and what’s more, other helpful reports.

Fig. 31.11 Simple structure of home automation

31 Home Automation Using IoT

365

A run of the mill keen home is outfitted with a lot of sensors for estimating home conditions, for example, temperature, moistness, light, and vicinity. Every sensor is devoted to catching at least one estimations. Temperature and mugginess may be estimated by one sensor, different sensors figure the light proportion for a given region also, the good ways from it to each item presented to it. All sensors permit putting away the information and picturing it so the client can see it anyplace and whenever. To do in this way, it incorporates a sign processor, a correspondence interface and a host on a cloud framework. Makes the cloud services for overseeing home apparatuses which will be facilitated on a cloud foundation. The overseeing administration permits the client, controlling the yields of savvy actuators related to home apparatuses, for example, lights and fans. Shrewd actuators are gadgets, for example, valves and switches, which perform activities, for example, turning things on or off or modifying an operational framework. Actuators give an assortment of functionalities, for example, on/off valve administration, situating to rate open, adjusting to control changes on stream conditions, crisis shutdown. To initiate an actuator, an advanced compose order is issued to the actuator. Various technologies have been discussed to design a wireless home automation system and analyzed technologies based on the microcontroller to control, monitor and secure the home. This system provides support to elderly persons [65]. Home access innovations are regularly utilized for community entryways. A typical framework utilizes a database with the distinguishing proof traits of approved individuals. At the point when an individual is moving toward the entrance control framework, the individual’s distinguishing proof qualities are gathered right away and contrasted with the database. On the off chance that it coordinates the database information, the entrance is permitted, something else, the entrance is denied. For a wide conveyed establishment, we may utilize cloud service for midway gathering people’s information and preparing it. Some utilization attractive or vicinity ID cards, other uses face acknowledgment frameworks, fingerprint, and radiofrequency identification. Sensors are used to gather inside and outside home information and measure home conditions. These sensors are associated with the home itself and to the appended home gadgets. These sensors are not the web of things sensors, which are appended to home machines. The sensors’ information is gathered and persistently moved by means of the nearby system, to the savvy home server. Processors for performing special and incorporated activities. It might likewise be associated with the cloud for applications requiring broadened assets. The sensor’s information is then handled by the nearby server forms. A gathering of programming segments wrapped as APIs, permitting outer applications to execute it, given it pursues the pre-characterized parameters group. Such a the programming interface can process sensors information or oversee fundamental activities. Actuators to the arrangement and execute directions in the server or other control gadgets. It makes an interpretation of the expected move to the direction sentence structure; the gadget can execute. During preparing the got sensors’ information, the task checks if any standard turned out to be valid. In such a case, the framework may dispatch orders to the appropriate gadget processor. The database is used to stock the prepared data gathered from the sensors. It will likewise be utilized for information investigation, information introduction, and perception. A home computerization architecture

366

S. Tayyaba et al.

will switch the light, environment, refrigeration architecture, and machines. It might integrate home safety such as to get to switch and attentive system. At the point when linked with the Internet, devices are a significant component of the Internet of Things. A home robotization architecture generally interfaces measured the gadgets to a main focus or “portal”. The user interface is used to control the system uses either tablet or personal computers, mobile phone-based, web-based application, that may be open from far using the Internet. Various things in-home or building can be controlled. Home automation system based on microcontroller. Different systems are connected with microcontroller. Microcontroller programmed in a sense to control and monitor the things in the home. LCD is used as an output device to show the status of the appliance. The extra feature of automation is used as flash memory to check the status of things off line if required. Air conditioning is a significant piece of private structures such that, individual personal homes, condo architecture, inns, and older existing workplaces, average to enormous places of business, like high buildings and hospitals. Ventilating is the best approach to exchanging or expelling the air in any space to give high indoor air quality which incorporates temperature switch, oxygen reestablishment, and clearing of mugginess, scents, burn, heat, dust, floating organisms, and various gases. Ventilation empties bothersome fragrances and over the top clamminess, exist outside air, keeps inside the framework air orbiting, and maintains a strategic distance from the stagnation of within air. A lighting switch framework is a canny structure that depends on a lighting system that imparts among different wellsprings of information and produces the identification with lighting control with the use of central preparing devices. Lighting control architectures are commonly used in both inside and outdoor lighting of business, current, and private spaces. Lighting control structures serve to give the ideal proportion of light where and when it is required. The significant favorable place of a lighting switch structure on independent illumination panels or traditional manual switching is used to control separate illuminations or gatherings of illuminations from a private UI appliance. This capacity is used to switch various illuminations wellspring from a user gadget enables composite illumination sections to be made. A room may have various scenes pre-set, everybody made for different activities in the room. A critical bit of leeway of lighting control systems is reduced imperativeness use. Longer light life is moreover gotten when lessening and killing lights when not being utilized. Wireless illumination switch architecture gives additional points of interest including diminished foundation costs and extended versatility over where switches and sensors may be set. An indoor movement distinguishing gadget is used to recognize the nearness of an individual to naturally control lights or temperature or ventilation frameworks. The sensors utilize infrared, ultrasonic, microwave, or other innovation. The term pocket gadgets as various as PIR sensors, lodging keycard bolts, and shrewd meters. Inhabitance sensors are ordinarily used to spare vitality, give programmed control, and agree to construction laws. An electrical matrix that integrates a variety of action and vivacity procedures with shrewd meters, shrewd devices, maintainable power source, and vivacity cost. Electronic

31 Home Automation Using IoT

367

power molding and control of the generation and dispersion of power are significant parts of the shrewd matrix. Take off of keen framework innovation likewise infers a key re-building of the power administrations industry, albeit run of the mill use of the term is centered around the specialized foundation. An electronic gadget that records the operation of electrical vivacity and conveys the data to the power provider for checking and charging. Shrewd meters normally record vitality hourly or all day and generates the report every day. It empowers two-way correspondence among the meter and the central framework. Exchanges from the meter to the system can be wireless or wired, like programmable logic control. Remote correspondence choices in like manner use incorporate cell interchanges, Wi-Fi, wireless work stations, low power wireless system, ZigBee. The objective of the system is to distinguish disruption, unauthorized passage—into a framework or other zone. Security cautions are used in remote, commercial, mechanical, and armed assets for assurance against theft or harm the assets, just as near to house insurance against interlopers. Security cautions in local locations demonstrate a relationship with diminished robbery. Vehicle alerts in like manner help to safe automobiles and their material. Detainment facilities likewise use safety structures for the control of theft. Interruption caution system may be combined with close-circuit TV inspection to catch the movements of interlopers and may interface to switch architecture for an electric bolted gate, System is used to work independently with PC examination and regulate. It can be a two-way process that permits communication among the panel and the control system. Electronic entree switch uses personal computers to explain the restrictions of solid keys. A wide scope of accreditations can be used to displace solid keys. The electronic entree switch architecture awards entree dependent on the certification introduced. When the entree is without a doubt, the gate is unlocked for less time and the data is recorded. If the entree is can’t, the gate remains locked and the unauthorized entree is logged. The system will show the gate and aware if the gate is constrained to unlock or held unlock too long subsequent to being opened. The type of home robotization centers around making it workable for more seasoned grown-ups and individuals with handicaps to stay at home, sheltered and agreeable. Home robotization is turning into a practical alternative for more seasoned grownups and individuals with incapacities who might like to remain in the solace of their homes as opposed to move to a social insurance office. This field utilizes a great part of a similar innovation and gear as home robotization for security, amusement, and vitality preservation however tailors it towards more seasoned grown-ups and individuals with handicaps as shown in Fig. 31.12.

31.7 Embedded System for Home Automation Various components can be integrated together to accomplish a precise task is called an embedded system. It is used to automate the things in the home with less human efforts. Hardware components are connected with the microcontroller also embedded with software for special functionality. Embedded system is used in various areas

368

Fig. 31.12 Home automation system

S. Tayyaba et al.

31 Home Automation Using IoT

369

Fig. 31.13 Simple structure of embedded system

like control and automation, security systems, monitoring and tracking system, smart greenhouse systems and smart energy management system. Embedded system is used in home automation to regulate and display the home appliances. Embedded system consists of three basic components. • Hardware • Software • Communication Protocols. The microcontroller is an essential part of any embedded system to automate the things in industry or home as well. The microcontroller takes the decision on the specific set of instructions to perform the specific functionality of an embedded system. Input devices are connected with microcontroller and microcontroller processed the input data to perform the operation. The output displays on output devices as shown in Fig. 31.13.

31.7.1 Hardware Components Smart home system is used to regulate the home applications. A simple block diagram is depicted the input and output control system as shown in Fig. 31.15. A home computerization architecture has the capability of monitoring various parameters like temperature, humidity, light intensity, heat, and ventilation. The main objective of home automation is to monitor and control the appliances. Hardware components are required to design and implementation of the whole system such that sensors, microcontroller, and wireless module as shown in Fig. 31.14. A sensor is a device used to measure stuff, such as pressure, position, temperature, or acceleration, and respond with feedback. Different things can be measured at home and sensors are placed at a different position at home.

370

S. Tayyaba et al.

Fig. 31.14 Architecture of home automation

31.7.1.1

Temperature and Humidity Sensor

A device that is used to measure the room temperature and mugginess in the air are called temperature and humidity sensor. It is constructed by two metal plates. If there is any change in temperature then the electrical voltage produced. The extent of sogginess perceptible all around to the most critical proportion of clamminess at a particular air temperature is called relative moisture.

31.7.1.2

Motion Detection

Motion locator is a device to check the movement of objects, particularly for only persons. Such a contraption is habitually joined as a piece of a structure that normally plays out an endeavor or alerts a customer of development in a domain. This structure a crucial portion of safety, mechanized illumination switch, household switch, vivacity efficiency, and other valued architecture. Different type of technologies is used to detect the moving objects like PIR (passive infrared sensor), microwave sensor, an ultrasonic sensor and TMD (tomographic motion detector). The home model is broke down to exhibit a vitality proficient IoT based brilliant home. A few Multiphysics reproductions have completed concentrating on the kitchen of the home model. A movement sensor with an observation camera utilized as a major aspect of the home security framework. Combined with the

31 Home Automation Using IoT

371

Fig. 31.15 Home automation system

household illumination and heating, airing, and refrigeration switch architecture, the savvy framework can wirelessly switch the illumination and heating or refrigeration when a tenant comes in or leaves the galley [66].

31.7.1.3

Fire Detection

Fire finders sense at least one of the items or sensations coming about because of flame, for example, smoke, warmth, infrared as well as bright light radiation, or gas. The objective of a fire detector is to identify fire. It responds with an alert and disabling the source of the fire such as gas or petroleum line and activating a fire coverup architecture. Exactly when used applications, for instance, mechanical warmers, their main responsibility is to give confirmation that the radiator is working fittingly. A flame discoverer can consistently respond speedier and more absolutely

372

S. Tayyaba et al.

than a smoke or warmth identifier as a result of the frameworks it uses to recognize the fire.

31.7.1.4

Light Level

It is used to identify the voltage and current level such that how splendid/dull it is. There is a choice of different light detectors like ‘Photo resistors’, ‘Photodiodes’.

31.7.1.5

Door Sensor

These sensors are a basic section of your home security framework: they let you know when somebody is entering your home. These gadgets are comprised of two sections, which structure a circuit when they’re held parallel to one another. When somebody opens the entryway, the two sections independent and break the circuit, which triggers the control panel to sound an alert.

31.7.1.6

Microcontroller

A microcontroller is a single designed circuit. It has its own central processing unit combined with a memory unit and programmable input/output devices. Randomaccess memory or read-only memory is likewise comprised on-chip. Microcontrollers are used for special purpose applications to control the things, devices such as automobile switch structure, wearable medical devices, wireless panels, office apparatuses, and other implanted architecture. With respect to the snare of things, microcontrollers are reasonable and common techniques for data assembling, identifying and prompting the physical world as edge devices. Various type of microcontroller is used in home automation like Arduino, pic microcontroller and MSP430.

31.7.1.7

Wireless Modules

The system will have a gadget called a remote switch which physically joins to the approaching system and consequently the internet through speedy broadband or link. The remote switch takes the physically transmitted information and changes over it into a radio signal, which it sends to reception devices. Remote correspondence assumes a noteworthy job in everyday life. Other than correspondence, remote innovation has turned into a vital piece of our day by day exercises. The broadcast of data moving with one location then onto the next location is indicated to as wireless communication. This gives a trade of information with no wire over radio frequency and signal. The data is communicated over the gadgets over certain meters to many kilometers through well-characterized channels. Various

31 Home Automation Using IoT

373

sorts of the signals are utilized in correspondence between the gadgets for remote transmission of information. The diverse electromagnetic signal is utilized relying upon their wavelength and recurrence. Remote correspondence innovation is ordered into various kinds relying upon the separation of correspondence, the scope of information and the sort of gadgets utilized.

31.7.2 Software Requirement The microcontroller does not know what to do. Set of instructions are used to programed the microcontroller. Various software is used for different microcontroller like The Keil 8051 is a programming tool to support the designer. Arduino IDE is a programming interface that is running on your system that enables you to compose outlines for various Arduino sheets. C language is used to program the Arduino microcontroller and it is open source.

31.7.2.1

Database Server

The use of a database server is to provide services to store information about the customer framework. Clients get to a database server via a “UI” working on the customer’s PC—which displays the information getting from “back end”, which runs on the server. Instances of restrictive database applications incorporate Oracle, DB2, and Microsoft SQL Server. Instances of free programming database applications incorporate PostgreSQL, and under the GNU General, Public License incorporates Ingres and MySQL. Each server utilizes its very own inquiry rationale and structure. The SQL (Structured Query Language) inquiry language is pretty much the equivalent on all social database applications.

31.7.2.2

Web Application/Mobile Application

A web application is continuously running on a remote server. Web projects are used to access the website, over a framework, for instance, the Internet. Some web applications are used in associations and schools. Sites destined to be alluded to as “web applications” are those which have comparable usefulness to a work area programming application, or to a portable application. Hypertext markup language 5 is used to support for making applications that are stacked as website pages, however, can store information locally and keep on working while disconnected. An examination led including web building practice uncovered that web application advancement has a few attributes that must be tended to and these include: short improvement life-process durations; distinctive plans of action; multi-disciplinary

374

S. Tayyaba et al.

improvement groups; little improvement groups dealing with comparative errands; business investigation and assessment with end-clients; express prerequisite and thorough preparing against necessities; and, support. Some of the languages are used: active server page, cascading style sheet, hypertext markup language, Java, JavaScript, the personal home page, and python. Figure 31.16 is depicted about web interface for end-user. Authentication provides security for end-user by using password and user name. User data is stored in the database. If this data is matched then opened the web interface. There can be a different tab in the web interface and perform different functions. The status tab shows the status of the home appliance that already stored in the database. Control tab links to all the things that have to be controlled like the light. The automatic tab is used to control the appliances automatically like a smoke detection system. Savvy home and IoT are rich with sensors, which create huge information streams as messages or occasions. Handling this information is over the limit of a person’s capacities. Henceforth, occasion handling frameworks have been designed and used to react quicker to arranged occasions. The client can characterize the occasion activated guideline and control the best possible conveyance of services. A standard is made out of occasion conditions, occasion example and connection related data which can be consolidated for demonstrating complex circumstances. The framework can process a lot of occasions, execute capacities to screen, explore and upgrade forms progressively. It finds and breaks down oddities or exemptions and makes receptive/proactive reactions, for example, admonitions and avoiding harmful activities. Circumstances are displayed by an easy to use demonstrating

Fig. 31.16 Web interface for user end

31 Home Automation Using IoT

375

interface for occasion activated principles. Whenever required, it separates them into straightforward, reasonable components. The assessment procedure is activated by occasions conveying the latest state, what’s more, data from the important condition. The result is a choice chart speaking to the standard. It can separate complex circumstances to basic conditions, what’s more, consolidate them with one another, making complex conditions. The yield is a reaction occasion raised when a standard flames? The terminated occasions might be utilized as a contribution to different standards for further assessment. Occasion examples are found when numerous occasions happen and coordinate a pre-characterized design. Due to the graphical model and measured methodology for building rules, guidelines can be effectively adjusted to space changes. New occasion conditions or occasion examples can be included or expelled from the standard model. Guidelines are executed by occasion services, which supply the standard motor with occasions and procedure the assessment result. To guarantee the accessibility of reasonable handling assets, the framework can keep running in a dispersed model, on numerous machines and encourage the incorporation with outer frameworks, too. The meaning of connections and conditions among occasions that are pertinent for the standard handling are performed utilizing succession sets, created by the standard motor. The standard motor develops arrangements of occasions pertinent to an explicit principle condition to permit partner occasions by their setting information. Principles consequently, perform activities accordingly when expressed conditions hold. Activities produce reaction occasions, which trigger reaction exercises. Occasion examples can coordinate transient occasion groupings, permitting the depiction of home circumstances where the events of occasions are significant. For instance, when the entryway is kept open excessively long. Occasion preparing is worried about ongoing catching and overseeing predefined occasions. It begins from dealing with the receptors of occasions directly from the occasion the event, even distinguishing proof, information gathering, process affiliation, and enactment of the reaction activity. To permit quick and adaptable occasion taking care of, an occasion handling language is utilized, which permits the quick design of the assets required to handle the normal arrangement of exercises per occasion type. It is made out of two modules, ESP and CEP. ESP effectively handles the occasion, breaks down it and chooses the proper event. CEP handles totaled occasions. Occasion dialects portray complex occasion types connected over the occasion log.

31.8 Home Automation Using IoT There are three components used in IoT based home automation as listed under: • Hardware • Software/Applications

376

S. Tayyaba et al.

• Communication Protocols. Each part is paying a significant role in the framework a genuinely savvy house understanding for customers. Having the right hardware enables the ability to develop IoT model iteratively and respond to advancement turns easily. Frugal Labs IoT Platform (FLIP) for structure IoT empowered Smart Home has been discussed. The elements of Smart Home and its applications have depicted and presented the FLIP design with the usage of Smart Home services utilizing FLIP through a designed framework. The designed framework can be utilized for checking and controlling the Smart Home condition [67]. Raspberry pi based automatic system demand framework has implemented for shrewd home security related to a web server or the cloud through the Internet utilizing the remote system. The framework utilized distributed storage. This framework also utilized for observing the status of the home by utilizing various sensors. The sensor information can be handled by raspberry pi and on the off chance that any irregularity discovered, at that point, it can consequently send a system solicitations to the concerned individual with respect to the variation from the normal [68]. An Internet of Things-based savvy home framework has been structured and executed for home solace, relaxation, and security. The installed framework, 3G, and ZIGBEE advances are utilized by this framework to defeat the disadvantages of current savvy home frameworks, for example, discrete capacities, poor transportability, feeble refreshing ability, and PC reliance [69]. The savvy remote home security framework has been designed and implemented which can send the message on the owner mobile phone by using the Internet in the incident of any intrude and generates a caution repeatedly. The equipment can be utilized for house mechanization by using the precise planning of detectors [70]. The IoT Shrewd Home architecture has been designed which can give the wireless switch to the savvy house over multipurpose, infrared wireless switch with personal computers [71]. The smart house system has been designed by using the virtue of it. The framework is utilized to watch out for the youngster, overseeing the illumination at household, space and Porch Vinicultural using the ideals of the internet of thing [72]. Revamping buyer needs, household computerization has been designed to achieve the goal for the new progressive customer. A point of interest where consumers can optimism to see household modernization using Internet of Things permitted network are: Illumination switch, Central air, Grass/Gardening the board, Savvy Home Appliances, Better Home security and safety, Household air and water level detecting, voice control, Savvy buttons, Savvy Locks, and Savvy Energy Meters. Major parts of IoT based home automation can be broken down: • • • • •

Sensors Gateway Protocols Firmware Cloud and database.

31 Home Automation Using IoT

377

IoT sensors associated with home mechanization are in thousands, and there are many home robotization gateway also. A large portion of the software can be written in C, Python, or some other language. The largest service provider on the internet of thing cloud can be categorized as a platform as a service and infrastructure as a service. Largest IoT platform providers are listed as amazon web service, Thingspeak, Xivelym, inter-board machine bluemix. These platforms are amazingly separated over the Internet of Things appliances and safety associated highlights that they give. A couple of these stages are open source. A specific functionality provided by the IoT platform must meet the rules of the home automation system. Gadget security and validation, Message merchants and message lining, Gadget organization, Backing towards conventions like CoAP (constraint application protocol), MQTT, and HTTP, Information gathering, perception, and straightforward investigation abilities, combined with other network services, Flat and vertical versatility, and WebSocket application programming interface for continuous data stream are services provided by IoT platform. The protocol was developed for secure transmission, reduce the latency, packet loss, increasing the throughput and achieved data rate during transmission. The route optimization is achieved by the developed protocol [73]. The residence energy control system has been designed that can be used for energy saving for home appliances. IoT and the wireless socket is used in this system to reduce energy usage without using sensors. This has acquired that the designed system can save energy up to 43% [74]. There are most likely a great many such detectors that can be a part of this domain, however since this is a presentation towards savvy home innovation. We will stall Internet of Things detectors for household computerization by their detecting capacities: Temperature sensors, Lux sensors, Water level sensors, Air composition sensors, Video cameras for surveillance, Voice/Sound sensors, Pressure sensors, Humidity sensors, Accelerometers, Infrared sensors, Vibrations sensors, and Ultrasonic sensors. Each sensor performs special functionality and commonly used in home computerization to monitor the change. One of the most significant pieces of structure a home robotization item is to consider conventions. Conventions that your gadget will use to convey to portals, servers, and sensors. A couple of years back, the best way to do as such was by either utilizing Bluetooth, Wi-Fi, or GSM. In any case, due to included costs cell SIM cards and low execution of Wi-Fi, most such arrangements didn’t work. Bluetooth endures and later developed as Bluetooth Smart or Bluetooth Low Energy. This acquired a ton of network the “versatile server controlled economy.” Essentially, your telephone would go about as a middleware to get information from Bluetooth low energy, fueled sensors and transmit it to the web. The significant home computerization conventions are given as Bluetooth Smart: remote convention with work capacities, safety, information hiding methods, and substantially more. Zigbee: cost-effective, work organized, and less power signal depends on the convention for the internet of thing. Diverse Zigbee renditions don’t converse with one another. X10: A heritage convention that uses high voltage wiring for flagging and switch. Insteon:

378

S. Tayyaba et al.

Connects with gadgets both remotely and with wires. Z-wave: Focusses in household robotization with an accentuation on safety. Wi-Fi: No need for clarification. UPB: Uses existing electrical cables introduced in a home. Diminishes costs. String: An eminence free convention for shrewd home mechanization, utilizes a 6lowpan (low power wide area network). Subterranean insect: A ultra-low-control convention helping designers construct low-fueled sensors with work dissemination abilities. For building up a home robotization item, frequently an independent item sending information to a server isn’t sufficient. Because of battery and convention confinements, the information from detectors that are existing in a household has been steered using the internet of thing passage. To choose the ideal portal for IoT home mechanization, think about a portion of these elements: Communication protocol upheld, Ongoing capacities, message queuing telemetry transport, constraint application protocol, and hypertext transfer protocol, Safety, and design. For instance, a passage with an awful systems administration line may bring about traffic clog, or it may not bolster the required conventions that you wish to utilize. Further, rotating with these entryways to some other innovation stack may turn out to be extremely troublesome. It ought to be underlined that they are very useful for strong prototyping needs. The accompanying programming dialects rule the household mechanization space: Embedded C, Python, and JavaScript. This has, for the most part, occurred because of the sheer streamlining of the dialects for comparative use cases. The remote patient wellbeing checking framework has been structured and executed in savvy homes by utilizing the idea of mist processing at the keen entryway. The model utilized propelled methods and services, for example, inserted information mining, disseminated capacity, and notice administrations at the edge of the system. Event activating based information transmission system has been established to process the patient’s constant information at the haze layer. Transient mining idea utilized to dissect the occasions difficulty by ascertaining the fleeting wellbeing record of the patient [75]. Various leveled savvy home control framework has been planned and executed, which controls the residential machines. There are three controllers used to perform various undertakings. This framework can be utilized to decrease vitality utilization at home [76]. A low-cost home robotization framework has been designed to utilize IoT. All the home apparatuses and electronic machines can be controlled and saw through a site in all respects effectively. The metering strategy for a home can likewise be managed to utilize this framework. A web-based charging framework likewise incorporated into the structured framework [77]. The software application has been developed to control home appliances. Software is used to control equipment by using a set of instructions. Use’s Input received by software and forwarded this input to the host server. Software is also used to display controlling information about the house condition. The AMQP is a communication protocol that is used by the smart home system to ensure the security of transmitted data by the system [78].

31 Home Automation Using IoT

379

A voice control home automation system has been designed and implemented through IoT, artificial intelligence and natural language processing. Hardware system has interfaced with home appliance and programmed as implement through natural language. Mobile devices have used to transmit vocal sound instructions to switch the household appliance from the far end. Software Application has used to transmit vocal sound instructions to switch the appliances through the internet [79]. A wireless home security system has been discussed. Advanced encryption standard is used for secure transmission over the network. An alert message sent to the user through the internet if any intruder at home and alarm started. Resparry Pi can be used to perform the task of controlling home appliances and authentication is also provided by the controller to the user [80]. A Client-Server administration and gadget have been discussed that is a neighborly methodology for Home robotization. A run of the mill home computerization work process comprises of 4 phases. Understanding the client condition by detecting, revealing the occasions to a concentrated element, unified element examinations and triggers the work process, the work process can execute and refresh clients by any intuitive channels or even exercise over a home gadget (impelling) [81]. The android mobile has utilized to send directions to the Arduino to control all the home machines. The primary element of this framework is to control the voltage levels of home machines in a home like the speed of the fan dependent on temperature, the power of light-dependent on the light force. what’s more, another component is used to get the status of home apparatuses from android cell phone [82]. A lightweight time synchronization algorithm has been developed for the CoAPbased home robotization framework systems. The CoAP choice field and a shim header can be utilized to incorporate time-stamps in the home computerization framework. The planned algorithm can be connected to both IP-based and non-IP-based home computerization frameworks [83]. A novel approach, gadget-free, and protection saving WiFi-empowered Internet of Things stage for inhabitance detecting, which can advance a horde of developing applications [84]. A low cost, secure, pervasively, open, auto adjustable, wirelessly switched planning. The process has conversed that is novel and has accomplished the objective to switch household apparatuses wirelessly utilizing the WiFi innovation to associates framework parts, fulfilling client needs and prerequisites. WiFi innovation skilled arrangement has confirmed to be measured wirelessly, gave household safety and is financially savvy when contrasted with the already existing frameworks [85]. Home automation system has designed and implemented for old people who can not work easily. The command button is used for automatic control. LED and LCD is used to display an alert message. Radiofrequency signals are used to understand the communication and automatic control of the appliance by XBee transmitter and receiver [86]. A structure of DNS Name auto configuration has been created, (called DNSNA) for IoT gadgets in both IPv6 and IPv4 IoT systems [87]. Figure 31.17 portrays the propelled savvy home principle segments and their interconnectivity. On the left rectangular box, the savvy home condition, we can see the commonplace gadgets associated with a neighborhood [LAN]. This empowers

380

S. Tayyaba et al.

Fig. 31.17 IoT based home automation system

correspondence among the gadgets and outside of it. Associated with the LAN are a server and its database. The server controls the gadgets, logs its exercises, gives reports, answers questions and executes the suitable directions. For increasingly exhaustive or then again regular errands, the smart home server moves information to the cloud and remotely enact assignments in it utilizing APIs, application programming interface forms. In expansion, IoT home machines are associated with the web and to the LAN, thus extends the savvy home to incorporate IoT. The association with the web permits the end-client, occupant, to speak with the savvy home to get present data and remotely actuate undertakings. A progressively pragmatic model is the place a few isolates machines, for example, a broiler, a moderate cooker and a skillet on the gas stovetop, are dynamic in satisfying the inhabitant demand. The occupant is getting an earnest telephone call and leaves home quickly, without closing off the dynamic apparatuses. In the event that the important IoTs have been tuned to consequently close down dependent on a predefined rule, it will be dealt with at the IoT level. Something else, the savvy home understands the occupant has left home (the home entryway has been opened and after that bolted, the carport has been opened, the occupant’s vehicle left, the main door was opened and afterward shut, nobody was at home) and will close down every single dynamic gadget named hazard if

31 Home Automation Using IoT

381

there should be an occurrence of nonappearance. It will send a suitable message to the mailing rundown characterized for such an event.

31.9 Advantages of IoT for Home Automation The cell phones we bear in our pockets are incredible assets that make life simpler, and each progression in innovation improves their amazing abilities. The following huge advance forward for this sort of “savvy” innovation is into our homes. Using incorporated mechanical frameworks in your house is one of the most critical new patterns in advanced development. At the present time is the best time to begin receiving the rewards of these abilities. Changing to a more astute home can improve your command over each part of how your home works, and increment the wellbeing and availability of it too. Also, you can receive the rewards of a progressively proficient home, prompting investment funds in your vitality and upkeep costs. • By introducing machines in your home, for example, a smart broiler, you would now be able to utilize applications on your cell phone to appreciate unlimited oversight of your home’s capacities from anyplace on the planet. Did you leave your home and neglect to kill your stove? A climate control system as yet running at home while you’re in the midst of some recreation? No compelling reason to pressure. You can rapidly and effectively power off these apparatuses in seconds from the separate going with applications. • There are no impediments to the applications that are coming to showcase consistently. Engineers have made applications and gadgets to control home sound systems, water utilization, lighting, yard care, carport entryways, your canine’s nourishment dish, and even shopping for food—all effectively and promptly constrained by the telephone you as of now have in your pocket. • Savvy homes don’t just empower us to dismiss unexpected flames from unattended stoves. These homes in like manner empower us to keep our loved ones ensured. Safety architectures can be acquainted that empower owners with the screen the comings and goings of guests and alert you when suspicious development is recognized. Entrances can be locked, safety structure outfitted, and cameras detected from the mobile phone, taking a safer and increasingly protected condition for family. • On the off chance that you have companions or relatives who are older or incapacitated, you know how troublesome even the most essential ordinary undertaking can be for them. savvy home innovation can incredibly build their personal satisfaction, and using voice directions can make the expectation to learn and adapt a lot simpler for somebody new to PCs. Setting up robotized frameworks for exercises like grass care expels pointless worry from the lives of these people. • As the innovation pushes ahead, a While numerous advantages of a savvy home incorporate simplicity and openness, there are significantly more advantages to

382

S. Tayyaba et al.

appreciate. Savvy home innovation enables apparatuses to work with minimal measure of vitality required. For instance, enlistment cook-top stoves presently have the insight to warm only when a metal dish is put over it. No more burners running revealed, and no more skillet being overheated. Stovetops can even deal with an ideal bubble while utilizing a minimal measure of vitality conceivable. Ever-increasing a number of troublesome errands will wind up available, improving adaptability and freedom in lodging for individuals who probably won’t be completely equipped for dealing with their homes without anyone else. Getting a good deal on that water bill has additionally never been simpler. • you can build vivacity productivity by monitoring electrical installations using the internet of thing. In the occasion that you are ambiguous whether your youngster has left illuminations ON before departure, you can check and switch it using the telephone. • With IoT home robotization you are less worried over household safety. You can switch the safety of your household using the mobile phone. On account of anything turns out gravely, you may get sees on your phone and you may in all probability work you lights or dashes through your phone.

31.10 Discussion and Recommendations The home computerization architecture is used to switch and monitor household applications and security systems. Household computerization is an emerging technology to secure homes. Various types of home automation systems have been designed and implemented to support the disable and elderly persons. Global system for mobile communication-based control architecture is used to switch the household applications. GSM is used to send and receive commands in term of SMS on mobile phones for ON/Off home appliances such as light, fan, and heater. Technology increased rapidly day by day and Bluetooth based household computerization architecture has been designed and implemented. This structure is used Bluetooth as a wireless module to control the home appliances but can not be monitored the appliances. The range of Bluetooth is short as 10 meters. Bluetooth based system used at home for those people who can not speak or walk. Voice control home computerization structure is designed and implemented for disabling and old persons to control the appliance through voice. This system can be used from far away. From two decades technology has been very fast. Every person is using the internet for sharing, retrieving and storing meaningful information. The gadgets that are linked to the internet for gathering information are called the Internet of Things. Laptop, mobile phone, PC, and tablet PC are connected to the internet and used for sharing information. The home computerization system is connected to the internet. Devices that can be connected to the internet and used for monitoring and controlling the home appliances and security systems at home.

31 Home Automation Using IoT

383

Fig. 31.18 Structure of home automation using IoT

Internet of Things-based home security systems can be used from anywhere and anytime in the world. Structure of IoT based home computerization architecture as shown in Fig. 31.18. detectors are used to sense any change in physical parameters. Various type of sensors can be used and placed to control household applications. Sensors are linked to the microcontroller. The microcontroller is working on a set of instructions to control the home appliances. Sensors transform the signal and send to the microcontroller. Microcontroller received the data and process it on the basis of instruction. The WiFi module is a wireless module that is used to connect the user end devices with embedded system. IoT devices are connected to the WiFi module through the internet to monitor and controlling home appliances. A web application is developed and installed in IoT gadgets to switch and monitor household applications on a real-time basis. The advantage of IoT based home automation system is to secure, energy-efficient, low cost and less time-consuming. Every person who has an IoT device can be used far from home. Cloud computing is a database system used for storing and retrieving information. IoT based home automation system is a commercial application that is accessible for all users, provides security and easily manageable.

384

S. Tayyaba et al.

References 1. Ziegler, S.: Considerations on IPv6 scalability for the Internet of Things—towards an intergalactic internet. In: 2017 Global Internet of Things Summit (GIoTS), Geneva, Switzerland (2017) 2. Forouzan, B.A.: Network model. In: Data Communication and Networking, McGraw-Hill, pp. 38–41 3. Zhu, J., Xie, P., Xuan, P., Zou, J., Yu, P.: Renewable energy consumption technology under energy internet environment. In: 2017 IEEE Conference on Energy Internet and Energy System Integration (EI2), Beijing, China (2018) 4. Chenghui Tang, F.Z.: Combined electricity-gas-heat energy internet scheduling with power-togas and renewable power uncertainty. In: 2018 2nd IEEE Conference on Energy Internet and Energy System Integration (EI2), Beijing, China (2018) 5. Yang, Y., Fang, Z., Zeng, F., Liu, P.: Research and prospect of virtual microgrids based on energy internet. In: 2017 IEEE Conference on Energy Internet and Energy System Integration (EI2), Beijing, China (2017) 6. Liu, S., Gu, Y., Wang, C., Guo, F., Li, S.: Multi-energy coordination based on multi-agent system in energy internet. In: 2018 2nd IEEE Conference on Energy Internet and Energy System Integration (EI2), Beijing, China (2018) 7. Cheng, L., Ji, X., Zhang, F., Liang, C., He, H.: Internet information applied in the energy internet planning: a review and outlook. In: 2017 IEEE Conference on Energy Internet and Energy System Integration (EI2), Beijing, China (2017) 8. JayavardhanaGubbi, R.S.M.: Internet of Things (IoT): a vision, architectural elements, and future directions. Future Gener. Comput. Syst. 29(7), 1645–1660 (2013) 9. Stojkoska, B.L.R., Trivodaliev, K.V.: A review of Internet of Things for smart home: challenges and solutions. J Cleaner Prod. 140(1), 1454–1464 (2017) 10. Narang, S., Nalwa, T., Choudhury, T., Kashyap, N.: An efficient method for security measurement in Internet of Things. In: 2018 International Conference on Communication, Computing and Internet of Things (IC3IoT), Chennai, India, India (2018) 11. Yan, Z., Li, H., Zeadally, S., Zeng, Y., Geng, G.: Is DNS Ready for ubiquitous internet of things? In: Urban Computing and Well-being in Smart Cities: Services, Applications, Policymaking Considerations, vol. 7, no. 29, pp. 28835–28846 (2019) 12. Cheng, L., Qi, N., Zhang, F., Kong, H., Huang, X.: Energy internet: concept and practice exploration. In: 2017 IEEE Conference on Energy Internet and Energy System Integration (EI2), Beijing, China (2017) 13. Li, B., Yu, J.: Research and application on the smart home based on component technologies and Internet of Things. Procedia Eng. 15, 2087–2092 (2011) 14. Raza, S., Magnússon, R.M.: TinyIKE: Lightweight IKEv2 for Internet of Things. IEEE Internet Things J. 6(1), 856–866 (2018) 15. Wang, Q., Wang, Y.G.: Research on power Internet of Things architecture for smart grid demand. In: 2018 2nd IEEE Conference on Energy Internet and Energy System Integration (EI2), Beijing, China (2018) 16. Celia, L., Cungang, Y.: (WIP) authenticated key management protocols for Internet of Things. In: 2018 IEEE International Congress on Internet of Things (ICIOT), San Francisco, CA, USA (2018) 17. Lenk, A., Marcus, P., Povoa, I.: GeoFPE: Format preserving encryption of geospatial data for the Internet of Things. In: 2018 IEEE International Congress on Internet of Things (ICIOT), San Francisco, CA, USA (2018) 18. Muhal, M.A., Luo, X., Mahmood, Z., Ullah, A.: Physical unclonable function based authentication scheme for smart devices in Internet of Things. In: 2018 IEEE International Conference on Smart Internet of Things (SmartIoT), Xi’an, China (2018) 19. Mengistu, T., Alahmadi, A., Albuali, A., Alsenani, Y., Che, D.: A no data center “solution to cloud computing”, In: 2017 IEEE 10th International Conference on Cloud Computing (CLOUD), Honolulu, CA, USA (2017)

31 Home Automation Using IoT

385

20. Sahl, R., Dupont, P., Messager, C., Honnorat, M., La, T.V.: High-resolution ocean winds: hybrid-cloud infrastructure for satellite imagery processing. In: 2018 IEEE 11th International Conference on Cloud Computing (CLOUD), San Francisco, CA, USA (2018) 21. Ma, Y., Wang, F., Wang, Z.: Intelligent laboratory management system based on Internet of Things. In: 2017 12th International Conference for Internet Technology and Secured Transactions (ICITST), Cambridge, UK, 2017 22. Kim, H.J., Kim, H.I., Chang, J.W.: A privacy-preserving kNN classification algorithm using Yao’s garbled circuit on cloud computing. In: 2017 IEEE 10th International Conference on Cloud Computing (CLOUD), Honolulu, CA, USA (2017) 23. Joshi, M., Joshi, K., Finin, T.: Attribute based encryption for secure access to cloud based EHR systems. In: 2018 IEEE 11th International Conference on Cloud Computing (CLOUD), San Francisco, CA, USA (2018) 24. Lufei, Z., Zuoning, C.: vStarCloud: an operating system architecture for cloud computing. In: 2017 IEEE 2nd International Conference on Cloud Computing and Big Data Analysis (ICCCBDA), Chengdu, China (2017) 25. Geetha, P., Robin, C. R.: A comparative-study of load-cloud balancing algorithms in cloud environments. In: 2017 International Conference on Energy, Communication, Data Analytics and Soft Computing (ICECDS), Chennai, India (2017) 26. Byrne, J., Svorobej, S., Gourinovitch, A., Elango, D.M., Liston, P., Byrne, P.J., Lynn, T.: RECAP simulator: simulation of cloud/edge/fog computing scenarios. In: 2018 Winter Simulation Conference (WSC), Las Vegas, NV, USA (2017) 27. Zhou, B., Dastjerdi, A.V., Calheiros, R.N., Srirama, S.N., Buyya, R.: mCloud: a contextaware offloading framework for heterogeneous mobile cloud. IEEE Trans. Serv. Comput. 10(5), 797–810 (2015). (Sept. 2017) 28. Fowley, F., Pahl, C., Jamshidi, P., Fang, D., Liu, X.: A classification and comparison framework for cloud service brokerage architectures. IEEE Trans. Cloud Comput. 6(2), 358–371 (2016) 29. Shirazi, S.N., Gouglidis, A., Farshad, A., Hutchison, D.: The extended cloud: review and analysis of mobile edge computing and fog from a security and resilience perspective. IEEE J. Sel. Areas Commun. 35(11), 2586–2595 (2017) 30. Awad, A., Matthews, A., Qiao, Y., Lee, B.: Chaotic searchable encryption for mobile cloud storage. IEEE Trans. Cloud Comput. 6(2), 440–452 (2015). (April 2018) 31. Wang, W., Bao, X., Zhao, T.: A research for embedded system software accident mechanism. In: 2017 2nd International Conference on System Reliability and Safety (ICSRS), Milan, Italy (2017) 32. Gorobetz, M., Alps, I., Potapov, A., Levchenkov, A.: Reliability improvement of embedded train anti-collision system by multi-processor technology. In: 2017 IEEE 58th International Scientific Conference on Power and Electrical Engineering of Riga Technical University (RTUCON), Riga, Latvia (2017) 33. Pašali´c, D., Cviji´c, B., Bundalo, D., Bundalo, Z., Kuzmi´c, G.: Embedded systems for user identification in access to objects and services using mobile phone. In: 2017 6th Mediterranean Conference on Embedded Computing (MECO), Bar, Montenegro (2017) 34. Maret, Y., Oberson, D., Gavrilova, M.: Real-time embedded system for gesture recognition. In: 2018 IEEE International Conference on Systems, Man, and Cybernetics (SMC), Miyazaki, Japan, Japan (2018) 35. Ramachandran, R.: The analysis of different types of IoT sensors and security trend as quantum chip for smart city management. IOSR 20(1), 55 (2018) 36. Díaz, Á., Peña, D., Villar, E.: Short and long distance marker detection technique in outdoor and indoor environments for embedded systems. In 2017 32nd Conference on Design of Circuits and Integrated Systems (DCIS), Barcelona, Spain (2017) 37. Nugraha, M.B., Taufiqurrahman, I., Trilaksono, B.R., Hidayat, E., Sagala, F.: Embedded system integration on GaneshBlue underwater Glider. In 2018 International Conference on Signals and Systems (ICSigSys), Bali, Indonesia (2018) 38. Huang, J., Li, R., An, J., Ntalasha, D., Yang, F., Li, K.: Energy-efficient resource utilization for heterogeneous embedded computing systems. IEEE Trans. Comput. 66(9), 1518–1531 (2017)

386

S. Tayyaba et al.

39. Aminifar, A., Eles, P., Peng, Z.: Optimization of message encryption for real-time applications in embedded systems. IEEE Trans. Comput. 67(5), 748–754 (2017) 40. Chang, W., Goswami, D., Chakraborty, S., Ju, L., Xue, C.J., Andalam, S.: Memory-aware embedded control systems design. IEEE Trans. Comput. Aided Des. Integr. Circuits Syst. 36(4), 586–599 (2016) 41. Ansari, M., Safari, S., Yeganeh-Khaksar, A., Salehi, M., Ejlali, A.: Peak power management to meet thermal design power in fault-tolerant embedded systems. IEEE Trans. Parallel Distrib. Syst. 30(1), 161–173 (2018) 42. Boubriak, A., Cooper, A., Hossack, C., Permogorov, D., Sherratt, R.S.: SlimFS: a thin and unobtrusive file system for embedded systems and consumer products. IEEE Trans. Consum. Electron. 64(3): 334–338 (2018) 43. Willmann, S., Krätzig, M., Rauchhaupt, L.: Methodology for holistic assessment of dependability in wireless automation. In: 202017 22nd IEEE International Conference on Emerging Technologies and Factory Automation (ETFA), Limassol, Cyprus (2018) 44. Sun, H., Liu, D., Pei, X., Sun, J., Zhou, J.: A novel variable step-size LMS algorithm for process controlling of shale gas components automation detection. In: 2018 2nd IEEE Advanced Information Management, Communicates, Electronic and Automation Control Conference (IMCEC), Xi’an, China (2018) 45. Sangeetha, N., Umanand, L., Radhaswamy, G., Anandi, V.: Development of SCADA automation system as a testing platform at IISC (Indian Institute of Science) Campus. In: 2018 International Conference on Inventive Research in Computing Applications (ICIRCA), Coimbatore, India (2019) 46. Rentschler, M.: Roaming in wireless factory automation networks. In: 2017 22nd IEEE International Conference on Emerging Technologies and Factory Automation (ETFA), Limassol, Cyprus (2017) 47. Nafees, M.: RFID based prepaid energy meter and home automation with reporting. In: 2017 23rd International Conference on Automation and Computing (ICAC), Huddersfield, UK (2017) 48. Kodali, R.K., Jain, V., Bose, S., Boppana, L.: IoT based smart security and home automation. In 2018 Fourth International Conference on Computing Communication Control and Automation (ICCUBEA), Pune, India, India (2018) 49. John, A., Varghese, R., Krishnan, S.S., Thomas, S., Swayambu, T.A., Thasneem, P.: Automation of 11 kv substation using raspberry pi. In: 2017 International Conference on Circuit, Power and Computing Technologies (ICCPCT), Kollam, India (2017) 50. Ivanovi´c, S., Milivojša, S., Eri´c, T., Vidakovi´c, M.: Collection and analysis of system usage data in smart home automation systems. In: 2017 IEEE 7th International Conference on Consumer Electronics—Berlin (ICCE-Berlin), Berlin, Germany (2017) 51. Tomov, P.: Possibilities for Implementing Production “Automation Islands” in an Automatic Production System. In: 2018 International Conference on High Technology for Sustainable Development (HiTech), Sofia, Bulgaria (2018) 52. Hassija, V., Chamola, V., Saxena, V., Jain, D., Goyal, P., Sikdar, B.: A survey on IoT security: application areas, security threats, and solution architectures. IEEE Access. 7(20), 82721–82743 (2019) 53. Alaa, M., Zaidan, A.A., Zaidan, B.B., Talal, M., Kiah, M.L.M.: A review of smart home applications based on Internet of Things. J Netw. Comput. Appl. 97(1), 48–65 (2017) 54. Wu, F., Redouté, J.M., Yuce, M.R.: WE-Safe: a wearable IoT sensor node for safety applications via LoRa. In: 2018 IEEE 4th World Forum on Internet of Things (WF-IoT), Singapore, Singapore (2018) 55. Gunathilake, N.A., Buchanan, W.J., Asif, R.: Next generation lightweight cryptography for smart IoT devices:: implementation, challenges and applications. In: 2019 IEEE 5th World Forum on Internet of Things (WF-IoT), Limerick, Ireland, Ireland (2019) 56. Vo, H.M.: A double regulated footer and header voltage technique for ultra-low power IoT SRAM. In: 2018 IEEE 4th World Forum on Internet of Things (WF-IoT), Singapore, Singapore (2018)

31 Home Automation Using IoT

387

57. Miladinovic, I., Schefer-Wenzl, S.: NFV enabled IoT architecture for an operating room environment. In: 2018 IEEE 4th World Forum on Internet of Things (WF-IoT), Singapore (2018) 58. Roy, D.S., Behera, R.K., Reddy, K.H.K., Buyya, R.: A Context-aware fog enabled scheme for real-time cross-vertical IoT applications. IEEE Internet Things J. 6(2), 2400–2412 (2018) 59. Wang, S., Hou, Y., Gao, F., Ji, X.: A novel IoT access architecture for vehicle monitoring system. In: 2016 IEEE 3rd World Forum on Internet of Things (WF-IoT), Reston, VA, USA (2016) 60. Ling, R.W.C., Gupta, A., Vashistha, A., Sharma, M., Law, C.L.: High precision UWB-IR indoor positioning system for IoT applications. In: 2018 IEEE 4th World Forum on Internet of Things (WF-IoT), Singapore (2018) 61. Govindraj, V., Sathiyanarayanan, M., Abubakar, B.: Customary homes to smart homes using Internet of Things (IoT) and mobile application. In: 2017 International Conference On Smart Technologies For Smart Nation (SmartTechCon), Bangalore, India (2017) 62. Kum, S.W., Moon, J., Lim, T.B.: Design of fog computing based IoT application architecture. In: 2017 IEEE 7th International Conference on Consumer Electronics—Berlin (ICCE-Berlin), Berlin, Germany (2017) 63. Fan, Q., Ansari, N.: Application aware workload allocation for edge computing-based IoT. IEEE Internet Things J. 5(3), 2146–2153 (2018) 64. Perumal, T., Chui, Y.L., Ahmadon, M.A.B., Yamaguchi, S.: IoT based activity recognition among smart home residents.: In: 2017 IEEE 6th Global Conference on Consumer Electronics (GCCE), Nagoya, Japan (2017) 65. Vaidya, V.D., Vishwakarma, P.: A comparative analysis on smart home system to control, monitor and secure home, based on technologies like GSM, IOT, Bluetooth and PIC Microcontroller with ZigBee Modulation. In: 2018 International Conference on Smart City and Emerging Technology (ICSCET), Mumbai, India (2018) 66. Salman, L., Salman, S., Jahangirian, S., Abraham, M., German, F., Blair, C., Krenz, P.: Energy efficient IoT-based smart home. In: 2016 IEEE 3rd World Forum on Internet of Things (WFIoT), Reston, VA, USA (2016) 67. Malche, T., Maheshwary, P.: Internet of Things (IoT) for building smart home system. In: 2017 International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC), Palladam, India (2017) 68. Madupu, P.K., Karthikeyan, B.: Automatic service request system for security in smart home using IoT. In: 2018 Second International Conference on Electronics, Communication and Aerospace Technology (ICECA), Coimbatore, India (2018) 69. Bing, K., Fu, L., Zhuo, Y., Yanlei, L.: “Design of an internet of things-based smart home system. In: 2011 2nd International Conference on Intelligent Control and Information Processing, Harbin, China (2011) 70. Kodali, R.K., Jain, V., Bose, S., Boppana, L.: IoT based smart security and home automation system. In: Conference: 2016 International Conference on Computing, Communication and Automation (ICCCA), Noida, India (2016) 71. Khan, A., Al-Zahrani, A., Al-Harbi, S., Al-Nashri, S., Khan, I.A.: Design of an IoT smart home system. In: 2018 15th Learning and Technology Conference (L&T), Jeddah, Saudi Arabia (2018) 72. Yadav, V., Borate, S., Devar, S., Gaikwad, R., Gavali, A. B.: Smart home automation using virtue of IoT. In: 2017 2nd International Conference for Convergence in Technology (I2CT), Mumbai, India (2017) 73. Shin, D., Sharma, V., Kim, J., Kwon, S., You, I.: Secure and efficient protocol for route optimization in PMIPv6-based smart home IoT networks. Secur. Priv. Appl. Serv. Future Internet Things. 5(1), 11100–11117 (2017) 74. Tsai, K.L., Leu, F.Y., You, I.: Residence energy control system based on wireless smart socket and IoT. IEEE Access. 4(27), 2885–2894 (2016) 75. Verma, P., Sood, S. K.: Fog assisted-IoT enabled patient health monitoring in smart homes. IEEE Internet Things J. 5(3), 1789–1796 (2018)

388

S. Tayyaba et al.

76. Parsa, A., Najafabadi, T.A., Salmasi, F.R.: A hierarchical smart home control system for improving load shedding and energy consumption: design and implementation. IEEE Sens. J. 19(9), 3383–3390 (2018). (May 2019) 77. Mahmud, S., Ahmed, S., Shikder, K.: A smart home automation and metering system using internet of things (IoT). In: 2019 International Conference on Robotics, Electrical and Signal Processing Techniques (ICREST), 2019 International Conference on Robotics, Electrical and Signal Processing Techniques (ICREST) (2019) 78. Adiono, T., Manangkalangi, B.A., Muttaqin, R., Harimurti, S., Adijarto, W.: Intelligent and secured software application for IoT based smart home. In: 2017 IEEE 6th Global Conference on Consumer Electronics (GCCE), Nagoya, Japan (2017) 79. Rani, P.J., Bakthakumar, J., Kumaar, B.P., Kumaar, U.P., Kumar, S.: Voice controlled home automation system using Natural Language Processing (NLP) and Internet of Things (IoT). In: 2017 Third International Conference on Science Technology Engineering & Management (ICONSTEM), Chennai, India (2017) 80. Somani, S., Solunke, P., Oke, S., Medhi, P., Laturkar, P.P.: IoT based smart security and home automation. In: 2018 Fourth International Conference on Computing Communication Control and Automation (ICCUBEA), Pune, India, India (2018) 81. Brundha, S.M., Lakshmi, P., Santhanalakshmi, S.: Home automation in client-server approach with user notification along with efficient security alerting system. In: 2017 International Conference On Smart Technologies For Smart Nation (SmartTechCon), Bangalore, India (2017) 82. Reddy, P.S.N., Reddy, K.T.K., Reddy, P.A.K., Ramaiah, G.K., Kishor, S.N.: An IoT based home automation using android application. In: 2016 International Conference on Signal Processing, Communication, Power and Embedded System (SCOPES), Paralakhemundi, India (2016) 83. Son, S.C., Kim, N.W., Lee, B.T., Cho, C.H., Chong, J.W.: A time synchronization technique for coap-based home automation systems. IEEE Trans. Consum. Electron. 62(1), 10–16 (2016) 84. Yang, J., Zou, H., Jiang, H., Xie, L.: Device-free occupant activity sensing using WiFi-Enabled IoT devices for smart homes. EEE Internet Things J. 5(5), 3991–4002 (2018) 85. ElShafee, A., Hamed, K.A.: Design and implementation of a WiFi based home automation system. World Acad. Sci. Eng. Technol. Int. J. Comput. Inf. Eng. 6(8), 1074–1080 (2012) 86. Ghazal, B., Al-Khatib, K.: Smart home automation system for elderly, and handicapped people using XBee. Int. J. Smart Home. 9(4), 203–210 (2015) 87. Lee, K., Kim, S., Jeong, J.P., Lee, S., Kim, H., Park, J.S.: A framework for DNS naming services for Internet-of-Things devices. Future Gener. Comput. Syst. 29, 617–627 (2018). (2019)

Chapter 32

Artificial Intelligence: State of the Art Bhaskar Mondal

Abstract Artificial Intelligence (AI) is the most fascinating and discussed technology in the current decade for its nature of mimic human intelligence. As John McCarthy defines it is “The science and engineering of making intelligent machines, especially intelligent computer programs”. AI simply means the study of building machines with human like sense (perceiving), analysis or understand and response. Precisely, it’s the Weak AI, the AI systems are capable to do a specific kind of job for which it is trained. Even, the journey of AI was started back in 1950s, it become popular and started using in recent years for three reasons. First, the availability of big data; the gigantic amount of data generated by the e-commerce, social networks and businesses, second the machine learning algorithms are improved and more reliable, third the cloud and high-performance computer systems become cheap. The AI is changing the personal, social, and business landscape with every new day. It is used to develop products ranging from general to specific, such as playing music, playing chess, Painting, self-driving cars, proving theorems, etc. AI is widely used in automobile, logistic, healthcare, stock-trading, robotics, finance, transport, education like industries. This chapter starts with defining AI and its relationship with machine learning and deep learning followed by a brief time-line of the evaluation of AI, advantages and challenges of AI in today’s world. Then discuss about the three fundamental techniques problem solving, knowledge and reasoning, and learning, artificial neural networks and natural language processing (NLP) are presented.

B. Mondal (B) Xavier School of Computer Science & Engineering, Xavier University Bhubaneswar, Bhubaneswar, Orissa, India e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_32

389

390

B. Mondal

32.1 Introduction I propose to consider the question, ‘Can machines think?’—Alan Turing

32.1.1 What Is It? Artificial (made by human) Intelligence (power of thinking) is the study of machines which can sense, make decision and act like human beings. The meaning of intelligence is “the ability to acquire and apply knowledge and skills.”; in Merriam-Webster intelligence is defined as “the ability to learn or understand or to deal with new or trying situations”. So, an intelligent entity must be able to acquire knowledge through various ways like by observations, learning from experience, reading information (data) and processing text, by discussing with others. It should be able to reason this acquired knowledge to make decisions, summaries, setting and following goals, understand text and images etc. Therefore, AI is the study of science and engineering to build artifacts which can develop knowledge by learning from experience, reading and processing text written in natural languages, reason with the acquired knowledge (able to perform tasks such as explaining, planning, diagnosing, etc.) and acting rationally. A machine is intelligent if it can learn, can do reasoning, and solve problems. In AI the machines are not programmed to solve a single problem but they can learn and solve more complex problems. So, the machines are programs to be learned. In other word a machine can be called intelligent if it pass the Turing test [42]. The test was named after its creator Alan Turing, the father of theoretical computer science, cryptanalyst and great mathematician (Fig. 32.1).

Fig. 32.1 Foundation of AI

Control Theory Computer and Cybernetics

Linguistics

Psychology Artificial Intelligence

Philosophy

Mathematics

Neuroscience

Economics

32 Artificial Intelligence: State of the Art

391

The foundation and development of AI is build based on an amalgamation of several subjects like • Philosophy: The idea of knowledge acquisition, understanding and taking action based on the knowledge in the human mind which manifest from the physical brain. It gives the concept of “How some conclusion can be drawn by a machine from the formal rules”. • Mathematics: It involves computation, representation of logic, probability theory, and decision making. Mathematics is used to determine computability, forming model for knowledge representation and reasoning the knowledge. • Economics: Is used for optimization and understanding of payoff quantity and duration. • Neuroscience: How physical human brain works for logical reasoning. The working method in the brain for any particular action taken. • Psychology: It gives the idea about how human mind thinks and takes action and intersects. • Computer Engineering: How to build system to function like AI artifact. Improve their capacity and efficiency. • Control Theory and Cybernetics: How can the AI artifacts can function or act itself. It involves adopt and feedback to adjust with the environment. • Linguistics: The study of understanding natural languages by AI artifacts. The natural language processing (NLP) is used widely.

32.1.2 A Short History of AI The term AI was introduced by McCarthy in 1956. But, the funding stone was kept in 1943 by McCulloch and Pitts when they had designed first artificial neuron [31]. The history of AI can be divided in to few time frames. • Gestation Period: The time during 1943–1955 is considered as gestation period for AI development. It was started with the McCulloch and Pitts’s model of first artificial neuron and followed by the artificial neural network consist of 40-neurons designed by Minsky [32]. In the mean time Alan Turing in his Computing Machinery and Intelligence article presented the question “Can machine think” in 1950 [43] and proposed the famous Turing test. So the thought of developing machine to rationally act like human was flourished during this time. • The birth of AI: In 1956 the name was introduced by McCarthy during an 2 month conference involved 10 men at Dartmouth College in Hanover, New Hampshire, and accepted [29]. • Early enthusiasm: During 1952–1969 there were a great expectations from AI but very little bit of that was fulfilled. GPS was introduced by Newell et al. in 1959 [35], a symbolic language called Lisp was invented by McCarthy and Levin [27] (Fig. 32.2).

392

B. Mondal

Fig. 32.2 A time line of important events in AI

• Facing the Reality: The expectation was mounting up now face the reality during 1966–1979. Many of the expectation become real but it also started realizing the limitation in interacting withe ambiance, only syntax analysis is no more enough to understand natural languages and so on. • Knowledge-based systems: during 1969–1979 the early knowledge systems were evolved. The first Expert systems MYCIN was introduce by in 1975 [28, 40]. • AI in Industry: Since 1980 AI entered in the industry and many successful application have been seen. It started using to save human labor, save money and increase productivity.

32 Artificial Intelligence: State of the Art

393

Fig. 32.3 Graphical representation of Turing test

• A Come back of AI: During 1986 AI made a come back with the PDP book by Rumelhart and McClelland [30] and the Connectionist models versus symbolic models [37]. By 1987 a rigorous study on AI was started and new techniques like data mining and machine learning flourished. • Intelligent Agents: in 1995 the Intelligent agents were invented. These can continuously use sensors to response to the environment [45]. In 1997 the AI enabled chess deep blue defeated the world chess champion Gary Kasparov. • Deep Learning: is one of the latest advancement in AI introduced by Ian Goodfellow in 2012 [13].

32.1.3 The Turing Test The Turing test consist of one human questioner (C) and two answering body one of is a machine (A) and another is a human (B), who are isolated from each other like in Fig. 32.3. The questioner asks same question to the machine and a human at the same time and the questioner is unaware which answer is answered by whom. In this situation if the questioner becomes unable to detect the machine for more than 50% of the time then the machine will be called as intelligent [42].

32.2 Applications of AI AI enabled devices can solve complex problems with more accuracy. Some of the use of AI in current world are (Fig. 32.4) • Virtual Assistance: Cortana, Google Assistant and Siri and IBM Watson are some popular virtual digital personal assistance. These are typically understand the user’s

394

B. Mondal Natural language processing (NLP)

Sentiment analysis Language recognition Machine translation Supervised learning Unsupervised learning

Machine Learning

Reinforcement learning Deep learning Data mining Recommander systems

Artificial Intelligence -

Perceiving Reasoning Acting Interacting with human

Decision Making

Computer Vision

Expert systems Case based reasoning

Facial recognition Object recognition Intelligent APIs Process automation

Responding Natural language generation Speech synthesis

Fig. 32.4 Applications ans fields of AI

voice instructions and take action accordingly. Cortana is introduced by Microsoft, Google Assistant by Google, and Siri by Apple [16]. In 2010 Watson defeats champion Jeopardy on an TV show. • Robots are automated AI enabled machines which can work in an environment where survival of humans can be at risk. An annual international competition of robotics named RoboCup was started in 1996 [21] and in 2005 robots play Soccer without headbutting once. • Self Driving cars are the most expected product in near future. These are autonomous vehicles can detect the road condition and take real time decisions. Some successful test of self deriving car was done by Waymo, GM, Tesla, and Uber. In 2005 Stanley, the autonomous car developed by the Stanford university and 3 self deriving cars covers 132 miles on mountain road as a part of project DARPA [24]. As per Elon Musk the CEO, Tesla by the end of 2020 we will have future complete self deriving car. • Vision is the techniques to extract information out of digital images and use it like human visual system. Its widely used in object detection by robots, self deriving cars and medical diagnostics [9].

32 Artificial Intelligence: State of the Art Fig. 32.5 Relationship among AI, machine learning and deep learning

395

Artificial Intellegence

Machine Learning

Deep Learning

32.2.1 AI, Machine Learning and Deep Learning A relationship among AI, Machine learning and Deep learning is presented in Fig. 32.5. AI refers to the systems which can seance, reason and react like human being. In the previous discussion AI is already defined.

32.2.1.1

Machine Learning

These are the systems build to learn from the historical data or from the environment. These system are not program for a task rather programed to learn a task [34].

32.2.1.2

Deep Learning (DL)

DL is a emerging part of machine learning. In DL a vast neural network is trained with huge historical data to do a specific task [12].

32.3 Solving Problems by Searching Search is the innate in AI to find a solution to a problem. Human finds solution by applying perception and experience, innate cognitive abilities, pattern recognition, invariably must turn to search [26]. Each search problem is associated with An initial state: The current state or a set of conditions. State space or search space: Total possible states or combination of conditions. Each state may be labeled as Explored, Frontier, and Unexplored state. Moves or a set of actions: By a move based on some defined strategies the current state may be changes to a new state in the search space.

396

B. Mondal

Fig. 32.6 Classification of some popular search techniques

C C A

B B

Initial state

A goal state

Goal test criteria: A set of criteria given to determine if an state is a goal state or solution. A goal state: Is the target state; definition of a need. A path cost: Every change of state involve some cost. Its calculated by a cost function. Search is used to solve problems like route planning to robot movements in AI. A small example can be demonstrated by the given figure in Fig. 32.6 where we want to move from the initial state to the goal state. Single agent path finding games like 8-tile 3 × 3, 15 tile 4 × 4 Puzzles, 8 Queen puzzles are some popular example of search used in problem solving. Some other important problems are VLSI chip designing, protein formation, route finding for robots. Four important properties of a search technique are: • • • •

Completeness Optimality Time Complexity Space Complexity.

The time and space complexity measured in term of maximum branching factor (b) depth of the solution (d), and maximum depth of the state space (D). Some search techniques does not use any domain knowledge and try to search a goal state using brute force and some used domain knowledge to optimize the search. The search techniques can be classified in two types: uninformed (blind) search and informed (heuristic) search as given in Fig. 32.7 discussed below.

32.3.1 Uninformed Search Techniques 32.3.1.1

Breadth-First Search (BFS)

BFS search for a goal or destination node on a search tree/graph level by level as presented in Fig. 32.8. It traverse each node in a level sequentially until finds the goal node. It use a first in first out (FIFO) data structure for this purpose. BFS always finds

32 Artificial Intelligence: State of the Art

397 Search

No domain knowledge LIFO Depth-first Search (DFS)

Uninformed Search

Informed Search

FIFO Breadth-first Search (BFS)

Uniform Cost Search (UCS)

Best First Search

use domain knowledge

Generate and Test

Priority Queue Depth-limited Search

fixed depth limit

Iterative deepening DFS)

gradually fixed depth limit

A* Search

Priority Queue f (n) = h(n) + g(n)

Fig. 32.7 Classification of some popular search techniques Fig. 32.8 Breadth-first search

A

B

C

D

H

E

I

G

F

J

K

L

M

the shortest path but memory inefficient. Therefore, its not used when the memory is costly [23]. If the branching factor of the search tree is b at level l then maximum possible nodes at level l is l × b. The maximum number of possible node in a search l+1 tree can be 1 + b + b2 + · · · + bl = b b−1−1 nodes. So, the worst case time complexity is O(bl ) and space complexity is also O(bl ). Its used where memory is not an consideration but find the shortest path. Some examples are GPS system, minimum spanning tree, network analysis, bipartite test of graphs.

398

B. Mondal

Fig. 32.9 Depth-first search

A

B

C

D

H

32.3.1.2

E

I

F

J

K

L

G

M

Depth-First Search

DFS search for a goal or destination node on a search tree/ graph level by level as presented in Fig. 32.9. It traverse towards the deepest node first until finds the goal node. It use a last in first out (LIFO) data structure for this purpose. DFS is slower and memory inefficient. Therefore, its not used when the memory is costly. DFS takes exponential time. The worst case time complexity will be O(bk ) and space complexity will be O(bN ) where N is the deepest level in the search space. To avoid the exponential growth the DFS is modified with a depth limit called Depth Limited Search (DLS). This version of DFS can only traverse the nodes within the given depth limit [23].

32.3.1.3

Iterative Deepening DFS (ID-DFS)

ID-DFS apply DFS up to kth level at a time where k = 0, 1, . . . , n. The value of k is increased by 1 at the end of each round of DFS on kth level until it finds the goal node. ID-DFS is designed to accumulate the benefits of BFS (find the goal in minimum depth) and DFS (less memory required). This method gives optimal result among all the uninformed search techniques. The worst case time complexity is O(bl ) [22].

32.3.1.4

Uniform Cost Search (UCS)

UCS was introduced by Dijkstra in 1959. It explore the nodes with minimum cost from the origin. The route cost is usually termed as the total of the costs of the preceding steps (g(n)). Therefore, the newly explored nodes are put into a priority queue in assenting order (the order of (g)) to ensure the cheapest path always explored first. UCS has a time and space complexity O(bl ) which is exponential [7] (Fig. 32.10).

32 Artificial Intelligence: State of the Art Fig. 32.10 Uniform cost search

399 Origin

A

1

B

3

D

C 5

4 1

E

1

F

6

3

G

2

2

H

4 1

I

Goal

32.3.2 Bidirectional Search Assume that the search graph is bidirectional and path from origin to goal and goal to origin is available. Fifteen puzzle, 8 queen problem allow such moves. In this technique one can start form the origin state towards the goal state and from the goal state towards the start state simultaneously; finally meets in some intermediate state. The time and space complexity of this search is O(bl/2 ) where b is the branching factor and the foreword and backward search meet at level l. The BFS would take O(bl ) for the same [3].

32.3.3 Informed or Heuristic Search Techniques The uninformed search techniques are inefficient in most of the cases as it follows a brute force approach. The informed search methods perform better as it uses the problem specific domain knowledge. The informed search techniques apply heuristics before take any move in the search space. So, it involves a heuristic function given by h(n), the optimum cost feron current node n the goal state. For example if our current state is Delhi and goal state is Mumbai the h(n) may be h(n) = euclidean Distance(Delhi, Mumbai) [14].

32.3.3.1

Best-First Search

The Best-first Search is an extended version of UCS. It maintains the a cost function f which returns the cost from the current node to the preceding nodes. The newly explored nodes are kept in a priority queue sorted by f [25].

400

32.3.3.2

B. Mondal

Greedy Search

In this a heuristic function h(n) is used to estimation of cost from the current node to the goal node. Greedy search methods usually finds a good solution in less time. But, sometimes it becomes unable to find any solution even if there exist some solution. So, its a incomplete and not optimal algorithm [18].

32.3.3.3

A* Algorithm

This search algorithm was introduced by Hart Nilsson and Rafael in 1968. A* is an version of Best-first search which defines the cost function f (n) = g(n) + h(n). It can be seen that the search is optimum is the h(n) ≤ h ∗ (n) is admissible and there are finite numbers of nodes, every edge is associated with a cost  > 0. There are hybrid versions of A* like best-first with A*, iterative deeping with A*, Hill Climbing with A*, MA* and SMA* [6].

32.3.3.4

Local Search

This methods only considers a few adjacent nodes to find a local optimum solution. The global optimality of the solution is highly depends on the step size and very hard to achieve as given in Fig. 32.11. Some widely used local search in optimization techniques are Gradient descent, Simulated annealing, Hill climbing, etc [6] (Table 32.1).

Fig. 32.11 A situation in local search

Local maxima

cost h(n)

Local maxima (acts as barrier) Local minima Local minima

Global minima time

32 Artificial Intelligence: State of the Art

401

Table 32.1 Complexity, completeness and optimality of different search algorithms Algorithm Time Memory Complete Optimal BFS DFS ID-DFS Bidirectional Best first A* ID A* Beam search Means end Learning real time A* Real time A*

O(bl ) O(bl ) O(bl ) O(bl/2 ) O(bl ) O(bl ) O(bl ) O(bl ) O(bl ) O(bl )

O(bl ) O(l) O(l) O(bl/2 ) O(bl ) O(bl ) O(l) O(bl ) O(bl ) O(bl )

Y N Y Y Y Y Y N N Y

Y N Y Y N Y Y N N Y

O(bl )

O(bl )

Y

N

32.4 Adversarial Search The best example of adversarial search is game playing. The game playing problems are nontrivial, formal and good reasoning problems. It helps to compare human intelligence with machine intelligence. Most importantly it creates the base for reinforcement learning. In compare to search games problems has a time limit and unpredictable opponents [39]. Strategic games involves moves, set of rules for the moves, cost of each move and minimize cost and maximize profit are the most preferable for AI studies which is represented in Fig. 32.12. Games are involve a finite set of states, an initial state, a function to calculate the successors, a test to decide end of the game, an utility function to track total cost. In this chapter we will discuss about two-players games where one player will tray to maximize the chance to win and other will minimize the chance the loss as demonstrated in Fig. 32.13 (Table 32.2).

32.4.1 Min–Max Min–Max procedure is a suitable process to develop deterministic two players games where both the player have moves, a utility function is used at the terminal or leaf of the game tree. It search in several levels for the solution. The basic idea is to choose the maximum value of highest payoff against the opposition as presented in Fig. 32.14. Min–Max is a complete, optimal for an optimal opponent but if the opponent (human) doesn’t choose the optimum (Min/ Max) then the computer can be

402

B. Mondal Opponent’s Move

Generate new position

True Game over

Return

False Generate Successors

Evaluate Successors

False

Move to highest valued successor

True Game over

Return

Fig. 32.12 Two player game (computer versus opponent (human))

0 0 × × 0 0 × 0 0 × × × −1 0 × −1 +1 0 0 × 0 0 × 0 × × × × 0 × 0 × 0

0 0 × × −1 0 × × −1 0 0 0 × 0 0 × 0 × × 0 0 × × 0 × ×

0 0 × × × × 0 × 0 +1

0 0 × × × 0 0 × × 0

computer’s move (start state) 0 0 × × × 0 human’s move (In0 × termediate state) +1 0 0 0 × 0 0 × × × 0 × × 0 × 0 computer’s move 0 × 0 0 × × × 0 0 × × 0

Fig. 32.13 Game tree for a two player Tic-Tok-Toe game

0 0 × × × 0 human’s move (ter0 × 0 minate/ Game +1 over)

32 Artificial Intelligence: State of the Art Table 32.2 Different types of games Deterministic Perfect information Imperfect information

Chess, checkers, go, othello Battleships, blind, tic-tac-toe

403

Chance moves Backgammon, monopoly Bridge, poker, scrabble, nuclear war

Fig. 32.14 A Min–Max tree for a two player game

exploited. It has a time complexity O(bk ) and space complexity O(bk) considering a BFS way of node exploration.

32.4.1.1

Alpha-Beta Pruning

This is used to prune the nodes from the game tree for which a move will not change the min or max value. For this purpose two variables α and β are initiated to all the nodes. The α always tracks the Max or highest value along the path and β tracks the Min or lowest value along the path. The α–β pruning never affects the result but gives exact same result like full min– max algorithms. The number of nodes to be pruned is depends in the move orders. It reduces the time complexity of min–max to O(bl/2 ).

32.5 Knowledge Representation, Reasoning and Problem Solving Any intelligent system performs task by perceiving data from the environment, then reasoning the data and finally respond accordingly. But, most important is represent the perceived data as knowledge for proper reasoning. This section discuss about different formal ways of knowledge representation like propositional logic, first order predicate logic, and rule based systems. It helps to understand syntax, semantics, va-

404

B. Mondal

Representations:

Sentences

entails

Semantically refers

Real World:

Facts

Sentences

Semantically refers

follows

Facts

Fig. 32.15 Sentences are entails from sentence and facts follows facts

lidity satisfiability, interpretations, models and entailment. Therefor, the knowledge representation is the most important in any AI system [44]. Usually an intelligent agent works based on a knowledge-base which is a set of sentences. The intelligence systems infers new sentence from the knowledge base based on some logic. Main challenge is that logic cant handle uncertain conditions like in reasoning natural language involves some contexts. Logic handles only syntax and semantic reasoning easily. Facts are real inference of any sentence, can be true or false like in Fig. 32.15.

32.5.1 Propositional Logic (PL) We can define a set of variables (like x, Y , etc.) or propositional symbols with some semantic meaning; like A means “its cold”, B means “foggy weather”, and C means “Have tea” etc. In PL a sentence is represented formally as a propositional symbol A which is closed under below logical operations • If A is a sentence then ¬A must be a sentence. • If A and B are two sentences then A ∨ B (OR), A ∧ B (AND), A ⇒ B (implies) and A ⇔ B (if and only if) are also sentences. A sentence can be a finite combination of the above constructs. Lets assume that a knowledge base (Bk ) has sentences A, B and C with propositional logic B ⇒ A, A ⇒ ¬B and B ∨ C. Then some possible inferences are presented in Table 32.3. The truth table shows that the Bk |= C or Bk entails B but Bk not entails A when (A, B, C) = (F, F, T ); i.e. C ⇒ A is false. B ⇒ C is true for every Bk or Bk |= B ⇒ C.

32 Artificial Intelligence: State of the Art

405

Table 32.3 Truth table for the given propositional logic A B C B⇒A A ⇒ ¬B T T T T F F F F

T T F F T T F F

T F T F T F T F

T F F T

F F T

B ∨C

C⇒A

B⇒C F T

T F F

T

T F

F

F T

32.5.2 First Order Predicate Logic This another way of knowledge representation in AI. FOL is also known as predicate logic, allow the to convert sentences in natural language to first order logic (FOL) using some variables, constants and logical symbols. FOS is capable to represent almost all natural language sentences. It can find the objects and relation among them in simple way and can represent the syntax and semantics of sentences. The elements of FOL are • Constant: are noun, pronoun and object name and values like A, X , 1, 45, J im, Delhi, dog, etc. • Variables: are symbols or string which can hold some value like a, x, y, m, p, etc. If some variable belongs in scope them it called as bound variable otherwise a variable will be called as free bound variable. • Predicates: these are relationship among the objects like sister, doctor, adviser, teacher, etc. • Function: are usually verbs which performs some operation on variables and objects like LeftShift, power, etc. • Connectives: these are mainly binary operator like AND (∧), OR (∨), NOT (¬), implies (⇒) and if and only if (⇔). • Equality: it represent the balance between left side and right side (==). • Quantifiers: There two quantifiers universal or for all/ each/ every (∀) and existential quantifier or there exist (∃).

32.5.2.1

Atomic Sentence and Complex Sentence

The atomic sentence are the simplest form of sentences in FOL. It is formed by a predicate and a list of constants or objects in parenthesis. For example “lion and tiger are animals” can be represented as animals(lion, tiger ). The complex sentences are built with multiple atomic sentences connected by connectives.

406

B. Mondal

Like English grammar the sentences in FOL can be divided into two parts: subject and predicate. Like in “Jim is a good man”; “Jim” is the subject and “is a good man” is predicate. A sentence like “All kids like milk” can be represented using quantifiers as ∀qkid(q) → like(q, milk). But if the sentence is “Some kids like milk” the representation will be ∃qkid(q) → like(q, milk).

32.5.3 Rule Based Systems The expert systems infers new knowledge from the existing sentences in Bk . This is done by a inference engine works in either mode foreword chaining and backward chaining. The forward chaining is and bottom up process in which the inference engine starts with an atomic sentence and infers towards to root until finds a goal. The backward chaining is an top down approach in which the inference engine starts from the root and infers knowledge from the Bk towards the bottom, until it finds a goal.

32.5.3.1

Clause

The inference engines needs the sentences in knowledge base in FOL definite clause format. A disjunction of minimum one positive literal is known as horn clause. For example in (¬a ∨ ¬b∀x), x is only positive literal or horn clause which is equivalent to a ∧ b → x. Where as a disjunction with strictly single positive literal is called as strict horn clause or definite clause.

32.5.4 Semantic Nets It is a graph with objects at its nodes and relationships are edges. In this the relationships are mostly represented by is a, has a, part of, instance of etc. An is a represents a inheritance and specialization of objects, A has a used for description of attribute, part of represents a generalization or aggregation of objects, instance of represents membership of a class [15]. Its flexible and simple as uses the natural way of knowledge representation but sensitive to exception and hard for representing procedural knowledge. A semantic net representing the mammals and its instances and objects is given in Fig. 32.16. The semantic net can be used as knowledge base and infer answers for questions like “is rat a animal?”.

32 Artificial Intelligence: State of the Art Fig. 32.16 A semantic net representing the mammals and its instances and objects

407

has

Animal

is an

has

Dog

Vertebra

Fur

has

is a

Mammal

is a

Cow

is a is an

Rat lives on

birds

lives on

Tree

32.5.5 Planning Agents These are problem solving agents enabled with knowledge base and inference engines. These agent first generate a target or goal to reach then search the possible path or a sequence of actions to achieve the goal. Finally follows sequence of actions [38].

32.6 Reasoning Using Statistics The most important part of AI is the reasoning based on probability. Probability is the ratio of number of wanted outcomes and number of possible outcomes. Like if we consider a dice move the probability of each outcome will be 16 . The basic axioms of probability are 0 ≤ P(ri ) ≤ 1 for any outcome ri , The sum of probabilities of all outcomes  P(ri ) = 1, and P(R) + P(¬R) = 1. The concept is highly depends on joint probability and conditional probability.

32.6.1 Joint Probability If we consider two events E 1 and E 2 then the joint probability means the probability of happening E 1 and E 2 at the same same time independently and mutually exclusively. It is given as Eq. 32.1 P(E 1 , E 2 ) = P(E 1 ∩ E 2 ) = P(E 1 ) × P(E 2 |E 1 )

(32.1)

408

B. Mondal

32.6.2 Conditional Probability Considering two events E 1 and E 2 then the conditional probability means the probability of happening E 1 when E 2 already happened. It is given as Eq. 32.2 P(E 1 |E 2 ) =

P(E 1 , E 2 ) P(E 2 )

(32.2)

32.6.3 Chain Rule The conditional probability can be generalized as Eq. 32.3 which is known as chain rule. P(E 1 , E 2 ) P(E 2 ) P(E 1 |E 2 )P(E 2 ) =P(E 2 , E 1 ) P(E 1 |E 2 , E 3 )P(E 2 |E 3 ) =P(E 1 , E 2 |E 3 ) P(E 1 |E 2 ) =

P(E 1 , E 2 , E 3 , . . . , E n ) =P(E 1 |E 2 , E 3 . . . , E n )P(E 2 |E 3 , E 3 . . . , E n ) . . . P(E n )  n    n i−1    P Ei = P E i t| Ej i=1

i=1

k=1

(32.3)

32.6.4 Bayes’ Theorem This theorem is used to find out the probability of an outcome from some already happened outcomes [4]. This theorem is based on the conditional probability given by Eq. 32.4. P(E 1 )P(E 2 |E 1 ) (32.4) P(E 1 |E 2 ) = P(E 2 ) For example a famous problem is stated as: at the time of leaving for summer vacation for seven days, Tony ask his neighbor Roni to water his young plant. There is a 90% probability of dying without water and also 20% probability of dying is there even after proper watering. The probability of Roni will forget to water is 30%. Then what is the chance that the plant will alive till Toni returns? The problem can be well represented in Fig. 32.17. The probability of remain the tree alive will be the sum of probability of being alive with water and probability of being alive without water. As calculated in Eq. 32.5 the chance is 59% but if Roni forgets to water then the chance of alive is only 10%.

32 Artificial Intelligence: State of the Art

409

Fig. 32.17 Decision tree for the dying tree problem

P (D

P(

= W)

0 .7

|W )

= 0.

2

D

W

P (¬

D |W

)=

0 .8

¬D

Start

P (¬

W)

=0

.3

|¬ P (D

W)

= 0.

9

D

¬W

P (¬

D|¬

P(A) = P(W ) × P(¬D) + P(¬W ) × P(¬D) = (0.7 × 0.8) + (0.3 × 0.1) = 0.59

W)

= 0. 1

¬D

(32.5)

Now, If Tony finds the plant is dead after return then the probability that Roni was forget to water P(¬W |D) will be calculated using Bayes’ theorem as in Eq. 32.6. P(D|¬W )P(¬W ) P(D|¬W )P(¬W ) + P(D|W )P(W ) 0.9 × 0.3 = (0.9 × .3) + (0.2 × 0.7) = 0.66

P(¬W |D) =

(32.6)

32.6.5 Bayes’ Net Bayes network is a pictorial representation of conditional probability of related events, represented as directed acyclic graph (DAG). Each node of the DAG represents a variable and associates a table of conditional probability. If a edge or arc exists between two nodes E i and E j then that means the joint probability P(E j |E i ) is factor for E 2 node’s conditional probability table. The joint probability table for a large number of random variable may be very large and hard to use. But a Bayes’ net handles numbers of small tables which increase the efficiency of calculation. The basic concept is derived from the chain rule in Eq. 32.3. The Bayes’ net follows the local Markov property which means a vertex E k is conditionally not dependent K nd to its non-decedents if parent E p is given, that is P(E k |E p , E nd ) ≡ P(E k |E p ) [11]. So, it can be simplified as Eq. 32.7

410

B. Mondal

Fig. 32.18 A Bayes’ network

P(E 1 , E 2 , E 3 , . . . , E n ) = =

n  i=1 n 

P(E i |E 1 , E 2 , E 3 , . . . , E n ) (32.7) P(E i |Par ents(E i ))

i=1

For example lets consider the Bayes’ network in Fig. 32.18.

32.7 Machine Learning ML is the study and techniques to build machine which can learn from the previous data. The ML systems are not programmed explicitly for a particular task but trained for some specific tasks. The learning techniques can be classified in three classes supervised learning, unsupervised learning and reinforcement learning.

32.7.1 Supervised Learning Supervised learning is typically ML with labeled data. The supervised learning techniques involves a training phase and a test phase. The original data set is divided into two parts namely training data set and test data set. Once the machine is trained the test data set is used to test the performance and accuracy of the machine as presented in Fig. 32.19.

32 Artificial Intelligence: State of the Art

411

Fig. 32.19 A flowchart representation of supervised learning

Data Set

test set

training set

not satisfied

learn model

test model

satisfied Trained Model

32.7.1.1

Regression

Regression is the process of estimating discrete value from a continuous variable. For example estimation rent of house, cost of property, age of a person or number of visitor to a event. There are two types of regression: linear regression and nonlinear regression [10]. Linear regression a statistical method to best fit a straight line over all the available data. For estimation of new value it finds the amplitude of a point on the line plotted. For example if we have a data set of flat size in square foot and corresponding cost then we can estimate cost of a flat in that area using liner regression [10]. Logistic regression is a statistical method to best fit a curve line over all the available data, mainly categorical data. For estimation of new value it finds the amplitude of a point on the curve plotted. It is used for classification problems like spam detection, malignant tumor etc. The classification problems may be binary, multinomial, or ordinal logistic regression [10]. For example say for email spam detection output will be 1 if an email is detected as spam or 0 if the email is not spam. We may take a hypothesis that u = wx + b such that hθ (x) = σ (u) the σ is given by Eq. 32.8 which is shown in Fig. 32.26. σ =

1 1 − e−u

(32.8)

412

B. Mondal

where u varies 0 to ∞ predicts output o = 1 other wise output o = 0. The hypothesis can be presented as (32.9) h θ = P(o = 1|x; θ ) where P(o = 1|x; θ ) + P(o = 0|x; θ ) = 1. The calculated probability P is compared with some threshold values defined for each class known as decision boundary. It may be linear or non-linear. The logistic regression is associated with some cost function given by Eq. 32.10.  C(h θ , o(r eal)) =

−log(h theta (x)) if o = 1 −log(1 − h theta (x)) if o = 0

(32.10)

To optimize the cost gradient decent is used widely. 32.7.1.2

Classification

Classification of grouping of data into a finite number of classes based and label them. Classification is possible if the data set is labeled with classes like for a patient data set the disease may be treated as level. The classes are usually categorical unlike continuous in regression. Some widely used classification algorithms are decision tree, Random forest, Naive-Bayes’ classifier, Support Vector machine (SVM), k-nearest neighbor (KNN) [36]. Decision tree are tree built based on if-else rules. A set of if-else rules are extracted from the data set to build the tree and leaves of the tree represents classes. In Fig. 32.20 a Decision tree for bank loan precessing is presented. The selection of splitting attribute in each level is very important and the efficiency of the algorithm highly depends on it [36].

s

Ye Age > 40

Credit score > 850

Y es

Loan granted

No Loan not granted

No Credit score > 750

Y es

Loan granted

No Loan not granted

Fig. 32.20 Decision tree for bank loan precessing

32 Artificial Intelligence: State of the Art

413

Random forest This is an ensemble of multiple decision trees. It classify a new object by all the trees and makes the final decision by vote of majority [36]. Naive-Bayes’ classifier This is a statistical method of classification based on the Bayes’ theorem with a assumption that the features are conditionally independent. It is given by Eq. 32.11 P(k|v) =

P(v|k)P(k) P(v)

(32.11)

where P(k|v) is posterior probability of v belongs to class k, P(v|c) is likelihood, P(k) is class prior probability and P(v) is predictor prior probability. The classification is highly depends on the likelihood computation. From Eq. 32.11 we can have Eq. 32.12. P(k|V ) = P(v1 |k) × P(v2 |k) × · · · × P(vn |k) × P(k)

(32.12)

Support vector machine (SVM) SVM is a technique to separate the classes by computing a vector (line) as a separators. The vector works as a threshold. The new objects are classified based what side of the vector ill falls. A good vector or hypothesis will separate the classes with maximum possible distance as presented in Fig. 32.21 [5]. K-nearest neighbour (kNN) k-NN determines class of a new object based on majority votes among k number of neighbours of the objects as demonstrated in Fig. 32.22. This is a slow and simple

=

·x

0

+

w

b

2

=

1

y

w

w

·x

+

b

w

=

−

·x

1

+

b

w

Fig. 32.21 Support vector machine

x b w

414

B. Mondal

Fig. 32.22 K-nearest neighbour

ClassA ClassB kN N

2

nearest neighbour

y

1 0 −1 −2 −2

−1

0

1

2

x

technique. Determining the suitable value of k is a challenge and computing distance from each neighbour to the new object v consumes extra time every time.

32.7.2 Unsupervised Learning When the historical data are not labeled well and the numbers of possible classes are not preciously known then unsupervised learning is used for grouping the data set into clusters. It is widely used in market segmentation, recommendation engines, social network analysis, medical imaging, image segmentation, search result grouping, and so on. Anomaly detection Clustering techniques like k-means, DBSCAN, c-means etc. are some example of clustering algorithms [17]. There are mainly two types of clustering techniques • Hard clustering: All the objects or elements are strictly member of a single cluster. • Soft clustering: The objects or elements may have membership of a multiple cluster and the membership value is calculated based on the distance between the object and each clusters. Based on the behavior of data and algorithm • Partition based: The tanning data set is partitioned into k numbers of clusters and each cluster is identified by its centroid. Examples are k-means, PAM, k-medoids, CLARANS, CLARA, etc [33]. • Hierarchy based: The tanning data set is divided into clusters based on hierarchy of the objects. It is possible if specialization and generalization are possible among the objects. Examples are ROCK, BRICH, CURE, etc. • Fuzzy theory based: Its like partition based techniques but the objects may have membership to multiple clusters. It produces higher accuracy but there are chance to trapped into local optima. Some examples are FCS, FCM, MM, etc.

32 Artificial Intelligence: State of the Art

415

• Distribution based: It creates clusters based on distribution of the objects. Therefore, when the partition based clusters are surrounded around the centroid, this techniques allow the cluster grow in any direction. Some examples are GMM, DBCLASD. • Density based: These techniques allows the cluster to be of any shape based of the density of similar objects. It used widely in special data segmentation. Some examples are DBSCAN, Mean-shift, OPTICS [1]. • Graph based: These techniques are applicable when data is represented as graph. The nodes are representing the objects and the edges are connection among objects. Social network analysis is an example. Scalability is a challenge for this techniques. Some examples are MST and CLICK. • Grid based: Data is represented as grid. Some examples are CLIQUE and STING. • Fractal based: Based on different geometric shapes the data is clustered. These process has very high scalability and efficiency. One examples is FC. • Model based: Different models are used to generate different clusters. ANN and statistical learning are two popular models. Some examples are GMM< COBWEB.

32.7.2.1

Semi-supervised Learning

The supervised learning needs labeled data and classification of unlabeled data using supervised learning is not possible. But, most of the available data are unlabeled. On the other hand clustering the unlabeled data using unsupervised learning has a big challenge of describing the clusters and determining optimum number of clusters. Figure 32.23 shows the benefit of using unlabeled data with labeled data [8]. In this situation semi-supervised learning becomes useful. In this the model is trained on labeled and unlabeled data simultaneously. Usually large amount of unlabeled data and some labeled data to identify the cluster and determining a optimum numbers of clusters. The main benefit of this is labeling all the data is not required and some unknown pattern or clusters may be discovered. Some of applications are protein sequence analysis, web page classifications and speech classification.

Fig. 32.23 Benefit of using unlabeled data with labeled data

416

B. Mondal Action (Aτ −1 )

Agent

Environment State (Sτ )

Reward (Rτ )

Fig. 32.24 Process of reinforcement learning

32.7.3 Reinforcement Learning In this unlike supervised learning no labeled data will be provided, instead the agent will learn by interacting with the environment. This learning involves a critic or value function who either rewords or penalize the agent for each of its move. The agent try to get high reward and reduce lose every time. Figure 32.24 shows a simple agents action and reword process [41]. One of the popular reinforcement learning technique is Q-learning given by Eq. (32.13). Where L is the learning rate, rτ is reward, δ is discount function, Q(sτ , L τ ) is previous value and Q(sτ , L τ ) is updated value. Q(sτ , L τ ) ← (1 − L) · Q(sτ , L τ ) + L · (rτ + δ · maxa (sτ +1 , L))

(32.13)

Reinforcement learning is used in AI game playing like AlphaGo Zero, in robotics, control system, text symmetrization systems, etc.

32.8 Introduction to ANN The McCulloch-Pitts model was an extremely simple artificial neuron. The inputs could be either a zero or a one. And the output was a zero or a one. And each input could be either excitatory or inhibitory (Fig. 32.25). Where X = [x1 . . . xn ] are inputs or features and b is bias and the variables W = [w1 . . . wn ] indicate which input is excitatory, and which one is inhibitory. These are called weights. So, in this model, if a weight is 1, it is an excitatory input. If it is −1, it is an inhibitory input.  is given by u=

n  i=1

xi wi + b = x1 w1 + x2 w2 + x3 w3 + · · · + xn wn + b

(32.14)

32 Artificial Intelligence: State of the Art

417

Fig. 32.25 A simple model of perceptrons

u = WT X

(32.15)

The activation function may be of two types Unit step function and Sigmoid function. Inputs are mainly feature vector for prediction systems. It may be anything based on the problem.

32.8.1 Unit Step Function (Heaviside Step Function) In this if final sum  or u is less than some threshold T value, then the output is zero. Otherwise, the output is a one. The activation function f (Fig. 32.26)  y = f (u) =

1 u≥T 0 u new> python file shown by Fig. 34.17 • A file named conversation.py has been created as depicted by Fig. 34.18 • Creating the text file for AMANDA CHATBOT. Amanda Chatbot> new> file. Figure 34.19 • This is how our text file appeared as shown by Fig. 34.20 • Installing the package for the AMANDA CHATBOT (Fig. 34.21). Importing the modules of chatterbot. It is a python library that generates automated response for the users input. Following commands will be used From chatterbot import Chatbot

Fig. 34.17 Create a new python file

34 Ai Chatbots: Transforming the Digital World

477

Fig. 34.18 Conversation.py is the python file

Fig. 34.19 Creation of text file

From chatterbot.trainers import List Trainer Bot = Chatbot(“Chatbot”) Open the conversation file with the command conversation open(‘Textfile.txt’,’r’).readlines() • Finally train the bot and put the conversation together.

34.22 Chatbot in Finance Artificial Intelligence has completely changed the outlook of the people looking at the things around. It now depends upon the organization how they can explore the capabilities of the chat bot to explore their business on the front end. Chat bots are

478

S. Paliwal et al.

Fig. 34.20 Content of the text file

preferred by the customer themselves as there is no longer a need to be in a queue at the helpline, one can easily get in touch through the messaging. Financial sector have transformed completely with the entrance of artificial intelligence as chat bots have become a major technology transforming the customer service automation. Below are the advantages that chat bots have provided in the financial industry. Reduction in call centre agents Increase in margins and cost savings. Easy to handle Personalization of content Scalable FinChat bot is an AI powered chat bots for the financial industry. Holly is the virtual assistant that helps in the customer interaction. Holly can interact with several potential clients in different languages anytime. It can be deployed across multiple platforms. Below we present a short interaction with the chat bot Holly in the Fig. 34.22.

34 Ai Chatbots: Transforming the Digital World

479

Fig. 34.21 List of available packages

34.23 Chat Bot in Healthcare Healthcare is another field where chat bots and health are working great together. Chat bots for healthcare can either be a counsellor or a care giver depending upon the functionality of the chat bot. They are helping in providing augmentation and diagnosis. As the healthcare services are becoming patient-centric, providing personalised and satisfactory experience has become a priority for the health care providers. As the medical knowledge keeps on updating so the need of proper understanding of the technology is required. The chat bot SafedrugBot provide with the right information about the drug dosage. Florence another bot launched in 2017 helps in reminding the patients to take pills, track their weight, periods and moods. It also provides the location of nearby pharmacy or a doctor depending upon your disease. Below is the representation of the chat with Florence chat bot by Fig. 34.23.

480

S. Paliwal et al.

Fig. 34.22 Holly the bot

34.24 Conclusion Chat bots have made a great impact on each and every industry and operational domain. Customer service is the major area hit by the chat bot. With the introduction of devices such as Google Home and Amazon Echo, artificial intelligence has already hit a major area of our daily lives, thereby growing the space of artificial intelligence in workspace [16]. Chat bots are providing a seamless engagement experience to customers across all the platforms. They have enhanced interaction and also helped in managing large amount of information. Continuous improvement in the natural language processing has enabled the bots to have a conversation like the human being. Performing simple jobs in a repetitive and efficient manner have made the chat bots to build an organization fatigue-free. They have made the human agents to handle complex queries. The global Chatbot market is expected to increase by $1.34 in valuation by 2024. Chat bots have come up

34 Ai Chatbots: Transforming the Digital World

481

Fig. 34.23 Florence—Medical assistant bot

with a maximum number of opportunities that offers personalization. For ensuring 100% of customer satisfaction, we can not rely in technologies completely at some point human intervention is required, But still chat bots provide a logical, transparent and clear communication. So if we take into account the current market trends then we can say that chat bot have a great future ahead.

References 1. 2. 3. 4. 5. 6.

7. 8. 9. 10.

11.

Shawar, B.A., Atwell, E.: Chatbots: are they really useful? Ldv forum 22(1), 29–49 (2007) https://www.alexa.com https://www.apple.com/in/siri/ Dale, R.: The return of the chatbots. Nat. Lang. Eng. 22(5), 811–817 (2016) Khan, R., Das, A.: Introduction to chatbots. Build Better Chatbots, pp. 1–11. Apress, Berkeley, CA (2018) Klopfenstein, L.C., Delpriori, S., Malatini, S., Bogliolo, A.: The rise of bots: a survey of conversational interfaces, patterns, and paradigms. In Proceedings of the 2017 Conference on Designing Interactive Systems, pp. 555–565. ACM (2017, June) Ferrara, E., Varol, O., Davis, C., Menczer, F., Flammini, A.: The rise of social bots. Commun. ACM 59(7), 96–104 (2016) Cahn, J.: CHATBOT: architecture, design, and development. University of Pennsylvania School of Engineering and Applied Science Department of Computer and Information Science (2017) AbuShawar, B., Atwell, E.: ALICE chatbot: trials and outputs. Computación y Sistemas 19(4), 625–632 (2015) Rahman, A.M., Al Mamun, A., Islam, A.: Programming challenges of chatbot: current and future prospective. In 2017 IEEE Region 10 Humanitarian Technology Conference (R10-HTC), pp. 75–78. IEEE (2017, December) Følstad, A., Brandtzæg, P.B.: Chatbots and the new world of HCI. Interactions 24(4), 38–42 (2017)

482

S. Paliwal et al.

12. Burden, D.J.: Deploying embodied AI into virtual worlds. In International Conference on Innovative Techniques and Applications of Artificial Intelligence, pp. 103–115. Springer, London (2008, December) 13. Shum, H.Y., He, X.D., Li, D.: From Eliza to XiaoIce: challenges and opportunities with social chatbots. Front. Inf. Technol. Electron. Eng. 19(1), 10–26 (2018) 14. Hill, J., Ford, W.R., Farreras, I.G.: Real conversations with artificial intelligence: a comparison between human–human online conversations and human–chatbot conversations. Comput. Hum. Behav. 49, 245–250 (2015) 15. Brandtzaeg, P.B., Følstad, A.: Why people use chatbots. Int. Conf. Internet Sci., pp. 377–392. Springer, Cham (2017) 16. Khan, R., Das, A.: Build better chatbots: a complete guide to getting started with chatbots (2017) 17. Abdul-Kader, S.A., Woods, J.C.: Survey on chatbot design techniques in speech conversation systems. Int. J. Adv. Comput. Sci. Appl. 6(7) (2015) 18. Bird, S., Klein, E., Loper, E.: Natural language processing with Python: analyzing text with the natural language toolkit. O’Reilly Media, Inc. (2009) 19. Kerlyl, A., Hall, P., Bull, S.: Bringing chatbots into education: towards natural language negotiation of open learner models. In International Conference on Innovative Techniques and Applications of Artificial Intelligence, pp. 179–192. Springer, London (2006, December) 20. McTear, M.F.: Spoken dialogue technology: enabling the conversational user interface. ACM Comput. Surv. (CSUR) 34(1), 90–169 (2002) 21. Pearl, C.: Designing Voice User Interfaces: Principles of Conversational Experiences. O’Reilly Media, Inc. (2016) 22. Radziwill, N.M., Benton, M.C.: Evaluating quality of chatbots and intelligent conversational agents. arXiv preprint arXiv:1704.04579 (2017) 23. Janarthanam, S.: Hands-on chatbots and conversational UI development: build chatbots and voice user interfaces with Chatfuel, Dialogflow, Microsoft Bot Framework, Twilio, and Alexa Skills. Packt Publishing Ltd. (2017) 24. Paikari, E., van der Hoek, A.: A framework for understanding chatbots and their future. In Proceedings of the 11th International Workshop on Cooperative and Human Aspects of Software Engineering, pp. 13–16. ACM (2018, May) 25. Zumstein, D., Hundertmark, S.: Chatbots–an interactive technology for personalized communication, transactions and services. IADIS Int. J. WWW/Internet 15(1) (2017) 26. Chung, K., Park, R.C.: Chatbot-based heathcare service with a knowledge base for cloud computing. Cluster Computing 1–13 (2018) 27. Zamora, J.: Rise of the chatbots: finding a place for artificial intelligence in India and US. In Proceedings of the 22nd International Conference on Intelligent User Interfaces Companion, pp. 109–112. ACM (2017, March)

Chapter 35

Applications of Smart Devices Prabhsimar Kaur and Vishal Bharti

Abstract In this chapter we have tried to explore how IOT, connected devices and automation shall bring about a revolution in the field of agriculture and tremendously improve nearly every facet of it. IOT in agriculture is the amalgamation of Information technology, telecommunications and sensor technology. Agriculture has been a neglected field in India as far as automation and technological use and applicability are concerned due to lack of funds for technological expansion and limited technical expertise for implementation of the available technologies. Agro based economy like India being perfect for implementation of smart agriculture; Organizations are focusing on improving accessibility of low cost IOT sensors and providing more scalable and cost effective solutions. Smart farming techniques encompass a number of applications such as Precision Farming, Smart Greenhouses, Livestock Management, and Agriculture Drones and Farm Management Systems. The use of smart & connected devices in agriculture have helped farmers gain better control over their farm produce, help make climatic predictions and raise their livestock’s more efficiently. The use of IOT for digitization of farms has caught the attention of Government of India and has been included in its policy for Digital India. A recent study by Statistic shows that smart agriculture shall take up $26.76 billion of global market size by 2020. According to a report by NASSCOM, India has around 40 startups dealing in smart agriculture.

35.1 Introduction The recent studies and statistics reveal that with the global population on a rise, it shall be difficult to feed the world population. As per the reports by Food and P. Kaur (B) · V. Bharti Department of Computer Science and Engineering, DIT University, Mussoorie-Diversion Road, Village Makkawala, Dehradun, Uttrakhand, India e-mail: [email protected] V. Bharti e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_35

483

484

P. Kaur and V. Bharti

Agriculture Organization, the food grain requirement shall be touching 3 billion tons by 2050 [1]. To feed such a huge population there is a need for the agriculture industry to be automated, by adopting IoT in the different areas and aspects of farming and agricultural methods shall make the whole process more efficient and effective. The food production needs to be raised by nearly 50% in the next 50 years to maintain the per capita supply. Water quality and climatic degradation being a global concern now also demands modernization and automation of agriculture and farming techniques [2]. IoT is a network of things, helping in clear identification of elements with help of sensors, software intelligence and connectivity to the internet. IoT reduces the human interaction, collects and processes the data using sensors, controllers and actuators [3]. Traditional approaches of weather predictions, cattle management, soil monitoring, environmental impact etc. are not very automated and involves inefficiencies, IoT helps in automation of these traditional approaches, and helps eradicating the challenges thus enhancing the overall farm yield. Rapid growth in cloud technology combined with Internet connectivity has led to rapid increase in implementation of IoT applications in farming and agriculture. IoT has proved to be a driving force behind increased agricultural production at lower cost. If we go by the recent reports the use of IoT applications and installation of IoT devices shall see a compounded annual growth rate of 20% in the agriculture industry. The number of connected devices in agriculture shall grow to 225 million by 2025 [4]. Smart Farming is the breakthrough implementation of science and technology in the agriculture sector. Smart farming involves the implementation of technologies like IoT, Big Data Systems and analytical systems on the agricultural fields. Technologies like Internet of Things, cloud computing, Machine learning, and Big Data are used to enable farmers to have deep insights on the consequences of their actions, hence enabling them to take much better and informed decisions on the farming practices. Smart Agriculture systems are currently being used for recording and collection of data in relation to various environmental parameters such as soil levels, climatic conditions, moisture levels, nutrient levels etc. Efficient and accurate data collections through the deployment of wide range of sensors are making farm processes more efficient. Sensor Technology also reduces human interaction in farming techniques, thus improving efficiency. IoT based farm management systems are essential for any agriculturally based country. IoT based water management systems, through the use of installed sensors help collect data on environmental attributes such as temperature, water levels, and humidity levels thereby providing accurate irrigation timings. Crop Management using IoT provide systems of monitoring of temperature, humidity, soil thus providing farmers with data to manage their crops efficiently. Climatic changes and water scarcity becoming major challenges for countries around the world, Wireless Sensor Networks (WSN), are being used in water management applications, owing to their design and ease of deployment [5].

35 Applications of Smart Devices

485

35.1.1 This Chapter Explains Why there is a Need to Study How Smart Farming is Transforming Agriculture. Why Should the Farmers Make a Shift from Traditional Methods of Farming and Adopt IOT in Farming. The 5 Key Aspects IOT Can Transform Agriculture Are Described Below 1. Improves Data Collection: Tons of data can be collected through installation of smart sensors, data pertaining to weather conditions, soil health, cattle health, monitoring of crops health. This data helps improve state of your business and increase efficiency. 2. Better Internal Control and Low Production Risks: IOT provides an ability to foresee the output of your production. This provides the farmers an opportunity to better plan their product distribution. If the farmers know how much crop they shall be harvesting, they shall plan their production distribution in such a manner that the harvest doesn’t lie around unsold. 3. Waste Reduction and Cost Management: The use of technology provides an ability to foresee any anomalies in crop yield and livestock health thus mitigating the risks of losing yield. 4. Process Automation: The use of smart devices automates multiple processes across the production cycle. Automation of irrigation, pest control, soil monitoring, fertilizing etc. helps enhance overall farming efficiency. 5. Production Quality and Volume gets enhanced: The deployment of smart devices in the fields provides the farmers the ability to gain better control over their production processes, help maintain higher standards of crop quality and increase the overall crop production. In developing countries like India, where 70% of the population are connected with agriculture and its allied sectors, and agriculture being a major contributing to their GDP (Gross Domestic Product), the farmers just receive 10–23% of the price the Indian consumer pays for the produce, the difference being taken up by losses, inefficiencies, poor production distribution, middlemen. Whereas, farmers in developed countries like Europe, USA receives 64–81% of the price [6]. IOT through smart farming is transforming the agriculture industry rapidly and with advanced technologies in software and hardware developing solutions to reduce wastages and cost, and enable proper production distribution. Smart farming offers the following technologies [7] to the present day farmers that are represented in the below diagram and explained below (Fig. 35.1): 1. Sensing Technologies: This technology provides soil monitoring, water monitoring, prediction of climatic conditions, temperature management. 2. Software Applications: Analyzing the needs of the farm software applications are developed to help farmers keep track of their crop production, cattle health, livestock management etc.

486

P. Kaur and V. Bharti

SENSING TECHNOLOGIES

TELEMATICS, POSITIONING TECHNOLOGIES

SOFTWARE APPLICATIONS

COMMUNICATION SYSTEMS - CELLULAR

HARDWARE AND SOFTWARE SYSTEMS

DATA ANALYTICS SOLUTIONS

Fig. 35.1 Smart farming technologies

3. Communication Technologies: The data collected by sensors are actuators are collected and analyzed through vide range of Communication technologies. 4. Positioning Technologies: These systems help in field mapping, soil mapping, crop scouting etc.GPS systems and tools are types of positioning technologies that are helping farmers to work in farms during low visibility conditions such as rains, dust, darkness and foggy conditions. 5. Hardware and Software Systems: IOT in agriculture comprises 30% Hardware, 30% Software and 40% Communication Protocols. Hardware’s square measure set of devices that square measure responsive in nature and have the potential to retrieve knowledge. IOT Software’s square measures are programs that permits knowledge assortment, storage and process manipulating and instructing to and from IoT hardware. Operating Systems, apps or firmware are examples of IOT software. 6. Data Analytics: Big Data and Machine Learning technologies are being used to analyze data to make action oriented conclusion-able intelligence. Data analytics helps farmers to make data based decisions such as which crops to plant, providing data on rainfall percentage, soil health etc.

35.2 What Essential Things the Farmers Should Take into Consideration Before Adopting the Smart Farming Solutions In this Chapter we have been explaining about how the Smart Farming Techniques are transforming the Agriculture Industry. There are many ways smart devices can help in increasing in the farm’s performance and revenue. But the setting up of the IOT App’s development system is not an easy task. There are certain things which the farmer should take into consideration before investing in smart farming [8]: 1. Hardware: The most essential thing for building an IoT based structure for agriculture is to choose the right hardware, i.e. choosing the right sensors of

35 Applications of Smart Devices

2.

3.

4.

5.

487

your device depending upon the needs of the farm. Sometimes these sensors are readily available but sometimes they need to customize according to farm needs. The choice of sensors depends upon the type of information to be collected and the purpose of solution. The accuracy of data collected and its reliability is crucial for the success of the IOT structure. The Brain: Big data and Machine Learning technology systems are being used for analysis of data. The collected data will make sense only if we can analyze it properly. Thus predictive algorithms and machine learning techniques are being applied in order to make action oriented conclusion-able intelligence. Maintenance: While choosing the hardware, one must ensure that the hardware used in IOT based structure for agriculture must be durable and easy to maintain as the sensors used in the fields can easily be damaged. If the sensors are not chosen wisely one may end up paying high replacement cost for sensors. Mobility: Smart farm applications should be linked and connected in such a manner so that they can be remotely accessed via a smartphone or desktop computer by the farm owners. The framework provides enough wireless range to the server which makes ease for several devices to get connected with each other and exchange information and data with other devices. Infrastructure: A solid internal infrastructure is essential to ensure that your smart farming applications perform well. The internal systems should be secured enough so that they decreases the likeliness of anyone hacking into the system, stealing the data or even taking control of the applications [9].

35.3 Components of Smart Farming Smart Farming comprises of several network technologies which are interdependent, interdisciplinary and complementing each other. Components of Smart farming comprises of the following:

35.3.1 Management Information Systems MIS are systems where all data collected from multiple sensors and actuators are captured, stored and analyzed for action. Optimal MIS systems offers information on: (a) Crops: Crop Stress, Crop Yield, Population of Crop, Weed patches, Crop Health, Crop Nutrients, Fungal or Insect infestation. (b) Soil: Soil Moisture, Soil Nutrients, Soil Texture, Physical Condition of Soil & More. (c) Climatic Conditions: Humidity levels, Wind Speed, Rainfall, and Temperature.

488

P. Kaur and V. Bharti

35.3.2 Devices ‘Smart’ in Smart Farming denotes technology, technologies like Machine Learning, Big Data and Data Analytics makes the whole process and setup of Smart Farming meaningful. The ‘Smart’ devices comprises of sensors, professional agriculture Drones/UAV’s, cameras transmitting real time images and data and other devices using the remote sensing technology. Geographic Information systems for providing better positioning. IoT systems with devices having capability to communicate with each other in the network and provide real time updates on water levels, crop health, soil conditions, moisture levels etc.

35.3.3 Application of Smart Devices in Farming IOT systems generates enormous amount of data, known as big data, in varying data quality. The accurate and correct analysis of this data using the machine learning technology is the key to advancing smart IoT utilization. Some of the vital Agriculture and farming sub-verticals where IOT applications are being used in order to enhance productivity and efficiency can be listed as below: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

Water Management Crop Management Smart Farming Livestock Management Irrigation and Soil Management Greenhouse Management Weather Monitoring Green House Management Smart Lightening Waste Management etc.

The IOT applications and smart devices being used in agriculture and farming sub-verticals are being explained in detail: A. Irrigation Management During the last few years, the scarcity of water levels and resources in most of the developed and developing countries has made the sub-vertical of water management of utmost importance. Irrigation management systems influences crop production with its central focus being to increase productivity [10]. Irrigation management systems helps determine future irrigation expectations. The irrigation management systems aims at using the water resources in the most profitable and sustainable manner [11]. Many methods have been developed for the conservation of water. Agriculture is a field where water is required in massive quantities. The most major problem faced

35 Applications of Smart Devices

489

in agriculture is lack of water management and excess of water being given to the fields. Many systems have been developed to control the wastage of water like Ditch Irrigation, Terraced Irrigation, and Sprinkler System Rotary Systems. Automatic Smart Irrigation Decision Support System, SIDSS, are determined for managing irrigation in agriculture. These systems evaluates both the soil and climatic variables through several autonomous sensors deployed in the field and determines the irrigation needs of a plantation. These systems help farmers to know the status of their fields at their home or in any part of the world. Automation not only provides comfort but also focuses on reducing energy, efficiency and time saving. Some of the IOT based Agriculture sensor systems are detailed below: 1. Automated hydroponics: Bitponics Hydroponics systems simplifies and automates the plant growing process, it’s an innovative approach that is ideal for environments where plants can’t grow naturally like areas with harsh climates or indoor spaces of apartments. It is a strategy for developing plants without soil, the supplements are given to the foundations of the plant through a supplement arrangement. The root frameworks don’t go far, so more plants can be developed in a similar measure of room. Bitponics is your alternate way to a green thumb. It’s intended to be an extra existing part to any current water system framework, and it will disentangle and computerize your developing procedure [12]. Bitponics will automatize anything which will be controlled by outlet, like water pumps and lights. Bitponics is a task that is comprised of two sections a sensor gadget and backend web administration. The framework works by first entering the subtleties of sort of plant and hydroponic framework that we need to develop, at that point the framework catches this data to produce a custom developing arrangement. A custom growing plan is a step by step guide of how to take care of our plants i.e. (i) how many hours of light a plant needs per day, (ii) The pH range safe for your plants, balance of nutrients required by the pants, when to replace the nutrient solution. The Bitponics then making use of its sensors, such as water & air, humidity, light and pH sensors shall collect data and log them to the online account of the user. The user can even log data manually and upload pictures to track the growth of the plant. The gadget highlights two electrical plugs that can be controlled over web. Bitponics gadget wants an expense of around $250 and up. One buy of the gadget one become an individual from the online Bitponics people group. The client will probably store 1 years of developing history and many photographs on the Bitponics site and furthermore can impart their developing arrangement to the Bitponics Community (Fig. 35.2). Sensor Used: Water/Air Temperature, Humidity, pH, Brightness

490

P. Kaur and V. Bharti

Fig. 35.2 Bitponics device along with its dashboard. Source https://www.kickstarter.com/projects/ 1498890810/bitponics-your-shortcut-to-a-green-thumb

2. Botanicalls Botanicalls is an IOT supported device, it is a project about communication between the plants and humans. It opens up new Channels of Communication and is an effort to promote inter-specie understanding. The project has opened up avenues of interaction and the plants that were being neglected have been given the ability to call and text message people requesting for assistance [13] (Fig. 35.3). Botanicall kits helps human by giving an association with your plant through online Twitter update or cell phone. At the point when your plant needs water, it will post to tell you, and send its thanks when you show it adore. Botanicall units are empowered by AT Mega368 microcontroller and can be made by DIY way. Sensor Used: Air Temperature, Humidity, Brightness, Soil Moisture

Fig. 35.3 Botanicall kit along with its components. Source https://www.botanicalls.com/kits/

35 Applications of Smart Devices

491

Fig. 35.4 Edyn: Smart garden tracking system. Source Edyn, https://www.kickstarter.com/projects/ edyn/edyn-welcome-to-the-connected-garden

3. Edyn Edyn is a smart garden tracking system that monitors and keep track of your garden, keeping you connected to your garden at all times. Edyn tracking systems has sensors which are inserted in the soil capturing and providing data about weather and soil conditions. One can always keep track of its garden through the Edyn app, which provides a real time snapshot of the farms and helps in maximizing the plant health. These systems provides information to the farmers about the fertilizers to be used, water requirements of the plants thus helping the plants thrive [14]. Edyn water valves gets connected to the hose or sprinkler systems and smartly manages your water systems, maintaining the moisture level of plants and avoiding overwatering. The sensors captures the information and steams the data over the internet to the Edyn Cloud. The Edyn app provides information about which plant to grow, optimal climatic conditions and soil conditions and water needs of the plant. Edyn app continually monitors conditions and alert you to changes that require immediate action (Fig. 35.4). 4. Plantlink Plantlink is a wireless plant sensor that monitors soil humidity for both indoor and outdoor plants. It makes watering easy for the gardener as all the measured data is sent to a gardener’s phone and just by 5–10 min of reading, he can easily track plant’s conditions, set alerts when they need water and can schedule automatic watering times. Plantlink system has a built in catalog of over 50,000 plants with which it automatically knows how much water needs to be supplied for particular plant. Its installation is very easy just as Wi-Fi in our house and monitoring can be done from anywhere just by installing Plantlink App in your phone [15]. Algorithms that are used in the system are plant-specific. It has a long battery life and can monitor up to 64 plants to 100 yd range with a single base station (Fig. 35.5).

492

P. Kaur and V. Bharti

Fig. 35.5 Plantlink: plant sensor device. Source http://thewifigarden.com/plantlink-soil-sensorreview

5. HarvestGeek Harvest geek is an open-source, Internet of Things Greenhouse Monitoring system for garden. It is an easy system that provides automation and optimization features for the farmers. On HarvestGeek Software, one can easily upload pictures of the plants, ask any questions, and can get detailed analysis of your plants. Farmers can also get alerts through email or notifications on mobile phones or tablets. HarvestBot Is an electronic device that is used in HarvestGeek system to monitor the key environmental conditions to improve the yields. It uses sensors like soil, moisture, temperature, ambient light, etc. to monitor environmental condition of the garden and communicate the real-time gathered information back to the cloud servers [16]. Each sub unit in Harvest Geek system is specialized to perform its own task that makes the system versatile to be used in large commercial indoor operations (Fig. 35.6). 6. Wi-Fi Sprinkler System Iro Iro or Rachio is a wireless smart sprinkler controller designed to automatically adjust according changing local weather. It can be easily monitored by using smartphones from anywhere. The Rachio app is used to easily control and monitor your outdoor water use that serves as your personal watering assistant [17]. The installation of Rachio smart sprinkler controller takes less than 30 min to install and is very easy as it is compatible with almost any irrigation system. You just need to swap your old controller with Rachio controller without changing existing sprinkle heads and pipes. No special expertise, digging or plumbing is required in its installation. Last step after installation is to connect it with your Wi-Fi and you are ready to use it. One can also add leak detection and continuous flow monitoring from the app (Fig. 35.7). 7. Spruce Irrigation Controller Just like Rachio, Spruce is also a smart irrigation controller device but it is zone specific. Using Spruce, plants get appropriate amount of water when needed according

35 Applications of Smart Devices

493

Fig. 35.6 HarvestGeek system. Source https://www.kickstarter.com/projects/2077260917/ harvestgeek-brains-for-your-garden

Fig. 35.7 Rachio device: smart sprinkler controller. Source WiFi Sprinkler System: Iro, https:// rachio.com/

to specific time. It combines both real time soil moisture sensors and temperature sensors to automatically best schedule for the garden, with which it saves up to 60% of water usage which is wasted using conventional time-based schedule [18]. Spruce controllers are compatible with zigbee and are inserted at depth of 4–6 inches below the surface so that sensor can measure the soil moisture 3–5 inches below the surface which helps in promoting healthier plants. User can easily check the status by just clicking on their smart phones (Fig. 35.8).

494

P. Kaur and V. Bharti

Fig. 35.8 Spruce device: smart irrigation controller. Source http://www.spruceirrigation.com/

B. Weather Monitoring Monitoring of the weather and climatic conditions is another important sub-vertical of smart farming. Weather stations gadgets are the most popular in smart agriculture. These weather stations combining various sensors are installed in the fields. The technology used in weather stations is cloud technology which collates data from the environment and sends it to the cloud. The data collected helps in mapping of the climatic conditions and help choose the appropriate crops, and take appropriate measures to increase crop production. According to studies 25–30% of the food losses in developing countries occurs during food production. Climatic conditions and dependency on rainfalls being a major factor for these production losses. Thus, the importance of weather monitoring in farming all the more increases. A farmers can’t battle the climate. Be that as it may, he can receive the given circumstance and take extra homestead the board practices to limit crop misfortunes. In this way, precise data in regards to the climate is significant so ranch exercises can be arranged without antagonistic occasions. Monitoring constant climate conditions like air and dew temperature, precipitation, and dampness is the most ideal approach to ensure crops and secure a high and sound yield. Outrageous climate, for example, dry season, flood, hail, or ice can cause moment plant pressure, in this way prompting

35 Applications of Smart Devices

495

bombed creation and expanded expense. The best way to oversee and guarantee beneficial harvest creation is precise climate estimates. B.1 Climate Monitoring Determines Farmer’s Success Yield development is firmly identified with the climate. A few yields require certain high or low temperatures to begin their germination and proceed with further advancement. Then again, temperatures in mix with moistness are regularly used to foresee different creepy crawly irritation and sickness events. In light of this data, ranchers can plan times for planting, insurance, collecting, and other field exercises so as to stay away from negative climate impacts and yield misfortunes [19]. Climate observing offers a lot of estimation alternatives to give helpful data about soil and yield conditions, some of which are: Air and soil temperature, Relative mugginess, Soil dampness, Rainfall, Wind speed/course, Evapotranspiration. Cultivating dependent on climate information is essential to effective ranch the board. All the more critically, it guarantees maintainable cultivating, consequently securing the earth. B.2 IOT Based Weather Monitoring Systems In the agriculture zone the monitoring of the weather conditions plays a very crucial role in planning & determining the crop growth. The old age wired and analog devices are not suggestible to use during extreme weather conditions, thus sensor based devices using the technology Internet of Things (IoT) plays a very crucial role in making the climatic predictions. 1. allMETEO allMETEO are weather management systems, using state of the art sensor technology and Internet of Things (IOT) for making weather predictions. The technology used in weather management systems is cloud technology which collates data from the environment and sends it to the cloud. The data collected helps in mapping of the climatic conditions and provides access to the farmers, the farmers by logging into their account, through their phone or pc can monitor the weather predictions and plan their crops accordingly [20]. weather.allMeteo.com local weather portal provide the farmers with features such as: • • • •

data privacy and data sharing settings tagging of weather station sharing and exchange of data amongst the farmers, friends and neighbors provides access to live data as they arrive from the weather stations.

allMETEO, weather management systems make use of the following Sensors & Products: (a) Radiation Shields (b) Temperature and Humidity Sensors

496

P. Kaur and V. Bharti

Fig. 35.9 allMETEO weather station with its portal. Source https://www.allmeteo.com/agricultureiot-weather-station

(c) (d) (e) (f)

Wind Sensors Sun Sensors Rain Measurement Sensors Pressure Sensors.

With the METEOHELIX IOT Weather Stations installed in the fields, the famers can track temperature changes every 10 min on weather.allmeteo.com through their phone or computer (Fig. 35.9). 2. Smart Elements Keen Elements are climate the board frameworks giving a scope of items, boosting the time by disposing of the need of the ranchers to physically check their most significant resources on the fields. A wide scope of sensors are introduced on the field, which are associated with an online dashboard. The sensors gather information and transmits to these online dashboards, which further procedures it and makes it accessible for the ranchers. Setting up these systems on the field is an easy process, requiring Base Stations & Blue Node [21]: (a) Base Station The Base Station are basically the heart of the connected farms. These base stations are being connected to the web via 4G/LTE, Wi-Fi and actively listens for any Smart Elements sensors reporting via LoRa long range, low power radio technology. Simple to install and set up these base stations help the farmers in creating their own networks across the farms (Fig. 35.10). (b) Blue Node The Blue Node could be outstanding piece of engineering that enables you to line your sensors out in the field, and then not worry about checking them again for years.

35 Applications of Smart Devices

497

Fig. 35.10 Base station device. Source https:// smartelements.io/

With low power design, rugged body and connectors, and utilizing the latest in long range communications, the Blue Node reports your sensor data back to the Cloud via the Existing LoRaWAN network with no need of any complex setup—simply insert your sensor into plug and tighten the bracket bolt (Fig. 35.11). Smart Elements make use of the following Sensors and Products: (i) (ii) (iii) (iv) (v) (vi) (vii) (viii) (ix) (x) (xi) (xii)

Rain Gauge Sensors Wind Speed and Wind Direction Sensors Temperature and Humidity Sensors Soil Monitoring Probes Electric fence Monitors Ultrasonic Distance Sensors Leaf Wetness Sensors Water Level Pressure Sensors Flow Rate Sensors Soil Moisture Probes Air Temperature and Humidity Probes Compass and Inclinometer

498

P. Kaur and V. Bharti

Fig. 35.11 Blue node. Source https://smartelements.io/

Smart Elements Systems are helping the ranchers set aside cash and settling on convenient and educated choices dependent on constant conditions. These frameworks have empowered ranchers to screen their water tanks remotely, measure the dirt substance and patterns in temperature, Record exact climate conditions, Store and fare their homestead information to different frameworks.

35 Applications of Smart Devices

499

Fig. 35.12 Pycno sensor kit. Source https://pycno.co/ sensors

3. Pycno Systems Pycno are sensor systems that are installed in the fields, these sensors with full autonomy, solar panels and internal battery, can be easily installed in the fields and are cost effective. The sensors form a network with each other [22]. There are two type of sensors: 1. Master Sensors: Master sensor goes about as a passage to the homesteads, it gathers information from all the sensor readings and pushes them to the web utilizing cell systems (Sim Cards). 2. Node Sensors: Node Sensors has similar functionalities with the Master Sensor, but instead of pushing the data to the internet using cellular networks, it works in conjunction with the master. This sensor system acts as a gateway for the master sensors (Fig. 35.12). It takes a bunch of Node Sensors utilizing one ace to drive the information out of the homesteads. The greatest number of groups can be introduced to send the information as one ranch. These Sensor Systems captures and provides the following: 1. The Solar panels installed in the systems captures the solar radiation i.e. the amount of sunlight the plants are getting. 2. Humidity levels and Air Temperature are measured through these systems, the readings of which are further used in evapotranspiration and disease models. 3. Soil temperatures are recorded through these sensor systems which are useful for root health of the plant and sensitive plants like flower buds. 4. Moisture levels of the soils can be measured and monitored through these sensor systems, can be daisy-chained up to 1.2 meters. C. Greenhouse Automation In recent scenario of climate change and its effect on the environment has motivated the farmers to install greenhouses in their fields. Automation helps to survey an

500

P. Kaur and V. Bharti

environment for your crops that can supply reactive solutioning to outside influences that will otherwise jeopardize the yield of your crops. If watering manually, it can be easy to overwater or underwater your plants. To be accurate you would need to stick to a good schedule to make sure there were no inconsistencies. Ensure accuracy by setting a schedule which can be adjusted based on type of crop, quality, yield and local weather station parameters [23]. Your water run-off can tell you about the health of your plant, an automated control system uses sensors to monitor and measure your run-off for you. The data gathered can then be used to ensure compliance with local legislation if required and give you an accurate measurement of expenditure. A great automation system will also monitor your root zone, your climate, send alerts when there’s a problem and allow you to manage everything by remote access on any device. A good Green House Automation System will manage factors like: • • • • • • •

vents heating cooling lighting temperature CO2 irrigation

C.1 The benefits of having an Automated Greenhouse systems are many 1. Helps reducing labor costs and using the manpower more efficiently. By reducing the amount of work that needs to be done manually, labor costs can be reduced significantly and the workers can be freed to focus on other areas of importance and thus add new skills. 2. The use of technology and equipment’s help the farmers to take more data driven decisions and not just take decisions based upon assumptions. 3. The use of technology and equipment’s helps in providing information to farmers, this knowledge helps the farmers to plan their crops and thus increase the quality and yield of their crops. C.2 IOT Based Green House Automation Systems Companies across the globe with a group of dedicated software designers, engineers and horticultural technology experts are working hard to make automated systems available to farmers at affordable costs. Autogrow Systems Ltd. with its offices across 40 countries is one such Company providing fully automated greenhouses ranging from a single compartment environments through to large-scale systems. C.3 Climate Controllers Climate Controller frameworks oversees almost all parts of the developing condition for example temperature, lighting, CO2 , air, supplements, dampness and so forth.

35 Applications of Smart Devices

501

Regularity has and will consistently be a factor in the cultivating part, Climatic and Environmental Controllers gives the information and visibility of the progressing conditions and adjusts and acclimates to the harvest needs, along these lines helping ranchers to have the option to design and strategize adequately for future yields [24]. 1. Intelliclimate A product offering of Autogrow Systems Ltd., Intelliclimate is a complete climate controller, using every peripheral available to meet the farms climatic needs. Intelliclimate enables the ranchers to deal with all parts of the atmosphere from temperature to CO2 , lighting to stickiness, all through one straightforward controller. IntelliGrow software, allows the users to set up the entire schedule up in one fell swoop. Change in lightening systems, adjustment of CO2 all can be controlled automatically (Fig. 35.13). Features: (a) Intelliclimate systems comprises of built in fail-safes which ensures that the system continues to function in case of equipment failures. (b) Intelliclimate systems combined with magnetic door switch acts as an intruder alarm. These alarm systems alerts when someone enters the room and also protects against dangerous CO2 levels [25].

Fig. 35.13 Intelliclimate device. Source https://autogrow.com/products/intelliclimate

502

P. Kaur and V. Bharti

(c) Intelliclimate enables the ranchers to deal with all parts of the atmosphere from temperature to CO2 , lighting to stickiness, all through one straightforward controller. (d) Intelliclimate systems also provide the users with daily CO2 and light integration stats. 2. Intellidose Another product offering of Autogrow Systems Ltd., Intellidose helps in managing the nutrient and pH levels, of farms. These systems can run up to 4 irrigation stations allowing the growing spaces to be watered at different schedules. The users can set up their systems in ways to irrigate during a particular time of the day. Liquid or Powder these systems has the ability to work with any nutrient line. Depending upon the needs of the farmers these systems can easily be connected with Laptops or PC. One can easily configure these systems and manage the ph and nutrient levels and set remote alarms. 3. Growlink Growlink One Controller are frameworks intended to convey a definitive ranch understanding to the clients. These frameworks involves the most astounding quality sensors and have the preparing capacity to arrange between many sensors and gadgets all through the ranch. Growlink One controllers give expanded execution, limit, efficiency, and security to help fulfill the developing needs of high thickness sensors organizes and complex gear control in modern smart farm [26]. Features: • • • • • • • • • •

Monitors and control the Temperature regimes Monitors and control Humidity and VPD Helps in maintaining the Light levels PAR Monitors and maintains the CO2 levels Water quality observing and control Crop dampness status observing Multi-Feed supplement infusion frameworks Intelligent, demand-based irrigation Water temperature, pressure and flow monitoring Air exchange and circulation (Fig. 35.14).

Growlink systems can easily be linked through the Growlink App and can be remotely monitored and controlled. The users can easily connect to the system and can manage all of the system’s features, such as lights, CO2 , cameras, temperature, humidity and much more. 4. Growtonics Growtonics Systems are an all in one Grow Facility Controller. The systems are extremely configurable and have the ability to monitor, Temperature Sensors, Humidity Sensors, CO2 Sensors, PH Sensors, EC/TDS Sensors, Light Detectors, PAR Light

35 Applications of Smart Devices

503

Fig. 35.14 Growlink system. Source https://growlink.com/control/

Sensors, Soil Temperature Sensors, Flood Detectors, Smoke Detectors, Motion Sensors, IP/Web cams and much more. Growtronix systems are really quick and easy to setup. The system can be customized to meet the user needs. The user can add up to 500 add-on sensors or controls to the system. The Growtronix system can control multiple rooms/grow spaces. The Growtonics systems are modular in design and the Growtonics hardware can be easily located depending upon its need. All Growtronix frameworks start with a Growtronix Base System. The base framework incorporates the Growtronix programming that is introduced onto a Windows based PC. The PC at that point turns into the cerebrums of the framework. The Growtonic Network Interface Module helps in conveying between the Computer and the Growtonics Hardware. Setting up the Growtonics frameworks are straightforward and not an intricate procedure. The product can be introduced on a Smartphone, tablet or a PC. When the product is introduced the client simply needs to connect an extra equipment thing, (for example, a temperature sensor) and select “Include New Hardware”. Select the sensor type and hit the submit catch, the sensor will at that point begin getting readings from that specific sensor [27]. Similarly the ‘Controllable Units’ can also be added to the systems. While selecting Add new Hardware the user just have to select Controllable outlet as the type. These Controllable outlets can be controlled from any web enabled device (Fig. 35.15). 5. Motor Leaf Systems Motor Leaf Systems are an emerging technology systems, by harnessing the power of artificial intelligence, Motorleaf are helping in building the latest automated greenhouse technologies. Motor Leaf Systems are providing the growers with the automated technology which help in making accurate harvest forecasts. Grower gain better capacities to monitor and control and have insights on the future yield of their harvests. These systems helps the agronomists to focus on business development rather than worrying about the future yield of their harvests [28]. Accurate predictions of the future yield help farmers to focus on marketing and sales campaign,

504

P. Kaur and V. Bharti

Fig. 35.15 Growtonic system along with mobile app. Source https://www.growtronix.com/cart/ blog/how-growtronix-works-n5

Fig. 35.16 Motor leaf system. Source https://motorleaf.com/

planning of labor needs, reduce wastages and cut costs. Using artificial intelligence (AI) these systems allow the users to analyze the farm data and make predictions about their future greenhouse outputs (Fig. 35.16).

35 Applications of Smart Devices

505

D. Crop Management The most basic thing for the people and ranchers is to keep precise records of their yields and homestead rehearses. It is indispensable for the ranchers to know how much sum they are spending on work or the amount they are spending on other crucial assets. Regardless of whether the contributed assets will deliver the ideal outcomes. In addition, ranchers ought to have the option to make arrangements anticipating the future, deal with its assets and group of laborers, and give applicable documentation respects to sanitation and other yield related exercises in the homesteads.For maintaining accurate crop records, managing farm operations on daily basis and taking crucial decisions, the most effective solution is to have proper Crop Management software’s and systems in place. According to World Bank, we shall need to produce 50% more food by 2050 as the global population is increasing at pace rate. But due to climatic changes, crop yield is falling by more than a quarter. Thus the need of precision farming is increasing exponentially [29]. In order to solve this problem, Crop management systems are used. Crop management starts with the sowing of seeds, then crop maintenance is done during growth of the plants, and finishes with crop harvest, storage and distribution. Soil fertilization is an important component of crop management that provides sufficient nutrition to the plants. Crop Management Software systems are systems helping the farmers to systemize and organize all their crop information at a centralized location. Through the help of these systems every aspect of farm management can easily be coordinated and help farmers keep check and tabs on the activities of the farm, crop conditions, workers and farm equipment’s. These systems help farmers to organize information about harvesting, planting seasons, packaging, and spray records etc. D.1 A good crop management system includes the following: 1. 2. 3. 4. 5. 6. 7. 8.

Soil management and crop nutrition Crop protection Waste and pollution management Crop rotation Energy consumption Monitoring and auditing Wildlife and landscape management Organization management

D.2 Some Benefits of Crop Management Systems can be listed as follows: 1. Crop Management Systems allows farmers to take appropriate decisions based upon real time data. Farmers can use these systems to monitor crop health, allocate resources and help manage their cost of production. 2. Through these systems all financial records can be kept at a centralized location. Farmers can allocate financial resources to each crop production and also keep track of its cost on sales, distribution, investments etc.

506

P. Kaur and V. Bharti

3. These systems help farmers to monitor the soil nutrients and the fertilizer requirements of crops. 4. Accurate weather predictions through the use of systems helps farmers to plan their activities based upon the weather conditions. Rainfall predictions helps farmers to prevent crops from spoilage during rain. 5. These systems allows remote controlled processes to perform tasks like: spraying, weeding and crop monitoring. 6. The systems can also be used to keep track of the laborers, record the working hours of laborers. Monitor their daily activities and schedules. Manage their pay schedule and helps in allocation of resources to farm workers. D.3 IOT Based Green House Automation Systems: IoT devices has enabled the crop management easy and efficient to enhance crop productivity and hence profits for the farmers. These devices uses smart sensors are placed in the fields to gather the information about conditions like humidity, temperature, moisture and overall crop health. IoT devices used in crop management helps the farmers to improve farming practices and allows easiness to get connected to his farms from anywhere and anytime [30]. Below are the devices explained that gives a representation of how IoT is applied for crop management. 1. Semios Semios is the leader tool in predictive analytics solution which provides total coverage of your fields with 24 × 7 monitoring service for agricultural crops. Automatic camera-traps are installed in the device that captures daily images of the crops and provide up-to-date analysis and alerts using a simple dashboard. Semios also allows the farmers to manage frost conditions like temperature and humidity trend analysis, etc. For any disease recognized, Semios alerts potential disease triggers including dew, watering events or rainfall [31]. It also works with soil and temperature levels to understand water events and do correct scheduling (Fig. 35.17). 2. Arable The Arable Mark is the first device built by Arable Labs to link global weather data with-in filed observations. It acts as both weather station and crop monitor for your crops. Arable is calibrated against top-of-the line scientific equipment with very easy installation. Hourly and daily predictions up to 10 days ahead can be retrieved through these devices [32]. The main features of Arable includes the following: 1. Provide insights on Dew/Rainfall detection. 2. Helps in tracking temperature to within 0.75 °C with 3% margin of error. 3. The devices provides Cellular connectivity through Sim networks, thus keeping the farmers connected to their farms. 4. Constantly monitors the plant health and sends alerts regarding the same. 5. Measures the solar radiations and provide information on radiations considered bad for the crops.

35 Applications of Smart Devices

507

6. Provides useful information to the farmers regarding best harvesting/growing days. Arable devices are only devices on the planet that synthesis both climate and plant measurements. These devices help configure and integrate weather and crop data from your business (Fig. 35.18). E. Livestock/Cattle Monitoring and Management The livestock sector is a supplier of protein needs to more than one-third of human population and is a major provider of livelihood in almost all developing countries. The livestock sector provides immense benefits to the population. Poor Livestock management can have many harmful effects at the local, regional, and national levels, but still this sector has been ignored and adequately handled in many emerging economies. Cows often referred to as cattle are being raised as livestock for primary purpose of meat and milk. To meet the needs of the growing population around 1.4 billion heads of cattle are being reared. With the growing demand for meat every year, the requirements of food safety and quality are also rising constantly. Cattle Management and tracking systems helps the livestock producers cater to the challenges of increased scale and also ensure the wellbeing of the animals. On site tracking systems provide reliable information about the location and the physical conditions of the animals [33]. Implementation of Cattle Management systems helps in enabling the animals to maintain optimal body conditions.

Fig. 35.17 Working of semios system. Source https://semios.com/network/

508

P. Kaur and V. Bharti

Fig. 35.18 Arable device. Source https://www.arable. com/

E.1 Cattle Management Systems provides the following features 1. The important health stats and other vitals of the cattle such as blood pressure, health rate, and digestion can be monitored by farmers through the use of sensors. 2. Helps farmers to address issues like feeding problems and illness by analyzing their health. 3. Increase in milking session with more successful outcomes. 4. Tracks animal location which help farmers to optimize their grazing patterns. 5. In cases of anomalous development e.g. Loss of weight etc. the affected animals can be treated and feeded individually. E.2 IoT Based Cattle Management Sensors Deployment Similar to crop monitoring, there are IoT based wireless agricultural sensors that are attached to the animals. There are two ways of deploying these sensors [34]: 1. Static networks-These sensors are deployed through tracking and measuring of cow movement, soil moisture, etc. 2. Mobile sensors-These sensors are deployed through monitoring and studying the animal behavior, health, temperature, etc. E.3 IoT Based Cattle Management System 1. SCR by Afflex SCR systems provides advanced cow monitoring tools designed to capture and analyze critical data points, these system helps tracking every movement of the cows like grazing patterns, rumination etc. These systems provide insights to the farmers about the heat, health and nutrition needs of the cattle. SCR systems make use of agricultural sensors such as ‘collar tags’ to provide insights on each individual cow and provide collective information about the herd. These systems provides information about the health, activities, temperature and nutrition insights.

35 Applications of Smart Devices

509

Fig. 35.19 SCR milking system. Source http://www.scrdairy.com/

SCR Milking Systems SCR milking systems helps optimizing the milking operations with best in class milk measurement technologies, which helps in streamlining the milking process and saves time, thus improving accuracy and overall efficiency (Fig. 35.19). SCR milking components can be either implemented as standalone solutions or can be implemented with the milking systems of several other Companies [35]. The SCR milking components being scalable they can be integrated with any kind of milking parlors i.e. they are suited for dairies of every type and sizes. 2. Cowlar Cowlar is another state of the art IOT& AI based smart device that helps monitor and track every movement of the cow. Cowlar has a simple strap like design, which comfortably fits around the neck of the cow [36]. The strap has a small box attached to it which measures the temperature, activity and cow behavior. All activities of the cow whether it’s eating, sleeping, running or showing lameness etc. can be tracked and monitored through the Cowlar. The data collected is further interpreted to study the behavior patterns of the cow in a bid to prevent diseases, detect heat cycles and increase milk yield

510

P. Kaur and V. Bharti

Fig. 35.20 Cowlar device. Source https://www.cowlar.com/

Cow Routers are used to collect data from the Cowlars and pass it on to the servers. Cow Routers are solar powered and are easy to install. These Routers can connect to all Cowlars within a 2 mile range and makes use of cellular technology to pass on the information to the servers. Then using machine learning & dairy science techniques sense is added to this data and interpretations are done for each individual breed of cow (Fig. 35.20). F. Agricultural Drones Agriculture Drones are another type of Smart Device that is bringing about a technological revolution in the farming sector. Agriculture Drones are unmanned aerial vehicle applied to farming in order to help increase crop production and monitor crop growth. Sensors and digital imaging capabilities provides farmers a richer picture of their fields. Drones also referred to as UAV’s have been mostly associated with military, industry and other specialized operations but with recent developments in information technology and the area of sensors the scope of drones have been widened to the area of Agriculture. Professional Drones/UAV’s gives the farmers a holistic view of their crop’s growth enabling the farmers to precisely identify issues and better target their field scouting. Use of Drone Technology provide farmers with the real time data, the farmers can review and analyze this data over a period of time thus allowing better planning and monitoring of improvements. Drone Technology can also be employed to analyze soil characteristics which includes soil moisture, temperature, elevation and more [37]. The technology helps provide more accurate soil sampling and thus enhance production. Agronomists are also using the data collected from professional drones to understand which plants emerge and study the population and spacing metrics. The data obtained are being used to replanting decisions, thinning and pruning activities and also improve the crop models. The data collected from Drones helps in assessing of crop vigor at different stages of growth, which helps assessing the application of right rates of fertilizers, reduce wastages and optimize crop health and production. Drones equipped with RGB and/Thermal

35 Applications of Smart Devices

511

Fig. 35.21 Ag360 agriculture drones. Source https://www.sensefly.com/solution/ag-360agricultural-drone/

Infrared Cameras are being used across farms to help famers use and troubleshoot their irrigation systems and manage the water flow across their farms. Mostly two types of Drones are being used across farms [38]: • Ag360 Agriculture Drones • EBeeSQ Agriculture Drones (Fig. 35.21).

35.4 Smart Farming in the Indian Agriculture Industry Perspective 35.4.1 Introduction: Agriculture in India The population of India being of over a billion people, Agriculture is the backbone of the Indian Economy. With over 40% of the country’s workforce being associated with the agriculture industry, this industry is a major influencer of the Indian economy. But despite this the contribution by the agriculture sector to the $2.3 trillion economy is just a small proportion of 16% of the entire GDP. As compared to other developed nations of the world agriculture in India is really backward and lacks technological advancement. India is an agriculture heavy economy but unlike other sectors like communication, transportation, education, finance etc., agriculture has not been blessed with the latest tech advancements [39]. Agriculture in India also lacks institutional attention and support. The banks are hesitant in lending out any loans to farmers and the farmer schemes formulated by the Government lacks implementation and the real benefits of the scheme are not passed on to the farmers. Agriculture in India also suffers from a myriad of disasters. Unpredictable monsoons, depleting groundwater, climatic changes, lack of proper warehousing facilities, unfair pricing and dependency on middlemen are some of the factors responsible for the backwardness Indian Agriculture Industry. Advancements in the Indian Agriculture are necessary to balance the demand and

512

P. Kaur and V. Bharti

supply as the population is increasing day by day. With the introduction of modern technologies and replacement of traditional, inefficient and time consuming farming practices the efficiency of farming practices can dramatically be improved.

35.4.2 ‘DIGITAL INDIA’ Campaign and BIG Data Bringing Technological Revolution in Indian Agriculture Government of India under the leadership of the Prime Minister, Shri Narendra Modi, has launched a campaign of ‘DIGITAL INDIA’. The purpose of this campaign is to make India a technological advanced nation. The DIGITAL INDIA initiative includes plans to connect rural areas with high-speed internet networks. Digital India consists of three core components: the development of secure and stable digital infrastructure, delivering government services digitally, and universal digital literacy. The revolutionary technology, by the names of Big Data is making waves in the Indian Agriculture Sector. Apart from Major Companies several startup companies have started attracting the use of Big Data for farming. According to a report by NASSCOM, India has around 40 startups dealing in smart agriculture.

35.4.3 Satsure Satsure being one of the emerging startup Company in the country is making use of the Big Data and its allied technologies and IoT for making the lives of the farmers better. Satsure systems are helping dealing with parameters associated with the health of the soil and crop growth and making use of the Machine Learning and Big Data technologies for solving restrictions and giving insights on the crop phenology. Satsure are offering satellite based field monitoring systems, embedding sensor systems on crops and in fields, wind direction prediction systems, pest infestations systems, and notifications on fertilizer requirements. Tractors having GPS systems installed, water cycles and more are vital sources of rich data which are being studied and interpreted for better agriculture practices [40].

35.4.4 Cropin Cropin is another such startup Company having its headquarters in Karnataka, Cropin is an intuitive, intelligent, self-evolving system that delivers future-ready farming solutions to the entire agricultural sector [41]. Cropin helps in providing decisionmaking tools which provides consistency, dependability and sustainability to the agriculture based businesses. They provide live reporting analysis tools, digitalization

35 Applications of Smart Devices

513

of farms, data managing tools and provides interpretation and insight that span across geographies. These systems helps farmers to analyze the real time data, predict trends and help make blueprint of their businesses in times to come.

35.5 Challenges of Smart/Precision Farming 35.5.1 Right Resources Although smart/precision agriculture techniques have been contributing over years to maximize profits, optimize production, minimize input costs and minimize the harmful environmental impacts, there are still a number of challenges around such developments. The adoption of genetically modified seeds and precision machineries in farms are increasing at a rapid pace [42]. The use of genetically modified seeds have many compelling and visible values due to which there use have exceeded 50% in many continents. Though the use of genetically modified seeds and fertilizers have increased manifolds, the precise utilization of fertilizer and seed management continues to lag. Efficient use of fertilizers are a key to efficient farming for several reasons: (i) Fertilizers contribute to around 30–50% of the total farming costs. With the cost of fertilizers increasing each day the cost of production/farming are increasing. (ii) Use of Fertilizers triggers oxygen depletion due to the nitrogen and phosphorus levels present in fertilizers. Fertilizers if not used efficiently leads to runoffs which contaminates the water bodies and the living creatures living in the water. The contaminated water if consumed by humans can lead to deadly water borne diseases. Smart devices using machine learning technology, measuring the nutrient levels of the crops should be used with a lot of precision and accuracy. The farmers need to realize the right source and amounts of nutrient needed by the plant and must realize limiting the nutrient levels correctly and adequately. Depending upon the nature it is very difficult to measure the nutrient levels of some of the crops, the nutrient levels of such crops cannot be measured accurately through sensors and machine learning techniques. Most of the sensors used in the smart devices focuses on macro application. Sensors which identifies unique nutrient deficiencies in different species of crops should be used. (a) Right Place Smart devices and sensors deployed in the fields collecting data, relies on quite a number of sampling techniques that allows farmers to take decisions. The data procured from these farmers allows the farmers to quantify and characterize the pattern to the entire field. Sampling includes smart sampling, soil electrical conductivity measurement, site specific measurement etc. For a smart device to work effectively

514

P. Kaur and V. Bharti

the place for collecting samples should be identified accurately else the whole purpose of installation of smart sensors in fields shall go in vain. Thus, the identification of the right place is of utmost importance. (b) Cost The sensors used in precision agriculture methods does not come cheap and are not at all cost effective, especially for domestic farmers. Deployment of machinery using expensive sensors, driverless technology for drones and use of expensive Drones/UAV make practical sense for large scale farmers, but are not at all cost effective for domestic farmers. Efforts should be made to provide these Sensors at low costs or provide subsidies on use of such Smart Devices and Sensors [43]. Increased cost being a factor has slowed down the uptake of such technologies. (c) Lack of Awareness Small scale and domestic farmers are still not shifting to precision farming methods and use of smart devices due to the core factor being that of profitability. There is still a common belief amongst the farmers that precision farming methods are meant for large scale farms and if deployed in small fields it will lead to increase in input cost and hence effect the overall profitability. There is an urgent need to change this belief and create awareness amongst small scale farmers. (d) Education levels of farmers Precision farming methods involves software’s and technologies, which are not that straight forward to operate and service. The literacy and education level of a farmer thus plays a major factor in adoption of the smart farming techniques. In cases where the farmers are poor and resources are scarce the deployment of smart farming techniques cannot be possible [44].

35.6 Limitations of Smart/Precision Farming • Technologies, including software, develop evolve very fast; this may be a challenge to farmers or users who have a lot to catch up with, and the amount of money that comes along with such developments or upgrades are not cheap; • Some lighter unmanned aerial vehicles are still relatively unsteady, especially during adverse weather conditions like in wind, smoke or heavy rains resulting in poor quality images. • Data collected might be sophisticated for farmers. Therefore, tools for analysis and interpretation may further be required which is an added cost. • Moreover, the data may be too large, especially if it has a wider look-back period. This would imply that more storage space should be availed, which is costly. • The existence of trade-offs and high complexity has reduced the adoption pace of precision agriculture techniques.

35 Applications of Smart Devices

515

• The farmers’ perception of the usefulness has also reduced the adoption rate. This should be enhanced by proper farmer education and awareness. • Cost-benefit analysis ratio is another key point [45]. Farmers are not able to confidently establish at what point would they break-even and beat their costs and eventually get better pay off as the result of precision agriculture technologies. • Agriculture being a natural phenomenon relies mostly on nature, and man can never predict or control nature let it be rain, drought, sunlight availability, pests control, etc. So even implementation IoT system we cannot implement SMART agriculture [46]. • Faulty sensors or data processing engines can cause faulty decisions which may lead to over use of water, fertilizers and other wastage of resources.

35.7 Conclusion Through the help of this chapter we have tried to list that how the Internet of thingsconnected devices are making its way in the agriculture sector, automating and transforming the agriculture processes. The smart agricultural technology have helped farmers gain better controls over their processes of growing crops, rearing livestock, cutting down on costs bringing economies of scale and achieving proper allocation of scare resources. With the world population on a rise and the essential resources such as water becoming scarce, smart agricultural technology is playing a pivotal role enabling the farmers around the globe to act diligently and manage the resources efficiently. The adoption of IoT solutions for agriculture have helped farmers to take the driving seat and gain control over the internal processes reducing the production risks. Big data and Machine learning data analysis techniques have enabled farmers to foresee the actual output of their production and better their production distribution. Yes, the adoption of Smart agricultural technology and use of IoT connected devices involves deployment of a lot of Sensors across fields which does not come cheap and are still beyond the reach of small scale farmers. Constant Research and Developments are being carried out in these area and policies are being farmed by the Government to help reduce the costs of these Sensors and make this technology affordable for the small scale and individual farmers.

References 1. http://www.fao.org/ 2. Jury, W.A., Vaux, H.J. Jr.: The emerging global water crisis: managing scarcity and conflict between water users. Adv. Agron. 95 (2007) 3. Arvind, G., Athira, V., Haripriya, H., Rani, R., Aravind, S.: Automated irrigation with advanced seed germination and pest control. In: IEEE Technological Innovations in ICT for Agriculture and Rural Development (TIAR (2017)

516

P. Kaur and V. Bharti

4. https://www.biz4intellia.com/blog/5-applications-of-IoT-in-agriculture/ 5. Ahmad, Z., Pasha M., Ahmad A., Muhammad, A., Masud, S., Schappacher, M. Sikora, A.: Performance evaluation of IEEE 802.15.4-compliant smart water meters for automating largescale waterways. In: 9th IEEE International Conference on Intelligent Data Acquisition and Advanced Computing Systems: Technology and Applications (IDAACS) (2017) 6. Providing smart agricultural solutions/techniques by using IoT based toolkit. In: International Conference on Trends in Electronics and Informatics (ICEI) (2017) 7. https://medium.com/sciforce/smart-farming-or-the-future-of-agriculture-359f0089df69 8. https://dzone.com/articles/IoT-in-agriculture-five-technology-uses-for-smart 9. Perera, S.: IoT analytics: using big data to architect IoT solutions, WSO2 White Paper, http:// wso2.com/whitepapers/IoT-analytics-using-big-data-to-architect-IoT-solutions/ (2015) 10. Rajkumar, M., Abinaya, S. Kumar, V.: Intelligent irrigation system—an IOT based approach. In: International Conference on Innovations in Green Energy and Healthcare Technologies (IGEHT) (2017) 11. Salvi, S., Framed Jain, S.A., Sanjay, H.A., Harshita, T.K., Farhana, M., Jain, N., Suhas M.V.: Cloud based data analysis and monitoring of smart multi-level irrigation system using IoT. In: International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC) (2017) 12. https://www.kickstarter.com/projects/1498890810/bitponics-your-shortcut-to-a-green-thumb 13. https://www.botanicalls.com/kits/ 14. https://www.kickstarter.com/projects/edyn/edyn-welcome-to-the-connected-garden 15. http://thewifigarden.com/plantlink-soil-sensor-review 16. https://www.kickstarter.com/projects/2077260917/harvestgeek-brains-for-your-garden 17. WiFi Sprinkler System: Iro, https://rachio.com/ 18. Spruce. http://www.spruceirrigation.com/ 19. https://blog.agrivi.com/post/importance-of-weather-monitoring-in-farm-production 20. https://www.allmeteo.com/agriculture-iot-weather-station 21. https://smartelements.io/ 22. https://pycno.co/sensors 23. Vimal, P.V., Shivaprakasha, K.S.: IOT based greenhouse environment monitoring and controlling system using Arduino platform. In: 2017 International Conference on Intelligent Computing, Instrumentation and Control Technologies (ICICICT), pp. 1514–1519. IEEE (2017, July) 24. https://autogrow.com/our-products-solutions/climate-control 25. https://autogrow.com/products/intelliclimate 26. https://growlink.com/control/#1543704227636-bda7b431-b1ff 27. https://www.growtronix.com/cart/blog/how-growtronix-works-n5 28. https://motorleaf.com/ 29. ShyamSundar, S., Balan, B.: Sensor based smart agriculture using IOT. Int. J. MC Square Sci. Re. (2017) 30. InfantialRubala, J., Anitha, D.: Agriculture field monitoring using wireless sensor networks to improving crop production. Int. J. Eng. Sci. Comput. (2017, March) 31. https://semios.com/network/ 32. https://www.arable.com/ 33. https://www.telit.com/industries-solutions/agriculture/crop-livestock-monitoring/ 34. http://www.sol-chip.com/applications_livestock.asp 35. http://www.scrdairy.com/ 36. https://www.cowlar.com/ 37. Puri, V., Nayyar, A., Raja, L.: Agriculture drones: a modern breakthrough in precision agriculture. J. Stat. Manage. Syst. 20(4), 507–518 (2017) 38. https://www.sensefly.com/solution/ag-360-agricultural-drone/ 39. Roy, S., Ray, R., Roy, A., Sinha, S., Mukherjee, G., Pyne, S., Hazra, S.: IoT, big data science and analytics, cloud computing and mobile app based hybrid system for smart agriculture. In: 2017 8th Annual Industrial Automation and Electromechanical Engineering Conference (IEMECON) (2017)

35 Applications of Smart Devices

517

40. https://www.satsure.co/ 41. https://www.cropin.com/ 42. Bacco, M., Berton, A., Ferro, E., Gennaro, C., Gotta, A., Matteoli, S., Paonessa, F., Ruggeri, M., Virone, G., Zanella, A.: Smart farming: opportunities, challenges and technology enablers. 2018 IoT vertical and topical summit on agriculture—Tuscany (IOT Tuscany) (2018) 43. Venkatesan, R., Tamilvanan, A.: A sustainable agricultural system using IoT. In: International Conference on Communication and Signal Processing (ICCSP) (2017) 44. Jayaraman, P., Yavari, A., Georgakopoulos, D., Morshed, A., Zaslavsky, A.: Internet of things platform for smart farming: experiences and lessons learnt. Sensors 16(11), 1884 (2016). https://doi.org/10.3390/s16111884 45. Roselin, A., Jawahar, A.: Smart agro system using wireless sensor networks. In: International Conference on Intelligent Computing and Control Systems (ICICCS) (2017) 46. Saraf, S., Gawali, D.: IoT based smart irrigation monitoring and controlling system. In: 2nd IEEE International Conference on Recent Trends in Electronics, Information and Communication Technology (RTEICT) (2017)

Chapter 36

Fundamental Concepts of Convolutional Neural Network Anirudha Ghosh, Abu Sufian, Farhana Sultana, Amlan Chakrabarti and Debashis De

Abstract Convolutional neural network (or CNN) is a special type of multilayer neural network or deep learning architecture inspired by the visual system of living beings. The CNN is very much suitable for different fields of computer vision and natural language processing. The main focus of this chapter is an elaborate discussion of all the basic components of CNN. It also gives a general view of foundation of CNN, recent advancements of CNN and some major application areas. Keywords Computer vision · Convolutional neural network · CNN · Deep learning · Image classification · Image understanding

36.1 Introduction Among different deep learning [11] architecture, a special type of multilayer neural network for spatial data is Convolutional Neural Network (or CNN or ConvNet.). The architecture of CNN is inspired by the visual perception of living beings. Though it is become popular after the record breaking performance of AlexNet [20] in 2012 but it is actually initiated in 1980. After 2012, the CNN got the pace to take over different fields of computer vision, natural language processing and many more. The foundation of convolutional neural network started from the discovery of Hubel and Wisel in 1959 [16]. According to them, cells of animal visual cortex recognize light in the small receptive field. In 1980, inspired by this work, Kunihiko Fukusima proposed neocognitron [7]. This network is considered as the first theoA. Ghosh · A. Sufian (B) · F. Sultana Department of Computer Science, University of Gour Banga, Malda, W.B, India e-mail: [email protected] A. Chakrabarti A.K. Choudhury School of Information Technology, University of Calcutta, Kolkata, W.B, India D. De Department of Computer Science & Engineering, M.A.K.A.U.T., Kolkata, India © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_36

519

520

A. Ghosh et al.

retical model for CNN. In 1990, LeCun et al. developed the modern framework of CNN called LeNet-5 [21] to recognize handwritten digits. Training by backpropagation [35] algorithm helped LeNet-5 in recognizing visual patterns from raw images directly without using any separate feature engineering. But in those days despite of several merits, the performance of CNN in complex problems was lacked by the limited training data, lack of innovation in algorithm and insufficient computing power. Recently we have large labeled datasets, innovative algorithms and powerful GPU machines. In 2012, with these up-gradation, a large deep CNN, called AlexNet, designed by Krizhevsky et al. [20] showed excellent performance on the ILSVRC [36]. The success of AlexNet paved the way to invent different CNN models [41] as well as to apply those models in different fields of computer vision and natural language processing. A traditional convolutional neural network is made up of single or multiple blocks of convolution and pooling layers, followed by one or multiple fully connected (FC) layers and an output layer. The convolutional layer is the core building block of a CNN. This layer aims to learn feature representations of the input. The convolutional layer is composed of several learnable convolution kernels or filters which are used to compute different feature maps. Each unit of feature map is connected to a receptive field in the previous layer. The new feature map is produced by convolving the input with the kernels and applying elementwise non-linear activation function on the convolved result. The parameter sharing property of convolutional layer reduces the model complexity. Pooling or sub-sampling layer takes a small region of the convolutional output as input and downsamples it to produce a single output. There are different sub-sampling techniques as example max pooling, min pooling, average pooling, etc. Pooling reduces the number of parameters to be computed as well as it makes the network translation invariant. Last part of CNN is basically made up of one or more FC layers typically found in feedforward neural network. The FC layer takes input from the final pooling or convolutional layer and generates final output of CNN. In case of image classification, a CNN can be viewed as a combination of two parts: feature extraction part and classification part. Both convolution and pooling layers perform feature extraction. As an example of dog’s image, different convolution layers from lower level to higher level detect various features such as two eyes, long ears, four legs, etc for further recognition. On top of this features, the FC layers are added as classifier, and a probability is assigned for the input image being a dog. Beside the layer design, the improvement of CNN depends on several different aspects such as activation function, normalization method, loss function, regularization, optimization and processing speed, etc. After the success of AlexNet, CNN got huge popularity in three major fields namely image classification [41], object detection [42] and segmentation [45], and many advance models of CNN has been proposed in those areas in successive years. Major application areas that apply CNN to achieve state-of-the-art performance includes image classification, object tracking, object detection, segmentation, human pose estimation, text detection, visual saliency detection, action recognition, scene labelling, visual question answering, speech and natural language processing, etc. Though current CNN models work very well for various applications, it is yet to know

36 Fundamental Concepts of Convolutional Neural Network

521

why and how it works essentially. So, more efforts on investigating the fundamental principles of CNNs are required. The chapter will provide a better understanding of CNN as well as facilitates for future research activities and application developments in the field of CNN. The rest of the chapter consist of foundation of CNN in Sect. 36.2, main contributions of the chapter i.e. detailed concepts of CNN in Sect. 36.3 with different subsections, the learning process of CNN with different aspects for improvements in Sect. 36.4, recent advancements of CNN in Sect. 36.5, and some major application areas in Sect. 36.6. The last section i.e. Sect. 36.7 will conclude the chapter.

36.2 Foundation of Convolutional Neural Network In 1959, two neurophysiologists David Hubel and Torsten Wiesel experimented and later published their paper, entitled “Receptive fields of single neurons in cat’s striate cortex” [16], described that the neurons inside the brain of a cat are organized in layered form. These layers learn how to recognize visual patterns by first extracting the local features and then combining the extracted features for higher level representation. Later on, this concept is essentially become one of the core principle of Deep Learning. Inspired by the work of Hubel and Wiesel, in 1980, Kunihiko Fukushima proposed Neocognitron, which is a self-organizing Neural Network, containing multiple layers, capable of recognizing visual patterns hierarchically through learning and this architecture became the first theoretical model of CNN as in Fig. 36.1. A major improvement over the architecture of Neocognitron was done by LeCun et. in 1989 by developing a modern framework of CNN, called LeNet-5, which successfully recognized the MNIST handwritten digits dataset. LeNet-5 was trained using error back-propagation algorithm and it can recognize visual patterns directly from raw input images, without using any separated feature engineering mechanism. After discovering LeNet-5, because of several limitation like lack of large training data, lack of innovation in algorithm and inadequate computing power, CNN did not performs well in various complex problems. But nowadays, in the era of Big Data we have large labeled datasets, more innovative algorithms and especially powerful GPU machines. With these type of up-gradation, in 2012, Krizhevsky et al. designed AexNet, which achieved a fantastic accuracy on the ImageNet Large Scale Visual Recognition Challenge(ILSVRC [36]). The victory of AlexNet paved the way to invent several CNN models [41] as well as to apply those models in different field of computer vision and natural language processing.

522

A. Ghosh et al.

Fig. 36.1 Schematic diagram illustrating the interconnections between layers in the neocognitron, Kunihiko Fukushima [7]

36.3 Concepts of Convolutional Neural Network Convolutional Neural Network (CNN), also called ConvNet, is a type of Artificial Neural Network(ANN), which has deep feed-forward architecture and has amazing generalizing ability as compared to other networks with FC layers, it can learn highly abstracted features of objects especially spatial data and can identify them more efficiently. A deep CNN model consists of a finite set of processing layers that can learn various features of input data (e.g. image) with multiple level of abstraction. The initiatory layers learn and extract the high level features (with lower abstraction), and the deeper layers learns and extracts the low level features (with higher abstraction). The basic conceptual model of CNN was shown in Fig. 36.2, we will discuss different types of layers in subsequent sections. Why Convolutional Neural Networks is more considerable over other classical neural networks in the context of computer vision? • One of the main reason for considering CNN in such case is the weight sharing feature of CNN, that reduce the number of trainable parameters in the network, which helped the model to avoid overfitting and as well as to improve generalization.

36 Fundamental Concepts of Convolutional Neural Network

523

Fig. 36.2 Conceptual model of CNN [41]

• In CNN, the classification layer and the feature extraction layers learn together, that makes the output of the model more organized and makes the output more dependent to the extracted features. • The implementation of a large network is more difficult by using other types of neural networks rather than using Convolutional Neural Networks. Nowadays CNN has been emerged as a mechanism for achieving promising result in various computer vision based applications like image classification, object detection, face detection, speech recognition, vehicle recognition, facial expression recognition, text recognition and many more. Now description of different components or basic building blocks of CNN briefly as follows.

36.3.1 Network Layers As we mentioned earlier, that a CNN is composed of multiple building blocks (known as layers of the architecture), in this subsection, we described some of these building blocks in detail with their roll in the CNN architecture.

36.3.1.1

Convolutional Layer

Convolutional layer1 is the most important component of any CNN architecture. It contains a set of convolutional kernels (also called filters), which gets convolved with the input image (N-dimensional metrics) to generate an output feature map.

What is a kernel? A kernel can be described as a grid of discrete values or numbers, where each value is known as the weight of this kernel. During the starting of training process of an 1 Notable

thing: CNN uses a set of multiple filters in each convolutional layers so that each filter can extract the different types of features.

524

A. Ghosh et al.

Fig. 36.3 Example of a 2 × 2 kernel

CNN model, all the weights of a kernel are assigned with random numbers (different approaches are also available there for initializing the weights). Then, with each training epoch, the weights are tuned and the kernel learned to extract meaningful features. In Fig. 36.3, we have shown a 2D filter.

What is Convolution Operation? Before we go any deeper, let us first understand the input format to CNN. Unlike other classical neural networks (where the input is in a vector format), in CNN the input is a multi-channeled image (e.g. for RGB image as in Fig. 36.4, it is 3 channeled and for Gray-Scale image, it is single channeled). Now, to understand the convolution operation, if we take a gray-scale image of 4 × 4 diamension, shown in Fig. 36.5 and an 2 × 2 kernel with randomly initialized weights as shown in Fig. 36.6. Now, in convolution operation, we take the 2 × 2 kernel and slide it over all the complete 4 × 4 image horizontally as well as vertically and along the way we take

Fig. 36.4 Example of a RGB image

Fig. 36.5 An 4 × 4 gray-scale image

36 Fundamental Concepts of Convolutional Neural Network

525

the dot product between kernel and input image by multiplying the corresponding values of them and sum up all values to generate one scaler value in the output feature map. This process continues until the kernel can no longer slide further. To understand the thing more clearly, let’s do some initial computations performed at each step graphically as shown in Fig. 36.7, where the 2 × 2 kernel (shown in light blue color) is multiplied with the same sized region (shown in yellow color) within the 4 × 4 input image and the resulting values are summed up to obtain a corresponding entry (shown in deep blue) in the output feature map at each convolution step. After performing the complete convolution operation, the final output feature map is shown in Fig. 36.8 as follows. In the above example, we apply the convolution operation with no padding to the input image and with stride (i.e. the taken step size along the horizontal or vertical position ) of 1 to the kernel. But we can use other stride value (rather than 1) in convolution operation. The noticeable thing is if we increase the stride of the convolution operation, it resulted in lower-dimensional feature map. The padding is important to give border size information of the input image more importance, otherwise without using any padding the border side features are gets washed away too quickly. The padding is also used to increase the input image size, as a result the output feature map size also get increased. Figure 36.9 gives an example by showing the convolution operation with Zero-padding and 3 stride value. The formula to find the output feature map size after convolution operation: 

 h− f +p +1 h = s   w− f + p  +1 w = s 





Where h denotes the height of the output feature map, w denotes the width of the output feature map, h denotes the height of the input image, w denotes the width of the input image, f is the filter size, p denotes the padding of convolution operation and s denotes the stride of convolution operation. Main advantages of convolution layers are: • Sparse Connectivity: In a fully connected neural network each neuron of one layer connects with each neuron of the next layer but in CNN small number of weights are present between two layers. As a result, the number of connection or

Fig. 36.6 A kernel of size 2 × 2

526

Fig. 36.7 Illustrating the first 5 steps of convolution operation

A. Ghosh et al.

36 Fundamental Concepts of Convolutional Neural Network

527

Fig. 36.8 The final feature map after the complete convolution operation

weights we need is small, and the amount of memory to store those weights is also small, so it is memory efficient. Also, the dot(.) operation is computationally cheaper than matrix multiplication. • Weight Sharing: In CNN, no dedicated weights are present between two neurons of adjacent layers instead of all weights works with each and every pixel of the input matrix. Instead of learning new weights for every neuron we can learn one set of weights for all inputs and this drastically reduces the training time as well as the other costs.

36.3.1.2

Pooling Layer

The pooling2 layers are used to sub-sample the feature maps (produced after convolution operations), i.e. it takes the larger size feature maps and shrinks them to lower sized feature maps. While shrinking the feature maps it always preserve the most dominant features (or information) in each pool steps. The pooling operation is performed by specifying the pooled region size and the stride of the operation, similar to convolution operation. There are different types of pooling techniques are used in different pooling layers such as max pooling, min pooling, average pooling, gated pooling, tree pooling, etc. Max Pooling is the most popular and mostly used pooling technique. The main drawback of pooling layer is that it sometimes decreases the overall performance of CNN. The reason behind this is that pooling layer helps CNN to find whether a specific feature is present in the given input image or not without caring about the correct position of that feature. In this way the CNN model looses contextual information of an image (Fig. 36.10). The formula to find the output feature map size after pooling operation:  h− f h = s   w− f  w = s 

2 “The



pooling operation used in convolutional neural networks is a big mistake and the fact that it works so well is a disaster.” –Geoffrey Hinton.

528

A. Ghosh et al.

Fig. 36.9 The computations performed at each step, where the 3 × 3 kernel (shown in light blue color) is multiplied with the same sized region (shown in yellow color) within the 6 × 6 input image (where we applied zero-padding to the original input image of 4 × 4 dimension and it becomes of 6 × 6 dimensional) and values are summed up to obtain a corresponding entry (shown in deep green) in the output feature map at each convolution step

36 Fundamental Concepts of Convolutional Neural Network

529

Fig. 36.10 Illustrating an example that shows some initial steps as well as the final output of maxpooling operation, where the size of the pooling region is 2 × 2 (shown in orange color, in the input feature map) and the stride is 1 and the corresponding computed value in the output feature map (shown in green)

530

A. Ghosh et al.

Fig. 36.11 Sigmoid





Where h denotes the height of the output feature map, w denotes the width of the output feature map, h denotes the height of the input feature map, w denotes the width of the input feature map, f is the pooling region size and s denotes the stride of the pooling operation.

36.3.1.3

Activation Functions (Non-Linearity)

The main task of any activation function in any neural network based model is to map the input to the output, where the input value is obtained by calculating the weighted sum of neuron’s input and further adding bias with it (if there is a bias). In other words, the activation function decides whether a neuron will fire or not for a given input by producing the corresponding output. In CNN architecture, after each learnable layers (layers with weights, i.e. convolutional and FC layers) non-linear activation layers are used. The non-linearity behavior of those layers enables the CNN model to learn more complex things and manage to map the inputs to outputs non-linearly. The important feature of an activation function is that it should be differentiable in order to enable error backpropagation to train the model. The most commonly used activation functions in deep neural networks (including CNN) are described below (Fig. 36.11).

Sigmoid: The sigmoid activation function takes real numbers as its input and bind the output in the range of [0,1]. The curve of the sigmoid function is ‘S’ shaped. The mathematical representation of sigmoid is: f (x)sigm =

1 1 + e−x

36 Fundamental Concepts of Convolutional Neural Network

531

Tanh: The T anh activation function is used to bind the input values (real numbers) within the range of [−1, 1]. The mathematical representation of T anh is (Figs. 36.12): f (x)tanh =

e x − e−x e x + e−x

ReLU: The Rectifier Linear Unit (ReLU) [25] is the most commonly used activation function in Convolutional Neural Networks. It is used to convert all the input values to positive numbers (Fig. 36.13). The advantage of ReLU is that it requires very minimal computation load compared to others. The mathematical representation of ReLU is: f (x) ReLU = max(0, x) But sometimes there may occur some major problems in using ReLU activation function. For example, consider a larger gradient is flowing during error back-propagation algorithm, and when this larger gradient is passed through a ReLU function it may cause the weights to be updated in such a way that the neuron never gets activated again. This problem is known as the Dying ReLU problem. To solve these types of problems there are some variants of ReLU is available, some of them are discussed below.

Leaky ReLU: Unlike ReLU, a Leaky ReLU activation function does not ignore the negative inputs completely, rather than it down-scaled those negative inputs. Leaky ReLU is used

Fig. 36.12 T anh

532

A. Ghosh et al.

Fig. 36.13 ReLU

to solve Dying ReLU problem. The mathematical representation of Leaky ReLU is (Fig. 36.14):  x, if x > 0 f (x) Leaky ReLU = mx, x ≤ 0 where m is a constant, called leak factor and generally it set to a small value (like 0.001).

Noisy ReLU: Noisy ReLU is used Gaussian distribution to make ReLU noisy. The mathematical representation of Noisy ReLU is (Fig. 36.15): f (x) N oisy ReLU = max (x + Y ), with Y ∼ N (0, σ (x))

Fig. 36.14 Leaky ReLU

36 Fundamental Concepts of Convolutional Neural Network

533

Parametric Linear Units: It is almost similar to Leaky ReLU, but here the leak factor is tuned during the model training process. The mathematical representation of Parametric Linear Units is:  f (x) ParametricLinearU nits =

x, if x > 0 ax, x ≤ 0

where a is a learnable weight.

36.3.1.4

Fully Connected (FC) Layer

Usually the last part (or layers) of every CNN architecture (used for classification) is consist of fully-connected layers, where each neuron inside a layer is connected with each neuron from it’s previous layer. The last layer of Fully-Connected layers is used as the output layer (classifier) of the CNN architecture. The Fully-Connected Layers are type of feedforward artificial neural network (ANN) and it follows the principle of traditional multi-layer perceptron neural network (MLP). The FC layers take input from the final convolutional or pooling layer, which is in the form of a set of metrics (feature maps) and those metrics are flattened to create a vector and this vector is then fed into the FC layer to generate the final output of CNN as shown in Fig. 36.16.

36.3.2 Loss Functions In Sect. 36.3.1, has described different types of layers used in CNN architecture. Now, we know that the last layer of every CNN architecture (classification based) is the output layer, where the final classification takes place. In this output layer, we

Fig. 36.15 Noisy ReLU

534

A. Ghosh et al.

calculate the prediction error generated by the CNN model over the training samples using some Loss Function. This prediction error tells the network how off their prediction from the actual output, and then this error will be optimized during the learning process of the CNN model. The loss function uses two parameters to calculate the error, the first parameter is the estimate output of the CNN model (also called the prediction) and the second one is the actual output (also known as the label). There are different types of loss functions used in different types of problem. Some of the most used loss functions are briefly described in next subsections.

36.3.2.1

Cross-Entropy or Soft-Max Loss Function

Cross-entropy loss, also called log loss function is widely used to measure the performance of CNN model, whose output is the probability p ∈ {0, 1}. It is widely used as an alternative of squared error loss function in the multi-class classification problems. It uses softmax activations in the output layer to generate the output within a probability distribution, i.e. p,y ∈ R N , where p is the probability for each output category and y denotes the desired output and the probability of each output class can be obtained by: eai pi = N  a e k k=1

where N is the number of neurons in the output layer and eai denotes each unnormalized output from the previous layer in the network. Finally, cross-entropy loss can be define as:

Fig. 36.16 The architecture of fully connected layers

36 Fundamental Concepts of Convolutional Neural Network

H ( p, y) = −



535

yi log ( pi )

i

where i ∈ [1,N]

36.3.2.2

Euclidean Loss Function

The Euclidean loss also called mean squared error is widely used in regression problems. The mean squared error between the predicted output p ∈ R N and the actual output y ∈ R N in each neuron of the output layer of CNN is defined as H ( p, y) = ( p − y)2 . So, if there are N neurons in the output layer then, the estimate euclidean loss is defined as: 1  ( pi − yi )2 2N i=1 N

H ( p, y) =

36.3.2.3

Hinge Loss Function

The Hinge loss function is widely used in binary classification problems. It is used in “maximum-margin” based classification problem, most notably for support vector machines (SVMs). Here, the optimizer tries to maximize the margin between two target classes. The hinge loss is defined as: H ( p, y) =

N 

max(0, m − (2yi − 1) pi )

i=1

where m is the margin which is normally set equal to 1, pi denotes the predicted output and yi denotes the desired output.

36.4 Training Process of Convolutional Neural Network In the previous Sect. 36.3, we have described the basic concepts of convolutional neural network (CNN) as well as the different key components of CNN architecture. Here in this section we try to discuss the training or learning process of a CNN model with certain guidelines in order to reduce the required training time and to improve model accuracy. The training process mainly includes the following steps: • Data pre-processing and Data augmentation. • Parameter initialization. • Regularization of CNN.

536

A. Ghosh et al.

• Optimizer selection. We will discuss those steps in the next subsections.

36.4.1 Data Pre-processing and Data Augmentation Data pre-processing refers to some artificial transformations to the raw dataset (including training, validation and testing datasets) in order to make the dataset more clean, more featurefull, more learnable and in a uniform format. The data preprocessing is done before feeding the data to the CNN model. In a convolutional neural, network it is a fact that the performance of CNN is directly proportional to the amount of data used to train it, i.e good pre-processing, always increases accuracy of the model. But on the other side, a bad pre-processing can also reduces the performance of the model. The general pre-processing techniques that are mostly used are given in the following subsections.

36.4.1.1

Mean-Subtraction (Zero Centering)

Here we substruct the mean from every individual data point (or feature) to make it zero-centered as shown in Fig. 36.17. The operation can be implemented mathematically as:  X = X − x∗ 1 xi N i=1 N

And, x ∗ =

where N denotes the size of the training datset.

Fig. 36.17 Mean-subtraction (Zero centering the data). (Source http://cs231n.stanford.edu/)

36 Fundamental Concepts of Convolutional Neural Network

537

Fig. 36.18 Normalization of data. (source http://cs231n.stanford.edu/)

36.4.1.2

Normalization

Here we normalize the data sample’s (belonging from both train, validation and test dataset) dimension by divideing each dimension by its standard deviationas shown in Fig. 36.18. The operation can be implemented mathematically as: X  = 

X N  i=1



(xi −x ∗ )2 N −1



where N , X and X ∗ are the same as discussed in Sect. 36.4.1.1 Data Augmentation is a technique used to artificially increase or expand the size of the training dataset. Here we apply different operations to the data samples (belonging to training datset only) and artificially transform it to one or many new data samples (new version), which is then used in training process. Data Augmentation is important because, sometimes there is a very limited sized training data set is available for most of the real-life complex problems (e.g. medical datasets) and the true fact is that the more training data samples can resulted in a more skillful CNN model. There are several data augmentation operations are available such as cropping, rotations, flipping, translations, contrast adjustment, scaling, etc. We can apply those operations separately or in combination to make several new versions from a single data sample. Another reason to use it is that the data augmentation is also able to enforce regularization in the CNN model by avoiding over-fitting problem as shown figures (Figs. 36.19, 36.20, 36.21 and 36.22).

36.4.2 Parameter Initialization A deep CNN consists of millions or billions number of parameters. So, it must be well initialized at the begin of the training process, because weight initialization directly

538

A. Ghosh et al.

determines how fast the CNN model would converge and how accurately it might end up. Here in this section, we discuss some mostly used parameter initialization techniques used in CNN as follows: The most easiest way to doing it is by initializing all the weights with zero. However, This turns out to be a mistake, because if we initialize weights of all layer to zero, the output as well as the gradients (during backpropagation) calculated by every neuron in the network will be the same. Hence the update to all the weights would also be the same. As a result, there is no disparity between neurons and the network will not learn any useful features. To break this disparity between neurons, we do not initialize all weights with the same value, rather than, we use different techniques to initialize the weights randomly as follows.

36.4.2.1

Random Initialization

As the name suggests, here we initialize the weights (belonging from both convolutional and FC layers) randomly using random matrices, where the elements of that matrices are sampled from some distribution with small standard deviation (e.g., 0.1 and 0.01) and with zero mean.

Fig. 36.19 An example of a raw training data sample

Fig. 36.20 Three new data samples, which are created by applying random cropping augmentation technique

36 Fundamental Concepts of Convolutional Neural Network

539

Fig. 36.21 Three new data samples, which are created by applying random flipping augmentation technique

Fig. 36.22 Three new data samples, which are created by applying random cropping and flipping augmentation technique in combination

But the key problem of random initialization is that it may potentially lead to vanishing gradients or exploding gradients problems. Some popular random initialization methods are: • Gaussian Random Initialization: Here weights are randomly initialized using a random matrices, where the elements of that matrices are sampled from an Gaussian distribution . • Uniform Random Initialization: Here weights are randomly initialized using a random Uniform matrices, where the elements of that matrices are sampled from an uniform distribution. • Orthogonal Random Initialization: Here weights are randomly initialized using random orthogonal matrices, where the elements of that matrices are sampled from an orthogonal distribution.

36.4.2.2

Xavier Initialization

This technique is proposed by Xavier Glorot and Yoshua Bengio in 2010, it tries to make the variance of output connections and input connections to be equal for each layer in the network. The main idea is to balancing of the variance the activation functions. This technique is turned out to be very useful at that time and since Xavier Glorot and Yoshua Bengio designed this tecnique for logistic sigmoid activation function, thus it does not perform well with ReLU (mostly used in CNN architecture nowadays) activation function. Later on, He initialization technique proposed by Kaiming He et al. is used to work with ReLU activation on the same idea.

540

36.4.2.3

A. Ghosh et al.

Unsupervised Pre-training Initialization

Here in this technique, we initialize a convolutional neural network with another convolutional neural network (that was trained using an unsupervised technique), for example, a deep autoencoder or a deep belief network. This method can sometimes workes very well by helping to handle both the optimization and overfitting issues. • Vanishing Gradient Problem: During Back-propagation over a deep convolutional neural network (CNN with many layers e.g.,1000 ), we need to calculate gradients of loss (Error) with respect to corresponding weights in each layer’s neurons for updating those weights. So, here we used the derivative operation to perform this task and it leads the gradients to gets smaller and smaller as we keep on moving backward in the network. Because of that, the earlier layer’s neurons receive very small gradients (sometimes gradients may become almost zero), as a result, the earlier layer’s weights get very mild update and those layers learn very slowly and inefficiently. This problem is known as Vanishing Gradient Problem. Few Solutions: 1. We can use ReLU/ leaky ReLU instead of others (tanh,sigmoid) as the activation function in CNN architecture, that helps to avoid this problem. 2. Residual networks provide another solution for this type of problem. Figure 36.23 shows basic block of the Residual network 3. Batch normalization layers can also resolve the issue. • Exploding Gradient Problem: This is the exact opposite of vanishing gradient problem, here large error gradients accumulate during back-propagation and it resulted in very large updates to network’s weights and make the model unstable and then this unstable model can not able to learns efficiently. The explosion in weight update occurs through the exponential growth of gradient by repeatedly multiplying gradients through the network layers during backpropagation when the gradients move backward in the network. At the extreme point, the values of weights (after updated with large gradients) may become so large as to overflow and resulted in NaN (Not a Number) values. Few Solutions: 1. We can use different Weight Regularization techniques to avoid this problem. 2. We can re-design the Network Model architecture to resolve the issue.

36.4.3 Regularization to CNN The core challenge of deep learning algorithms is to adapt properly to new or previously unseen input, drawn from the same distribution as training data, the ability

36 Fundamental Concepts of Convolutional Neural Network

541

Fig. 36.23 The basic block of the residual network

Fig. 36.24 The examples of over-fitting, under-fitting and just-fitting model’s hypothesis with respect to binary classification

to do so is called generalization. The main problem for a CNN model to achieve good generalization is over-fitting. When a model performs exceptionally well on training data but it fails on test data (unseen data), then this type of model is called over-fitted. The opposite is an under-fitted model, that happen when the model has not learned enough from the training data and when the model performs well on both train and test data, then these types of models are called just-fitted model. Figure 36.24 tries to show the examples of over-fitted, under-fitted and just-fitted models. Regularization helps to avoid over-fitting by using several intuitive ideas, some of which are discussed in the next subsections.

36.4.3.1

Dropout

Dropout [39] is one of the most used approach for regularization. Here, we randomly drop neurons from the network during each training epoch. By dropping the units (neurons) we try to distribute the feature selection power to all the neurons equally and we forced the model to learn several independent features. Dropping a unit or neuron means, the dropped unit would not take part in both forward propagation or backward propagation during the training process. But in the case of testing process, the full-scale network is used to perform prediction. With the help of Figs. 36.25 and 36.26, we try to show the effect of dropout in network’s architecture during training.

542

Fig. 36.25 An normal neural network

Fig. 36.26 After applying dropout

A. Ghosh et al.

36 Fundamental Concepts of Convolutional Neural Network

36.4.3.2

543

Drop-Weights

It is very much similar to dropout (discuss in Sect. 36.4.3.1). The only difference is instead of dropping the neurons, here we randomly drop the weights (or connections between neurons) in each training epoch.

36.4.3.3

The ı 2 Regularization

The ı 2 regularization [26] or “weight decay” is one of the most common forms of regularization. It forces the network’s weights to decay towards zero (but not equal to zero) by adding a penalty term equal to the ‘squared magnitude’ of the coefficient to the loss function. It regularizes the weights by heavily penalize the larger weight vectors. This is done by adding 21 λw2 to the objective function, where λ is a hyperparameter, which decides the strength of penalization and w denotes the matrix norm of network weights. Consider a network with only a single hidden layer and with parameters w. If there are N neurons in the output layer and the prediction output and the actual output are denoted by yn and pn where n ∈ [0, N ]. Then the objective function will be as follows: 1 Cost Function = loss + λw2 2 In the case of euclidean objective function: Cost Function =

M  N 

1 ( pn − yn )2 + λw2 2 m=1 n=1

where M is number of training examples. Now the weight incrementation rule with the ı 2 regularization will be as: W eight I ncr ement = argmin w

36.4.3.4

N M  

( pn − yn )2 + λw

m=1 n=1

The ı 1 Regularization

The ı 1 regularization [26] is almost similar to the ı 2 regularization and also widely used in practice, but the only difference is, instead of using “squared magnitude” of coefficient as a penalty, here we used the absolute value of the magnitude of coefficients as a penalty to the loss function. So the objective function with ı 1 regularization as: Cost Function = loss + λw

544

A. Ghosh et al.

where λ is a hyper-parameter, which decides the strength of penalization and w denotes the matrix norm of network weights.

36.4.3.5

Data Augmentation

The easiest way to avoid over-fitting is to train the model on a large amount of data with several varieties. This can be achieved by using data augmentation, where we use several techniques to artificially expand the training dataset size. Section 36.4.1 have described the data augmentation techniques in more detail.

36.4.3.6

Early Stopping

In early stoping, we keep a small part (maybe 20–30%) of the train dataset as the validation set which is then used for Cross-Validation purposes, where we evaluate the performance of the trained model over this validation set in each training epochs. Here we use a strategy that stops the training process when the performance on the validation set is getting worse with the next subsequent epoch as shown in Fig. 36.27. As the validation error gets increased the generalization ability of the learned model also gets decreased.

36.4.3.7

Batch Normalization

Batch Normalization [17] explicitly makes sure that the output activations of a network will follow a unit Gaussian distribution by normalizing the output at each layer by subtracting the mean and dividing by the standard deviation. It can be imagined as the preprocessing step at every layer of the network, but it can be differentiable and integrated with the networks. Batch Normalization is used to reduce the “internal covariance shift” of the activation layers. The internal covariance shift can be explained as the change in the distribution of activations in each layer. Due to

Fig. 36.27 A typical illustration of early stopping approach during network training

36 Fundamental Concepts of Convolutional Neural Network

545

continuous weight updation during training, the “internal covariance shift” can may become very high (it may happen when the training data samples are taken from several different distribution e.g., day-light images and night-vision images) and with this high “internal covariance shift” the model takes more time to converge and training time will increase. To solve this problem Batch Normalization operation is implemented as a layer in a CNN architecture. The benefits of using batch normalization are given below: 1. It also avoids the vanishing gradient problem. 2. It can handle bad weight initialization very efficiently. 3. It greatly improves the network convergence time (it becomes very helpful in case of large-scale dataset). 4. It tries to reduce training dependency over hyper-parameters. 5. It reduces the chances of overfitting because it has a slight regularization effect.

36.4.4 Optimizer Selection After successfully discussing the pre-processing steps with the data samples and enforcing regulization techniques to the CNN model, here we are going to discuss the learning process of the CNN model. The learning process includes two major things, first one is the selection of the learning algorithm (Optimizer) and the next one is to use several improvements (such as momentum, Adagrad, AdaDelta) to that learning algorithm in order to improve the result. The main objective of any supervised learning algorithm is to minimize the Error (difference between the predicted output and the actual output) or we can say the loss functions, based on several learnable parameters like weights, biases, etc. In case of learning to a CNN model the gradient-based learning methods come as a natural choice. To reduce the error the model parameters are being continuously updated during each training epoch and the model iteratively search for the locally optimal solution in each training epoch. The size of parameter updating steps is called the “learning rate” and a complete iteration of parameter update which includes the whole training dataset once is called a “training epoch”. Although the learning rate is a hyper-parameter, but we need to choose it so carefully that, it does not affect the learning process badly as shown in Fig. 36.28. Gradient Descent or Gradient-Based Learning Algorithm: As we discussed, the gradient descent algorithm update the model parameters continuously during each training epoch in order to reduce the training error. For updating those parameters in correct way, it first calculates the slope (gradient) of the objective function by using first-order derivative with respect to the model’s parameters, and then to minimize the error it update the parameter in the opposite direction of the slope (gradient) as shown in Fig. 36.29. This parameter updation process is done during back-propagation of the model, where the gradient at each neuron is back

546

A. Ghosh et al.

propagate to all the neurons belonging form it’s previous layer. This operation can be mathematically represented as: wi j t = wi j t−1 − wi j t wi j t = η ∗

∂E ∂wi j

where wi j t denotes the final weight in current t’th training epoch, wi j t−1 denotes the weight in previous (t − 1)’th training epoch, η is the learning rate, E is the prediction Error. There exist a number of variants of the gradient-based learning algorithm, out of them the most widely used are: 1. Batch Gradient Descent 2. Mini Batch Gradient Descent 3. Stochastic Gradient Descent. We are going to discuss them very briefly in the next subsections. Another widely used learning algorithm or optimization technique is Adam optimization, which uses second-order derivative represented by Hessian Matrix. We also try to discuss it very briefly in next subsection.

Fig. 36.28 The effect of different learning rate (LR) value on the training process, where we can see, a very low LR value need more training epochs to get the optimal solution and a very large LR value can overshoot the optimal solution

Fig. 36.29 The working principle of gradient-based learning algorithm

36 Fundamental Concepts of Convolutional Neural Network

36.4.4.1

547

Batch Gradient Descent

In Batch Gradient descent [34] the parameters of the network are updated only once after passing the whole training dataset through the network. That is, it computes the gradient on the entire training set and then updates the parameters using this gradient. With Batch gradient descent the CNN model produces more stable gradient and also converges faster for small-sized datasets. It also needs fewer resources, because the parameters are updated only once for each training epoch. But if the training dataset becomes large, then it takes more time to converge and it may converge in local optimal solution (in case of non-convex problems).

36.4.4.2

Stochastic Gradient Descent (SGD)

Unlike Batch Gradient descent, here the parameters are updated for each training sample separately [3]. Here it is recommended to randomly shuffle the training samples in each epoch before training. The benefit of using it over Batch Gradient descent is that it converges much faster in case of large training dataset and it is also memory efficient. But the problem is, due to frequent updates it takes very noisy steps towards the solution that make the convergence behavior very unstable.

36.4.4.3

Mini Batch Gradient Descent

Here we divide the training examples into a number of mini-batches in nonoverlapping manner, where each mini-batch can be imagined as the small set of samples and then update the parameters by computing the gradients on each minibatch. It carries the benefit of both Stochastic Gradient Descent and Mini Batch Gradient Descent by mixing them. It was more memory efficient, more computationally efficient and also has a stable convergence. Next are descriptions about different improvement techniques in Gradient-Based learning algorithms (typically, in SGD) which improves the training process of the CNN model more efficiently.

36.4.4.4

Momentum

Momentum is a technique used in the objective function of neural networks, it improves both training speed and accuracy by adding the gradient calculated at the previous training step weighted by a parameter λ called the momentum factor. The major problem of Gradient-Based learning algorithm is that it easily stuck in a local minima instead of global minimum, this mostly happens when the problem has non-convex solution space (or surface) as shown in Fig. 36.30. The following figure illustrates this problem.

548

A. Ghosh et al.

Fig. 36.30 Illustration the examples of optimizer stuck in a local minima problem

To solve this issue we used the momentum along with the learning algorithm. It can easily mathematically expressed as: wi j t = (η ∗

∂E ) + (λ ∗ wi j t−1 ) ∂wi j

where wi j t is weight increment in current t’th training epoch, η is the learning rate and λ is the momentum factor and (wi j t−1 ) is weight incrementation in previous (t − 1)’th training epoch. The value of momentum factor is staying in between 0 and 1 that increases the step size of weight update towards minima, for minimizing the error. The large value of momentum factor helps the model to converge faster and the very lower value of momentum factor can not able to avoid local minima. But if we use both LR and momentum factor value as high, it may also missed global minima by jumping over it. In case, if the gradient keeps changing its direction continuously during training the good momentum factor value makes smoothing out the variations of weight update. The momentum factor is a hyper-parameter.

36.4.4.5

AdaGrad

AdaGrad or adaptive learning rate method update each network parameter differently, based on their significance for the problem, that is, here we perform larger updates for infrequent parameters (by using larger learning rate value) and smaller updates for frequent parameters (by using smaller learning rate value). It is done by dividing the learning rate of each parameter with the sum of square of all the past gradients for each parameter wi j in each training epoch t. In practice AdaGrad is very useful, especially in case of sparse gradients or when we have sparse training data for a large scale neural networks. The update operation can easily mathematically expressed as:

36 Fundamental Concepts of Convolutional Neural Network

wi j t = wi j t−1 − 

η t 

δi j

549

∗ δi j t k2

+∈

k=1

where wi j t is the weight in current t’th training epoch for parameter wi j , wi j t−1 is the weight in previous (t − 1) ’th training epoch for parameter wi j , δi j t is the local gradient of parameter wi j in t ’th epoch , δi j t−1 is local gradient of parameter wi j in (t − 1)’th epoch, η is the learning rate and ∈ is a term contains very small value to avoid dividing by zero.

36.4.4.6

AdaDelta

AdaDelta can be imagined as the extension of AdaGrad. The problem with AdaGrad is that, if we train the network with many large training epochs (t), then the sum t  2 of square of all the past gradients ( δi j m ) become large, as a result, it almost m=1

vanishes the learning rate. To solve this issue the adaptive delta (AdaDelta) method divide the learning rate of each parameter with the sum of square of past k gradients (instead of using all the past gradients, which is done in the case of AdaGrad) for each parameter wi j in each training epoch t. The update operation can easily mathematically expressed as: wi j t = wi j t−1 − 

η t 

∗ δi j t δi j m 2 + ∈

m=(t−k+1)

where wi j t is the weight in current t’th training epoch for parameter wi j , wi j t−1 is the weight in previous (t − 1)’th training epoch for parameter wi j , δi j t is local gradient of parameter wi j in t’th epoch , δi j t−1 is local gradient of parameter wi j in (t − 1)’th epoch, η is the learning rate and ∈ is a term contains very small value to avoid dividing by zero.

36.4.4.7

RMSProp

Root Mean Square Propagation (RMSProp) is also designed to solve the Adagrad’s radically diminishing learning rates problem as discussed in the previous section. It was developed by Geoffrey Hinton’s group, it tries to resolve Adagrad’s issue by using a moving average over past squared gradient E[δ 2 ]. The update operation can easily mathematically expressed as:

550

A. Ghosh et al.

wi j t = wi j t−1 − 

η E[δ 2 ]t

∗ δi j t

And, E[δ 2 ]t = γ E[δ 2 ]t−1 + (1 − γ )(δ t )2 where for an efficient learning rate adjustment during training. Hinton suggests γ to be set to 0.9, with a good default initial learning rate value like 0.001.

36.4.4.8

Adaptive Moment Estimation (Adam)

Adaptive Moment Estimation (Adam [19]) is another learning strategy, which calculates adaptive LR for each parameter in the network and it combines the advantages of both Momentum and RMSprop by maintaining the both exponential moving average of the gradients (as like Momentum) and as well as the exponential moving average of the squared gradients (as like RMSprop). So the formulas for those estimators are as: E[δ]t = γ1 E[δ 2 ]t−1 + (1 − γ1 )[δ t ] E[δ 2 ]t = γ2 E[δ 2 ]t−1 + (1 − γ2 )(δ t )2 E[δ]t is the estimate of the first moment (the mean) and E[δ 2 ]t is the estimate of the second moment (the uncentered variance) of the gradients. Since at initial training epoch the both estimates are set to zero, they can remain biased towards zero even after many iterations, especially when γ1 , γ2 are very small. To counter this issue the estimates are calculated after bias-correction and the final formulas for those estimators become as: E[δ]t t = E[δ] (1 − (γ1 )t )  2 ]t = E[δ

E[δ 2 ]t (1 − (γ2 )t )

So the parameter update operation in Adam finally can easily mathematically expressed as: η t wi j t = wi j t−1 − ∗ E[δ]  2 ]t + ∈ E[δ • Adam is more memory efficient than others and also needs less computational power.

36 Fundamental Concepts of Convolutional Neural Network

551

36.5 Recent Advancement in CNN Architectures Till now it has covered the basic concepts of CNN along with different basic components or building blocks of CNN in Sect. 36.3. Then in Sect. 36.4, we discuss the learning process of CNN with several learning algorithms with guidelines in order to improve efficiency (including pre-processing, parameter initialization and regularization to CNN). In this section we try to explain some example of successful CNN architecture that shows the recent major advancements in CNN architecture in the computer vision field. Computer vision has three major subdomain where several CNN architectures (models) contribute a vital role to achieve excellent result, we are going to discuss those subdomains with related CNN models as follows.

36.5.1 Image Classification In image classification, we assume that the input image contains a single object and then we have to classify the image into one of the pre-selected target classes by using CNN models. Some of the major CNN architectures (models) designed for image classification are briefly described as follows:

36.5.1.1

LeNet-5

The LeNet-5 [21] is one of the earliest CNN architecture, which was designed for classifying the handwritten digits. It was introduced by LeCun et al. in 1998. The LeNet-5 has 5 weighted (trainable) layers, that is, three convolutional layer and two FC layers. Among them, each of first two convolution layer is followed by a maxpooling layer (to sub-sample the feature maps) and afterward, the last convolution layer is followed by two fully connected layers. The last layer of those fully connected layers is used as the classifier, which can classify 10 digits. The architecture of LeNet5 is shown in Fig. 36.31. Key Note:

Fig. 36.31 The architecture of LeNet-5 [21]

552

A. Ghosh et al.

• The LeNet-5 was trained on the MNIST digit dataset. • It used sigmoid non-linearity as the activation function. • It used stochastic gradient descent (SGD) learning algorithm with 20 training epoch. • It used 0.02 as the momentum factor value. • It reduced test error rate to 0.95% on MNIST data set.

36.5.1.2

AlexNet

Inspired from LeNet, Krizhevky et al. designed first large-scale CNN model, called AlexNet [20] in 2012, which is designed to classify ImageNet data. It consists of eight weighted (learnable) layers among which the first five layers are convolutional layers, and afterward, the last three layers are fully connected layers. Since it was designed for ImageNet data, so the last output layer classify the input images into one of the thousand classes of the ImageNet dataset with the help of 1,000 units. The architecture of AlexNet is shown in Fig. 36.32. Key Note: • The AlexNet used rectified linear unit (ReLU) non-linearity activation function after each convolutional and fully connected layer. • It used max-pooling layer after each LRN layer and the last convolutional layer. • Since it has a larger number of weights (learnable), so to avoid over-fitting it uses several regularization tricks like dropout and data augmentation. • The AlexNet was trained using stochastic gradient descent(SGD) learning algorithm with min-batch size 128, weight decay 0.0005 and momentum factor value 0.9. • The AlexNet was trained (on the ImageNet dataset) in two NVIDIA GTX 580 (with 3 GB memory) using cross-GPU parallelization and it takes around six days to complete. • AlexNet was the winner of ILSVRC-2012.

Fig. 36.32 The architecture of AlexNet [20]

36 Fundamental Concepts of Convolutional Neural Network

553

Fig. 36.33 The architecture of ZFNet [46]

36.5.1.3

ZFNet

ZFNet [46] was presented by Zeiler and Fergus in ECCV-2014, it has almost similar architecture as AlexNet except that here they used 7 × 7 filter with stride 2 in 1’st convolutional layer. In case of AlexNet, Krizhevky et al. use 11 × 11 filter with stride 4 in 1’st convolutional layer. As a result ZFNet becomes more efficient than AlexNet and become the winner of ILSVRC-2013. The architecture of ZFNet is shown in Fig. 36.33.

36.5.1.4

VGGNet

VGGNet [38] is one of the most popular CNN architecture, which is introduced by Simonyan and Zisserman in 2014. The authors introduced a total 6 different CNN configurations, among them the VGGNet-16 (configuration D) and VGGNet19 (configuration E) are the most successful ones. The architecture of VGGNet-16 is shown in Fig. 36.34. Key Note: • The reason behind the popularity of VGGNet is its architectural simplicity and the use of small-sized filters for convolutional operation. • It shows, a stack of 3 × 3 sized filters has same effective receptive field as the larger sized filters in convolution operation (e.g., two layers of 3 × 3 sized filters has same effective receptive field as the 5 × 5 filters in convolution operation, 7 × 7 filters with three layer of 3 × 3 sized filters, and so on). Most importantly, use of small sized filters decreases the number of parameters of the network. • The VGGNet was designed and trained on the ILSVRC dataset. • It is very deeper network compare to AlexNet.

Fig. 36.34 The Architecture of VGGNet [38] (VGGNet-16)

554

A. Ghosh et al.

Fig. 36.35 a Simple Inception Module [43], b Inception Module with dimensionality reduction [43]

• It also used rectified linear unit (ReLU) non-linearity activation function after each convolutional and fully connected layer. • It also uses several regularization tricks like dropout and data augmentation to avoid over-fitting.

36.5.1.5

GoogLeNet

The GoogleNet [43] architecture is different from all the previously discussed conventional CNN models, It uses network branches instead of using single line sequential architecture. The GoogleNet was proposed by Szegedy et al. in 2014. The GoogleNet has 22 weighted (learnable) layers, it used ‘Inception Module’ as the basic building block of the network. The processing of this module happens in parallel in the network, and each (a simple basic) module consist of 1 × 1, 3 × 3 and 5 × 5 filtered convolution layers in parallel and then it combines their output feature maps, that can resulted in very high-dimensional feature output. To solve this issue they used inception module with dimensionality reduction (as shown in Fig. 36.35b) in their network architecture instead of the naive (basic) version of inception module (as shown in Fig. 36.35a). Key Note: • Although the GoogLeNet has 22 layers, but it has 12 times lesser parameters than AlexNet. • It has auxiliary classifiers, that is use to combat vanishing gradient problem. • It also used rectified linear unit (ReLU) non-linearity activation function. • It used an average pooling layer instead of the fully connected layers. • The GoogLeNet used SGD learning algorithm with a fixed learning rate and with 0.9 as momentum factor. • The GoogLeNet was the winner of ILSVRC-2014.

36 Fundamental Concepts of Convolutional Neural Network

555

Fig. 36.36 a Mapping inside Residual block, b Simple direct mappings

36.5.1.6

ResNet

Since a deep CNN model suffers from vanishing gradient problems as we discussed earlier, He et al. from Microsoft, introduced the idea of ‘identity skip connection’ to solve vanishing gradient problem by proposing the ResNet [14] model. The ResNet’s architecture use residual mapping (H (x) = F(x) + x) instead of learning a direct mapping (H (x) = F(x)) and these bolcks are called residual bocks. The complete ResNet architecture is consist of many residual bocks with 3 × 3 convolution layers. Figure 36.36 illustrates the difference between the direct mapping and the residual mapping. Key Note: • The authors propose several version of ResNet with different depth, and they also used ‘bottleneck’ layer for dimensionality reduction in each ResNet architecture that has depth more than 50. • Although the ResNet (with 152 Layer) is 8 times deeper than VGGNets (22 layers), it has complexity lower than VGGNets (16/19). • The ResNet used SGD learning algorithm with the min-batch size of 128, weight decay of 0.0001 and momentum factor of 0.9. • The ResNet was the winner of ILSVRC-2015 with a big leap in performance, it reduces the top-5 error rate to 3.6% (the previous year’s winner GoogleNet has the top-5 error rate to 6.7% ).

36.5.1.7

DenseNet

DenseNet [15] extends the idea of residual mapping by propagating the output of each block to all subsequent blocks inside each dense block in the network. By propagating the information in both forward and backward directions during the training of the model it strengthens feature propagation ability and solve the vanishing gradient problem. DenseNet was introduced by Huang et al. in 2016 and it becomes the winner of ILSVRC-2016. Figure 36.37 shows a DenseNet based model.

556

A. Ghosh et al.

Fig. 36.37 A DenseNet based model with three dense blocks. [40]

36.5.2 Object Detection Unlike image classification, here an input image can contain more than one object. We try to detect those objects inside the input image with proper identification of each object along with their correct location inside that image by using CNN models. Some of the major CNN architectures(models) designed for object detection are briefly described as follows.

36.5.2.1

R-CNN

The first CNN model designed for object detection is Region-based CNN (R-CNN [10]), which uses sliding window based approaches for successfully detect the objects. Here the authors divides the complete task into three modules. The first module extracts the region (window) proposals from each input images that may likely contain some object (by using traditional selective search technique) and then in the second module, the authors used affine image warping to make all the extracted region proposals of fixed sized (or fixed aspect ratio) and then fed those warped region proposals through the AlexNet CNN model to extract the final features (fixed sized feature vectors). Finally, in the third module, the objects are classified from each region proposals by using category-specific linear support vector machine (SVM) classifier. The architecture of R-CNN is shown in Fig. 36.38.

36 Fundamental Concepts of Convolutional Neural Network

557

Fig. 36.38 Architecture of R-CNN [10]

36.5.2.2

SPP-Net

SPP-Net is almost similar to R-CNN, but SPP-Net [13] uses spatial pyramid pooling (SPP) on each variable length region (window) proposals to make them fixed sized output before feeding to the fully connected layers. SPP-Net removes the constraint of fixed sized input images from R-CNN and make itself more effective.

36.5.2.3

Fast R-CNN

Although R-CNN and SPP-Net works well, but they had some major problems like taking huge time to extract region proposals because of the traditional Selective Search technique, multi-stage pipeline process, etc. To solve those issues the Fast R-CNN [9], an CNN based detector uses convolution layer before the extraction of region proposals. This method substantiality reduce the number of region proposals, as the result Fast R-CNN became more efficient than R-CNN. The Fast R-CNN was designed by Ross Girshick, here both the objectives, identification of each object along with their correct location was performed in a single-stage process by using two sibling branch in the output layer. Fast R-CNN uses region of interests (RoI) pooling layer for reshape the variable length region (window) proposals into fixed sized output before feeding them to the fully connected layers. The architecture of Fast R-CNN is shown in Fig. 36.40.

36.5.2.4

Faster R-CNN

Faster R-CNN [32] is almost similar to Fast R-CNN, but, here the authors replaced previous traditional selective search technique with a region proposal network (RPN). The RPN is a fully convolutional network used to produce the high-quality region proposals. Faster R-CNN (combining RPN with Fast R-CNN) can be trained in an end-to-end fashion. By using RPN, Faster R-CNN achieves further speed-up in order to detect the objects inside the input image. The architecture of Faster R-CNN is shown in Figs. 36.39, 36.40.

558

A. Ghosh et al.

Fig. 36.39 A architecture of faster R-CNN [42]

Fig. 36.40 Architecture of fast R-CNN [9]

36.5.2.5

Mask R-CNN

Mask R-CNN [12] extend the concept of Faster R-CNN by locating exact pixels of each object (combinedly called object’s mask), whereas Faster R-CNN just predicts the bounding boxes of each object. Mask R-CNN use a RoIAlign layer instead of RoI pooling layer in same Faster R-CNN architecture. RoIAlign layer is used to align the extracted features of an object from region proposals to the location of that object in the final output. So, the Mask R-CNN generate three outputs: predicted class, the location of the object and a binary object mask. The architecture of Mask R-CNN is shown in Fig. 36.41.

36.5.2.6

YOLO

YOLO [31] (You Only Look Once) is a single pipeline based detection model, that can directly detect the objects as well as their location using an end-to-end trained CNN model.

36 Fundamental Concepts of Convolutional Neural Network

559

Fig. 36.41 Architecture of mask R-CNN [12]

YOLO splits the input image into a set of grides (fixed number) and then from each grides the network generate a fixed number of bounding box locations with a class probability. Then it uses a threshold value to select and locate the object within the image or not(when the class probability is less than the threshold value).

36.5.3 Image Segmentation CNN has shown excellence in the segmentation task also. After the record-breaking performance of AlexNet in 2012, we have got many state-of-the-art models of semantic segmentation and instance segmentation. Some of them are highlighted below.

36.5.3.1

Semantic Segmentation

Unlike Classification and object detection, semantic segmentation is a low level vision task. It is the process of associating each pixel of an image with a class label i.e. It detects all the objects present in an image.

Fully Convolutional Network (FCN): The author of FCN [37] have substituted the fully connected layer of AlexNet, VggNet and GoogLeNet (all three networks are pretrained on ILSVRC dataset) by 1x1 convolutional layers to make them dense FCN. The last layer is made up with a 1 × 1 convolution with channel dimension twenty one to predict scores for each of the PASCAL VOC [6] class (including background). To produce fine-grained seg-

560

A. Ghosh et al.

Fig. 36.42 End-to-end model of FCN [37]

mentation the authors have used bilinear interpolation and skip connection in their architecture. The end-to-end model of FCN is shown in Fig. 36.42.

DeepLab: Application of deep CNN in semantic segmentation has encountered two drawbacks: downsampling and spatial invariance. To handle the first problem the authors have used ‘atrous’ (with holes) algorithm. To get rid of the second problem they have used conditional random field (CRF) to capture fine details. DeepLab [4] achieved 71.6% Intersection over Union (IoU) accuracy in the test set of the PASCAL VOC 2012 semantic segmentation task. DeepLabv2 has one additional technology called Atrous Spatial Pyramid Pooling (ASPP), which is the main difference from DeepLabv1.

SegNet: The author has used encoder-decoder architecture [2] for semantic segmentation. The encoder part uses 13 layers of VGG16 network and the decoder part uses those 13 layers in reverse order. The last layer is a pixel-wise classification layer. The end-to-end model of SegNet is shown in Fig. 36.43.

Fig. 36.43 End-to-end model of SegNet [2]

36 Fundamental Concepts of Convolutional Neural Network

561

Fig. 36.44 End-to-end model of Deconvnet [27]

Deconvnet: This network [27] consists of 13 convolutional layers and 2 fully connected layers from VGG16 as convolutional network and those 15 layers in hierarchically opposite order are used in deconvolutional network. Convolutional layer extracts feature maps using convolution and pooling layers whereas the deconvolutional network uses deconvolution and unspooling to reconstruct the original size of activation. The endto-end model of Deconvnet is shown in Fig. 36.44.

U-Net: This architecture [33] consists of a u-shaped contracting and expansive path. Each step of the contracting path consists of two 3 × 3 convolutions, ReLU and 2 × 2 max-pooling. In contrast, expansive path consists of 2 × 2 upconvolution, 3 × 3 convolution and ReLU. In between upconvolution and convolution in expansive path, the feature map is concatenated with the cropped feature map from contracting path from corresponding layer.

36.5.3.2

Instance Segmentation

Instance segmentation takes the semantic segmentation one step ahead. It detects as well as distinguishes all the instances of an object present in an image.

DeepMask: The author used VGGNet in their architecture [28] but removed last max-pooling layer and fully connected layers. After VGGNet, the extracted feature map enters into two paths for class agnostic semantic mask and assigning a score corresponding to how likely the patch is to contain an object. The first path contains 1 × 1 convo-

562

A. Ghosh et al.

Fig. 36.45 End-to-end model of DeepMask [28]

lution layer, a nonlinear layer, two FC layers followed by a bilinear interpolation to mask single object. The second path contains 2 × 2 max pool layer followed by two FC layers and a prediction layer. The end-to-end model of DeepMask is shown in Fig. 36.45.

SharpMask: The architecture of SharpMask [29] is similar to DeepMask except it produces sharper, pixel-accurate object mask. This network consists of a series of convolution –pooling layer followed by fully connection layer to generate the object Mask.

PANet: This model [23] is based on the framework of Mask R-CNN and the Feature Pyramid Network. The main idea of PANet is to enhance information propagation through the network. The authors have used FPN based feature extractor associated with a new augmented bottom up pathway to improve the propagation of low layer features. To extract proposals from all level features, the feature maps are subsampled with a RoIAlign pooling layer. In each stage, an adaptative feature pooling layer processes the features maps with a fully connected layer. Then the network concatenates all the outputs. The output feature pooling layer goes to three branch for prediction of bounding box, prediction of object class and prediction of binary pixel mask.

TensorMask: This architecture uses dense sliding window approach instead of detecting object in a bounding box. The main concept of TensorMask [5] architecture is the use of structured high-dimensional tensors to present image content such as masks in a set of densely sliding windows. These models have two heads : One for generating mask in sliding window another for prediction of object categories.

36 Fundamental Concepts of Convolutional Neural Network

563

36.6 Applications Areas of CNNs In this section, we discuss some of the major application areas that apply CNN to achieve state-of-the-art performance including image classification, text recognization, action recognition, object detection, human pose estimation, image captioning, etc.

36.6.1 Image Classification Because of several capabilities like weight sharing, different level of feature extraction like classifiers, etc, the CNN have been achieving better classification accuracy [41] compared to other methods especially in the case of large scale datasets. The first breakthrough in image classification is comes with the development of AlexNet in 2012, which won the ILSVRC [36] challenge in that same year. After that, several improvements in CNN model have made by the researchers over the times, and that makes CNN as the first choice for image classification problem.

36.6.2 Text Recognition The text detection and text recognition inside an image has been widely studied for a long time. The first breakthrough contribution of CNN in this field begins with LeNet5, which recognized the data in MNIST [22] dataset with a good accuracy. After that in recent years, with several improvements, CNN contributes a vital role [44] to recognize the text (digits, alphabet and symbols belonging from several languages) inside the image.

36.6.3 Action Recognition Based on the visual appearance and motion dynamics of any human body, various effective CNN base methods are now able to predict the action or behavior of human subjects with a notable accuracy. This leads the CNN to the next level in the context of AI. It includes recognition of action from a video sequence or from the still images.

36.6.4 Image Caption Generation It means to obtaining a description about the target image, which includes detection and recognition of different objects inside that image with their status description.

564

A. Ghosh et al.

Here we used CNN to perform the first task [18] and we used several Natural Language Processing (NLP) techniques for a textual status description.

36.6.5 Medical Image Analysis With the advancement in CNN-based image analysis, CNN is rapidly proved to be a state-of-the-art foundation, by achieving enhanced performances in the diagnosis of diseases by processing medical images [1] like MRI, X-rays, etc. Nowadays, CNN based models can successfully diagnose the various health problems like breast cancer, pneumonia, brain tumour, diabetes, parkinson’s diseases and many others.

36.6.6 Security and Surveillance Nowadays, Security system with Computer Vision capabilities provides constant surveillance to houses, metro stations, roads, schools, hospitals, and many other places, that gives the ability to find or identify the criminals even in crowded areas [30]. CNN based models could be used for this purposes.

36.6.7 Automatic Colorization of Image and Style Transfer In the last few years, with the deep learning revolution, some popular CNN models give an automation way to convert black and white images or gray images to equivalent colorful RGB images [47]. As a result now we can see the old black and white movies in color format. On the other hand image style transfer is a concepts of representing an image in the style of other image, for that a new artificial image could be generated. This style transfer could be efficiently done using convolutional neural networks [8].

36.6.8 Satellite Imagery Nowadays, CNN contribute a vital role to detect different natural hazards [24] like tsunamis, hurricanes, floods, and landslides. By satellite image analysis we can do smart city plan, roadway and river extraction, land classification, crop pattern classification, prevention of deforestation and many more.

36 Fundamental Concepts of Convolutional Neural Network

565

36.7 Conclusion Convolutional Neural Networks(CNN) has become state-of-the-art algorithm for computer vision, natural language processing, and pattern recognition problems. This CNN has been using to build many use cases models from simply digit recognition to complex medical image analysis. This chapter tried to explain each components of a CNN, how it works to image analysis, and other relevant things. This chapter also gives a review from foundation of CNN to latest models and mentioned some applications areas.

References 1. Anwar, S.M., Majid, M., Qayyum, A., Awais, M., Alnowami, M., Khan, M.K.: Medical image analysis using convolutional neural networks: a review. J. Med. Syst. 42(11), 1–13 (Nov. 2018) 2. Badrinarayanan, V., Kendall, A., Cipolla, R.: Segnet: a deep convolutional encoder-decoder architecture for image segmentation. CoRR, abs/1511.00561 (2015) 3. Bottou, L.: Large-scale machine learning with stochastic gradient descent. In: Lechevallier, Y., Saporta, G. (eds.) Proceedings of COMPSTAT’2010, pp. 177–186. Heidelberg, Physica-Verlag HD (2010) 4. Chen, L., Papandreou, G., Kokkinos, I., Murphy, K., Yuille, A.L.: Semantic image segmentation with deep convolutional nets and fully connected crfs. In: 3rd International Conference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Conference Track Proceedings (2015) 5. Chen, X., Girshick, R.B., He, K., Dollár, P.: Tensormask: a foundation for dense object segmentation. CoRR, abs/1903.12174 (2019) 6. Everingham, M., Gool, L., Williams, C.K., Winn, J., Zisserman, A.: The pascal visual object classes (voc) challenge. Int. J. Comput. Vision 88(2), 303–338 (2010) 7. Fukushima, K.: Neocognitron: a self-organizing neural network model for a mechanism of pattern recognition unaffected by shift in position. Biol. Cybern. 36(4), 193–202 (1980) 8. Gatys, L.A., Ecker, A.S., Bethge, M.: Image style transfer using convolutional neural networks. In: The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2016) 9. Girshick, R.: Fast r-cnn. In: Proceedings of the 2015 IEEE International Conference on Computer Vision (ICCV), ICCV ’15, pages 1440–1448, Washington, DC, USA, (2015). IEEE Computer Society 10. Girshick, R.B., Donahue, J., Darrell, T., Malik, J.: Rich feature hierarchies for accurate object detection and semantic segmentation. CoRR, abs/1311.2524 (2013) 11. Goodfellow, I., Bengio, Y., Courville, A.: Deep Learning. MIT Press (2016). http://www. deeplearningbook.org 12. He, K., Gkioxari, G. Dollár P., Girshick, R.: Mask r-cnn. In: 2017 IEEE International Conference on Computer Vision (ICCV), pp. 2980–2988 (2017) 13. He, K., Zhang, X., Ren, S., Sun, J.: Spatial pyramid pooling in deep convolutional networks for visual recognition. CoRR, abs/1406.4729 (2014) 14. He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learning for image recognition. In: The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2016) 15. Huang, G., Liu, Z., Weinberger, K.Q.: Densely connected convolutional networks. CoRR, abs/1608.06993 (2016) 16. Hubel, D.H., Wiesel, T.N.: Receptive fields and functional architecture of monkey striate cortex. J. Physiol. (Lond.) 195, 215–243 (1968)

566

A. Ghosh et al.

17. Ioffe, S., Szegedy, C.: Batch normalization: accelerating deep network training by reducing internal covariate shift. CoRR, abs/1502.03167 (2015) 18. Karpathy, A., Fei-Fei, L.: Deep visual-semantic alignments for generating image descriptions. In: The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2015) 19. Kingma, D.P., Ba, J.: Adam: a method for stochastic optimization. arXiv preprint arXiv:1412.6980 (2014) 20. Krizhevsky, A., Sutskever, I., Hinton, G.E.: Imagenet classification with deep convolutional neural networks. In: Pereira, F., Burges, C.J.C., Bottou, L., Weinberger, K.Q. (eds.) Advances in Neural Information Processing Systems 25, pp. 1097–1105. Curran Associates, Inc. (2012) 21. Lecun, Y., Bottou, L., Bengio, Y., Haffner, P.: Gradient-based learning applied to document recognition. Proc. IEEE 86(11), 2278–2324 (1998) 22. LeCun, Y., Cortes, C.: MNIST handwritten digit database (2010) 23. Liu, S., Qi, L., Qin, H., Shi, J., Jia, J.: Path aggregation network for instance segmentation. CoRR, abs/1803.01534, (2018) 24. Maggiori, E., Tarabalka, Y., Charpiat, G., Alliez, P.: Convolutional neural networks for largescale remote-sensing image classification. IEEE Trans. Geosci. Remote. Sens. 55(2), 645–657 (2017) 25. Nair, V., Hinton, G.E.: Rectified linear units improve restricted boltzmann machines. In: Proceedings of the 27th International Conference on International Conference on Machine Learning, ICML’10, pp. 807–814. USA (2010). Omnipress 26. Ng, A.Y.: Feature selection, l1 versus l2 regularization, and rotational invariance. In: Proceedings of the Twenty-first International Conference on Machine Learning, ICML ’04, pages 78–, New York, NY, USA (2004). ACM 27. Noh, H., Hong, S., Han, B.: Learning deconvolution network for semantic segmentation. CoRR, abs/1505.04366 (2015) 28. Pinheiro, P.H.O., Collobert, R., Dollór, P.: Learning to segment object candidates. CoRR, abs/1506.06204 (2015) 29. Pinheiro, P.H.O., Lin, T., Collobert, R., Dollór, P.: Learning to refine object segments. CoRR, abs/1603.08695 (2016) 30. Rasti, P., Uiboupin, T., Escalera, S., Anbarjafari, G.: Convolutional neural network super resolution for face recognition in surveillance monitoring, vol. 9756, pp. 175–184 (2016) 31. Redmon, J., Divvala, S., Girshick, R., Farhadi, A.: You only look once: Unified, real-time object detection. In: 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 779–788 (2016) 32. Ren, S., He, K., Girshick, R., Sun, J.: Faster r-cnn: towards real-time object detection with region proposal networks. In: Cortes, C., Lawrence, N.D., Lee, D.D., Sugiyama, M., Garnett, R. (eds.) Advances in Neural Information Processing Systems, vol. 28, pp. 91–99. Curran Associates, Inc. (2015) 33. Ronneberger, O., Fischer, P., Brox, T.: U-net: Convolutional networks for biomedical image segmentation. In: Medical Image Computing and Computer-Assisted Intervention (MICCAI), volume 9351 of LNCS, pp. 234–241. Springer, 2015. Available on arXiv:1505.04597 [cs.CV] 34. Ruder, S.: An overview of gradient descent optimization algorithms. CoRR, abs/1609.04747 (2016) 35. Rumelhart, D.E., Hinton, G.E., Williams, R.J.: Learning internal representations by error propagation. In: Rumelhart, D.E., Mcclelland, J.L. (eds.) Parallel Distributed Processing: Explorations in the Microstructure of Cognition. Volume 1: Foundations, pp. 318–362. MIT Press, Cambridge, MA (1986) 36. Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bernstein, M., Berg, A.C., Fei-Fei, L.: ImageNet large scale visual recognition challenge. Int. J. Comput. Vis. (IJCV) 115(3), 211–252 (2015) 37. Shelhamer, E., Long, J., Darrell, T.: Fully convolutional networks for semantic segmentation. IEEE Trans. Pattern Anal. Mach. Intell. 39(4), 640–651 (2017) 38. Simonyan, K., Zisserman, A.: Very deep convolutional networks for large-scale image recognition. CoRR, abs/1409.1556 (2014)

36 Fundamental Concepts of Convolutional Neural Network

567

39. Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I., Salakhutdinov, R.: Dropout: a simple way to prevent neural networks from overfitting. J. Mach. Learn. Res. 15(1), 1929–1958 (2014) 40. Sufian, A., Ghosh, A., Naskar, A., Sultana, F.: Bdnet: bengali handwritten numeral digit recognition based on densely connected convolutional neural networks. CoRR, abs/1906.03786 (2019) 41. Sultana, F., Sufian, A., Dutta, P.: Advancements in image classification using convolutional neural network. In: 2018 Fourth International Conference on Research in Computational Intelligence and Communication Networks (ICRCICN), pp. 122–129 (2018) 42. Sultana, F., Sufian, A., Dutta, P.: A review of object detection models based on convolutional neural network. CoRR, abs/1905.01614 (2019) 43. Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., Erhan, D., Vanhoucke, V., Rabinovich, A.: Going deeper with convolutions. In: The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2015) 44. Wang, T., Wu, D.J., Coates, A., Ng, A.Y.: End-to-end text recognition with convolutional neural networks. In: Proceedings of the 21st International Conference on Pattern Recognition (ICPR2012), pp. 3304–3308 (2012) 45. Zaitoun, N.M., Aqel, M.J.: Survey on image segmentation techniques. Procedia Comput. Sci. 65, 797- 806 (2015). International Conference on Communications, management, and Information technology (ICCMIT’2015) 46. Zeiler, M.D., Fergus, R.: Visualizing and understanding convolutional networks. In: Fleet, D., Pajdla, T., Schiele, B., Tuytelaars, T. (eds.) Computer Vision - ECCV 2014. pp, pp. 818–833. Springer International Publishing, Cham (2014) 47. Zhang, R., Isola, P., Efros, A.A.: Colorful image colorization. CoRR, abs/1603.08511 (2016)

Chapter 37

Router Problems of Networking in Cloud Using SIEM Rajshree Srivastava

Abstract Cloud is a virtual space where a user can store, share and can access the data. In this chapter we have highlighted the major problem on cloud i.e., DOS Attack as well as other minor problems related to routers on networking. This chapter describes solution to some of major routing problem during an Attack. It contains statistical data as well as tools and techniques of major attacks performed on cloud. It also contains preventions from DOS attack on cloud server. The concept of accessing the data in cloud with the help of public and private key network is also explained with this survey report. Keywords SIEM · DOS attack · Data loss · Cloud security · Private key and public key · Routing network · OSI model layer · CDN · Bandwidth

37.1 Introduction SIEM is a networking tool used to monitor the networking using Log Table. This Log Table is not similar to the Logarithmic Table of Mathematics. The Log Table in SIEM is used for monitoring the network over a routing network. SIEM has two parts: SIM and SEM. SIM stands for Security Information Management. This helps in collecting information from different sources and storing them in a single location like a database. For example- collecting raw facts and figures from different sites and storing them at a single location to make it useful information [1]. On the other hand SEM stands for Security Event Management. This helps as a monitoring tool which monitors the information provided by SIM. It also protects the data of SIM. The paper is being divided into number of section, Sect. 37.1 introduction, Sect. 37.2 shows the working of SIEM, Sect. 37.3 shows the architecture of SIEM, Sect. 37.4 shows how to access on cloud, Sect. 37.5 shows an introduction to DOS R. Srivastava (B) Department of Computer Science and Engineering, DIT University, Dehradun, Uttrakhand, India e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_37

569

570

R. Srivastava

Attack, Sect. 37.6 shows how D-DOS Victims report cost in different categories, Sect. 37.7 shows some of the major case studies of D-DOS Attack, Sect. 37.8 shows prevention of D-DOS Attack, Sect. 37.9 shows routing and network concept, Sect. 37.10 shows major risk on cloud, Sect. 37.11 shows the prevention and Sect. 37.12 conclusion and future work.

37.2 Working of SIEM SIEM provides 3 types of detections on the basis of attacks performed on cloud network: 1. Supervised 2. Semi-Supervised 3. Unsupervised. Supervised: In this condition, the tool detects the attack just at the time of happening, so that attack can be prevented and there is a least loss of resources. Semi Supervised: In this condition, the tool detects when the attack is half performed. In this SIEM uses Cloud resources to prevent the attack. Unsupervised: In this condition, the attack is detected by SIEM when it is almost done. In this situation, almost all the resources of cloud are exhausted.

37.3 Architecture of SIEM SIEM Model works on three layers. Each layer has its own separate work to be performed individually [2]. Once the layer finishes its work it passes the work to another layer. Layer 1: The task of layer 1 is to collect all data from different sources and storing them at a single location in the form of logs. This information contains the following: – – – – –

Domain Controller Database Email Server Firewall System IDPS

Layer 2: The task of Layer 2 is to receive the information from Layer 1 and sends the information to the centralized server. Layer 3: Once the information reaches to centralized server, it starts working to process the Log data. Once the processing is finished, it sends the data to cloud for the storage.

37 Router Problems of Networking in Cloud Using SIEM

571

37.4 Accessing Information on Cloud On cloud we use the method of public and private key to secure data. When data is stored on cloud it is encrypted by public key and is decrypted by a private key which is only provided to user. Even the cloud service provider doesn’t have access to user data.

37.4.1 Concept of Public and Private Network Public and private both keys are used for encryption as well for decryption. Private key is created first in public key Crypto networking. Public key is generated in public domain where information is available for all to share but when a particular/specified user try to access that information it has to use private key which is attached with public key. Each individual has its own private key to decrypt the information which is encrypted by public key. These keys are generated using mathematical algorithm and just number figures.

37.5 Introduction to DOS Attack In today’s world full of technology, fast flow of information securely is at the upmost importance. Making this information safe is at the top of the chain with various algorithms and ways to encrypt data that are used and implemented by various sites. But that doesn’t stop hackers to intercept and hack the data, this make the site irresponsive. One of the techniques used by hackers to stop online services is the DOS attack [3]. It stands for Denial of Service. A DoS attack is a type of cyber-attack through which the attacker renders a machine or a network which is intended to be given to a user, unavailable. It is basically accomplished by flooding the host or the network with never ending requests resulting in the server being overloaded and making the functioning stops properly. A distributed denial-of-service attack (DDoS attack) is a type of DoS attack in which the requests intended to overload a server comes from different stops. As the requests are coming from multiple sources, it is next to impossible to stop the attack by simply blocking two or three sources [4]. There are some of the methods through which DOS Attack are done, these are as follows: 1. SYN-Flood It is a synchronized (SYN) message used to exploit the weaknesses found in the three-way handshake. In three-way handshake, the host receives a synchronized (SYN) message to begin the process; the server acknowledges it by sending a flag known as acknowledgment flag (ACK) to the sender, after that the connection is

572

2.

3.

4. 5.

R. Srivastava

closed. In SYN Flood, spoofed messages are sent which results in shutting down of the service. UDP-Flood UDP stands for User Datagram Protocol. It’s a session less protocol. UDP flood targets any port it likes on a computer or a network. The host or server checks if it catches this unusual traffic but it can’t be found. HTTP-Flood Here the traffic appears as legit GET or POST requests. Since they both are essential part of the web they are slided. However most of the time they are sent by a hacker and it forces the server to use maximum resources over this resource which in reality will require less resources. Ping-of-Death Malicious pings are sent to a system which in turn manipulates the IP protocols. Smurf-Attack A malware called smurf is used to exploit Internet Protocol (IP) and Internet Control Message Protocols (ICMP).

There are some of the symptoms through which D-DOS Attack can identified, these are as follows: 1. Server becoming laggy or crashing regularly. 2. Inability to access the web or a particular website. 3. A source continues to query certain data even after its Time to Live (TTL) has passed. 4. A data is taking an unusual amount to time to process. 5. Having unusual IP accessing your network using the net stat command. 6. One can’t simply see if traffic is originating from a single IP as it’s the main reason behind a DDoS.

37.6 How D-DOS Victims Report Cost in Different Categories The D-DOS Victims report in damage trust, business disruption, reactive speed, and professional services with their percentage of loss is given below, Table 37.1 shows Table 37.1 Damage trust

Damaged caused

Percentage (%)

Damage to company reputation

41

Loss of contacts or future business opportunities

35

Damage to their credit

33

Increase in insurance premiums

30

37 Router Problems of Networking in Cloud Using SIEM Table 37.2 Business disruption

Table 37.3 Reactive spend

Table 37.4 Professional services

573

Damaged caused

Percentage (%)

Temporary loss of access to critical information

63

Temporary loss of ability to trade

40

Hired PR/corporate image consultants

30

Damaged caused

Percentage (%)

Software or infrastructure costs

51

Staffing costs

50

Training costs

46

Damaged caused

Percentage (%)

IT security consultants

67

Lawyers

48

Risk management

43

Management consultants

42

Auditors/accountants

41

Physical security consultants

39

the damage trust, Table 37.2 shows the business disruption, Table 37.3 shows the reactive speed, and Table 37.4 shows the professional studies.

37.7 Major Case Studies Related to D-DOS Attack There are some of the major case studies which have occurred due to D-DOS Attack, these are as follows: 1. Dyn Attack It is considered as the second largest D-DoS attack [5]. It happened on October 2016 and was directed at Dyn, a major DNS provider. It led to disrupted services of various sites like Netflix, PayPal, Reddit and many more. A malware called Mirai was used to implement this attack. It creates a botnet and attaches itself to cameras, TV’s and other Internet of Things (IoT) devices. The compromised devices were programmed to send requests simultaneously to a single victim. Main suspicions were thrown towards WikiLeaks founder Julian Assange, to which he denied the claims.

574

R. Srivastava

Fig. 37.1 Statistic of D-DOS attack

2. GIT Hub Attack It was the largest ever D-DoS attack recorded in history [6]. The attack lasted several days and is believed to originate from China. The attack specifically targeted two projects at GitHub which were related to Chinese State Censorship. Main motivation behind the attack is thought to pressure the GitHub enough so that they remove those projects. 3. Mafiaboy Attack A 15 year old high schooler named Michael Calce, going by the name Mafiaboy implemented this attack by hacking into several of his nearby universities and forcing their servers to do the DDoS attack on websites which included CNN, Dell, eBay and Yahoo, which had devastating consequences (Fig. 37.1).

37.8 Preventions of D-Dos Attack There are some of the methods through which D-DOS Attack can prevented, these are as follows: • Many D-DoS protection services are available to monitor abnormal traffic and also to redirect unusual traffic away from your network.

37 Router Problems of Networking in Cloud Using SIEM

575

Fig. 37.2 OSI model

• Installing and maintaining a good antivirus and sufficient configurations on the firewall. • Regularly check your network traffic. Any unusual may mean you’re under a DoS attack. • Having a server with high capacity can overrule the abnormal traffic and requests sent from a DoS attack rendering them useless. • Intrusion Prevention System (IPS) is one of the counter attacks to D-DoS. • Having an extra amount of bandwidth won’t guarantee to stop DoS attacks but it will give you some extra time to remove the attack at its root. • Having multiple data centres will give you an extra edge even if one your centre is attacked and brought down. • Having sufficient protection against devices working on IoT will decrease the chances drastically. • High end companies can use a CDN (Content Delivery Network) to knock back or divert the DoS traffic somewhere else. • Having an experienced staff in your company is highly important. • Spoofed IP addresses should be blocked. Spoofed IP masks the botnet.

576

R. Srivastava

37.9 Routing and Network Concept in Stem Let us consider OSI Model Layer for diagnosis of Network related problems. Later in this survey, we will see the major methods provided by SIEM used to protect Router networking. Figure 37.3 presents the OSI Model (Fig. 37.2). Now some of the techniques to detect attacks on stem are as follows: Method 1(TCP/IP Flag) The best way to prevent attack on SIEM network is to distinguish between real user and fake user request. This can be done by using TCP/IP flag method. [7] We can set up flag when the request to server increases at sudden. At that time server will check the request and verify it, whether the packets at destination are same as sent from source or not. If the request is not from a real user than block that request and proceed to another request by raising TCP/IP flag. Method 2 (Sample Test Method) In this method, first we determine the root from source to the destination over networking. Between each individual routers, we make sample slots. After regular interval of time, we take the samples and will test them to check whether extra packets or some malicious packets have been inserted or not. If such packets are found we will block that routing chain to prevent the attack. Method 3(Awaiting Formula) In this method, when the packet is about to reach to the router from the source to the destination, we will check the packet with a pre-defined key. We check the key with the initial values assigned to them at source end. If the key matches then only the router will allow it to pass through it else it will return.

Fig. 37.3 Percentage of data loss

37 Router Problems of Networking in Cloud Using SIEM

577

37.10 Major Risk on Cloud The minor risks in Cloud Network which have very vast impact on Users using Cloud Services are: Data loss and malware injection. Data Loss Data loss is one of the most major issues we have in cloud. In this sometime the data of individual or group of individual get lost due to some reason. Either is lost because of overloading or by third party attack [8]. Data can be lost easily but is very difficult to recover. There is threat of losing customers on financial level as their private will be lost from cloud. The percentage of data loss is mentioned in Fig. 37.4 and stat of data breach is mentioned in Fig. 37.5. To prevent data loss new software used is CASB(cloud access security access broker). This software is placed between cloud server and the user and injects some safety resources to data on cloud. The second way to prevent data loss is by creating data backups. At regular intervals data backups should be made and have to be stored in secondary server. For this we can use DLP software. It monitors data as well gives an alert to prevent data loss. It even helps in identifying the unusual behavior during data sharing. Fig. 37.4 Stats of data breach

Fig. 37.5 Malware injection increase rate in year

578

R. Srivastava

Malware Injection It is process in which malicious software are injected in cloud to gain information related to its database. These are generally used to take some personal information. There are different types of malware injection sql injection, cross site scripting attack and man in the middle. Figure 37.5 represents the malware injection increase rate in year. The attack is being generated by certain steps, these are as follows: 1. 2. 3. 4.

Malicious user authenticates himself against cloud provider. Malicious user tries to access a vulnerable web application. Web application gets loaded onto the malicious users browser. Malicious user injects command by modifying HTTP request (GET or POST) parameters. 5. Victim user authenticates himself against cloud provider. 6. Victim user tries to access a web application. 7. Web interface gets loaded in the victim user’s browser and the user experiences long waiting time as a result of CI attack has considerable impact on user-toservice interface.

37.11 Prevention In order to prevent this attack there are certain software and tools which must be installed. These are as follows: • Anti-virus—This is a type of software which is used to detect any kind of incoming virus threats during the time of information exchange between the source and destination. For example—Storing or retrieving of data in a Cloud. • Anti-spy software—This is a type of software that provides protection on real time basis by incoming traffic scanning and by threats blockage. For example— Windows Defender in computers. • Spam Targeting—These are a special type of mails received by a user by some unknown sender. Once the user try to access the information of the mail by providing all its personal information will be transferred to third person on the network. For example—Fraud emails that is generally stored on spam folder. • IDS and Firewalls—Firewalls and intrusion detection systems provide traffic security for network activity and block anything strange. This is an enterprisegrade technology that gives protection to user computers, servers or networks from unauthorized access or malicious applications. Firewalls may not prevent installation of malware, but they can detect unwanted in-process tasks. • Security searches—They are generally used on websites which are related to Ecommerce in order to detect the harmful software or bug that can harm the application code. Most of these types of website developers use this security searches

37 Router Problems of Networking in Cloud Using SIEM

579

for hosting and selling things. These searches do not provide 100% security so another surveillance is also required. • Updated Patch—These try to keep your device connected to network so that it should be properly maintained and updated. As soon as the patch of the bug is available, it should be directly updated in the system settings so that attack threat can be minimized. • Common sense—The easiest way to deal with malware is to not get it in the first place. Experienced computer users avert potential disasters by practicing “skeptical computing,” which assumes that any new program is potentially harmful until proven safe.

37.12 Conclusion and Future Scope This paper gives a brief idea about SIEM Model and its layers that how it works and provides security over cloud network with a brief description of Cloud. It also consists of major attacks like DOS attack, Malware injection and Data Loss as well as their preventions. It also tells us about accessing the data with the help of public and private key concept. Further in future we can do survey on third security feature of SIEM which will be very useful as it can help us in improving networking bugs by providing security to routing devices. For example—we can make an alert system with the help of double socket layer that can work as an Alarming System which will raise a warning if an attack is being performed on a Networking Cloud.

References 1. Mchugh, J.: Testing intrusion detection systems: a critique of the 1998 and 1999 DARPA intrusion detection system evaluations as performed by Lincoln laboratory. ACM Trans. Inf. System Secur. (2000) 2. University of California: KDD Cup 1999 Data. http://kdd.ics.uci.edu/databases/kddcup99/ kddcup99.html (2011) 3. Shiravi, A., Shiravi, H., Tavallaee, M., Ghorbani, A.: Toward developing a systematic approach to generate benchmark datasets for intrusion detection. Comput. Secur. 31(3), 357–374 (2012) 4. Lincoln Laboratory MIT: DARPA intrusion detection evaluation. http://www.ll.mit.edu/mission/ communications/ist/corpora/ideval/index.html (2011) 5. Anastasov, I., Davcev, D.: SIEM implementation for global and distributed environments. IEEE (2014) 6. Nour, M., Slay, J.: UNSW-NB15: a comprehensive data set for network intrusion detection systems (UNSW-NB15 network data set). In: Military Communications and Information Systems Conference (MilCIS). IEEE (2015) 7. Chaure, T.M., Singh, K.R.: Frequent itemset mining techniques—a technical review. In: IEEE WCFTR (2016) 8. Lee, J., Lee, C., Cho, J.: A study on efficient log visualization using D3 component against APT how to visualize security logs efficiently? IEEE (2016)

Chapter 38

An Energy Efficient Clustered Routing Protocols for Wireless Sensor Networks Nitin Mittal and Rajshree Srivastava

Abstract Wireless sensor network (WSN) is a cost-effective networking solution for information updating in the coverage radius or in the sensing region. To record a real time event, large number of sensor nodes (SNs) need to be arranged systematically, such that information collection is possible for longer span of time. But, the hurdle faced by WSN is the limited resources of SNs. Hence, there is high demand to design and implement an energy efficient scheme to prolong the operational lifetime of WSN. Clustering based routing is the most suitable approach to support for load balancing, fault tolerance, and reliable communication to prolong performance parameters of WSN. These performance parameters are achieved at the cost of reduced lifetime of cluster head (CH). Inappropriate CH election may lead to more energy dissipation, overburden the CHs and thus degrades the network lifetime. So, there should be an appropriate selection of CHs using efficient routing algorithms to prolong the lifespan of network. To overcome this problem, many researchers make use of optimization algorithms for decision making of CH selection in WSN. This paper illustrates a survey of clustering hierarchical routing protocols along with clustering protocols based on optimization algorithms with possible future directions. Keywords WSN · Sensor nodes · Network lifetime · Stability period · Clustering

38.1 Introduction The vision of ambient intelligence with advancements in enabling technologies have led to the proliferation of new intelligent devices. Integrating all the capabilities of processing, sensing, storage, computation and communication into a tiny intelligent device opens the door to an important research subject termed as Wireless Sensor N. Mittal Electronics and Communication Engineering, Chandigarh University, Punjab 140413, India e-mail: [email protected] R. Srivastava (B) Computer Science Engineering, DIT University, Dehradun 248009, India e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_38

581

582

N. Mittal and R. Srivastava

Network (WSN) [1]. It comprises of numerous tiny elements known as sensor nodes (SNs) which can sense, compute and communicate with each other. The key benefit of WSNs is that these can be implemented almost anywhere without the need for any specific communication infrastructure [2]. SNs are networked in a self-organizing manner in applications that involve unattended operations. Current advances in wireless networking have empowered the development of daily life application of WSNs such as energy management, machine surveillance, home automation, disaster relief applications, health care, etc. [3]. The main objective of WSN deployment is to collect and report environmental data for data consumers. SNs measure/sense ambient events and then transmit data to the sink which usually has sufficient resources to process and store the information. As SNs usually have a limited communication range, they are not able to report data in single-hop to the sink if it resides out of reach (in terms of radio range). In this case, data packets are forwarded via a set of intermediate nodes (multi-hop paths) that relay the measured data from the event regions to the sink. WSN data reports are application dependent. This means that WSNs report ambient data according to the network applications and/or consumer requirements. These networks are basically designed for applications generally classified into detection of an event or any kind of phenomena, periodic measurements and tracking purposes [4]. In periodic monitoring applications, data is collected from all SNs at regular intervals and is generally delay tolerant. In event detection applications, the primary objective is to detect a Phenomenon of Interest (PoI) when it occurs—such as forest fire detection or flood estimation. Reports collected for these applications are delay sensitive [4]. Because of resource constraints in WSN (i.e. energy, bandwidth and/or communication range), none of the reporting schemes are a very good fit. Periodical reporting periodically transmit data even if data is redundant or the consumer is not interested in it. Thus, this scheme consumes network energy for collecting and reporting data that are irrelevant or useless. The cost of query-based reporting also is high as it needs several round communications during the reporting procedure until the sink receives data. The event triggered scheme is not flexible as the data reports cannot be controlled by the consumer. The nodes repeatedly measure and transmit the sensor data as soon as an event occurs without considering the consumer interests and/or queries. This model wastes network resources as SNs report the environmental data that the sink is not interested in. By and large, it seems that none of the reporting schemes are able to collect and report data in an energy efficient manner. Owing to this, and due to the fact that SNs have restricted resources, WSN needs to utilize an appropriate technique for data collection that has the potential to reduce the transmissions as much as possible [5]. This paper focuses on review of energy efficient data routing protocols for WSNs. The paper is organized as follows. Energy aware routing protocols are explained in Sect. 2. Reviewing the cluster-based routing protocols of WSNs is given in Sect. 3. Section 4 offers some conclusions and future directions.

38 An Energy Efficient Clustered Routing Protocols …

583

38.2 Energy Aware Routing in WSN The WSNs are increasingly in use due to low deployment cost and ability to sense environmental data. The network collects environmental data using ad hoc communications without any specific infrastructure or centralized control. SNs report the measured data to the consumer access points (sink) via the wireless channels. The wireless modules only allow the nodes to communicate over limited, short radio ranges. Owing to this, routing the network traffic from the event source regions to the sinks through single or multi-hop links is required. However, transmitting raw data samples requires the establishment of a number of communication links that consume significant amounts of network resources. Network layer in WSN protocol stack performs data routing and self-configuration of network. It finds the best route so that energy consumption is as minimum as possible. It is also responsible for updating of network topology, if any link failure occurs [5]. Energy-aware routing algorithms for network layer can be classified into different categories, and are given in Table 38.1. Data aggregation routing is one possible mechanism to transmit a summarized scheme of sensed data in a convergent fashion to the sink. For this reason, a number of data aggregation routing protocols ranging from sparse and small networks to large and dense ones with varying network application, topology and homogeneity are proposed. One of the energy efficient and widely studied classification of aggregation routing is cluster-based routing algorithms that offer outstanding advantages of scalability and energy efficient communication. It effectively organizes SNs into clusters, each cluster has a CH that is responsible of receiving and aggregating information from all CMs and sending the aggregated data to BS [6]. The research in this field mainly focuses on designing cluster-based data aggregation routing protocols that focuses on three objectives: energy saving, reducing delay and enhancing accuracy. However, these objectives are defined and optimized according to the consumer requirements. For example, a protocol that is able to reduce delay and increase accuracy is well-suited to real-time applications even if it does not optimize energy efficiency [24]. In addition, routing protocols should consider and balance the potential correlations between energy, delay and the number of collected data samples. Reducing delay by utilizing direct communication (instead of multi-hop) increases energy consumption, whereas decreasing energy consumption using multi-hop routing will increase data collection delay [25].

38.3 Cluster-Based Routing Protocols of WSN Much of the recent research work in the area of cluster-based routing has extensively focused on lifetime, stability, energy efficiency and scalability. Numerous energy efficient clustering algorithms have been proposed in this aspect for a wide range of

584

N. Mittal and R. Srivastava

Table 38.1 Network layer energy aware routing classifications in WSN [2] Classification

Key features

Classification of WSN routing protocols: mode of functions Proactive

The route is determined in advance before it is actually needed or required. The routing information is stored in routing table. If there is any change in predetermined route, then the routing table is updated with new route information. Each node has to keep and maintain a routing table with itself. Thus, it includes extra overhead on the nodes

Reactive

These on-demand routing protocols do not compute route unless it is required. There is no overhead occurs on node because the node need not to maintain a routing table

Hybrid

Combine characteristics of both proactive and reactive protocols

Classification of WSN routing protocols: participation style of nodes Direct

Allow nodes to send information directly to base station (BS)

Flat

All SNs play same role and functionality. Flat network structure provides an acceptable overhead so that infrastructure among nodes can be maintained [6]. As every node performs sensing and sending operation, therefore, all the data collected from nodes can be duplicated or same. For example, flooding and gossiping

Clustering

SNs are properly arranged and categorized into different clusters and a cluster head (CH) is selected from each cluster. CH node has higher energy as compared to other nodes. Member nodes forward data to CHs which further process and send it to BS

Classification of WSN routing protocols: network structure Data-Centric

These protocols are based on query. The data is diffused through SNs by using naming scheme of desired data in order to remove redundant transmissions. Example protocols are DD [7], ACQUIRE [8], SPIN [9]

Hierarchical

Low energy nodes send data to high energy nodes which further process the data and send it to BS. Hierarchical network structures provide an energy efficient routing in which data fusion is performed so that the number of messages transmitted to BS can be greatly reduced. Example protocols are LEACH [10], LEACH-C [11], PEGASIS [12], TEEN [13], APTEEN [14]

Location-based

Requires the location of SNs in order to determine the route. SNs keep the location information of itself as well as information of all its neighbors and BS. Example protocols are DREAM [15], GEAR [16], IGF [17], OGF [18], HGR [19]

QoS aware

QoS aware routing generally focuses on QoS parameters such as delay, bandwidth, latency and reliability. These protocols need to maintain a balance between energy consumption and QoS parameters. Example protocols are SAR [20], SPEED [21], MMSPEED [22], MGR [23]

38 An Energy Efficient Clustered Routing Protocols …

585

applications. This section presents some typical routing protocols proposed in the literature along with their limitation.

38.3.1 Classical Cluster-Based Routing Protocols Heinzelman et al. presented LEACH [10] that is the first energy-efficient routing scheme and is still used as a state-of-the art protocol in WSN. The basic idea of LEACH is to choose CH among a number of nodes by rotation so that energy dissipation from communication can be spread to all nodes in a network. LEACH has some disadvantages such as probabilistic approach using random number for CH selection, which might result in suboptimal CH node thus resulting in high energy consumption. Furthermore, the dynamic clustering overhead and non-uniform distribution of CH will consume more energy and lead to poor network performance. In LEACH, it is the possibility that a SN with less energy can be elected as CH; if that happened then it will become dead and consequently, that cluster becomes inaccessible. LEACH assumes even consumption of energy for every CH and does not guarantee proper CH distribution. LEACH-centralized (LEACH-C) [11] is the modified LEACH in which clusters are created by BS. BS receives all the information regarding residual energy and location of each SN. By doing so, BS determines the number of CH and arranges network into various clusters. However, due to lack of coordination among SNs, the CH count varies in each round. CHs are selected from the set of nodes to ensure that they should have sufficient energy. The BS selects lowest energy routing paths and forwards the information of clustering and CH to all nodes using a minimum spanning tree approach. However, due to centralized approach communication overhead will increase in the reselection of CH, because reselection decision has to be made by BS. In addition, every cluster will send request; thus energy consumption will be high. Energy aware LEACH (E-LEACH) exploits remaining energy at each nodes to elect CHs [26]. Lindsey and Raghavendra proposed a greedy algorithm called PEGASIS for chain formation and token passing scheme is used for communication between neighbors [12]. In this scheme, each node communicates with its neighbor node only, thus it reduces energy being spent or utilized per round. Each node gathers data from its neighbors, fuses it and forwards it to next node. Only a designated or leader node send fused data to BS. PEGASIS suffers from excessive delays because of single chain formation. To overcome these shortcomings, an algorithm named hierarchical PEGASIS (H-PEGASIS) was introduced in order to reduce the delays occurred when packets are transmitted to BS [27]. Hybrid energy efficient distributed clustering (HEED) [28] algorithm showed an improvement in network lifetime over LEACH. It utilizes a rotation-based clustering algorithm by considering four objectives: (1) utilizing a finite number of iterations for CH selection to reduce clustering/re-clustering overhead, (2) minimizing network traffic by reducing the number of control packets that are forwarded during

586

N. Mittal and R. Srivastava

re-clustering procedure, (3) forming compact and well-distributed clusters by considering CH connectivity degree, (4) uniform network load distribution over clusters to prolong network lifetime. HEED considers two metrics to select CHs and form the clusters: residual energy and node connectivity degree. It forms the cluster by selecting the nodes that have sufficient energy and are able to dominate a specific number of nodes as CHs. Chand et al. proposed a 3-level heterogeneous version of HEED [29]. CH selection in this protocol has been done using fuzzy logic. By applying fuzzy logic along with heterogeneity, the network lifetime of proposed protocol increases to a great extent in comparison to HEED protocol. Manjeshwar and Agrawal proposed an event-driven clustering approach called TEEN [13]. In this protocol, the sensed information is communicated to BS only if some event occurs, which is based on soft and hard thresholds. The disadvantage of this approach is that the node responds only if the change in the attributes crosses these threshold values, which makes the approach less applicable in dynamic environment, as the selection of two threshold values is very sensitive and difficult for real applications. The user may keep on waiting to get response and does not get any information about the status of the node, which makes this approach not suitable for the applications, where the periodic updates are required. Later, this approach has been enhanced and proposed as adaptive TEEN (APTEEN) [14]. APTEEN combines event-driven approach of TEEN and periodic approach of LEACH to address the problems occurring in TEEN. APTEEN is good for periodic applications, but the complexity of the approach increases due to inclusion of extra threshold function and count time. Stable election protocol (SEP) was introduced to prolong network stability period essential for some applications [30]. In this protocol, advance nodes (higher energy nodes) have greater chances to become CH that is good fairness constraint to ensure enhanced stability by using heterogeneity. This protocol has improvement over the existing LEACH as it increases the network epoch proportion to energy augmentation. An enhancement of SEP (SEP-E), proposed by Aderohunmu et al. [31], considers three level of heterogeneity. Three tier node hierarchy considers three types of nodes. Threshold-sensitive SEP (TSEP) is a reactive protocol with three-tier node classification [32]. In this algorithm, three types of SNs are considered. In TSEP, similar to TEEN, the data transmission depends upon threshold parameters. TSEP increases the stability, network lifespan and network throughput. Kumar et al. proposed an enhance TSEP for heterogeneous WSN (HWSN) [33]. It selects minimum number of clusters using remaining energy of SNs. ETSEP completes with other protocols in terms of operational lifetime and stability. Qing et al. investigated a new distributed energy-efficient clustering scheme (DEEC) for HWSNs [34]. In DEEC, all SNs are provided with different initial energy. The protocol selects CHs using a probability function such as higher energy nodes have greater probability being selected as a CH as compared to lower energy nodes. DEEC performs well in comparison with SEP and LEACH in terms of energy utilization and network lifespan.

38 An Energy Efficient Clustered Routing Protocols …

587

Elbhiri et al. presented an enhanced version of DEEC protocol named developed DEEC (DDEEC) [35]. DDEEC proposed a balanced way to distribute load equally between all nodes. These modifications increase the performance of DDEEC in comparison to DEEC. Javaid et al. introduced an enhanced DDEEC (EDDEEC) protocol for HWSNs [36]. EDDEEC considers the three level of heterogeneity. During initial rounds, advanced and super nodes have greater probability for being selected as CH than normal nods. Kang et al. proposed a protocol named LEACH with distance thresholds (LEACHDT) for CHs selection [37]. In LEACH-DT, CH selection probability depends upon its distance as a parameter from BS. BS determines the distance amongst all nodes and calculates the probability function. This information is broadcast by BS to all nodes. SNs make the decision on the basis of following information about CH selection without any centralized control. LEACH-DT also proposed multi-hop routing where sensors are grouped into different sensor groups (SGs). Within SG the single hop communication is used and all SNs within one SG send the data to its own CH. After this, data is transferred to the CH which is closer to BS. Using multi-hop transmission, energy consumption is reduced to some extent in which the data is relayed from the far-away SG to the closer ones. Golsorkhtabar et al. investigated a novel energy adaptive protocol (NEAP) for HWSNs [38]. NEAP selects CHs on the basis of confidence value. This confidence value is a function of some parameters and these parameters are: distance between SN and CH, CH current energy, and node degree i.e. number of members under one CH. After CHs election the remaining nodes join the CH that has enough battery power and less communication distance from other CHs. NEAP forms adaptive power efficient clustering that significantly improves energy management and operational lifetime of network. In Two-Level Hierarchy LEACH (TL-LEACH) [39], clusters are formed using randomized, self-configuring and adaptive technique. TL-LEACH uses two-level hierarchical structure for data transmission. CHs that are close to the BS are called primary CHs, and others are known as secondary CHs. The operation of TL-LEACH can be described using different phases. After election of primary and secondary CHs, primary CHs send the advertisement message to its leaf nodes. The leaf nodes sends the join request message to the secondary CHs on the basis of received signal strength. In communication phase, leaf node forwards the data to its respective CH according to allotted time slot. Secondary CHs forward aggregated information to BS using dual-hop communication with primary CHs. In TL-LEACH, primary and secondary CHs are randomly rotated to achieve better load distribution. Ye et al. proposed EECS algorithm in which SNs are organized into small clusters. Single-hop or direct communication is used for relaying the messages from CHs to the BS [40]. Each SN sends its remaining energy information to neighboring nodes. A particular SN compare it’s residual energy value with other neighboring nodes energy values. The higher residual energy node selects itself as CH for the current round. It optimizes the CHs selection process and also reduces the intra-cluster communication distance, as SNs select nearest CH in order to minimize transmission

588

N. Mittal and R. Srivastava

energy cost. EECS balances the energy consumption as it selects the optimal point between inter-cluster load distribution and intra cluster communication cost. Li et al. introduced EEUC protocol in which CHs are selected using localize competition [41]. To become a CH, every node generates a random number if generated number is greater than predefined threshold value, then node is selected as CH. CHs broadcast advertisement message within a competition radius and this competition radius is calculated by its distance from BS. Node’s competitive radius decreases as the node gets closer to BS. CHs that are closer to BS have smaller cluster size, therefore, their intra cluster communication energy cost is reduced. Within cluster, nodes uses multihop routing for relaying the message to the CH. EEUC solves the hot-spot problem to some extent, and it also reduces the energy consumption by introducing the relay nodes in between nodes and CH communication. Kumar proposed a single-hop EECP algorithm that uses three levels of node heterogeneity [42] and the similar WEP concept for CH selection as EEHC. However, the protocol considers the residual energy of SNs for calculating WEP of each node. In single-hop EECP (S-EECP), data is directly transmitted by nodes to BS without any relay node, therefore, SNs that are far from BS die out quickly. To avoid this problem, multi-hop communication is used in multi-hop EECP that uses a greedy scheme to discover shortest path from each CH to BS. Tarhani et al. introduced a scalable energy efficient clustering hierarchy (SEECH) suitable for periodic data transmission applications. It makes use of distributed approach in which CHs and relay nodes are selected separately [43]. The reason for different CH and relay node selection is to mitigate the energy burden of CHs. SEECH protocol performed well for large scale WSNs in comparison to competitive protocols. Mittal and Singh presented a reactive cluster-based approach called DRESEP which is ideal for event based applications [44]. Mittal et al. proposed SEECP [45] in that CHs are chosen in deterministic manner based on remaining energy of SNs to reduce the energy variance among sensors. A comparison of cluster-based routing protocols is presented in Table 38.2. Since major clustering protocols in WSNs are motivated by LEACH, a comprehensive comparison is based on LEACH and its extensions.

38.3.2 Heuristic-Based Clustering Protocols in WSN The objectives of hierarchal routing protocols are saving the dissipated energy, ensuring the network connectivity, and prolonging the lifetime of the WSNs. These objectives can be achieved via finding the optimal head nodes and hence forming optimum clusters in the WSNs. This is a difficult problem and can be considered as a Non-deterministic Polynomial (NP) optimization problem. To solve and find optimal solutions for this problem, researchers have developed the cluster based routing schemes with the optimization algorithms to achieve extended network lifespan.

References

Heinzelman et al. (2000)

Heinzelman et al. (2000)

Manjeshwar et al. (2001)

Manjeshwar et al. (2002)

Lindsey et al. (2002)

Smaragdakis et al. (2004)

Qing et al. (2006)

Golsorkhtabar et al. (2010)

Protocol

LEACH

LEACH-C

TEEN

APTEEN

PEGASIS

SEP

DEEC

NEAP

Distributed

Distributed

Distributed

Distributed

Centralized

Distributed

Centralized

Distributed

Method

Homogeneous

Heterogeneous

Heterogeneous

Homogeneous

Homogeneous

Homogeneous

Homogeneous

Homogeneous

Node type

Table 38.2 Comparison of cluster-based routing protocols

Constant

Variable

Variable

Cluster count

Proactive

Proactive

Proactive

Proactive

Variable

Variable

Variable

Variable

Reactive/Proactive Variable

Reactive

Proactive

Proactive

Nature

Direct

Direct

Direct

Multi-hop

Multi-hop

Multi-hop

Direct

Direct

Inter-cluster topology

Hybrid

Hybrid

WEP

Random

Deterministic

Random

Deterministic

Random

CH election

(continued)

Good

Excellent

Better

Good

Excellent

Excellent

Good

Good

Network lifetime

38 An Energy Efficient Clustered Routing Protocols … 589

References

Kumar et al. (2011)

Kang and Nguyen (2012)

Dilip Kumar (2013)

Dilip Kumar (2013)

Tarhani et al. (2014)

Mittal and Singh (2015)

Mittal et al. (2016)

Protocol

MDCA

LEACH-DT

S-EECP

M-EECP

SEECH

DRESEP

SEECP

Table 38.2 (continued)

Distributed

Distributed

Distributed

Distributed

Distributed

Distributed

Distributed

Method

Heterogeneous

Heterogeneous

Homogeneous

Heterogeneous

Homogeneous

Homogeneous

Heterogeneous

Node type

Reactive

Reactive

Proactive

Proactive

Proactive

Proactive

Proactive

Nature

Constant

Variable

Variable

Variable

Variable

Variable

Variable

Cluster count

Dual-hop

Dual-hop

Multi-hop

Multi-hop

Direct

Multi-hop

Multi-hop

Inter-cluster topology

Deterministic

WEP

Hybrid

Hybrid

Hybrid

Hybrid

WEP

CH election

Stable

Excellent

Excellent

Better

Good

Good

Better

Network lifetime

590 N. Mittal and R. Srivastava

38 An Energy Efficient Clustered Routing Protocols …

591

Hussain et al. [46] introduced GA based clustering approach named hierarchical cluster routing (HCR) protocol for CH selection with modified fitness function. At each generation the best chromosome corresponding to best fitness is selected. HCR prolongs the network lifetime in comparison to LEACH but failed to ensure a longer reliability period until first node dead. ERP [47], EAERP [48], SAERP [49] and STERP using DE [50], HSA [51], SMO [52] and GA [53] are recently developed optimization algorithms based clustering protocols. EAERP restructured substantial features of EAs that assures extended stability period and prolonged lifetime. ERP overcame the shortcomings of HCR algorithm [46] by improving the cluster quality of network. SAERP based routing schemes (DESTERP, HSSTERP, SMSTERP and GASTERP) are inspired from SAERP to achieve extended stability period [50–53]. Kuila et al. proposed a GA-based clustering approach [54] that minimizes the standard deviation of CH load to solve load balancing problem in WSNs. Kuila et al. [55] have also proposed a DE-based clustering approach to prolong network lifespan by balancing the lifespan of CHs by an efficient vector encoding scheme for CH election. Shokouhifar and Jalali proposed ASLRP algorithm to extend network lifespan of WSN [56]. CHs are selected using residual energy, distance of node from sink and distance between node and respective CH parameters using hybrid GASA optimizer. A centralized clustering algorithm using HSA, considers minimum energy dissipation and minimum intra-cluster mean distances as parameters for CH selection to improve network lifespan [57]. Kumar et al. suggested a learning automata-based efficient routing protocol for WSNs [58]. Two level of node heterogeneity is considered in the proposed scheme with an assumption that automaton is located on each node. CHs are elected by the automaton using the concept of WEP. Automaton at each node receives penalty or reward from the environment based upon WEP of different nodes. The proposed approach showed significant improvement in network lifetime and stability period. Sert et al. proposed MOFCA protocol in which CHs are selected using fuzzy logic approach [59]. It is designed primarily for two major factors, first it should be energy efficient and second it should be lightweight in order to be implemented in real scenario. MOFCA selects the tentative CHs using node competitive radius. Three parameters are considered for the election of CHs: node’s distance from the BS, node’s residual energy and node density. It uses fuzzy logic for calculation of above said parameters. Three fuzzy inputs are used for measurements of tentative CHs competitive radius. The main objective of MOFCA is to overcome the hotspot problem, which arises due to multi-hop communication. MOFCA is used for both stationary and mobile environments. The results of MOFCA show that it is better than the existing algorithms. Tomar et al. proposed a cluster-based routing protocol with fuzzy inference system [60]. Fuzzy-logic along with ACO is used for CH selection and to determine an optimal path between node and BS. Tamandani et al. introduced a clustering protocol named SEP based on fuzzy logic (SEPFL) [61]. To enhance the CH selection process in SEP, three variables namely distance of SNs from BS, remaining energy and density of nodes are used with fuzzy logic control.

592

N. Mittal and R. Srivastava

Zahedi et al. suggested a swarm intelligence based fuzzy clustering approach (SIF) for WSNs [62]. It considers remaining energy, distance to sink and distance from the cluster centroid as parameters for CH selection using fuzzy logic. A hybrid FA-SA optimizer is used to optimize the fuzzy rule base table of SIF. SIF achieves more energy efficiency and network lifetime in comparison to ASLRP. Table 38.3 Table 38.3 Metaheuristic algorithms based routing protocols in WSNs Protocol

References

Algo. used

Optimization criteria

Effect

HCR

Hussain et al. (2007)

GA

Energy consumption, number of clusters, cluster size, direct distance to sink and cluster distance

Increase in network lifetime

EAERP

Khalil et al. (2011)

GA

Min. total dissipated energy in the network

Well-distributed energy consumption

ERP

Attea et al. (2012)

GA

MIN. intra-distance and Max. inter-distance

Longer network life-time, lower energy consumption, but at the expense of less stability awareness

SAERP

Khalil et al. (2013)

GA

Min. total dissipated energy in the network

Prolong the stability period of the network

LBCP

Kuila et al. (2013)

GA

Standard deviation of gateway load

Better performance in terms of energy consumption, rate of convergence and execution time

HSACP

Panda et al. (2014)

HSA

Min. intra-cluster mean distance and Max. Network energy

Lower energy consumption and extended network life-time

ASLRP

Jalali et al. (2015)

Hybrid GA and SA

Distance from BS, residual energy, distance from other CHs

Maximize the network lifetime

SIF

Zahedi et al. (2016)

Fuzzy logic based Hybrid FA and SA

Distance from BS, residual energy, number of previously became CH, distance from other CHs

Efficiently balance the energy consumption of nodes and maximize the stability period of network (continued)

38 An Energy Efficient Clustered Routing Protocols …

593

Table 38.3 (continued) Protocol

References

Algo. used

Optimization criteria

TERP

Mittal et al. (2018)

DE/HSA/SMO Total energy dissipation, Min. intra-distance and Max. inter-distance, Round-trip delay

Uniform Cluster formation, longer network lifetime, lower energy consumption

STERP

Mittal et al. (2018)

DE/HSA

Prolong the stability period of the network

Total energy dissipation, Min. intra-distance and Max. inter-distance, Round-trip delay

Effect

indicates the above said meta-heuristic algorithms based routing protocols with an aim to prolong network lifetime along with the criteria for optimization.

38.4 Conclusion This paper reviews the recently hierarchical-based routing protocols that are developed for WSNs. In this survey, the hierarchical based routing protocols are grouped into classical-based routing and optimized-based routing. Also, a detailed classification of the reviewed protocols based on different metrics, such as control manner, network architecture, clustering attributes, protocol operation, path establishment, communication paradigm, protocol objectives, and applications, is presented in this paper. This survey can be helpful for designers of WSNs in selection of an appropriate hierarchical routing protocol for a specific application. The effort in this area should be continued in the area of the hierarchical routing to improve the performance of WSNs. Two open issues should be considered in the future research. First issue is the real-work implantations. Simulation of the routing protocols depends on modeling the hardware of SNs. However, the mathematic models do not exactly simulate the real world. Second issue is the overheads and computational time of clustering the network. A centralized routing is more energyefficient than the distributed routing. However in the centralized routing, all SNs should send their information such as their locations and energy to BS. This takes some time and increases the overheads of protocol especially in the dense networks and BS is far from the sensor field. Designers should develop semi distributed or semi centralized protocols that run within CH rather than sink.

594

N. Mittal and R. Srivastava

References 1. Karl, H., Willig, A.: Protocols and Architectures for Wireless Sensor Networks. Wiley Ltd., London (2005) 2. Pantazis, N.A., Nikolidakis, S.A., Vergados, D.D.: Energy-efficient routing protocols in wireless sensor networks: a Survey. IEEE Commun. Surv. Tutorials 15(2), 551–591 (2013) 3. Guo, W., Zhang, W.: A survey on intelligent routing protocols in wireless sensor networks. J. Netw. Comput. Appl. 38, 185–201 (2014) 4. Zin, SMd, Anuar, N.B., Kiah, L.M., Pathan, A.K.: Routing protocol design for secure WSN: review and open research issues. J. Netw. Comput. Appl. 41, 517–530 (2014) 5. Akyildiz, I., Su, W., Sankarasubramaniam, Y., Cayirci, E.: Wireless sensor networks: a survey. Comput. Netw. 38(4), 393–422 (2002) 6. Akyildiz, I.F., Su, W., Sankarasubramaniam, Y., Cayirci, E.: A survey on sensor networks. IEEE Commun. Mag. 40(8), 102–114 (2002) 7. Intanagonwiwat, C., Govindan, R., Estrin, D.: Directed diffusion: a scalable and robust communication paradigm for sensor networks. In: Proceedings of the 6th Annual International Conference on Mobile Computing and Networking. Boston, Massachusetts, pp. 56–67 (2000) 8. Sadagopan, N., Krishnamachari, B., Helmy, A.: Active query forwarding in sensor networks (ACQUIRE). Ad Hoc Netw. 3(1), 91–113 (2005) 9. Kulik, J., Heinzelman, W.R., Balakrishnan, H.: Adaptive protocols for information dissemination in wireless sensor networks. In: Proceedings of the 5th Annual ACM/IEEE International Conference on Mobile Computing and Networking, Massachusetts Institute of Technology, USA, Washington, pp. 174–185 (1999) 10. Heinzelman, W., Chandrakasan, A., Balakrishnan, H.: Energy-efficient communication protocol for wireless microsensor networks. In: Proceedings of the 33rd Hawaii International Conference on System Sciences, HI, USA, vol. 8, p. 110 (2000) 11. Heinzelman, W., Chandrakasan, A., Balakrishnan, H.: An application-specific protocol architecture for wireless microsensor networks. IEEE Trans. Wirel. Commun. 1(4), 60–70 (2002) 12. Lindsey, S., Raghavendra, C.: PEGASIS: power-efficient gathering in sensor information systems. In: Proceedings of the IEEE Aerospace Conference, USA, Montana, vol. 3, pp. 1125–1130 (2002) 13. Manjeshwar, A., Agrawal, D.: TEEN: a routing protocol for enhanced efficiency in wireless sensor networks. In: Proceedings of the 15th International Parallel and Distributed Processing Symposium (IPDPS’01) Workshops, USA, California, pp. 2009–2015 (2001) 14. Manjeshwar, A., Agrawal, D.: APTEEN: a hybrid protocol for efficient routing and comprehensive information retrieval in wireless sensor networks. In: Proceedings of the International Parallel and Distributed Processing Symposium, Florida, pp. 195–202 (2002) 15. Basagni, S., Chlamtac, I., Syrotiuk, V., Woodward, B.: A distance routing effect algorithm for mobility (DREAM). In: Proceedings of the 4th Annual ACM/IEEE International Conference on Mobile Computing and Networking, USA, Texas, pp. 67–84 (1998) 16. Yu, Y., Govindan, R., Estrin, D.: Geographical and Energy Aware Routing: A Recursive Data Dissemination Protocol for Wireless Sensor Networks. UCLA Computer Science Department Technical Report, pp. 1–11 (2001) 17. Blum, B., He, T., Son, S., Stankovic, J.: IGF: A State-Free Robust Communication Protocol for Wireless Sensor Networks. Technical Report CS-2003-11, Department of Computer Science, University of Virginia, USA (2003) 18. Chen, D., Varshney, P.: On-demand geographic forwarding for data delivery in wireless sensor networks. Comput. Commun. 30(14–15), 29542967 (2007) 19. Chen, M., Leung, V., Mao, S., Xiao, Y., Chlamtac, I.: Hybrid geographical routing for flexible energy-delay trade-offs. IEEE Trans. Veh. Technol. 58(9), 4976–4988 (2009) 20. Sohrabi, K., Gao, J., Ailawadhi, V., Pottie, G.: Protocols for self-organization of a wireless sensor network. IEEE Pers. Commun. 7(5), 16–27 (1999)

38 An Energy Efficient Clustered Routing Protocols …

595

21. He, T., Stankovic, J., Lu, C., Abdelzaher, T.: SPEED: A stateless protocol for real-time communication in sensor networks, In Proc. 23rd International Conference on Distributed Computing Systems, Torodo, pp. 46–55 (2003) 22. Felemban, E., Lee, C., Ekici, E.: MMSPEED: Multipath multi-SPEED protocol for QoS guarantee of reliability and timeliness in wireless sensor networks. IEEE Trans. Mobile Comput. 5(6), 738–754 (2006) 23. Chen, M., Guizani, M., Jo, M.: Mobile multimedia sensor networks: architecture and routing. In: Proceedings of the Mobility Management in the Networks of the Future World, Shanghai, pp. 409–412 (2011) 24. Akyildiz, I.F., Vuran, M.C.: Wireless Sensor Networks. Wiley Ltd, London (2010) 25. Hu, F., Cao, X.: Wireless Sensor Networks: Principles and Practice. CRC Press, Boca Raton (2010) 26. Xiangning, F., Yulin, S.: Improvement on LEACH protocol of wireless sensor network. In: International Conference on Sensor Technologies and Applications (Sensor Comm 2007), Valencia, Spain, vol. 4, pp. 260, 14–20 Oct 2007 27. Lindsey, S., Raghavendra, C.S., Sivalingam, K.M.: Data gathering algorithms in sensor networks using energy metrics. IEEE Trans. Parallel Distrib. Syst. 13(9), 924–935 (2002) 28. Younis, O., Fahmy, S.: HEED: a hybrid, energy-efficient, distributed clustering approach for ad hoc sensor networks. IEEE Trans. Mob. Comput. 3(4), 366–379 (2004) 29. Chand, S., Singh, S., Kumar, B.: Heterogeneous HEED protocol for wireless sensor networks. Wirel. Pers. Commun. 77, 2117–2139 (2014) 30. Smaragdakis, G., Matta, I., Bestavros, A.: SEP: a stable election protocol for clustered heterogeneous wireless sensor networks. In: International Workshop on SANPA (2004) 31. Aderohunmu, F.A., Deng, J.D.: An Enhanced Stable Election Protocol (E-SEP) for Clustered Heterogeneous WSN. Department of Information Science, University of Otago, Dunedin 9054, New Zealand 32. Kashaf, A., Javaid, N., Khan, Z.A., Khan, I.A.: TSEP: threshold-sensitive stable election protocol for WSNs. In: 10th International Conference on Frontiers of Information Technology (FIT), vol. 164, no. 168, pp. 17–19 (2012) 33. Kumar, S., Verma, S.K., Kumar, A.: Enhanced threshold sensitive stable election protocol for heterogeneous wireless sensor network. Wirel. Pers. Commun. 85, 2643–2656 (2015) 34. Qing, L., Zhu, Q., Wang, M.: Design of a distributed energy-efficient clustering algorithm for heterogeneous wireless sensor network. Comput. Commun. 29, 2230–2237 (2006) 35. Rathee, A., Kashyap, I., Choudhary, K.: Developed distributed energy-efficient clustering (DDEEC) algorithm based on fuzzy logic approach for optimizing energy management in heterogeneous WSNs. Int. J. Comput. Appl. 115(17), 14–19 (2015) 36. Javaid, N., Qureshi, T.N., Khan, A.H., Iqbal, A., Akhtar, E., Ishfaq, M.: EDDEEC: enhanced developed distributed energy-efficient clustering for heterogeneous wireless sensor networks. Procedia Comput. Sci. 19, 914–919 (2013) 37. Kang, S.H., Nguyen, T.: Distance based thresholds for cluster head selection in wireless sensor networks. IEEE Commun. Lett. 16(9), 1396–1399 (2012) 38. Mehdi, G., Nia, F.K., Mehdi, H., Yones, V.: The novel energy adaptive protocol for heterogeneous wireless sensor networks. In: 3rd IEEE International Conference on Computer Science and Information Technology (ICCSIT), pp. 178–182 (2010) 39. Loscri, V., Morabito, G., Marano, S.: A two-levels hierarchy for low-energy adaptive clustering hierarchy (TL-LEACH). In: 2nd IEEE Semiannual Vehicular Technology Conference, vol. 25, no. 28, pp. 1809–1813 (2005) 40. Ye, M., Li, C., Chen, G., Wu, J.: EECS: an energy efficient clustering scheme in wireless sensor networks. In: 24th IEEE International Performance, Computing, and Communications Conference (IPCCC), USA, vol. 7, no. 9, pp. 535–540 (2005) 41. Li, C., Ye, M., Chen, G., Wu, J.: An energy-efficient unequal clustering mechanism for wireless sensor networks. In: 2nd IEEE International Conference on Mobile Ad-hoc and Sensor Systems Conference (MASS), vol. 7, no. 10, pp. 596–604 (2005)

596

N. Mittal and R. Srivastava

42. Kumar, D.: Performance analysis of energy efficient clustering protocols for maximising lifetime of wireless sensor networks. IET Wirel. Sens. Syst. 4(1), 9–16 (2014) 43. Tarhani, M., Kavian, Y.S., Siavoshi, S.: SEECH: Scalable energy efficient clustering hierarchy protocol in wireless sensor networks. IEEE Sens. J. 14(11), 3944–3954 (2014) 44. Mittal, N., Singh, U.: Distance-based residual energy-efficient stable election protocol for WSNs. Arab. J. Sci. Eng. 40(6), 1637–1646 (2015). https://doi.org/10.1007/s13369-0151641-x 45. Mittal, N., Singh, U., Sohi, B.S.: A stable energy efficient clustering protocol for wireless sensor networks. Wirel. Netw. 23(6), 1809–1821 (2017). https://doi.org/10.1007/s11276-0161255-6 46. Hussain, S., Matin, A.W., Islam, O.: Genetic algorithm for hierarchical wireless sensor networks. J. Netw. 2, 87–97 (2007) 47. Attea, B.A., Khalil, E.A.: A new evolutionary based routing protocol for clustered heterogeneous wireless sensor networks. Appl. Soft Comput. 12, 1950–1957 (2012) 48. Khalil, E.A., Attea, B.A.: Energy-aware evolutionary routing protocol for dynamic clustering of wireless sensor networks. Swarm Evol. Comput. 1(4), 195–203 (2011) 49. Khalil, E.A., Attea, B.A.: Stable-aware evolutionary routing protocol for wireless sensor networks. Wirel. Pers. Commun. 69(4), 1799–1817 (2013) 50. Mittal, N., Singh, U., Sohi, B.S.: A novel energy efficient stable clustering approach for wireless sensor networks. Wirel. Pers. Commun. 95, 2947–2971 (2017) 51. Mittal, N., Singh, U., Sohi, B.S.: Harmony search algorithm based threshold-sensitive energyefficient clustering protocols For WSNs”. Ad Hoc Sens. Wirel. Netw. 36, 149–174 (2017) 52. Mittal, N., Singh, U., Sohi, B.S.: A boolean spider monkey optimization based energy efficient clustering approach For WSNs. Wirel. Netw. 24(6), 2093–2109 (2018) 53. Mittal, N., Singh, U., Sohi, B.S.: An energy aware cluster-based stable protocol for wireless sensor networks. In: Neural Computing and Applications (NCAA), pp. 1–18 (2018) 54. Kuila, P., Gupta, S.K., Jana, P.K.: A novel evolutionary approach for load balanced clustering problem for wireless sensor networks. Swarm Evol. Comput. 12, 48–56 (2013) 55. Kuila, P., Jana, P.K.: A novel differential evolution based clustering algorithm for wireless sensor networks. Appl. Soft Comput. 25, 414–425 (2014) 56. Shokouhifar, M., Jalali, A.: A new evolutionary based application specific routing protocol for clustered wireless sensor networks. Int. J. Electron. Commun. 69, 432–441 (2015) 57. Hoang, D.C., Yadav, P., Kumar, R., Panda, S.K.: Real-time implementation of a harmony search algorithm-based clustering protocol for energy-efficient wireless sensor networks. IEEE Trans. Ind. Inf. 10(1), 774–783 (2014) 58. Kumar, N., Tyagi, S., Deng, D.: LA-EEHSC: Learning automata-based energy efficient heterogeneous selective clustering for wireless sensor networks. J. Netw. Comput. Appl. 46, 264–279 (2014) 59. Sert, S.A., Bagci, H., Yazici, A.: MOFCA: multi-objective fuzzy clustering algorithm for wireless sensor networks. Appl. Soft Comput. 30, 151–165 (2015) 60. Tomar, G.S., Sharma, T., Kumar, B.: Fuzzy based ant colony optimization approach for wireless sensor network. Wirel. Pers. Commun. 84, 361–375 (2015) 61. Tamandani, Y.K., Bokhari, M.U.: SEPFL routing protocol based on fuzzy logic control to extend the lifetime and throughput of the wireless sensor network. Wirel. Netw. 22(2), 647–653 (2015) 62. Zahedi, Z.M., Akbari, R., Shokouhifar, M., Safaei, F., Jalali, A.: Swarm intelligence based fuzzy routing protocol for clustered wireless sensor networks. Expert Syst. Appl. 55, 313–328 (2016)

Chapter 39

Analysis of Different Detection and Mitigation Algorithm of DDoS Attack in Software-Defined Internet of Things Framework: A Review Naveen Kumar, Nitin Mittal, Palak Thakur and Rajshree Srivastava Abstract A decentralized type of network which senses the information from surrounding regions and then forwards it to the base station is known as Internet of Things (IoT). With the increasing demand of IoT applications, security is the major issue. Recently most popular attack in the world of internet is the distributed denial of service (DDoS) attack and for the safety management of IoT devices SDx paradigm can be used. In this paper, we will present the overview and applications of IoT and SDx paradigm, brief description of distributed denial of service attack as well as the different detection and mitigation algorithms. The study shows that security has extremely important for the reliable communication. Keywords Software-defined Internet of Things (SD-IoT) · Detection of attack algorithm · Mitigation of attack algorithm · Denial of service attack · Distributed denial of service attack

39.1 Introduction Internet of things (IoT) refers to the collection of things, objects which are being used in our daily life they can readable, recognizable, locatable and addressable. For example, GPRS services are used for the navigation purposes and it works only if the person has strong internet connection. So, these things are controlled by the N. Kumar · N. Mittal · P. Thakur Electronics and Communication Engineering, Chandigarh University, Punjab, India e-mail: [email protected] N. Mittal e-mail: [email protected] P. Thakur e-mail: [email protected] R. Srivastava (B) Computer Science Engineering, DIT University, Dehradun, India e-mail: [email protected] © Springer Nature Switzerland AG 2020 V. E. Balas et al. (eds.), Recent Trends and Advances in Artificial Intelligence and Internet of Things, Intelligent Systems Reference Library 172, https://doi.org/10.1007/978-3-030-32644-9_39

597

598

N. Kumar et al.

internet corresponding to their communication means. It is the new and advanced discovery in the field of technology. They have the power to access all the information which is collected from all the other things and the information are integrated into smaller pieces. All the collected information is stored on the databases in the cloud computing and have unlimited addressing power [1]. Network of physical devices connected with software and they can interchange information with each other. 5G technology includes concept of IoT. A system which enables the interlinking of objects present in the physical world with the help of using either wired or wireless modes is known as IoT. Building a new future in which each smart object present around is linked to a network that senses, calculates, shares and processes the information from various scenarios, is the prior objective of developing this technology. Thus, there has been an increase in demand of this technology which has resulted in introducing some advanced technologies that have been applied today in almost all the fields existing now [2]. For providing a connection amongst these smart devices and global network, researchers have now been developing new designs and tools as per the advanced technologies. Before storing this data, there is a need to analyze it on the cloud. The concept of smart home is the major key factor that has made increase in demand of IoT. Several services such as home monitoring, ensuring safety and providing central control to the home appliances, are offered by smart home. Connecting the home appliances as well as networks along with the usage of hew standard communication based protocols is the major idea on which smart homes work [3]. This technology also uses the smart sensors as well as cameras. Further, smart agriculture is another popular application of IoT in which the smart sensors and RFIDs are used to modify the traditional decision makings and provide few enhancements for easing human work. In order to improve the productivity and quality of crops, the farmers can now make in time decisions such that better quality and productivity can be achieved. Supply chain management was the first application in which IoT had been used ever. Each process and transaction being made can be seen with real time insight by using IoT within the supply chain system [4–6] (Fig. 39.1).

39.2 Architecture and Application of IoT IoT architecture mainly consists of five layers but in this architect we have discussed mainly three types of layers which are quite useful for data processing. Applications of IoT connect the all world with the internet connectivity.

39 Analysis of Different Detection and Mitigation Algorithm … Fig. 39.1 Connectivity of IoT

599

Smart Home

Market

Industry

Healthcare

IoT

Transportation

Vehicle

Agriculture School

39.2.1 IoT Architecture Several components that collect together to form an IoT scenario is heterogeneous is nature and thus require effective management. This can be done using IoT middleware. Various layers that generate an IoT middleware are presented in general view shown in Fig. 39.2. The application layer present in the IoT architecture and the upper layer that is a component have direct interaction amongst each other. The request messages are received by the application layer and the middleware sends the responses that are relevant to the services. There is an interaction amongst the lower layer and physical layer. Amongst the physical devices, the binary information and control commands Fig. 39.2 IoT architecture

Physical Layer

Middleware IoT Upper Layer

Lower Layer

Application Layer

600

N. Kumar et al.

are exchanged here [7]. Providing interoperability and adaptation amongst heterogeneous devices, scalability, handling huge amount of data, providing context awareness, discovering devices and ensuring security of IoT scenarios is ensured with the help of including abstractions and relevant services. The ability of hiding details of various technologies is of highest importance amongst all the middleware features. Thus, the control logic of IoT applications is developed by the programmers through this. Thus, for every kind of device or data format, one does not need to write a separate code. In order to simplify the development of new services and integrate the legacy of technologies into new approaches, the IoT middleware packages are gaining popularity. Here, the need of proficiency within the rich set of technologies being used by lower layers is not required by the programmer which eases its work to much extent. The module through which the data volume that is very huge is handled is the common IoT middleware functional component. Thus, within this context, identifying and transmitting the data is done by generating new techniques. In order to manage the data generally, the database management systems (DBMS) are used by necessary components. Further, huge volume of data is processed by particular software technologies [8, 9].

39.2.2 Applications of IoT IoT applications are used everywhere such as 5G, Seven sense, Smart cities, Agriculture, Healthcare, Smart building, Industrial control etc. [10, 11]. Smart healthcare: Monitoring of the human diseases with the help of special device and it is connected with internet. Connectivity of device with internet shows that extended version of the technology. Seven sense technology: In the seven sense technology, IoT include with robotics and six sense technology. Seven sense is the future technology which comes with the IoT. Figure 39.3 shows that the different IoT applications. Smart Agriculture: In the field of agriculture, sensors can be deployed for testing of soil. If we increase the production of pulses, vegetables then control the condition of micro climate and this can be done with the help of Green house. Smart city: Smart city includes the features of IoT. In smart city, all the devices connected with internet and data will be used smartly. Different devices send data to base station and after the processing will be held smartly with the help of internet connected devices.

39 Analysis of Different Detection and Mitigation Algorithm …

601

Fig. 39.3 Application of IoT

Home automation is one of the example in smart city.

39.2.3 Issues and Challenges in IoT IoT has various advantages and attractive features which make it beneficial and useful in almost every field. But it has certain challenges and issues which works as a disadvantages. Some of them are mentioned below: Privacy and Security: IoT is one of the advanced and developed technologies and it is very easily employed in almost every large scale, partially mission-critical systems. There are certain challenges IoT facing like inadequate trust and quality of information, it prevents [12] the usage of IoT in various applications. Also, the data stored is not securely exchange among the users. Cost versus Usability: Very strong internet connection is required for the interaction between the users and to share data in the entire network. IoT is very expensive technology, this creates problem to the users as it is unaffordable to them. Therefore, it is one of the issues of concern. Interoperability: It is the property of the IoT or any other software which exchanges the information and makes use of that information. The connected system should be able to understand the language of the protocol and this is one the basic requirement. With the availability of the huge amount of data and devices, the use of basic interaction between these various entities has become very difficult and important at the same time. Hence, it is very important to solve this issue so that the user can easily exchange and store the data on the IoT.

602

N. Kumar et al.

Data Management: Managing large amount of data is very difficult. As the data is stored and exchanged is very large on the internet on the daily basis. It is very impossible to recognize and identify the relevant and irrelevant data. Therefore, it is also very concerning issue.

39.3 Denial of Service and Distributed Denial of Service Attack Denial of service attack is general type of attack in which fake request sent by one user again and again to server. But in distributed denial of service attack, different fake users hit the server but controlled by one user.

39.3.1 Denial of Service Attack Denials of Service (DoS) attacks are very large in numbers and very devastation. There are various DoS attacks and most of them will directly affect the communication between the networks. It involves either the use of one computer or more than one computer and is commonly known as zombies. Initially, it is known as DoS attack and later it is referred to as DistributedDoS (DDoS) attacks. Large number of measures and techniques were proposed to stop this attack. it is the type of attack which is launched to create networks and systems resources unavailable for the usage of users so that nobody can access the information and all the data is hidden. This will [13] create such a situation that the organization have to face heavy loss. Web servers, default gateways and personal computers are the main target of such types of attacks. The hackers have three objectives that is they explore in such a way that they can easily get the secret data/ information, then they can get authorization to all the confidential data and able to change it. The first two steps are not to perform because it is quite difficult to access the confidential files. Hence, they will try to enter such files and folders for which they don’t need any authorization. Figure 39.4 shows that the denial of service attack occurs in the server. Effect of the attack directly on the access of website means that accessing speed will be low due to this attack. DDoS attack is advancement of denial of service attack. In the DDoS attack, identity of the malicious node can be hide easily so the attack popularity increases day by day [14].

39 Analysis of Different Detection and Mitigation Algorithm …

603

Fig. 39.4 Denial of service attack

39.3.2 Distributed Denial of Service Attack Attackers either make use of single computer or multiple computers to launch these attacks. The usage of multiple computers to perform the attack is known as DDoS attack. The machines which are affected with such attacks are called zombies whereas the machines which control the machines are called masters. The relationship between the master-zombie is quite similar to the client-server relationship. It is very difficult to identify the DDoS attacks because the attackers can be present anywhere. Therefore, they cannot be distinguished from the legitimate traffic [15, 16]. Figure 39.5 shows that the distributed denial of service attack as shown. Fig. 39.5 Distributed denial of service attack

604

N. Kumar et al.

39.3.3 Some Solutions to DoS Attacks DoS attack is very dangerous attack which will exploit all the smooth working of the network. It will send the entire packet node as it appears to be the genuine node. In order to remove and this attack from the network some serious steps have been taken and some solutions are discussed below: (1) The user need to make sure that the network has a firewall that aggressively keeps everything out except the legal matters. (2) Implement router filter. This will reduce the exposure to certain denial of service attacks. Moreover, it will helps in preventing users [16] on the network from the effective launching of various denial of service attack. (3) Install patches to guard against TCP/IP attacks. This will reduce the revelation to these attacks but the risk is not removed completely. (4) Disable any unused or unneeded network services. This will fix a limit on the interference of an intruder who takes advantage of those services which executes denial-of-service attacks.

39.4 Detection and Mitigation Techniques of DDoS Attack Table 39.1 shows that the DDoS attack detection and mitigation techniques which are quite popular in the recent year. The latest approach in the software defined Table 39.1 Detection and mitigation techniques Author’s names

Year

Description

Outcomes

Bekara [18]

2014

Presented that even though there is availability of access technology

To ensure that end-customer participates within energy consumption equilibrium

Mukherjee [19]

2015

A model is proposed for local IoT deployment that includes multiple sensor and data sub-networks

For the IoT sensing applications, the best suiting techniques that provide highest energy-efficiency

Asplund and Nadjm-Tehrani [20]

2016

Proposed a novel approach to recognize information security requirements

A review of the IoT security was done. in relation to the perceptions

Ge et al. [21]

2017

A novel model is proposed to provide security in IoT devices

The potential attack paths are identified and the effects of attacks are mitigated (continued)

39 Analysis of Different Detection and Mitigation Algorithm …

605

Table 39.1 (continued) Author’s names

Year

Description

Outcomes

Zhang et al. [22]

2017

Proposed a novel approach through which several threats can be handled

In comparison to traditional approaches, the performance of SLOT framework is better

Yin et al. [17]

2018

A novel framework based on SDx pattern is proposed

The performance of proposed algorithm is shown better

Yan et al. [23]

2018

A multi-level DDoS mitigation approach for IoT devices is proposed

It is seen through the simulation results that the DDoS attack issue IoT can be solved

Sicari et al. [24]

2018

The REATO approach is pro-posed to identify and prevent DoS attack within IoT

In the presence of malicious nodes, the evaluation of the proposed model was done

Diro et al. [25]

2018

A novel approach is proposed that is based on distributed deep learning

It is seen that due to the sharing of parameters, the cyber attacks can be detected

Liu et al. [26]

2018

A Middlebox-Guard (MG) is proposed that is based on SDN

The dataflow of SDN-based IoT systems is managed

Mavropoulos et al. [27]

2018

A class-based notation is pro-posed along with refining of the modeling language of apparatus framework.

A smart public transport system is available that included various hardware and software components

Sani et al. [28]

2018

A cyber security approach is proposed using which the EI can be secured and the energy management can be supported

It is seen through the experimental results that the IoT-based EI is provided with efficient level of security and privacy using the proposed approach

Zhang et al. [29]

2018

Presented a study related to physical layer security of NOMA

It is possible to prevent the information that is being communicated

Chu [30]

2018

Proposed case studies of three various commercially-provided products

There is a disconnection amongst the IoT to developers and security

Bany Salameh et al. [15]

2018

Proposed a novel technique to reduce the invalidity ratio of CR packet transmissions

It is seen that the network performance is improved by ensuring security

606

N. Kumar et al.

framework for the detection and mitigation of DDoS attack is cosine similarity algorithm [17]. In the proposed framework, controller pool increases the hardware design which increases the time delay for the detection and mitigation of DDoS attack.

39.5 Conclusion IoT is a decentralized type of network in which sensor devices can sense information and pass sensed information to the base station. The sensor devices are configured with sensing technologies and it will first transmit the information to its hub which later passes it to base station. Due to dynamic nature of the network malicious nodes enter the network which triggers various type of active and passive attacks. The DDoS attack is the active type of attack in which intermediate nodes are triggered by malicious nodes. In this paper we have presented the scenario of proposed work which is based on the cosine similarity. It is analyzed that extra hardware and software is required for the detection of malicious nodes. In future work we can on hardware reduction with Watch Dog Approach or Trust Based Model.

References 1. Gluhak, A., Krco, S., Nati, M., Pfisterer, D., Mitton, N., Razafindralambo, T.: A survey on facilities for experimental internet of things research. IEEE Commun. Mag. 49(11), 58–67 (2011) 2. Fevgas, A., Tsompanopoulou, P., Bozanis, P.: iMuse mobile tour: a personalized multimedia museum guide opens to groups. In: Proceedings of 2011 IEEE Symposium on Computers and Communications (ISCC) (2011) 3. Kanda, A., Arai, M., Suzuki, R., Kobayashi, Y., Kuno, Y.: Recognizing groups of visitors for a robot museum guide tour. In: Proceedings of 2014 7th International Conference on Human System Interactions (HSI) (2014) 4. Yu, N., Han, Q.: Context-aware community: integrating contexts with contacts for proximitybased mobile social networking. In: Proceedings of 2013 IEEE International Conference on Distributed Computing in Sensor Systems (2013) 5. Atzori, L., Iera, A., Morabito, G.: The Internet of Things: a survey. Comput. Netw. 54(15), 2787–2805 (2010) 6. Sethi, P., Sarangi, S.: Internet of Things: architectures, protocols, and applications. J. Electr. Comput. Eng. 2017, 1–25 (2017) 7. Guo, B., Zhang, D., Wang, Z., Yu, Z., Zhou, X.: Opportunistic IoT: exploring the harmonious interaction between human and the Internet of Things. J. Netwo. Comput. Appl. 36(6), 1531–1539 (2013) 8. Aazam, M., Khan, I., Alsaffar, A., Huh, E.: Cloud of things: integrating Internet of Things and cloud computing and the issues involved. In: Proceedings of 2014 11th International Bhurban Conference on Applied Sciences & Technology (IBCAST) Islamabad, Pakistan, 14th–18th January, 2014 (2014) 9. Zhang, D., Yang, L., Huang, H.: Searching in Internet of Things: vision and challenges. In: Proceedings of 2011 IEEE Ninth International Symposium on Parallel and Distributed Processing with Applications (2011)

39 Analysis of Different Detection and Mitigation Algorithm …

607

10. Miorandi, D., Sicari, S., De Pellegrini, F., Chlamtac, I.: Internet of things: vision, applications and research challenges. Ad Hoc Netw. 10(7), 1497–1516 (2012) 11. Movassaghi, S., Abolhasan, M., Lipman, J., Smith, D., Jamalipour, A.: Wireless body area networks: a survey. IEEE Communi. Surv. Tutorials 16(3), 1658–1686 (2014) 12. Kuo, Y., Li, C., Jhang, J., Lin, S.: Design of a wireless sensor network-based IoT platform for wide area and heterogeneous applications. IEEE Sens. J. 18(12), 5187–5197 (2018) 13. Vermesan, O., Friess, P.: Internet of Things—Global Technological and Societal Trends. River Publishers, Aalborg (2011) 14. Kargl, F., Maier, J., Weber, M.: Protecting web servers from distributed denial of service attacks. In: Proceedings of the Tenth International Conference on World Wide Web—WWW’01 (2001) 15. Bany Salameh, H., Almajali, S., Ayyash, M., Elgala, H.: Spectrum Assignment in cognitive radio networks for Internet-of-Things delay-sensitive applications under jamming attacks. IEEE Internet of Things J. 5(3), 1904–1913 (2018) 16. Gresty, D., Shi, Q., Merabti, M.: Requirements for a general framework for response to distributed denial-of-service. In: Proceeding of Seventeenth Annual Computer Security Applications Conference (2002) 17. Yin, D., Zhang, L., Yang, K.: A DDoS attack detection and mitigation with software-defined Internet of Things framework. IEEE Access 6, 24694–24705 (2018) 18. Bekara, C.: Security issues and challenges for the IoT-based smart grid. Procedia Comput. Sci. 34, 532–537 (2014) 19. Mukherjee, A.: Physical-layer security in the Internet of Things: sensing and communication confidentiality under resource constraints. Proc. IEEE 103(10), 1747–1761 (2015) 20. Asplund, M., Nadjm-Tehrani, S.: Attitudes and perceptions of IoT security in critical societal services. IEEE Access 4, 2130–2138 (2016) 21. Ge, M., Hong, J., Guttmann, W., Kim, D.: A framework for automating security analysis of the Internet of Things. J. Netw. Comput. Appl. 83, 12–27 (2017) 22. Zhang, P., Nagarajan, S., Nevat, I.: Secure location of things (SLOT): mitigating localization spoofing attacks in the Internet of Things. IEEE Internet of Things J. 4(6), 2199–2206 (2017) 23. Yan, Q., Huang, W., Luo, X., Gong, Q., Yu, F.: A multi-level DDoS mitigation framework for the industrial Internet of Things. IEEE Commun. Mag. 56(2), 30–36 (2018) 24. Sicari, S., Rizzardi, A., Miorandi, D., Coen-Porisini, A.: REATO: REActing TO denial of service attacks in the Internet of Things. Comput. Netw. 137, 37–48 (2018) 25. Diro, A., Chilamkurti, N., Nam, Y.: Analysis of lightweight encryption scheme for fog-to-things communication. IEEE Access 6, 26820–26830 (2018) 26. Liu, Y., Kuang, Y., Xiao, Y., Xu, G.: SDN-based data transfer security for Internet of Things. IEEE Internet of Things J. 5(1), 257–268 (2018) 27. Mavropoulos, O., Mouratidis, H., Fish, A., Panaousis, E.: Apparatus: a framework for security analysis in Internet of Things systems. Ad Hoc Netw. (2018) 28. Sani, A., Yuan, D., Jin, J., Gao, L., Yu, S., Dong, Z.: Cyber security framework for Internet of Things-based energy internet. Future Gener. Comput. Syst. 93, 849–859 (2019) 29. Zhang, S., Xu, X., Wang, H., Peng, J., Zhang, D., Huang, K.: Enhancing the physical layer security of uplink non-orthogonal multiple access in cellular Internet of Things. IEEE Access 6, 58405–58417 (2018) 30. Chu G.: Security and privacy analyses of Internet of Things children’s toys. IEEE Internet of Things J. 1–1 (2018)