Antennas : Parameters, Models and Applications [1 ed.] 9781608762859, 9781606924631

Citation preview

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ANTENNAS: PARAMETERS, MODELS AND APPLICATIONS

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

No part of this digital document may be reproduced, stored in a retrieval system or transmitted in any form or by any means. The publisher has taken reasonable care in the preparation of this digital document, but makes no expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of information contained herein. This digital document is sold with the clear understanding that the publisher is not engaged in rendering legal, medical or any other professional services.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ANTENNAS: PARAMETERS, MODELS AND APPLICATIONS

ALBERT I. FERRERO

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

EDITOR

Nova Science Publishers, Inc. New York

Copyright © 2009 by Nova Science Publishers, Inc. All rights reserved. No part of this book may be reproduced, stored in a retrieval system or transmitted in any form or by any means: electronic, electrostatic, magnetic, tape, mechanical photocopying, recording or otherwise without the written permission of the Publisher. For permission to use material from this book please contact us: Telephone 631-231-7269; Fax 631-231-8175 Web Site: http://www.novapublishers.com NOTICE TO THE READER The Publisher has taken reasonable care in the preparation of this book, but makes no expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of information contained in this book. The Publisher shall not be liable for any special, consequential, or exemplary damages resulting, in whole or in part, from the readers’ use of, or reliance upon, this material. Any parts of this book based on government reports are so indicated and copyright is claimed for those parts to the extent applicable to compilations of such works. Independent verification should be sought for any data, advice or recommendations contained in this book. In addition, no responsibility is assumed by the publisher for any injury and/or damage to persons or property arising from any methods, products, instructions, ideas or otherwise contained in this publication.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

This publication is designed to provide accurate and authoritative information with regard to the subject matter covered herein. It is sold with the clear understanding that the Publisher is not engaged in rendering legal or any other professional services. If legal or any other expert assistance is required, the services of a competent person should be sought. FROM A DECLARATION OF PARTICIPANTS JOINTLY ADOPTED BY A COMMITTEE OF THE AMERICAN BAR ASSOCIATION AND A COMMITTEE OF PUBLISHERS. LIBRARY OF CONGRESS CATALOGING-IN-PUBLICATION DATA Antennas : parameters, models and applications / Albert I. Ferrero (editor). p. cm. ISBN 978-1-60876-285-9 (E-Book) 1. Antennas (Electronics) I. Ferrero, Albert I. TK7871.6.A535 621.382'4--dc22

Published by Nova Science Publishers, Inc.  New York

2008 2008041744

CONTENTS Preface Chapter 1

Antennas for Wireless Biomedical Devices C.P. Figueiredo and P.M. Mendes

Chapter 2

Antenna-Lanthanide Complexes: A Growing Technology-Driven Research Silvio Quici, Lidia Armelao, Francesco Barigelletti, Marco Cavazzini, Gregorio Bottaro and Gianluca Accorsi

Chapter 3

Chapter 4

Chapter 5

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

vii

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless Communications D. Zhou, R.A. Abd-Alhameed, C.H. See and P.S. Excell

75

121

Multiple Antenna Coding & Signal Processing: Space-Time Coding For Wireless Communications S. Manioudakis and A.M. Maras

139

Resistive Rectangular Patch Antenna with Uniaxial Substrate Amel Boufrioua

Chapter 7

Antennal Sensillar Morphology, Structure and Function in Parasitic Wasps Yan Gao, Li-Zhi Luo and Abner Hammond

Chapter 9

41

Application of SMA Wire Actuators in Flatness Control of Membrane SAR Antennae Fujun Peng, Yan-Ru Hu , Xin-Xiang Jiang, and Alfred Ng

Chapter 6

Chapter 8

1

Measurement of Parameters of the Acoustic Antenna Arising at Braking and Stopping of the Proton Beam in Water and Research of Characteristics of Created Field V.B. Bychkov, V.S. Demidov and E.V. Demidova Optimization of Non-Uniform Linear and Circular Phased Arrays Using Genetic Algorithms to Provide Maximum Interference Reduction in Wireless Communication Systems Marco A. Panduro

163

191

213

243

vi Chapter 10

Time Domain Design of Radiating Structures Zbynek Raida

Chapter 11

Maximum Penetration Depth Study in a Ground Penetrating Radar Survey Giovanni Leucci

281

Synthesis of Aperiodic Linear Phased Antenna Arrays Using the Differential Evolution Method David H. Covarrubias and Marco A. Panduro

295

Chapter 12

Index

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Contents 263

307

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

PREFACE An antenna is a transducer designed to transmit or receive electromagnetic waves. In other words, antennas convert electromagnetic waves into electrical currents and vice versa. Antennas are used in systems such as radio and television broadcasting, point-to-point radio communication, wireless LAN, radar, and space exploration. Antennas usually work in air or outer space, but can also be operated under water or even through soil and rock at certain frequencies for short distances. Physically, an antenna is an arrangement of conductors that generate a radiating electromagnetic field in response to an applied alternating voltage and the associated alternating electric current, or can be placed in an electromagnetic field so that the field will induce an alternating current in the antenna and a voltage between its terminals. Some antenna devices (parabolic antenna, Horn Antenna) just adapt the free space to another type of antenna. This important book presents new research in this dynamic field. Chapter 1 - An e-healthcare system requires a group of sensors, which are placed on a patient in order to sense a set of physiological parameters. Several monitoring networks have been used during the last decades consisting of conventional wired equipment, hence not allowing the patients to move freely around. However, recent advances in wireless sensors technology are changing this scenario, where mobile and permanent monitoring of patients may be accomplished, even during theirnormal daily activities. Wireless biomedical devices for sensing and recording, ranging from neural prosthesis to video-capsule endoscopy (VCE) systems, are emerging innovative technologies and they are expected to originate significant business activity in the near future. The success of such systems is in part due to the advent of microtechnologies, which made possible the miniaturization of several sensors and actuators, as well their integration with readout and communication electronics. The availability of very small wireless biomedical devices is leading to solutions where it is possible to have a device sending information using a wireless link at each sensing place. In this way, e.g., for EEG recording, instead of having several leads connected to an acquisition system sending the information for a base station, a solution where each EEG electrode sends its own measured information through a wireless link could be feasible. This new approach for biomedical sensing requires also new solutions for antenna design and integration. One is that, once the antennas are required to be integrated inside the electrodes, they are required to be very close to a highly lossy material, the human body. The

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

viii

Albert I. Ferrero

other subject that must be well understood is the effect that many radiating devices will have on the biopotentials that are being measured, which requires the antennas to be designed accordingly. Another characteristic of many wireless systems is the need for node localization. Many methods are available, but one suitable to work very close to the human body would be based on low frequency electromagnetic signals, which suffer less attenuation close to the human body. The interference from radio frequency signal with the biopotentials to be measured will become more important as the frequency of such signals is decreased. Again, a challenge in antenna design must be overcomed. They must not interfere with the biopotential, and they must be small enough to be integrated inside the small electrodes, even when operating at small frequencies. In this chapter the advantages and disadvantages of using independent wireless modules to record each biopotential locally will be firstly discussed. This requires the use of very small modules with integrated antennas. After, the authors will discuss possible solutions for antenna placement very close to the human body will be discussed. Solutions based on antenna design, EBG substrates and metamaterials will be introduced. Finally, the problems associated with node localization very close to the human body will be addressed. The localization methods suitable for this type of wireless network will be presented, together with the challenges associated with the design of antennas to support those methods. Finally, a wireless biomedical device to operate inside the human body will be presented. Chapter 2 - Luminescent and stable lanthanide ions (Ln3+) complexes are of great interest because of their unique photophysical properties especially with respect to the generation and amplification of light. The emission properties of these complexes are notable and cover an exceptionally wide spectral range: near infrared (Yb3+, Nd3+, Er3+), orange (Sm3+), red (Eu3+), yellow (Dy3+), green (Tb3+) and blue (Tm3+). In our technology driven lives lanthanide complexes are ubiquitous. Indeed they find application as active components in many different kind of advanced materials and devices such as: diagnostic tools, sensors, optical fibers, lasers and amplifiers, electroluminescent and magnetic molecular materials among others. Unfortunately, due to the very low absorption coefficients (1 < ε < 10 M-1 cm-1), the emissive states of the metal ions cannot be efficiently populated by direct excitation. A suitable way to overcome this problem is to complex the Ln3+ ion with a ligand containing a highly absorbing chromophore which promotes the metal centred (MC) emission through a sensitization process. In this process the excitation energy absorbed by the chromophore is transferred to the metal core thus efficiently populating the emissive states of the latter and finally the metal centred light emission is obtained (antenna effect). This review will focus on the metal centred emission properties of Ln3+ complexes either in solution or anchored into inorganic matrices. Particular attention will be paid to: (i) the description of the photophysical properties of Ln3+ ions that are relevant for the optimization of the sensitized metal centred light emission; (ii) the coordination properties of Ln3+ ions and the parameters ruling the design of ligands affording complexes with high kinetic and thermodynamic stabilities together with a complete shielding of the coordinated metal ion; (iii) the introduction of suitable reactive groups that allow the covalent insertion of the Ln3+ complex into inorganic and/or organic matrices to afford emissive materials and devices. Some examples of applications of luminescent Ln3+ complexes will be selected from the recent literature and critically reviewed.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Preface

ix

Chapter 3 - The needs for global antenna design tools are always crucial due to the wide range of wireless applications appear nowadays for which the antennas should meet certain required performances. This poses numerous research challenges in this field to find the optimal solution and to overcome the limitations imposed by the design specifications. Thus a significant amount of a research work is needed to develop the scientific tools to state the art of the antenna analysis. Several types of optimizers are combined with these solutions tools such as the well known Genetic Algorithms (GA) that use binary based random search engine subject to various antenna design constraints. The work presented here includes several designed and optimised antennas using GA. The Genetic algorithm driver, written in FORTRAN, was adopted in this work in conjunction with the industry-standard NEC-2 FORTRAN source code, which was used to evaluate the randomly generated antenna samples. Design examples of antennas were successfully demonstrated in this GA method and their results were verified through individual hardware realisation. Four types of antennas were proposed in this study for various applications and they include (1) design of quadrifilar helical antenna in the presence of small handset for mobile satellite wireless communication systems; (2) design of folded loop balanced antenna for mobile handsets; (3) design of microstrip patch antennas with circular polarisation; (4) design of antennas for wide harmonic suppression for active integrated antennas. The needs for global antenna design tools are always crucial due to the wide range of wireless applications appear nowadays for which the antennas should meet certain required performances. This poses numerous research challenges in this field to find the optimal solution and to overcome the limitations imposed by the design specifications. Thus a significant amount of a research work is needed to develop the scientific tools to state the art of the antenna analysis. Several types of optimizers are combined with these solutions tools such as the well known Genetic Algorithms (GA) that use binary based random search engine subject to various antenna design constraints. The work presented here, includes several designed and optimised antennas using GA. The Genetic algorithm driver, written in FORTRAN, was adopted in this work in conjunction with the industry-standard NEC-2 FORTRAN source code, which was used to evaluate the randomly generated antenna samples. Design examples of antennas were successfully demonstrated in this GA method and their results were verified through individual hardware realisation. Four types of antennas were proposed in this study for various applications and they include (1) design of quadrifilar helical antenna in the presence of small handset for mobile satellite wireless communication systems; (2) design of folded loop balanced antenna for mobile handsets; (3) design of microstrip patch antennas with circular polarisation; (4) design of antennas for wide harmonic suppression for active integrated antennas. Chapter 4 - A membrane SAR(Synthetic Aperture Radar) antenna will be subjected to flatness problems during its lifetime in orbit due to the thermal variations in space. A pure passive control method may not be sufficient to maintain the membrane’s flatness since the thermal loads are changing. Therefore, an active control system, which is used to adjust the tensions according to the thermal variations, is developed. SMA wire actuators are selected to exert required tension, due to their unique properties such as high force, long stroke, small size, light weight, and silent operation, etc. First, the selected SMA wire actuators are tested in the ambient air environment as well as in a vacuum chamber to characterize their stability and controllability properties. Tests are conducted under a constant load and variant input currents are used to activate the actuator.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

x

Albert I. Ferrero

The results indicate that, in the ambient air environment, the displacement outputs of the SMA actuator are not stable if the full phase transformation is not achieved. This clearly reflects the influence of the natural air convection on the stability of the actuation. As a contrast, the displacement outputs of the SMA in the vacuum condition are much more stable at all given input currents. This implies that SMA actuators probably can be used directly in space without any assistance of control strategy. However, it is also noticed that, in vacuum environment, it mostly takes much more time to achieve a stable displacement output, and in some cases even more than 20 minutes are needed. This may limit SMA's application in space if no other means is utilized to increase its response speed. To overcome SMA’s poor stability and increase its response speed, an actuator control strategy is developed based on the idea of adjusting the SMA wire temperature as fast as possible. This strategy is simple, stable, and no hysteresis model or thermal model is required. Tests are performed and the results demonstrated that the proposed control strategy is very effective in controlling SMA actuators. Under this control strategy, SMA wire actuator can track square wave, ramp, and sinusoidal signal with very high accuracy ⎯ small steady error and overshoot. By increasing the input current, SMA response speed can be increased remarkably. In spite of an increase of rising overshoot and possibly the RMS error, the absolute values of them are still very small. It is also indicated that shorter sampling interval is helpful in getting higher tracking accuracy. After getting the above testing results with satisfactory accuracy, the actuator control strategy is integrated into a control system based on genetic algorithm, and used to control the flatness of a membrane SAR antenna model. Tests are performed 20 times at room temperature. Each time all tension combinations generated by the genetic algorithm are very well realized by the SMA actuators and the standard deviation of the membrane flatness goes down quickly from around 0.22mm to less than 0.05mm. Another 20 tests are then performed with local thermal load applied. Again optimal tension combinations are realized very well. The membrane flatness goes down quickly from around 0.33mm to less than 0.05mm. Chapter 5 - Wireless communication systems are experiencing a rapid growth in the number of subscribers and the range of services and hence, limitations on the availability of the radio frequency spectrum are growing as well. As a result, a substantial amount of research effort has focused on transmission techniques, along with sophisticated coding and signal processing algorithms in order to overcome the various detrimental effects of wireless propagation phenomena. These techniques must also allow for efficient use of the spectrum and provide high-performance, reliable and cost-effective wireless communication. In order to provide improved performance in wireless systems and to cope with various propagation impairments it is possible to design various channel codes or modulation schemes jointly or independent of each other [1]. In a joint manner, modulation and coding can be used in schemes like trellis-coded modulation (TCM) [2] resulting in increased performance due to the increased Euclidean distance. In an independent manner, channel codes can be used to boost system performance at the expense of increasing resource requirements such as transmission power or bandwidth. This results in decreasing the spectral efficiency of a system. Alternatively, various modulation schemes can be used for different applications giving different trade-offs between performance and implementation complexity [1]. Another popular alternative is to use diversity techniques. This usually (but not always) results in employing multiple transmit and multiple receive antenna systems.

Preface

xi

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Chapter 6 - The moment method technique based on Galerkin’s procedure is developed to examine the scattering properties of a rectangular microstrip patch antenna with non zero surface resistance and containing anisotropic substrate. The electric field integral equation for a current element on a grounded dielectric slab of infinite extent is developed by different basis functions, also the asymptotic forms of these basis functions are given in this study. The electric field integral equation which enforces the boundary condition must vanish on the patch surface, can then be discritized into a matrix form, the necessary terms for representing the surface resistance on the patch are derived and are included in the equation in the form of a resistance matrix, once the impedance matrix and the resistance matrix are calculated, the results form a system of simultaneous equations, the resulting system of equations is then solved for the unknown current modes on the patch. The complex resonant frequency, the radiation and the scattering radar cross section of a microstrip antenna including the effect of the surface resistance and the effect of uniaxial anisotropy in the substrate are analyzed. Also a theoretical analysis of a rectangular patch antenna excited by a microstrip line or an electromagnetic coupled feeding with a perfectly or an imperfectly conducting patch and isotropic or uniaxial anisotropic substrate are presented in this chapter, note that the currents on the feed line and the patch are expanded in terms of three types of modes which will be given in detail in this study. Numerical results show that the surface resistance significantly affects the radar cross section and radiation of the rectangular microstrip patches. Also it is worth noting that our calculated frequencies do not depend on the surface resistance. Moreover the results indicate that the resonant frequency is slightly increased due to the positive uniaxial anisotropy, on the other hand, decreased due to the negative uniaxial anisotropy, however the resonant frequency, the radiation and the radar cross section are slightly shifted due to the ε x permittivity change and drastically change due to the ε z permittivity change. Chapter 7 - Antennae of insects play an important role in their adult life and vary greatly in their morphology, structure, and function. This is particularly true in the parasitic wasps, greater understanding is still lacking in this area. Understanding antennal sensillar type and structures will, therefore, add to our knowledge of the variations in parasitic wasp behavior. In this paper, the morphology, structure, and function of antennae in parasitic wasps were reviewed and summarized based on our research results and published literature. Antennal length, diameter and shape between male and female in the major species/groups of parasitic wasps were described and compared. Function and structure and distribution of antennal sensilla elucidated by scanning and transmission microscopy had been presented separately as trichoid sensilla (SW), chaetica sensilla (TP), basiconic sensilla (SW), coeloconic sensilla (DW), and placoid sensilla (SW). Variations in type, number and size of antennal sensilla in the different species as well as sex’s differences, were also discussed in light of behavior. The significance of antennal patterns and ontogeny of sensilla as phylogenetic tool is discussed. Chapter 8 - The purpose of the present paper is the experimental research of properties of the acoustic antenna arising at the braking of an intensive beam of accelerated protons in the water environment. Research was conducted in a near-field zone that had allowed to allocate signals from separate elements of the antenna and to carry out the analysis of parameters of signals, such as amplitude, width and time of their propagation. As a source of protons, the external beam of the accelerator at the Institute of Theoretical and Experimental Physics (ITEP, Moscow), with energy of 200 MeV and a time duration of

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

xii

Albert I. Ferrero

70 ns, was used. The beam intensity was supported at the level of 4⋅1010 protons per pulse and supervised by the current transformer. An experiment was carried out in the parallelepiped plexiglass basin of a square section 95 cm in length and with a volume of 250 liters filled to 85% with water. Input of the proton beam inside the volume was realized through a pipe with a diameter of 59 mm, 46 cm length and wall thickness of 1.5 mm inserted into a lateral side of the basin and closed by a plug made from organic glass with thickness of 2mm. The average ionizing range of protons in water was 25.2 cm. So, the sizes of the basin and the applied equipment have allowed study of the non-deformed structure of a hydroacoustic field induced by the proton beam. Measurements of an acoustic field were made by means of a relocatable hydrophone in two mutual-perpendicular directions. Along the beam axis hydrophone movement was carried out with a step of 8.9 mm at a distance of 3.5 cm from the beam axis. In the cross-cut direction the trace passed in the horizontal plane, passing through the beam axis at a distance of 35.6 cm from the point of the entrance of the proton beam in water. In this case the scanning step was equal to 4.45 mm. According to thermoacoustic model in the area of beam action for time, comparable with the action time, an acoustic antenna arises. In the present work the problem of reconstruction of the form of the antenna using the experimental results is being solved. The technique of calculation of the hydrophone response to the radiation of separate elements of the acoustic antenna has been developed. The dependences of amplitude of the signals and their time parameters on the relative position of the antenna and the hydrophone have been obtained. The angular distribution of the field created by the terminal area of the radiation zone has been obtained. This characteristic, generally speaking, is similar to the directional diagram of an audio antenna. To test the experimental results, the full-scale simulation of set-up geometry and the physical processes accompanying the propagation of protons in water was carried out using GEANT-3.21 package. The simulation of the process of generation of an acoustic signal was performed as a first approximation in the assumption of proportionality of the signal intensity to the energy that is generated at the ionisation of atoms of water by a proton without taking into account heat conductivity and the elastic properties of the environment, leading to relaxation. The model calculations confirm the qualitative conclusions and the results obtained at the processing of experimental data. Chapter 9 - This chapter deals with the interference reduction capability of non-uniform antenna arrays with linear and circular geometry at a base station of a cellular system. The well-known method of genetic algorithms is used to determine the optimal non-uniform array geometry between antenna elements in order to provide maximum interference reduction. The performance criterion for interference reduction employed in this work is the interference suppression coefficient. Simulation results using different numbers of antenna elements for non-uniform linear and circular arrays are provided. Simulation results show that the optimization of the array provides a better behavior of the interference reduction capability with respect to the uniform separation case of half-wave length. In addition to, a comparison of the performance of non-uniform linear and circular arrays when the separation between antenna elements is optimized with genetic algorithms to provide maximum interference reduction is accomplished. Chapter 10 - The interest in the characterization and optimization of antennas operating in a wide frequency band is a strong motivation for the research of efficient and accurate

Preface

xiii

time-domain numerical analysis techniques and optimization tools. As a computational engine, the time domain integral equation method can be used producing time responses of currents on the antenna surface. In order to eliminate the necessity of Fourier transforming time-domain quantities to the frequency domain, equivalent time-domain parameters are introduced, and the multi-objective cost function for the antenna optimization is formulated in terms of time-domain equivalents of scattering parameters, radiation patterns, and gains; incorporating the antenna dimensions and polarization properties to the design process are also discussed. Pareto optimum front of the formulated cost function is computed by genetic algorithms and swarm intelligence methods. Chapter 11 - The use of high frequencies limits the penetration of the radar energy—this low penetration depends, at a parity of used frequencies, on the electromagnetic properties of the investigated material. The principal drawback is the elevated sensibility of the system to the variations of the EM properties of the environment immediately surrounding the antennas. The radar signals that arrive at the receiving antenna are attenuated and modified because of the selective absorption of the pulses by the ground, because of geometrical spreading and because of an alteration of the actual amplitude due to the instrument amplification. In this paper the absorption of the radar signal is taken into consideration, and the authors determine, respectively: • •

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

•

the radar signal attenuation in the ground; the electrical conductivity (σ) and the relative dielectric permittivity (RDP) of the ground; the maximum penetration depth of GPR in the ground.

Chapter 12 - In this chapter, a stochastic search method, the Differential Evolution algorithm and an analytical method, the Gauss Newton method are applied to synthesize a linear array factor focused in sidelobe level. Differential Evolution has proven to be an efficient method for real valued problems. On the other hand, Gauss Newton method has been applied as a generalization for error-based methods which use desired arrays factors. Due to physical considerations, a uniform amplitude excitation of elements is analyzed. For comparison purposes the results obtained are evaluated scanning the main beam over an angular sector to estimate the changes in parameters such as sidelobe level and half power beam width (HPBW). Simulation results show a promising performance of Differential Evolution, improving the Gauss Newton method. Computational experiments are provided and discussed.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

In: Antennas: Parameters, Models and Applications Editor: Albert I. Ferrero

ISBN 978-1-60692-463-1 © 2009 Nova Publishers, Inc.

Chapter 1

ANTENNAS FOR WIRELESS BIOMEDICAL DEVICES C.P. Figueiredo and P.M. Mendes Department of Industrial Electronics, University of Minho Campus de Azurém, 4800-058 Guimarães, Portugal

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ABSTRACT An e-healthcare system requires a group of sensors, which are placed on a patient in order to sense a set of physiological parameters. Several monitoring networks have been used during the last decades consisting of conventional wired equipment, hence not allowing the patients to move freely around. However, recent advances in wireless sensors technology are changing this scenario, where mobile and permanent monitoring of patients may be accomplished, even during theirnormal daily activities. Wireless biomedical devices for sensing and recording, ranging from neural prosthesis to video-capsule endoscopy (VCE) systems, are emerging innovative technologies and they are expected to originate significant business activity in the near future. The success of such systems is in part due to the advent of microtechnologies, which made possible the miniaturization of several sensors and actuators, as well their integration with readout and communication electronics. The availability of very small wireless biomedical devices is leading to solutions where it is possible to have a device sending information using a wireless link at each sensing place. In this way, e.g., for EEG recording, instead of having several leads connected to an acquisition system sending the information for a base station, a solution where each EEG electrode sends its own measured information through a wireless link could be feasible. This new approach for biomedical sensing requires also new solutions for antenna design and integration. One is that, once the antennas are required to be integrated inside the electrodes, they are required to be very close to a highly lossy material, the human body. The other subject that must be well understood is the effect that many radiating devices will have on the biopotentials that are being measured, which requires the antennas to be designed accordingly. Another characteristic of many wireless systems is the need for node localization. Many methods are available, but one suitable to work very close to the human body would be based on low frequency electromagnetic signals, which suffer less attenuation close to the human body. The interference from radio

2

C.P. Figueiredo and P.M. Mendes frequency signal with the biopotentials to be measured will become more important as the frequency of such signals is decreased. Again, a challenge in antenna design must be overcomed. They must not interfere with the biopotential, and they must be small enough to be integrated inside the small electrodes, even when operating at small frequencies. In this chapter the advantages and disadvantages of using independent wireless modules to record each biopotential locally will be firstly discussed. This requires the use of very small modules with integrated antennas. After, we will discuss possible solutions for antenna placement very close to the human body will be discussed. Solutions based on antenna design, EBG substrates and metamaterials will be introduced. Finally, the problems associated with node localization very close to the human body will be addressed. The localization methods suitable for this type of wireless network will be presented, together with the challenges associated with the design of antennas to support those methods. Finally, a wireless biomedical device to operate inside the human body will be presented.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

INTRODUCTION Nowadays, research on future solutions envisions an ambient where people are being monitored all the time. Many research programs undergoing under different designations share a common target: to improve the well-being of each citizen. Projects are being conducted on Ambient Assisted Living, e-Health, and Tele-Care. The developing technologies are based on sensing and actuating the human body using Information and Communication Technologies (ICT). Since people live, more and more, in a very dynamic world, in which the people are required to move from one place to another, most of the research aims the development of wearable and wireless technologies. An e-healthcare system requires a group of sensors, which are placed on a patient in order to sense a set of physiological parameters. Several monitoring networks have been in use during the last decades using conventional wired equipment, hence not allowing the patients to freely move around. However, recent advances in wireless sensor networks technology are changing this scenario, in which mobile and permanent monitoring of patients may be accomplished, even during patient normal daily activities. Since many potential micro wireless devices communicating from the human body (interior or surface) are desirable, research on antennas must be done to handle such task. This chapter will be dealing with the problem of wireless bio-monitoring. Many sensors are becoming available in very small sizes, with the ability to measure many parameters from many different test sites. This will raise the need to transform each sensor place in an individual sensing element, with its own power harvesting technology and wireless communications. This chapter will start with an insight to the acquisition of electrical biomedical signals, then the communications usually required for such devices. Two types of communications will be presented: data communications and communications to provide extra features to the wireless systems, particularly, the localization ability.

Antennas for Wireless Biomedical Devices

3

BIOMEDICAL SIGNALS ACQUISITION In a general way, any signal that is related with a human body condition may be referred as a biomedical signal. In this way, biomedical signals are related to body movement, cardiac activity, muscle activity, glucose level orbody temperature, to mention just a few. Moreover, the acquisition of the different signals requires different transducers to measure different physical quantities. In the next section, two of the most important electrical signals that require wireless acquisition systems will be introduced.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Electrical Biomedical Signals The measurement of biomedical electrical signals at different regions of the human body has since long time been a very common method to access organ functionality and for the diagnostic of pathologies. Electric biomedical signals, also known as biopotentials, have their origin in the electrochemical activity produced by excitable cells at neuronal, muscular and glandular tissues [1, 2], whose cells are capable of reacting to external stimuli, altering their membrane potential in accordance. While at resting state, the excitable cells maintain a steady electric potential across their membranes (typically between -70 to -90 mV, relative to the outside of the cell), due to an established balance between the selective ion permeability and a continuous active transport mechanism of ionic species against their diffusion gradient (Na+K+ pump) [1]. When an adequate excitatory stimulus is applied to a cell, with sufficient magnitude or duration to alter the membrane potential beyond its excitation threshold, the cell generates a characteristic “all or none” electric response defined as an action potential, which propagates through the cell membrane [1, 3]. The propagation of the action potential generates an electrical field, which, when combined with further electrical fields produced by the surrounding cells, also propagates through the biological medium of the human body, allowing electrical signals to be measured at its surface in a non-invasive manner. The biomedical electrical signals have typically small amplitudes and frequency ranges and are very susceptible to external interference (e.g. electromagnetic interference from power sources or radio devices). Therefore, the design of an acquisition system for biomedical electrical signals must include adequate electronic instrumentation to guarantee the integrity and quality of the acquired signal, since a corrupted signal could give misleading information to, e.g., the physician who performs an exam. Table 1 lists the most commonly measured biomedical electrical signals and their electrical properties [4, 5]. Table 1. Signal properties for the most commonly measured biomedical electrical signals. Signal Electrocardiogram (ECG) Electroencephalogram (EEG) Electromyogram (EMG) Electro-oculogram (EOG)

Amplitude Range 1 – 10 mV 1 – 100 µV 1 – 10 mV 10 – 100 µV

Frequency Range 0.05 – 100 Hz 0.5 – 40 Hz 20 – 2000 Hz DC – 10 Hz

4

C.P. Figueiredo and P.M. Mendes

Electroencephalogram The electroencephalogram (EEG) involves the recording of biomedical signals generated by the neuron cells at the brain. This signal is clinically relevant on the detection of brain pathologies such as sleep disorders or epilepsy, but it is also a widely used tool on neuroscience research. This bioelectrical signal can be measured with both invasive and noninvasive methods, the latter by placing electrodes on the scalp, usually after a time-costly preparation with a light abrasion and application of a conductive gel, in order to reduce the impedance between the electrodes and the subject’s skin [6]. The placement of the electrodes is specified by the International 10-20 System [7], which is represented in Fig. 1.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 1. Electrode positioning map specified by the 10-20 international system.

The designation of this system derives from the criterion of electrode placement, which are placed at intervals of 10% and 20% between two reference points on the sagital plane: the nasion (depression at the top of the nose) and the inion (protuberance at the base of the skull and on the back of the head) [1]. The coronal and axial planes are also divided in intervals of the same percentages, using the midpoint between the nasion and the inion as a reference (Cz electrode). The measured potentials at each electrode are referenced to the average of the potentials at the earlobe electrodes (A1 and A2 electrodes). With proper preparation and electrode placement, it is possible to record electrical signals of typically few microvolts. The frequency range is dependent on the state of awareness of the subject, and it tends to increase with its mental activity. Four groups of wave rhythm patterns have been found to describe normal conditions: alpha (8 - 13 Hz, awake and relaxed state), beta (14 – 30 Hz, active mental activity), theta (4 – 7 Hz – drowsiness and on young subjects) and delta (0.5 – 4 Hz – deep sleep) [2]. Fig. 2 shows examples of EEG signals where different brain rhythms may be observed.

Antennas for Wireless Biomedical Devices

5

Figure 2. Typical signals obtained duringan EEG recording.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Electrocardiogram The electrocardiographic (ECG) signal records the electrical activity of the human heart and it is useful for the diagnostic of arrhythmias and myocardial infraction as well as for monitoring the pulse rate, which is regarded as a vital sign. ECG signal acquisition comprises recordings between different sets of electrodes placed at the chest and also at the extremities of the body (wrists and ankles). In the context of ECG, a differential measurement between two given electrodes is commonly denoted as a lead. The modern ECG contains twelve leads: limb (three), augmented limb (three) and precordial leads (six) [1]. These leads can be also divided in unipolar and bipolar leads. The limb leads are bipolar and they define the Einthoven triangle, introduced by Willem Einthoven in 1908 [7]. Three vectors are defined, using the extremities of the body, as represented in Fig. 3. When measuring only a 3-lead ECG, the remaining lead at each dipole becomes the ground by default. In modern systems, the ground electrode is usually placed at the right leg. The remaining leads of the 12-lead ECG are unipolar. Unipolar recordings were firstly introduced by F. N. Wilson et al. [8], who proposed a common reference point from the average of the potentials at the limb leads, so that the potentials at each distinct site of the human body could be differentially measured to the same reference, defined as the Wilson Central Terminal (see Fig. 4a). At a later time, Goldberger proposed a new set of leads based on the Wilson Central Terminal. He observed that the removal of the resistor connected to the correspondent measuring electrode allows for higher potentials to be recorded [9]. Thus, three new leads were defined and denoted as augmented limb or Goldberger leads, resulting in a modification from the Wilson’s Central Terminal according to Fig. 4 (b-d).

6

C.P. Figueiredo and P.M. Mendes

RA – potential at the Right Arm LA – potential at the Left Arm RL – potential at the Right Leg (Ground) LL – potential at the Left Leg Lead I: LA – RA Lead II: LL – RA Lead III: LL – LA

Figure 3. Electrode positioning map for the limb leads of the standard 12-lead ECG.

a)

b)

c)

d)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 4. Circuitry and electrode positioning map for the Wilson Central Terminal (a) and the augmented limb leads of the standard 12-lead ECG (b-d).

Finally, the precordial limbs, named V1 to V6, were also proposed by F. N. Wilson to measure electrical signals close to the heart [10]. These leads are also unipolar and the Wilson’s Central Terminal is used as the negative electrode to each of the V1 to V6 electrodes placed in the chest, as Fig. 5 shows:

Antennas for Wireless Biomedical Devices

7

Figure 5. Electrode positioning schema and circuitry for the precordial leads of the standard 12-lead ECG. ECG1 [µV]

-1000

-800

-600

-400

-200

0

-250

-200

-150

-100

-50

0

50

100

150

200

250

300

350

[ms]

200

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

400

600

800

1000

1200

Figure 6. Typical ECG signal shape.

The ECG signal has a characteristic and recognizable waveform, represented in Fig. 6. In normal conditions, the ECG signal features a pattern of successive waves, which are denoted between P and U, in alphabetical order. Each wave or complex of waves is associated with an

8

C.P. Figueiredo and P.M. Mendes

activity event on the heart [2]. The P wave relates to the atrial depolarization, while ventricular depolarization causes the QRS complex. The ventricular repolarization is responsible for the T wave (atrial repolarization is not usually distinguishable because it superimposes with the QRS complex). The U-wave is not visible in most individuals, and its origin has not yet been fully explained [11].

Acquisition Systems The acquisition of biomedical electrical signals follows the same guidelines as for any other electrical signal. However, when acquiring signals with low amplitudes and frequencies on the human body, special care must be taken with environmental and biological interference, in order to preserve signal information, allowing a correct assessment or diagnostic. Therefore, amplifier and filter design are key issues on the development of an acquisition system for biomedical electrical signals [4]. Moreover, some of the information contained in a biological electric signal is not readily available from the raw signal, requiring further analog or digital processing [12]. A generic block diagram of an acquisition system for biomedical electrical signals is shown in Fig. 7.

Electrodes

Amplifier

Filter

Analog to Digital Converter

Digital Processing

Display

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 7. Generic block diagram of an acquisition system for biomedical electrical signals.

The electrodes, in the first block, provide the interface between the body and the electronic instrumentation while acting as sensors or transducers, changing ionic current into electrical current [12]. The electrical signals obtained at the electrodes are usually very small in amplitude; therefore analog processing is required, through low-noise amplification and filtering. Biopotential amplifiers increase the amplitude of the biomedical electrical signals to match the electrical requirements of the acquisition hardware, such as the analog-to-digital converter (ADC) [12]. For an amplifier to be suitable for biomedical application, it requires adequate features such as: high input impedance (at least 10 MΩ, to minimize the current draw from the acquired signal, and therefore minimize distortion), low output impedance (close to zero, in order to provide the power required by the load), high differential gain (typically 1000 or greater), low common mode gain and consequent high common mode rejection ratio (to reject the distortion cause by the common mode signal) and temperature stability [4, 13]. The analog filtering stage is also placed prior to the analog to digital conversion, in order to limit the bandwidth of the acquired signal to the range of interest of the correspondent biomedical signal. Thus, this process is responsible for reducing undesirable low-frequency or high-frequency noise. Interference from specific sources, such as power lines, can also be reduced by adding notch filters at 50 or 60 Hz. Although the filtering stage provides the removal of undesirable signal features and prevents aliasing caused by sampling at a later stage on the ADC, it is necessary to ensure that it does not introduce any distortion on the signal to be acquired. To minimize this risk, the filtering stage should be ideally designed to

Antennas for Wireless Biomedical Devices

9

have a constant modulus and a linear phase frequency response over the frequencies of interest [12]. After the analog processing is complete, the signal is digitalized using typical sampling and quantization methods such as, e.g., Sigma-Delta or successive approximations [14]. Once the signal is in digital format, it can be easily handled by any type of digital processor, such as a microcontroller. The signal can be further processed by means of digital techniques [12], saved in memory for further processing or transmission, or shown on a display to allow its observation.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Electrodes As previously mentioned, electrodes have a transducing function, by converting ionic current into electrical current. The description of this transducing mechanism is out of the scope of this chapter, but a thorough description on biopotential electrodes can be found in [15]. For the acquisition of biomedical electrical signals, non-polarizable electrodes are used. This type of electrodes allows the current to cross the electrode/electrolyte interface, thus acting as a resistor [15]. This makes them suitable for recording applications, in contrast with the polarizable electrodes, which doesn’t allow current to cross the aforementioned interface and are used for stimulation applications. Electrodes for biomedical signal recording are commonly made of conductive metals coated with a salt compound of that metal and applied to the body with an electrolytic gel, being the silver chloride (Ag/AgCl) electrode the most commonly used. The problem of using this type of electrodes is the requirement to be in contact with the human body using an electrolytic gel. The use of these electrodes is not an adequate solution to be used with wearable devices. When someone dresses a shirt with monitoring electrodes, it is not expectable that someone has to use gel to guarantee the electrical contact required for proper measurements. The solution for this problem is to develop an electrode that ensures adequate performance without the use of gel. These are already under development and it is currently possible to find dry electrodes with performances similar to the electrodes requiring the use of gel. The wearable technologies demand the use of electrodes integrated within the vests, and, if possible, requiring no contact with the human body. Moreover, the technologies will be successful only if all the acquisition system is integrated (not distinguishable) with the wearable device.

BIOMEDICAL WIRELESS ACQUISITION SYSTEMS Wireless communications is a very well known established concept. Nowadays, it is almost possible to send and receive information, virtually from anywhere humans can place a device to communicate with, and the human body is no exception. The use of wireless biomedical devices for sensing and recording, ranging from neural prosthesis to video-capsule endoscopy (VCE) systems, are emerging innovative technologies

10

C.P. Figueiredo and P.M. Mendes

and they are expected to originate significant business activity in the near future. The success of such systems is in part due to the advent of microtechnologies, which made the miniaturization of several sensors and actuators possible, as well their integration with readout and communication electronics. Typically, smaller devices require less power to operate. This requires the use of microtechnologies to fabricate microsensors and microactuators, integrated with readout electronics. Many times, these microdevices requires also the use of power electronics, conversion between analog and digital domains, and signal processing capabilities. Finally, the microdevices are provided with communication electronics. Fig. 8 shows a block diagram of a highly integrated microsensing system [16].

Signal Processing

D/A

Display

A/D

Actuators

Sensors

Signal

Power

Communications

Wireless/wired

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 8. Microsystem architecture.

The block diagram represents a system architecture with the ability to control a significant number of sensors and actuators, which implies the integration of amplifiers, integrators, filters, multiplexers, current and voltage sources, ADC and DAC. Since the signals associated with sensors and actuators can be very complex, the integrated electronics must be able to deal with several parameters as linearity, noise, bandwidth, temperature or offset. Moreover, all the system parameters must be as programmable as possible. And preferably, the system should be able to harvest its own energy from batteries, photocells, electromagnetic fields, mechanical vibrations, or whatever can provide energy for system operation as long as possible. To make everything even more difficult, the physical dimensions should be kept as small as possible. There are many devices to measure biosignals, where the registered signals are delivered to medical instruments for processing, analysis, and storage. The importance of wireless microdevices is increasing due to different aspects. One is an increase of the number of microsensors available to read different biosignals, which complicates the use of wired solutions, due to the increase of the required number of wires for power and communications. The other is the need to provide solutions that are compatible with the daily life. Many clinical tests require the patient to use a device for some time, before diagnostics can be made. Also, many people are using monitoring devices during their sport activity, or to help

Antennas for Wireless Biomedical Devices

11

them with some daily tasks. Obviously, a wired device is not a desirable solution and, if possible, a wearable should be used instead.

Wireless Communications

VHF

Far

Microwave

Radio

100 m

10 m

1m

Thermal

Near

Infrared

1 mm

0.1 mm 10 μm 1 μm

Ultraviolet

HF

Visible

Wireless devices are nowadays a part of our daily lives. Besides their higher popularity on communication devices, wireless data acquisition systems have also many advantages over wired systems. A wireless microsystem uses the surrounding environment (e.g., atmosphere, human body) as communicating channel. That channel is available in several spectral windows or frequency bands. The most common channel in wireless communications is the use of the atmosphere, which has different propagating characteristics for different spectral windows. The separation of the available spectrum into different bands does not mean the available spectrum is discontinuous. However, there is a need to classify the different spectral ranges and to group them considering similar spectral properties. Fig 9 shows a range of spectral windows that are widely adopted for communications. One is the region of the radio waves and the other is the optical region, used for fiber optic communications or wireless optical communications.

0.1 μm

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 9. Spectral windows used for communications.

The deep knowledge of radiation and propagation properties for the different bands as led to an efficient use of each single spectral region and today each spectral region is normalized all around the world. The standard use for the different spectral regions splits the world in three regions, where it is possible to find two kinds of spectral windows. One are the proprietary windows and the other are the free bands, labeled ISM (Industrial, Scientific and Medical). The ISM bands are very important since they provide a free space for the development of new wireless devices. However, those devices are required to meet the specifications allowed inside such bands, like transmitted power, bandwidth, and interference that they introduce in other devices and are able to support. The use of the different spectral windows is dependent on several factors as the availability of signal sources, compatible electronics and/or devices and techniques that allows the signal processing at the desired frequencies. The most well-known and widely adopted wireless devices are the RF/microwave devices. These are very popular due to their high flexibility, like possibility to communicate without the need of a line-of-sight link, the easiness of integration an interconnection of RF devices, when compared with, e.g., optical

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

12

C.P. Figueiredo and P.M. Mendes

devices, the costs associated with fabrication (today it is possible to have a full RF transceiver on a CMOS die). The advantage of using optical communications could be the potential device size reduction, since light sources and detectors can be made very small. The main drawback would be the need to communicate in line of sight conditions. No object can be placed between emitter and receiver. Moreover, it would be also difficult to integrate light sources with enough power to provide communications in some applications, as for devices required to operate behind bones or inside the stomach. Wireless RF/ microwave systems have been the subject of significant research and development under the concept of wireless sensor networks. One of the most well known, and maybe most successful standard is the ZigBee standard [17]. It was designed to allow a set of sensing modules to communicate and exchange data, at low cost and low power. Due to the large availability, Bluetooth was also being used as a technology to support wireless sensor communications [18]. More recently, due to the low power constrains, Wibree was also proposed as a technology to be used in wireless sensor networks. When using a RF link on a wireless device, the frequency of operation must be carefully selected in order to meet the required specifications. Apart from other specifications, two of them are always required: bandwidth and link range. To satisfy the bandwidth requirements is “easy”. Just increase the carrier frequency until the bandwidth available meets your requirements. However, the use of high frequency carriers requires more complex electronics and, more important, the power budget required for propagation increases. In this way, the carrier frequency should be as low as possible, to keep the transmitted power as low as possible, saving system energy. The signal attenuation in free space increases as the frequency increases, which create some problems when very small wireless devices are required. The problem is that the device miniaturization will only be fully accomplished if the antenna is also completely integrated within the system. To achieve this, a very small antenna is required and, since most of the antennas used in such devices are resonant antennas, the carrier frequency must be increased to reduce antenna size. Using this technique, together with some miniaturization techniques, is possible to obtain very small antennas operating at relatively low frequencies. Is possible to have antennas operating in the 5-6 GHz band as small as 1.5x1.5x0.5 mm3. The signal attenuation increases even further when the transmission environment is the human body, instead of free space. As an example, take for instance the penetration depth dependency on frequency [19]. What happens is that for high frequencies, the signal penetration depth can be as small as a few millimetres, which is not suitable for most wireless biomedical devices. The solution is to decrease the carrier frequency to the MHz range, which causes the enlargement of antenna dimensions. The main requirements for wireless sensor networks is that they can be deployed at very low cost (for environmental applications, thousands of devices may be required), they should require low or no maintenance, and, since they potentially operate on remote places, power consumption should be kept as low as possible. Due to these constrains, researchers have been working on low power technologies that can integrate as many functions as possible.

Antennas for Wireless Biomedical Devices

13

Wireless System Integration The starting goal of microtechnologies was the development of techniques, processes, and materials to allow the availability of integrated systems with new functionalities available at a very small scale. After being able to fabricate very advanced microsystems, a new challenge was born: the need for interaction between such systems and the surrounding systems. To reach such a goal it was necessary to start working on the concept of microystem integration. The first problem that was necessary to solve was to place the readout electronics as close as possible from the sensing devices to avoid noise and interference. This led to technologies to integrate the sensors, actuators and electronics in the same chip. Later on, it was also possible to integrate processing, power, memory, and communications in a single chip. Since then, many different technologies are available that attempt to solve different integration problems, leading to highly complex integrated microsystems [20]. The Fig. 10 shows how complex an integrated microsystem may be [21]. All device fabrication starts with the base material: silicon. The microsystem fabrication was possible with the combination of different techniques. It was used the silicon processing techniques, but also the technique of MCM (Multi-Chip Module), where some elements are separately fabricated and then are all placed together on a silicon substrate. The interconnection may be obtained using wire-bonding or flip-chip.

spiral inductor

BiCMOS

HRPS

solder ball

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 10. Example of a possible fully integrated wireless microsystem.

Despite the high degree of integration achieved, these systems still require the use of package pins to provide a connection for energy and communications. The use of microtechnologies has led to the availability of very small wireless biomedical devices, leading to solutions where it is possible tohave a device sending information though a wireless link at each sensing place. In this way, e.g., for EEG recording, instead of having several leads connected to an acquisition system sending the information for a base station, one solution where each EEG electrode sends its own measured information through a wireless link could be feasible. The wireless link may be established using different methods, from acoustic to optical communications, but the purpose is always the same: to send information from one place to another without the use of wires. To do so, a transmitter and a receiver are needed, and at each device, means to translate the electrical signal carrying information into a signal that is able to travel through the surrounding environment, must be provided. In order to use electromagnetic waves in the radio frequency range, an antenna is needed to achieve such goal.

14

C.P. Figueiredo and P.M. Mendes

Antenna Design All wireless communication systems, ranging from implantable wireless biomedical devices to the wireless networks, require the use of an antenna for interfacing between the propagating environment and the RF circuitry. The antenna converts the incoming electromagnetic wave, which propagates along the feeding line, into an electromagnetic wave into free space, and vice versa. Depending on the application, antenna design may be a very challenging task. And this challenge becomes even higher when the antenna must be designed for integration with some specific device. An antenna may take several different shapes, dimensions, and geometries. An antenna may be as simple as a piece of wire, as a monopole, up to as complex as a complex matrix of parabolic dishes, working as radio telescopes. Nowadays, many wireless devices are being pushed to operate very close or inside the human body. There is an increasing demand to sense everything associated with human body. Movements, positioning, localization, electrophysiological signals, all are subject for monitoring. This new approach for biomedical sensing requires also new solutions for antenna design and integration. Maxwell equations are the base of antenna analysis [22], however, when dealing with antennas, a set of parameters used to describe their operating characteristics must be highlighted. Fig. 11 shows a microstrip antenna to illustrate the two kinds of parameters associated to an antenna; the radiating parameters and the feeding parameters. Z

G, D, η, Prad

VSWR RL Zin

ρ

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Y

X

Figure 11. Feeding and radiating characteristics used to describe an antenna.

An antenna may be considered as a two-port network, where one port is connected to the feed line and the other is “connected” to the surrounding environment. In this way, two sets of properties are used to describe the antenna. From the feed line side the most common is to measure the S-parameters and from the radiating side, the radiation patterns. Firstly, the antenna microwave parameters will be described. To feed the antenna, it is possible to use different techniques, where the simplest techniques are the use of a microstrip, a coplanar waveguide, or a coaxial cable.

Antennas for Wireless Biomedical Devices

15

Independently from the feeding technique, the most important aspect of such transition is the amount of power transferred to the antenna, which is related to an adequate impedance matching. To characterize the antenna performance from the circuit perspective, the parameters that may be required are input impedance, Z L , reflection coefficient, ρ(z ) , and return loss, RL. When doing antenna characterization, the mostly used input parameters are the S-parameters. Those are widely accepted to describe a microwave device due to their convenience for measurement. S-parameters can then be used to determine the other required parameters:

⎛P RL = −10 log10 ⎜⎜ R ⎝ Pin

⎞ ⎟⎟ = −20 log10 | ρ |= −20 log S11 ⎠

(1)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

where Pr and Pin are the reflected and incident power at antenna input. Bandwidth is also a very important parameter when dealing with an antenna and it appears in different ways. The antenna bandwidth represents its ability to operate inside some frequency range. That frequency may be measured considering, e.g., antenna gain, VSWR (Voltage Standing Wave Ratio), return loss, cross polarization level, front to back ratio. One needs to be careful when one reads “Antenna bandwidth is …”. This may be the frequency range where the power gain is above –3dB, or the range where VSWR is less then 2:1, or the range where the return loss is below 10 dB. However, despite all the possibilities, the most common is to refer the bandwidth to the return loss (smaller then 10 dB, or 6 dB for some applications) or to VSWR (< 2:1). Moreover, the bandwidth may be expressed as an effective value in Hz or as ratio of the central operating frequency. The previous antenna parameters describe the antenna input characteristics and are essential to the circuit designer in charge of the interface antenna-RF circuitry. However, since the main goal of the antenna is to transfer the energy to the surrounding environment, the antenna radiation properties must also be overviewed. Fig. 12 shows the coordinate system that is frequently used when dealing with antenna radiation properties [22].

r φ θ

r

θ

Y φ

X

Figure 12. Spherical coordinate system.

16

C.P. Figueiredo and P.M. Mendes

When the antenna receives power through the input port, it is converted to an electromagnetic wave that will radiate from the antenna. The electromagnetic field can be separated in three regions: reactive, near field or Fresnel, and far field or Fraunhofer. The most common way to describe the antenna properties is to use the far-field properties, since the main goal of one antenna is to transmit, or receive, electromagnetic waves along the largest distance possible. However, it is very important to know that all these regions do exist, and that the antenna radiation properties may be affected if those fields are disturbed. Moreover, those regions have different electromagnetic field distribution and may be affected in a different way then the far field. Changing the antenna environment around the reactive or the near field region may lead to different far field properties and input parameters. The boundary between the near and far field regions is usually defined for r = 2L

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

2

λ (some

authors uses r >> L ou r >> λ), where L is the small sphere diameter that encloses the antenna and λ is the wavelength [22]. One of the most important antenna parameters is the antenna gain pattern. It is a measure of the antenna ability to radiate electromagnetic waves. One common antenna design target is to have a specific antenna gain, for a specific direction, satisfying a specific bandwidth. The gain of one antenna can be present as a single value, referring to the maximum gain of the antenna, or as a gain pattern, where it is possible to observe the antenna gain as a function of radiation direction. The radiation pattern is a plot with the information of some radiation properties related to the spherical coordinate system. The plot gives information about the antenna ability to radiate, in the three-dimensional space. The radiation pattern most frequently used is the radiating diagram, which is a plot for the radiation intensity as a function of radiating angle. However, it is possible to use a similar plot to display the gain, or the antenna directivity. Another important antenna parameter is the antenna efficiency, which is the ratio between antenna gain, G, and antenna directivity, D. This parameter is many times neglected since most of the antennas show efficiency close to 100%, and because it is difficult to accurately measure. But, when dealing with antennas for some specific applications, like onchip antennas or antennas that must operate close or inside to high-loss materials, like the human body, these parameters must be taken into account. The antenna efficiency for an onchip antenna may be as small as 5-10%. There are, obviously, other important antenna parameters, that must be taken into account for some designs [22]. At this moment, one must be aware that an antenna is a device that radiates electromagnetic waves on specific directions, has an operating frequency, a bandwidth, has gain dependent on each direction, and will have a certain efficiency. It is know time to look at the specific application on wireless biodevices. The biosignals are low frequency signals, which means a small bandwidth antenna is required/used. Sometimes, the problem may be the need to record several different channels, where the aggregated bandwidth can be high, but always below the MHz range. About the radiation gain, two distinct situations exist. One concerns devices operating outside the human body. In this case, the antenna must radiate in the direction opposite to the human body, to avoid exposure to radiation. The other concerns devices operating inside the human body. If possible, the antenna should concentrate the radiated energy on directions where it is easier to reach the outside free space. But also here, appropriate antenna geometry can be

Antennas for Wireless Biomedical Devices

17

selected to provide the right radiation pattern. As an example, to avoid back radiation, a patch antenna with a large ground plane ca be used. The problem starts when certain specific designs are required. To design an antenna for operation inside the human body is a challenge because the antenna should be as small as possible, raising the antenna operating frequency, which will result on high losses. It is difficult to design a very small antenna to operate at low frequencies. The other challenge is to design wearable antennas. Besides the problem of antenna integration with textiles, these antennas are required to operate very close to the human body, at a dynamic variable distance. But these problems will be further discussed on later dedicated sections.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Antenna Integration Antenna design and integration with biomedical devices is still an open challenge requiring innovative solutions, to allow the development of new wireless biomedical devices. The availability of integrated solutions has the immediate advantage of allowing a fully integrated and autonomous wireless biomedical device, with reduced dimensions. From the economic perspective, it shows a potential for costs reduction associated with traditional antenna assembly. From the antenna performance perspective, it allows optimization of impedance at the interface antenna/circuitry, and reduces interference and mismatch along the antenna feed line. The solution to obtain a fully integrated wireless biomedical device requires the development of an antenna with the necessary characteristics to be integrated within a biomedical device. Despite the possibility to obtain very complex antenna geometries, it is desirable to use as simple antenna geometries as possible, to minimize fabrication costs. Moreover, their geometry must be compatible with the fabrication processes (microtechnologies, textiles) associated with antenna integration, must operate on materials that were not designed to support antenna operation, and the integration should not increase production costs. The limitation of available room, due to restrictions imposed by integrated biomedical devices, requires the development of antennas with geometries and dimensions adequate for integration with such devices. Up to now, solutions are mainly based on planar antenna solutions (for wearable devices) and loop antennas (for operation inside the human body). To integrate an antenna with a biomedical device, the dimensions must be kept as small as possible. Since antenna dimensions are inversely proportional to the operating frequency, the higher the operating frequency, the easier the antenna integration. However, signal attenuation increases with operating frequency, meaning smaller link range for the same transmitted power. In this way, it is necessary to find a tradeoff between transmitted power and antenna dimension. Antenna dimension has two distinc meanings. One is the most obvious, that is measured in meters, the physical length. The other is the electrical length, which relates the antenna physical size with the operating wavelength. That is the electrical length and the one that is difficult to reduce. Since the wavelength changes with the material electric permittivity, the easiest change/alteration that can be done to reduce the size is to increase the electric permittivity of the antenna substrate. This solution has the big disadvantage of bandwidth reduction. However, the bandwidth may be increased with antenna resistive, capacitive, or inductive

18

C.P. Figueiredo and P.M. Mendes

loading, leading to efficiency reduction. The solution must concern a look at antenna geometries that conduce to an antenna size reduction. As mentioned before, the available room for antenna integration is limited by the biomedical device requiring the antenna, the local where the antenna must operate, or by an aesthetic circumstance. In this way, the antenna must be as small as possible, including the ground plane. This will have impact on antenna operation performance. Fig. 13 shows an example of antenna integration on an integrated circuit, that is part of a biomedical device. Due to the operating frequency, this device is mostly suited for wearable applications, where the antenna is not required to operate inside the human body. Antenna

RF circuitry

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 13. Antenna integration within a microsystem.

It is easy to understand that, due to very limited available room, the ground plane does not exceed antenna dimensions, or exceeds in a very limited way. This may have consequences on the antenna operating performance, as well on the nearby circuitry. Since the antenna is very close to the remaining circuitry, it is possible that some coupling can exist between antenna and passive elements, like inductors or transmission lines. The solution could be to further reduce antenna dimensions, to leaving some ground plane margins. However, consecutive antenna size reduction, leading to antennas electrically very small, will introduce antenna limitations, both for antenna input and output parameters. A common question asked when the antenna miniaturization is required is: “What is the theoretical limit for antenna size reduction?”. The first work attempting to answer this question dates back from 1947 [23]. In that work, Wheeler investigated the fundamental limits of electrically small antennas. An electrically small antenna was defined to have overall dimensions smaller than λ 2π . Sometimes, this is also referred under the relation ka < 1,

where k = 2π λ , λ is the free space wavelength and a the radius of the smaller sphere enclosing the antenna, also called the radian-sphere. Fig. 14 shows the concept widely adopted to characterize an electrically small antenna. In his work, Wheeler analysed which could be the highest efficiency of an electrically small antenna, considering it as a capacitive or inductive load. Some time later, Chu published his work about electrically small antennas [24]. In that work, spherical wave functions were used to describe the antenna field, gain, and quality factor. One research topic was to find the maximum achievable gain by an antenna of moderate complexity. This is

Antennas for Wireless Biomedical Devices

19

equivalent to find the minimum quality factor for the antenna. Other research topic was to find the maximum ratio gain/quality factor.

a Antena electricamente Electrically small antenna pequena

Figure 14. Electrically small antenna enclosed by the radian-sphere.

From there on, Chu results have been widely used as a reference in antenna size reduction. Several authors have since then studied the behaviour of electrically small antennas. From those studies it is possible to conclude that, theoretically, it is possible to obtain an antenna with an extremely high gain. However, there are other important parameters in an antenna, namely efficiency and quality factor (bandwidth). Fig. 15 shows the tradeoffs in antenna parameters as the antenna complexity, N, increases, for two different antenna sizes. It can be observed that the gain, G, and quality factor, Q, increases with the antenna complexity, and the efficiency, ηa, is reduced. It can also be observed that the larger antenna has a smaller quality factor and a larger efficiency. 5

10

4

10

3

10

G, Q , ηa

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

2

10

1

10

G Q|ka=1

0

10

Q|ka=0,5 ηa|ka=1

-1

10

ηa|ka=0,5 -2

10

1

2

3

4

5

N Figure 15. Tradeoffs in antenna parameters as a function of antenna complexity, N.

6

20

C.P. Figueiredo and P.M. Mendes

Other interesting analysis is to observe the behaviour of antenna quality factor, efficiency, and gain for an antenna operating in the fundamental mode. Fig. 16 shows the results of such analysis, where it can be observed that all antenna parameters are improved as the antenna dimension increases. It can also be observed the fast degradation of antenna efficiency for ka < 0.4. 3

10

2

Q , G, ηa

10

1

10

0

10

G Q ηa

-1

10

0,0

0,2

0,4

0,6

0,8

1,0

ka

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 16. Influence of antenna dimensions on gain, quality factor, and efficiency.

After a brief look to the tradeoffs associated with antenna size reduction, imposed by the very small modules with integrated antennas, possible solutions for antenna operation very close and inside to the human body will be discussed. The widespread of biomedical devices using wireless communications requires a closer look to these specific devices. Since the antennas for these devices are required to operate in significant different environments, these wireless devices can be grouped in two main groups. One is the group of devices required to operate inside the body and the other is the group required to operate outside the body.

WEARABLE WIRELESS LOCALIZATION DEVICES A personal healthcare system consists of a group of sensors attached non-invasively to a patient, in order to monitor his/her vital signals to detect any life threatening abnormality. These body sensor networks (BSN) have been used in hospitals during the last decades using conventional wired equipment, hence not allowing the patient to freely move around. However, recent advances in wireless sensors technology are changing this scenario by allowing mobile and permanent monitoring of patients, even during their normal daily activities, and without compromising their quality of life [25]. In such healthcare systems, the

Antennas for Wireless Biomedical Devices

21

sensed information at the patient´s body is transmitted to a wireless base-station, located no more than a few tens of meters away, and then delivered to a remote diagnosis centre through a communication infrastructure. Radio localization is a currently a very active research topic with promising applications. Wireless biomedical devices could benefit of using their own radio to allow, for example, tracking of a monitored patient or, for individual implantable or wearable devices, to determine their positions on or inside the human body. Although this would be theoretically possible, precise methods with the required resolution and range are still not available. In this section, localization techniques and algorithms suitable for biomedical applications will be presented.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Localization Techniques A fairly wide range of systems have been proposed over the years to locate people and objects, making use of devices from very different technologies such as radio-frequency, ultrasound, infrared, etc. A survey on localization systems and techniques can be found at [26]. However, most of these systems are very expensive and/or have high power consumption, thus making them unsuitable for ah-hoc and low cost solutions such as wireless sensor networks [27]. Nonetheless, wireless devices such as sensor networks can still use their own radio communication capabilities to estimate distances between pairs of devices, through different methods such as RSS (Received Signal Strength), time-of-arrival or timedifference of arrival (ToA/TDoA) and Angle-of-arrival (AoA). For information on methods and techniques for distance measurement and localization in wireless sensor networks, the reader should consult [28]. RSS consists on the measurement of the signal power at a receiving device, after the transmission of a beacon signal by another device. The distance between both devices is estimated from the attenuation that the beacon signal suffered along its path. This method requires no additional hardware to the system besides the radio itself, but it is severely affect by shadowing and multipath effects arising from reflection, refraction and non-uniform propagation phenomena, and therefore its accuracy is poor for indoor applications (e.g. within different rooms of the same building). ToA and TDoA use the measured one-way and round-trip propagation times of transmitted beacons to infer the distance between nodes, respectively. While ToA requires the synchronization between the clocks of all the devices in a network, TDoA avoids this since the measurements and calculations are all regulated by the same local clock. This method requires counters with very high resolution in order to measure the time of flight of RF signals. Another possibility is to measure the TDoA between both an ultrasound and a RF beacon, which requires an ultrasound transducer as additional hardware. AoA makes use of the amplitude and phase responses of antenna arrays. The reception pattern of the antenna array is evaluated in order to determine the direction of the incoming beacon. The required hardware for this method (antenna array) is usually more expensive compared to the other methods and also tends to be bulkier. Concerning localization of biomedical devices, RSS is probably the most suitable and easily implementable from the aforementioned methods since it does not involve any additional and bulky hardware, and also because ToA would require extremely high resolution to deal with small distances (millimeter-order to meter-order). However, for this

22

C.P. Figueiredo and P.M. Mendes

method to be suitable to provide proper results while working close to the human body, it should use low frequency electromagnetic signals as beacons, in order to minimize the attenuation that the human body causes to the propagating signal, allowing more accurate distance estimates. On the other hand, reducing the frequency of the signal transmitted by an antenna leads to bigger antennas, which are not desirable for biomedical applications such as, for example, implantable devices. Thus, the design and development of new and suitable antennas is an important issue for applications on biomedical devices and shall be discussed in the next sections.

Algorithms for Coordinates Recovery

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Localization algorithms are used to compute coordinates for each node of a network, from different types of data, such as the distance measurements between nodes, connectivity or probabilistic information, etc. A wide range of localization algorithms has been proposed over the past years, with distinct features and requirements. Algorithms can be classified according to different parameters such the distribution of the computational load in the network (centralized or distributed), the use of either measurement or connectivity data (range-based or range-free) and the use of a portion of nodes with knowledge of their own position in a relative or absolute map (anchor-based or anchorless). Reviews on different types of popular localization algorithms and their use can be found in [29-31]. For a network of biomedical devices to achieve self-localization, without the use of any additional and external hardware, range-based and anchorless localization algorithms are recommended. Otherwise, the use of GPS (Global Positioning System) or an external and manually deployed infrastructure of anchor nodes at fixed positions would be required, which are hard and expensive to configure and, in the case of GPS, not suitable for indoor environments. Two localization algorithms, which avoid the use of anchors nodes and are also range-based, were selected for presentation: Multidimensional Scaling (MDS) [30-34] and Self-Positioning Algorithm (SPA) [35].

Multidimensional Scaling MDS is a set of techniques originally developed for mathematical psychology [32]. The simplest of these techniques is known as classical metric MDS. Its goal is to, given a set of points whose coordinates in a multidimensional space are unknown and the measurements of distances between each pair of points, determine an appropriate set of relative coordinates so that the distances between every point are conserved and related by some transformation (a linear transformation model is assumed) while minimizing the error in the least-squares sense. This algorithm is based on the definition in:

M = D+E

(2)

A square matrix M containing the values of the distance measurements between every pair of points is defined as the sum of two matrices D and E, which respectively represent the real distance values and the associated measurement errors. The algorithm proceeds with the computation of a coordinate matrix X so that the sum of the squares of E is minimized.

Antennas for Wireless Biomedical Devices

23

The D matrix is usually double-centered into a matrix B applying the conversion on: 1⎛ 1 ⎞ ⎛ 1 ⎞ B = − ⎜ I − U ⎟D 2 ⎜ I − U ⎟ 2⎝ N ⎠ ⎝ N ⎠

(3)

where I and U respectively represent the identity matrix and a matrix filled entirely with ones. Both the I and U have the same size as D. Note that D2 represents a matrix where every element of D is squared, instead of the matrix product of D by itself. The resulting B matrix is symmetric and positive definite. Next, singular value decomposition (SVD) is applied to B and, since B is square and symmetric, the following applies:

B = USV T = USU T

(4)

Finally, the coordinate matrix of the nodes in the network in the transformed space is obtained from the first 2 or 3 columns of the X matrix (for 2D or 3D), where X is given by (5). The order of the nodes in D is maintained in the X matrix:

X = US

1

2

(5)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The result of the algorithm is an arbitrarily rotated and flipped version of the true network map, since no absolute coordinates are used. Registration techniques can be used to convert the relative map into an absolute map, if absolute coordinates for at least four nodes are provided. MDS tolerates error well, even though every coordinate computation takes the distance values to every other node into account, and the error on each single measurement affects the calculations of the coordinates of every node on the network [31]. However, it has the disadvantages of requiring a centralized station with enough computational power to perform the algorithm and the availability of the pairwise distance measurements between every pair of nodes in the network, which means that every node must be in radio range of every other, which is a reasonable assumption concerning the size of the human body.

Self-Positioning Algorithm The SPA was proposed by Capkun et al. as a distributed localization algorithm for ad hoc networks, using distances between nodes to build a relative coordinate system in which the node positions are computed in a bidimensional map. Through SPA, every node computes its local coordinate system, where each node is at the center of its own system. The algorithm then proceeds with the alignment of the directions of every local coordinate system into a network coordinate system. However, the application of the algorithm on a three-dimensional environment is not straightforward, because the extra dimension produces too many degrees of freedom [36]. Still, the first phase allows the computation of a local coordinate system with reference to one of the nodes in the network. The reference node can be either arbitrarily chosen or adopting a criterion such as, e.g., choosing the node with the least sum of distances to the others, since it is more likely to be at a center position in the network’s topology.

24

C.P. Figueiredo and P.M. Mendes

The computation of the local coordinate map is based on the application of geometrical relations on the distance measurements between the reference node and three additional neighbouring nodes, one for each axis. Considering a generic reference node a and three neighbour nodes b, c and d, a coordinate system is defined so that a is at its origin, b lies on the positive part of the x-axis and c and d have respectively positive y and z components, therefore determining the orientation of the three axis. The coordinates of the a, b, c and d nodes are attributed according to:

a x = 0, a y = 0, a z = 0 bx = Dab , b y = 0, bz = 0 c x = Dac cos γ , c y = Dac sin γ , c z = 0 Dad2 − Dcd2 + c x2 + c y2 d x c x D −D +b dx = , dy = , d z = Dad2 − d x2 − d y2 − 2bx 2c y cy ⎛ D 2 + Dab2 − Dbc2 ⎞ ⎟ γ = arccos⎜⎜ ac ⎟ 2 Dac Dab ⎠ ⎝ 2 ac

2 bd

2 x

(6)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

where D stands for the distance measurement between a pair of nodes. Note that, in ideal conditions, dz has two solutions, which are symmetrical, and thus, only the positive root is considered. The γ value should be tested to avoid the selection of a, b and c as collinear points. As the γ value tends to zero, the a, b and c nodes tend to collinearity and therefore cannot define the x0y plain. When this happens, a new selection of neighbour nodes should be made. Once the coordinates of the first four nodes are computed, the position of the remaining nodes can be determined using multilateration. This technique consists on solving the system of sphere equations that generates from the distance measurements of the a, b, c and d nodes to a remaining node. The system of equations is solved through a standard least-squares approach ((ATA)-1ATb) after linearization of the sphere equations, considering the constraints in (6). The coordinates of a generic remaining node e are computed according to:

⎡2(a x − d x ) 2(a y − d y ) 2(a z − d z )⎤ ⎢ ⎥ A = ⎢ 2(bx − d x ) 2(b y − d y ) 2(bz − d z ) ⎥ ⎢ 2(c x − d x ) 2(c y − d y ) 2(c z − d z ) ⎥ ⎣ ⎦ ⎡a x2 − d x2 + a y2 − d y2 + a z2 − d z2 + Dde2 − Dae2 ⎤ ⎥ ⎢ b = ⎢ bx2 − d x2 + b y2 − d y2 + bz2 − d z2 + Dde2 − Dbe2 ⎥ ⎢ c x2 − d x2 + c y2 − d y2 + c z2 − d z2 + Dde2 − Dce2 ⎥ ⎦ ⎣ ⎡e x ⎤ x = ⎢⎢e y ⎥⎥ = ( AT A) −1 AT b ⎢⎣ e z ⎥⎦

(7)

The computation described in (7) can be performed for every remaining node based solely and repeatedly on the information of the a, b, c and d nodes. In alternative, the

Antennas for Wireless Biomedical Devices

25

information on the coordinates of each node can be added to the A and b matrices, right after its computation, for higher accuracy. As a consequence, the matrices grow in size and the calculations and computational load increase with every computed node. Although SPA is based on simple geometric relations and computations, it does not achieve so accurate results or tolerates error as well as MDS. This occurs because SPA is an incremental algorithm, which tends to propagate the errors on the distance measurements with each subsequent coordinate computation. Moreover, the node coordinates are computed in a one at a time fashion, thus only taking a portion of the all the possible distance measurements between nodes into account when computing the position of a node, which results on a weaker accuracy. Comparing both MDS and SPA algorithms, it is clear that MDS achieves better results, but has a higher computational cost. Performance studies of localization algorithms recommend MDS for networks for high connectivity and reduced amount of anchors while multilateration based algorithms such as SPA should be used when connectivity is scarce or when a reasonable portion of anchors is available [37].

Antenna Integration with Wearable Devices The use of localization techniques is important in many domains. For biomedical applications, it can be used for patient localization or for device localization. Either way, the success of such devices is associated with the ability to integrate the localization devices with textiles, antenna included. To look for an antenna integration solution, it is important to try to understand the environment where the antenna needs to operate. One research field related to neurosciences very active nowadays attempts to decode the neuro activity, the electric signals generated by brain operation. One research branch deals with Brain Computer Interface (BCI), a research field under intensive development. Fig. 17 shows an example of a possible BCI setup.

Se r ial

Link

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ZigBee

Figure 17. Wireless “wearable” BCI system.

26

C.P. Figueiredo and P.M. Mendes

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

It is trusted to be a very useful tool for impaired people, both in invasive and noninvasive modalities. Although subjects using invasive approaches usually show evidence of better device control than non-invasive way users, it is barely preferred due to the high risk involved in its research and practical implementation. BCI has the potential to enable people to control a device with their brain signals. Several studies have been attempting different BCI approaches that enable impaired people to communicate and control specific devices [38]. However, the available brain caps required for EEG recording are far from what users would like to have. Most solutions are still based on wired systems. The available wireless systems are not suitable for many channels and, more important, are not wearable. They are based on many electrodes for signal acquisition, each one requiring wires for connection to the central unit. This unit has an antenna attached, providing the required wireless link. Another very popular concept is the use of an e-shirt for cardio-respiratory function monitoring. The goal of such shirt is to provide an easy way to record several biomedical signals without the need of trained people. The shirt has several sensors embedded to record the required signals. In this way, it is possible, e.g., to recognize qualitatively and quantitatively the presence of respiratory disorders, both during wake and sleep-time in freeliving patients with chronic heart failure, providing clinical and prognostic significance data. The e-shirt approach is to route all the signals to a base station, with wires also embedded on the shirt, and then use a wireless link to send all the data to the central station. But the use of many wires is prone to system failure, leads to less wearable capability and brings problems for system cleaning. One better approach is to use sensing modules for each desired signal to record the information and use a wireless link to send the data to the central station. Fig. 18 shows the concept, where the purpose is to have an electrode with all the electronics integrated, from signal acquisition to the wireless link.

Figure 18. Wearable electrode for data acquisition.

The two applications described before are becoming more and more realistic than ever. An emerging new field of research that combines the strengths and capabilities of electronics and textiles into one: electronic textiles, or e-textiles are opening new opportunities. Etextiles, also called smart fabrics, not only have wearable capabilities like any other garment, but also enable/provide local monitoring and computation, as well as wireless communication capabilities. Sensors and simple computational elements are embedded in e-textiles, as well as

Antennas for Wireless Biomedical Devices

27

built into yarns, with the goal of gathering sensitive information, monitoring vital statistics, and sending them remotely (possibly over a wireless channel) for further processing [39]. However, before full integration of antennas with wearable biomedical devices becomes a reality, many problems must be solved, from antenna integration with textiles to proper antenna design to meet the requirements of wearable devices. And what is the problem of designing antennas to operate close, or inside, the human body? The radiation and propagation of electromagnetic waves can be described using the Maxwell equations. Using the differential form, it can be written [22]:

∇ × E(r , t ) = −

∂ B(r , t ) ∂t

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

∇ × H(r, t ) = J (r, t ) +

∂ D(r, t ) ∂t

(8)

(9)

∇ ⋅ D(r, t ) = ρ (r, t )

(10)

∇ ⋅ B(r, t ) = 0

(11)

where E(x, y, z, t) is the electric field intensity (V/m), H(x, y, z, t) is the magnetic field intensity (A/m), D(x, y, z, t) is the electric flux density (C/m2), B(x, y, z, t) is the magnetic flux density (T = Wb/m2), J(x, y, z, t) is the current density (A/m2) and r(x, y, z, t) is electric charge density (C/m3). Looking to these equations, it can be observed that the radiation close or inside the human body faces certain problems. The radiation is supported by an electromagnetic wave, this is, and electric component and a magnetic component. However, human tissues show a significant conductibility (up to 10 S/m at 10 GHz), which means that they do not support an electric field very well. In this way, an electromagnetic wave propagating inside the human body will suffer a severe attenuation. The radiation close the human body will be also perturbed, similarly as if it was close to perfect ground plane. Moreover, the design of antennas will be also difficult because the human body electric permittivity is very high, 40 to 60, and, e.g., the design by simulation may be very time consuming. Fig. 19 shows two possible options to integrate the antenna with a wearable device. The first option (with a cross on it) is to use a monopole antenna, where the human body is perpendicular to the antenna and acts as the ground plane. This solution is not very desirable because it is difficult to maintain the antenna properly positioned, and, more important, it is not fashionable. No one will appreciate wearing a device with pillars coming out of it. The second solution represents the option of using the dipole (or a patch, or any other type of planar antenna) parallel to the ground plane (human body). This is a more desirable solution because, from textile point of view, it is easier to integrate this type of antenna. However, when the antenna is integrated in a wearable electrode, as described previously, corresponding to the third situation represented in the figure. The antenna starts to be very close to the human body and the radiation properties become very poor. Image theory (image antennas are also shown) can be used to arrive to this conclusion or simply to think that in the

28

C.P. Figueiredo and P.M. Mendes

limit the antenna is coincident with the ground plane, leading to no radiation. Therefore, the closest to the human body, the worst is the radiation ability.

P(x, y, z)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 19. Antenna radiation close to the human body.

To reduce the effect of human body proximity, antenna design solutions based on EBG substrates or metamaterials may be used. These materials have the ability to resonate at a specific frequency and if they are designed coincident with the antenna operating frequency, they may be candidates to improve the radiating properties. Another problem to integrate the antennas with small recording electrodes is the fact that such electrodes must provide an electrical interface with the human body. In this way, they have an electrical resistance in the order of kΩ, being far from a good dielectric. The other problem is that to design antennas on such electrode material, acting as a substrate, will require prior characterization of material properties at the operating frequencies. After solving all technical problems of antenna integration with the wearable devices, one challenge to be solved still remains that must be well understood, which is the effect that many radiating devices will have on the biopotentials that are being measured. The radiating antennas must not interfere with the biopotential that is being measured (may be on the order of μV), requiring the antennas to be designed accordingly. This problem can be more significant when using low frequency radiating elements, because the human electrical signals are all low frequency signals. Finally, there is also the safety point of view. It is not desirable to have an antenna radiating towards the human body, situation that can be avoided only when considering devices outside the human body. For devices inside the human body, the radiation levels should be kept as small as possible.

Antennas for Wireless Biomedical Devices

29

Interference with Bio-Potentials

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The use of RF links to establish wireless communications, or to be used on localization devices, operating very close to the human body will create extra electromagnetic fields that will give origin to electric fields on the human body [40]. When using high frequency signals the potential to interfere with the bioelectric signals is reduced. However, when low frequency signals are considered, it is necessary to analyze if there is any possibility for interference. As it is well known, the power lines are a major source of interference for bioelectric signals, since the 50 Hz or 60 Hz fall inside the frequency range of many bioelectric signals (Table 1). The other sources of interference are the so called artifacts. Those are bioelectric signals that interfere with the signal to be measured, due to the different amplitude ranges. Since EMG and ECG have higher amplitudes, they can interfere with the EEG recording. Now, the electric signals originated by the RF links should be also considered. When using a low frequency signal on a localization device used for, e.g., BCI, that signal may fall inside the desired frequency range. Nowadays, the highest frequencies of interest for a biopotencial are in the kHz range. In this way, before using localization devices based on links operating at these frequencies, it is required to understood what is the transmitted power that will give raise to a significant interference. Despite the biopotential frequencies of interest are limited to a few kHz, this does not mean that no problems do exist for higher frequencies. Typically, the instrumentation used to record the biopotentials has a bandwidth that can fall in the MHz range. In this way, the RF links may cause saturation in the instrumentation amplifier, if some power limit is exceeded, turning any filtering to remove those components from the recorded signals useless. When using devices radiating inside the human body, care must also be taken to avoid the interference from the RF link on the recorded biopotential, and, since the device could be operating very close to electrically active tissue, care should also be taken to avoid the trigger, or inhibition, of action potentials. To avoid the interference with biopotentials, the antennas for devices operating close to the human body should radiate only in directions opposite to the human body, and for devices operating inside the human body, the antennas must also radiate in directions opposite to the recording spots and in such a way the wireless path does not include electrically active tissue.

WIRELESS BIODEVICES INSIDE THE HUMAN BODY The most challenging wireless biomedical devices are the ones required to operate inside the human body, as already discussed in the previous section. Moreover, there are two distinct operating situations. The implantable devices that are placed inside the human body for long periods of operation and the devices that are introduced through human cavities to collect data with a low degree of invasiveness. To overcome the associated problems, there are two distinct solutions to establish a wireless link to communicate with devices inside the human body. Fig. 20 shows the two most common approaches used to communicate with wireless devices placed inside the human body. The first option is based on a solution with two modules that are connected by

30

C.P. Figueiredo and P.M. Mendes

wires. One module is the antenna and the other is the sensor or actuator. The wires between them may be used to transport data signals only, or power signal also. The second solution uses only one module, where the antenna is part of the whole system. Transmission cables

RF link

RF link

Biodevice

Biodevice

Figure 20. Wireless links for operation inside the human body.

The first solution has the advantage of allowing the antenna placement very close to the human body surface. In this way, it is possible to deliver high values of power to the internal antenna since the losses are only due to the skin layer. However, it is not a highly desirable solution since the use of internal cables is prone to failure, leading to surgical interventions for repairing. Moreover, the use of cables may imply extra recovery time for the patient and higher risk for pos-surgical complications, leading to cost increase. Moreover, for temporary, or small period examination like endoscopic capsule, devices are introduced inside human body cavities. Consequently, it is not possible to bring the antenna close the human body surface. The solution is to attach an antenna to the biomedical device.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Implantable Wireless Biodevice Fig. 21 shows the commonly adopted system architecture to overcome the need of propagation inside the human body. The example shown is used to control the inferior urinary system [41]. The system has a signal generator that generates the appropriate stimulus to activate, e.g., the bladder. That stimulus is transmitted to the external coil, which induces the signal in the internal coil. Reaching the biologic environment, a receiver module delivers the stimulus through the transmission cables that carry the signal to the cuff electrode. Since the internal coil is placed in the frontal region and the electrodes are in the back, the transmission cables must go through the body and constitute the main cause of system failure. Moreover, the existence of these cables requires a small opening in the duramater, which is not desirable to maintain the spinal cord integrity. Instead of using this complex

Antennas for Wireless Biomedical Devices

31

system, it would be favorable to have one microsystem with all functionalities integrated that could be implanted in the desired region.

Signal generator

Inductors Cuff electrode

Transmission cables

Figure 21. Schematic view of an overall system used to communicate with devices inside the human body.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Low Frequency Wireless Link Antenna integration is a hard task to accomplish since it requires the multidisciplinary knowledge from antennas, microwaves, circuit design, and materials. Moreover, the on-chip antenna integration requires an electrically small antenna, due to wafer cost and devices size constrains, and operating on a substrate that was not initially intended for that purpose [42]. Despite its use to transmit data, or to power the device, it is necessary to have a wireless link operating at one frequency. As is well known, the human body shows higher attenuation to higher frequencies. This means that, the lower the frequency, the higher the power that may be received at the implant. Moreover, the attenuation is highly dependent on water content in tissues. The water content depends on the type of tissue and part of the body. In this way, the only way to evaluate the attenuation is to measure it. This was done using a pair of inductors (Fig 22) [43]. The calculation of inductance, coupling factor, and mutual inductance was made for a situation where it was predicted that an adequate level of signal could be transmitted in the absence of any tissue. To compute the human body attenuation, the setup was used to obtain the relation between the received signal when the inductors are separated by air and by a body phantom. It is intended to measure the human body attenuation in the implant region. To simulate the possible implant location, a tissue depth of about 4 cm can be assumed.

32

C.P. Figueiredo and P.M. Mendes Tissue (4 cm)

Signal generator

External coi ll

Internal coil Tissue (4 cm)

Display device

Figure 22. Schematic of the setup used for tissue characterization.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The problem in performing the measurements is the impossibility to perform them directly in the human body, except in very special situations. The solution was to use porcine tissue, with similar characteristics and thickness to perform the tests. The setup was used to measure the tissue attenuation and the effect of inductance misalignment. The inductor used to transmit the sinusoidal signal is shown in Fig. 23. The solenoid is placed on top of the porcine tissue, which is constituted by skin, muscle and fat. Inside the tissue shown in Fig. 23 is a box with a planar inductor that was used to receive the signal. Fig. 24 shows the used coil and how it was placed inside the porcine tissue.

Figure 23. Photograph of the setup, view from outside.

The box with the receiving inductor was covered, all around, by 4 cm of tissue. The values used for measurements are shown in Table 2.

Antennas for Wireless Biomedical Devices

33

Figure 24. Photograph of the setup, showing also the inside coil.

Table 2. Values used in the measurement setup

Number of Turns, N Diameter Inductance, L

External Coil 20 5 cm 18,74 μH

Internal Coil 10 3 cm 5,65μH

The following equation was used to compute the effect of tissue in the signal attenuation.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

% Attenuation =

VWithout _ tissue − VWith _ tissue VWithout _ tissue

*100

(12)

Fig. 25 shows the obtained results obtained for a frequency range from 110 kHz to 2.2 MHz. From last figure, it can be seen that tissue power attenuation is around 30% - 35%. Also, as expected, increasing the signal frequency results in a higher level of attenuation. The efficiency was also obtained at 10 MHz with a value close to 40%. Fig. 26 shows the signal attenuation versus the transmitting and receiving coils misalignment. This measure is important since, after implanting a microsystem, it is difficult to know exactly where it is, in case that it is necessary to communicate with it. From the above figure it can be observed that the raise in lateral misalignment induces a higher value of attenuation. This is due to two factors: − −

The misalignment reduces the efficiency of the inductive coupling; There is a thicker layer of tissue to cross before the signal reaches the receptor coil;

34

C.P. Figueiredo and P.M. Mendes

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 25. Measured tissue attenuation.

Figure 26. Measured effect of lateral misalignment between coils.

It is essential to assure a proper alignment between coils or the efficiency of radiofrequency connection will be very poor, requiring higher power needs. The use of a carrier frequency of 13,56 MHz, an unlicensed band (ISM), is recommended, since it is not severely attenuated by the adipose and muscular tissue. The implant should be as small as possible, but not excessively small, because, in this case, the strength and reliability of the implanted device would be jeopardized. The problem of using this carrier frequency is on how to design an antenna small enough to fit inside a microsystem. The next sections will show a possible solution.

Antennas for Wireless Biomedical Devices

35

Low Frequency MEMS Antenna Micro-Electro-Mechanical Systems (MEMS) are becoming an available option for RF communication systems since they can offer, simultaneously, devices with improved performance and they use IC-compatible materials, allowing their integration in a silicon chip, side by side with semiconductor circuits. Up to now, MEMS have been used for antenna applications to obtain non-conventional front-ends with improved, or new characteristics. However, some preliminary tests have shown that some MEMS structures could have the ability to operate as an antenna itself and this solution would have the potential to be smaller than conventional antennas. The basic principle of micromachined cantilevers offers an interesting possibility to measure a variety of physical parameters [44]. When used as a sensor, a MEMS structure requires the use of a sensing mechanism and the most widely used is the capacitive method. The moving structure, and a fixed plate, forms a parallel plate capacitor, where the structure movement is translated into a capacity change. The principle of operation is presented in Fig. 27. The Lorentz-force is the mechanism being used to sense the incoming electromagnetic waves. In this simple sensing device, the current carrying bar is used to detect a time-varying magnetic field. To measure the magnetic fields, the Lorentz-force applied on a current carrying lead is used as a sensor. L

u

I

X B

FL y

k

x

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 27. Simplified model of a mechanical device working as an antenna.

A device of this type measures only the magnetic flux density in the direction perpendicular to the current carrying bar, i. e., the out-of-the-plane direction in Fig. 27. The Lorentz-force acting on the bar is used to move the equivalent mechanical spring. Deflections, which are small compared to the length of the cantilever, are a directly proportional measure of the applied force. To reach the highest possible sensitivity it is advisable to use a resonant mechanism, where the bar is excited by an AC current with a frequency equal to an eigenfrequency of the elastic structure. Due to the high quality factors of Si structures, which are at least several hundred, this is an efficient way to enhance the sensitivity. The mechanical displacement can be converted using an optical, capacitive, or piezoelectric transducer [45]. The most attractive options are capacitive and piezoelectric since these solutions are easily available in MEMS technology and have the potential for low

36

C.P. Figueiredo and P.M. Mendes

power consumption. Since the desirable displacement depends on structure dimensions and material properties, capacitive sensing can be used as the sensing mechanism for MEMS micro-antennas. However, if large displacements are required, or if the MEMS structure area becomes too small for capacitive detection, the use of a piezoelectric material can be the solution since it can act both as sensor and actuator. Moreover, it produces a voltage in response to a deflection, leading to simple readout electronics. Despite the highly desired detection mechanism based on the electroactive mechanism, to test the potential of the envisage antenna, the device must be first designed for fabrication on an already available process. Fig. 28 shows a proposed electrically small MEMS antenna [46]. The structure was designed to provide enough sensitivity to low power magnetic fields [47].

Figure 28. Proposed MEMS antenna.

Mems Antenna Modelling The operation of the previous structure as an antenna implies that it must be able to receive a modulated signal. The structure proposed, can be modelled using the simplified model of Fig. 27. The electrical behaviour is modelled by the beam length L and the current I, whereas the mechanical behaviour can be described using the spring constant k. Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

r

Considering that a magnetic field, B , is applied perpendicularly to the current I, through

r

the length L, the resulting Lorentz force, FL , will be given by:

r r r FL = ( B × I ) L

(13)

The maximum force, which means maximum deflection, will be obtained when the magnetic field is applied perpendicularly to the plane containing the MEMS device.

r

Due to the applied force FL , the structure will move and the following equation can be written [48]:

Antennas for Wireless Biomedical Devices

FL = (B × I )L = ku = Felast

37 (14)

where Felast stands for elastic force, u for displacement and k is the spring constant of the structure. The MEMS structure will move until the elastic force equals the Lorentz force. From the previous equation, the displacement can be calculated as:

u=

(B × I ) L

(15)

k

Independently of the detection method, piezoelectric or capacitive, the displacement will give origin to a received voltage, VR, which will be proportional to the applied magnetic field, B, and so:

VR ∝ u ∝ B

(16)

If the magnetic field is originated by a modulated signal:

B(t ) = A(t )cos(2π ft + φ (t ))

(17)

the received voltage is:

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

VR (t ) = G

LI A(t )cos(2π ft + φ (t )) k

(18)

where G is the system gain. From (6) it is possible to conclude that the proposed device will behave as a signal receiver if the spectral properties of the modulated signal are chosen to fall inside the device operating frequency limit. Even if a modulated signal can be detected, a few other aspects need clarification. A parameter requiring analysis is the force that can be produced to move the structure in the presence of a magnetic field. Using (2), and considering I = 10 mA, L = 1,5 mm and B = 10 mT, the resulting elastic force would be 0.15 μN. This is enough force to produce movement in this kind of structure. A key advantage of this solution is that it can deliver gain through/due to the increase of the current I. If the resistivity of the beam material is kept low, the voltage drop will be small resulting in a low power device.

CONCLUSIONS Throughout this chapter the problematic of wireless communications for biomedical devices was discussed. Two situations where wireless communications can improve the device functionality were presented. The possibility of antenna integration within wearable devices was analyzed and the challenges of designing such devices were introduced. Besides the technological problem of placing an antenna on a wearable device, it is also difficult to have antennas radiating very

38

C.P. Figueiredo and P.M. Mendes

close to the human body. The antennas tend to not radiate, since they are very close to a ground plane, and the radiating fields may start interfering with the recorded biopotentials. The integration of antennas in implantable wireless devices requires the development of new antenna solutions to operate at low frequencies. One MEMS antenna, with the ability to operate in the kHz range and potential for operation in the MHz range was presented. The presented antenna was fabricated on a commercial fabrication process and can be integrated with the remaining circuitry. To design antennas to operate on or inside the human body is a challenge, which requires several relevant factors to be taken into account. The main challenge is to design as small and as integrated as possible, without degrading antenna radiation properties.

ACKNOWLEDGEMENT This work was supported by FCT - SFRH/BD/40341/2007 grants.

REFERENCES

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

[1] [2]

J. Malmivuo, R. Plonsey. Bioelectromagnetism, Oxford University Press, 1995. J. W. Clark Jr., “The Origin of Biopotentials”. In J. G. Webster, editor, Medical Instrumentation, John Wiley, New York, pp. 121-182, 1998. [3] R. C. Barr, “Basic Electrophysiology”. In J. D. Bronzino, editor, The Biomedical Engineering Handbook, CRC Press LLC, Boca Raton, pp. 101-118, 2000. [4] N. V. Thakor, “Biopotentials and Electrophysiology Measurement”. In J. G. Webster, editor, Measurement, Instrumentation and Sensors Handbook, CRC Press LLC, Boca Raton, chapter 74, 1999. [5] A. Cohen, “Biomedical Signals: Origin and Dynamic Characteristics; FrequencyDomain Analisys”. In J. D. Bronzino, editor, The Biomedical Engineering Handbook: Second Edition, pp. 805-827. CRC Press LLC, Boca Raton, 2000. [6] J. D. Bronzino. “Basic Electrophysiology”. In J. D. Bronzino, editor, The Biomedical Engineering Handbook. CRC Press LLC, Boca Raton, pp.201-212, 2000. [7] H. H. Jasper, “The ten-twenty electrode system of the International Federation”, Electroencephalography and Clinical Neurophysiology, Vol. 10, pp. 371-375, 1958. [8] F. N. Wilson, F. D. Johnston, A. G. Macleod, P. S. Barker, “Electrocardiograms that represent the potential variations of a single electrode”, Am. Heart J., Vol. 9, pp. 447471, 1934. [9] E. Goldberger, “A simple, indifferent, electrocardiographic electrode of zero potential and a technique of obtaining augmented, unipolar, extremity leads”, Am Heart J ., Vol. 23, pp. 483-492, 1942. [10] F. N. Wilson, F. D. Johnston, F. F. Rosenbaum, H. Erlanger, C. E. Kossmann, H. Hecht, N. Cotrim, R. Menezes de Olivieira, R. Scarsi, P. S. Barker, “The Precordial Electrocardiogram”, Am. Heart J., Vol. 27, pp. 19-85, 1944. [11] E. J. Berbari. “Principles of Electrocardiography”. In J. D. Bronzino, editor, The Biomedical Engineering Handbook, CRC Press LLC, Boca Raton, pp.181-190, 2000.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Antennas for Wireless Biomedical Devices

39

[12] L. T. Mainardi, A. M. Bianchi, S. Cerutti. “Digital Biomedical Signal Acquisition and Processing”. In J. D. Bronzino, editor, The Biomedical Engineering Handbook, CRC Press LLC, Boca Raton, pp. 828-852, 2000. [13] M. R. Newman. “Biopotential Amplifiers”. In J. G. Webster, editor, Medical Instrumentation, John Wiley, New York, chapter pp. 233-286, 1998. [14] E.B. Loewenstein, “Analog-to-Digital Converters”. In J.G. Webster, editor, Electrical Measurement, Signal Processing, and Displays, CRC Press LLC, Boca Raton, chapter 25, 2006. [15] M. R. Newman. “Biopotential Electrodes”. In J. G. Webster, editor, Medical Instrumentation, John Wiley, New York, pp. 183-232, 1998. [16] http://www.glennan.org/content/psindex.html. [17] http://www.zigbee.org/. [18] Bluetooth Specification, Version 2.0, Vol. 0-3, November 2004. [19] Ronald Kitchen, RF Radiation Safety Handbook, Butterworth-Heinemann, 1993. [20] Rao R. Tummala, “SOP: What Is It and Why? A New Microsystem-Integration Technology Paradigm-Moore’s Law for System Integration of Miniaturized Convergent Systems of the Next Decade,” IEEE Transactions On Advanced Packaging, Vol. 27, No. 2, pp. 241-249, May 2004. [21] M. Bartek, A. Polyakov, S.M. Sinaga, P.M. Mendes, J.H. Correia, J.N. Burghartz, Characterization of High-Resistivity Polycrystalline Silicon Substrates for Wafer-Level Packaging and Integration of RF Passives, ASDAM 2004, pp. 277-230, 17-21 October 2004, Smolenice Castle, Slovakia. [22] C. A. Balanis, Antenna Theory: Analysis and Design, Second Edition, John Wiley, 1997. [23] H. A. Wheeler, “Fundamental limitations of small antennas”, Proc. IEE, Vol. 35, pp 1479-1484, December 1947. [24] L. J Chu, “Physical Limitations of Omnidirectional Antennas”, Technical report nº 64, Research Laboratory of Electronics, MIT, May 1948. [25] E. Kyriacou et al., “e-Health e-Emergency Systems: Current Status and Future Directions”, IEEE Antennas & Propagation Mag., 49(1), 2007. [26] J. Hightower, G. Boridello, “Location Systems for Ubiquitous Computing”, Computer, Vol. 34, no. 8, pp. 57-66, Aug. 2001. [27] K. Langendoen, N. Reijers, “Distributed Localization Algorithms“. In R. Zurawski, editor, Embedded Systems Handbook, CRC Press, Boca Raton, chapter 36, 2006. [28] G. Mao, B. Fidan, B.D.O. Anderson, “Wireless sensor network localization techniques”, Comput. Netw., Vol. 51, no. 10, pp.2529-2553, 2007. [29] L. M. P. Leão de Brito, L. M. R. Peralta, “Collaborative Localization in Wireless Sensor Networks”, SENSORCOMM 2007, Valencia, Spain, pp. 94-100, Oct. 2007. [30] J. Bachrach, C. Taylor, “Localization in Sensor Networks”. In I. Stojmenović, editor, Handbook of Sensor Networks: Algorithms and Architectures, Wiley-Interscience, chapter 9, 2005. [31] R. Virrankoski, “Localization in Ad-Hoc Sensor Networks”, Postgraduate Seminar Report, Control Engineering Laboratory, Helsinki University of Technology, 2003. [32] I. Borg, P. Groenen, Modern Mutlidimensional Scaling, Theory and Applications. Springer-Verlag, New York, 1997.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

40

C.P. Figueiredo and P.M. Mendes

[33] Y. Shang, W. Ruml, Y. Zhang and M. Fromherz, “Localization From Mere Connectivity”, MobiHoc'03, Annapolis, Maryland, pp. 201-212, June 2003. [34] Y. Shang, W. Ruml, “Improved MDS-Based Localization”, INFOCOMM 2004, Hong Kong, Vol. 4, pp. 2640-2651, March 2004. [35] S. Čapkun, M. Hamdi, J.-P. Hubaux, “GPS-Free positioning in mobile and ad hoc networks”, HICSS-34, Maui, Hawaii, pp. 3481-3490, Jan. 2001. [36] A. Magani, K. K. Leung, “Self-Organized, Scalable GPS-Free Localization of Wireless Sensors”, WCNC 2007, Hong Kong, pp. 3801-3806, Mar. 2007. [37] Y. Shang, H. Shi, A.A. Ahmed, “Performance Study of Localization Methods for AdHoc Sensor Networks”, IEEE Conference on Mobile Ad-hoc and Sensor Systems, Fort Lauderdale, Florida, pp.184-193, Oct. 2004. [38] R.Wolpaw, D.J.McFarland, and T.M.Vaughan, "Brain–Computer Interface Research at the Wadsworth Center," IEEE Transactions on Rehabilitation Engineering, Vol. 8, No. 2, pp. 222-226, June 2000. [39] Marculescu, D., et al., “Electronic Textiles: A Platform for Pervasive Computing,” Proceedings of the IEEE, Vol. 91, No. 12, pp. 1995-2018, 2003. [40] Ronold W. P. King, “Electric Current and Electric Field Induced in the Human Body When Exposed to an Incident Electric Field Near the Resonant Frequency,” IEEE Transactions on Microwave Theory And Techniques, Vol. 48, No. 9, September 2000. [41] T. Rua, Paulo Vale, P. M. Mendes, "Implantable Wireless Microsystem for Physiological Functions Control," ISIE 2007 2007 IEEE International Symposium on Industrial Electronics, June 4-7, Vigo, Spain, 2007. [42] P. M. Mendes, J. H. Correia, "MEMS Micro-Antennas for Wireless Biomedical System," Book Chapter in Wireless Communications Research Trends, Nova Publisher, ISBN: 1-60021-674-9, 2007. [43] Jordi Parramon Piella, Energy management, wireless and system solutions for highly intregated implantable devices, PhD Thesis, Universitat Autónoma de Barcelona, Barcelona, 2001. [44] Lange, D., Brand, O., and Baltes, H., CMOS Cantilever Sensor Systems: Atomic Force Microscopy and Gas Sensing Applications, Springer, 2002. [45] S.D Senturia, Microsystem Design, Kluwer Academic Publishers, 2001. [46] L. A. Rocha, P. M. Mendes, "Electrically Small MEMS Antenna for Wireless Biomedical Microsystems", 38th European Microwave Conference 2008 (EuMC), Amsterdam, 28-29 October 2008. [47] L. A. Rocha, P. M. Mendes, "A SOI Microfabricated Antenna", 22nd International Conference EUROSENSORS, Germany, Dresden, 07-10 September 2008. [48] Rocha, L. A., Cretu, E., Wolffenbuttel, R. F., “Analysis and Analytical Modeling of Static Pull-In With Application to MEMS-Based Voltage Reference and Process Monitoring,” Journal Of Microelectromechanical Systems, Vol. 13, No. 2, pp. 342-354, 2004.

In: Antennas:Parameters, Models and Applications Editor: Albert I. Ferrero

ISBN 978-1-60692-463-1 ©2009 Nova Science Publishers, Inc.

Chapter 2

ANTENNA-LANTHANIDE COMPLEXES: A GROWING TECHNOLOGY-DRIVEN RESEARCH Silvio Quici,a Lidia Armelao,b Francesco Barigelletti,c Marco Cavazzini,a Gregorio Bottarod and Gianluca Accorsi.c a

Istituto di Scienze e Tecnologie Molecolari (ISTM), Consiglio Nazionale delle Ricerche (CNR), Via C. Golgi 19, I-20133 Milano, Italy. b Istituto di Scienze e Tecnologie Molecolari (ISTM), Consiglio Nazionale delle Ricerche (CNR) e INSTM, Dipartimento di Scienze Chimiche, Università di Padova, Via Marzolo 1, I-35131 Padova, Italy. c Istituto per la Sintesi Organica e Fotoreattività (ISOF), Consiglio Nazionale delle Ricerche (CNR), Via P. Gobetti 101, I-40129 Bologna, Italy. d Istituto di Metodologie Inorganiche e dei Plasmi (IMIP), Consiglio Nazionale delle Ricerche (CNR) e INSTM, Dipartimento di Chimica, Università di Bari, Via Orabona 4, I-70126 Bari, Italy

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ABSTRACT Luminescent and stable lanthanide ions (Ln3+) complexes are of great interest because of their unique photophysical properties especially with respect to the generation and amplification of light. The emission properties of these complexes are notable and cover an exceptionally wide spectral range: near infrared (Yb3+, Nd3+, Er3+), orange (Sm3+), red (Eu3+), yellow (Dy3+), green (Tb3+) and blue (Tm3+). In our technology driven lives lanthanide complexes are ubiquitous. Indeed they find application as active components in many different kind of advanced materials and devices such as: diagnostic tools, sensors, optical fibers, lasers and amplifiers, electroluminescent and magnetic molecular materials among others. Unfortunately due to the very low absorption coefficients (1 < ε < 10 M-1 cm-1) the emissive states of the metal ions cannot be efficiently populated by direct excitation. A suitable way to overcome this problem is to complex the Ln3+ ion with a ligand containing a highly absorbing chromophore which promotes the metal centred (MC) emission through a sensitization process. In this process the excitation energy absorbed by the chromophore is transferred to the metal core thus efficiently populating the

42

Silvio Quici, Lidia Armelao, Francesco Barigelletti et al. emissive states of the latter and finally the metal centred light emission is obtained (antenna effect). This review will focus on the metal centred emission properties of Ln3+ complexes either in solution or anchored into inorganic matrices. Particular attention will be paid to: (i) the description of the photophysical properties of Ln3+ ions that are relevant for the optimization of the sensitized metal centred light emission; (ii) the coordination properties of Ln3+ ions and the parameters ruling the design of ligands affording complexes with high kinetic and thermodynamic stabilities together with a complete shielding of the coordinated metal ion; (iii) the introduction of suitable reactive groups that allow the covalent insertion of the Ln3+ complex into inorganic and/or organic matrices to afford emissive materials and devices. Some examples of applications of luminescent Ln3+ complexes will be selected from the recent literature and critically reviewed.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

INTRODUCTION Stable complexes of luminescent trivalent lanthanide ions have been since long under scrutiny in view of important applications [1], which are based on the exploitation of their magnetic [2-5] or luminescence features [6-9]. In particular, when the Ln(III) complexes exhibit intense luminescence, applications ranging from biomedical [1] to sensing areas [10] and luminescence imaging [5,10] become possible. In most cases, the complexes of Eu(III) and Tb(III) have been studied because of their long-lived (ms timescale) and visible line-like emission. The photophysical properties of near infrared (NIR) lanthanide emitters such as Sm(III), Dy(III), Pr(III), Ho(III), Yb(III), Nd(III), and Er(III) have been less well investigated [11-15]. The interest in these complexes stems from their possible use in biomedical and telecommunication fields and for various photonic applications [16-22]. Actually, given that longer-wavelength emissions are more efficient to penetrate the human tissue than visible light, convenient medical diagnostic procedures can be conceived based on long-wave emitters. Similarly, NIR luminescence from ions such as Nd(III), Yb(III) and Er(III) proves very useful when employed in telecommunication network optical signal amplifiers [23,24]. Regarding Nd(III) species, since long it has found applications within laser systems [25,26]. Likewise, useful lasing properties are observed also for other Ln(III) centres [27]. Various lighting applications are also conceivable by using those Ln(III) centers that emit in the Vis region. These are Tb(III), Eu(III), Dy(III), and Sm(III) that, coupled with suitable antennas, can be incorporated in stable and transparent inorganic hosts, among others silica layers [28,29] and will be discussed in the following in some details. A schematic representation of the type of emission and the major fields of application of lanthanides is evidenced in Figure 1. In this review we describe the basic principles of ligands design consisting of a coordination site for the metal cation bearing covalently bonded suitable chromophores as sensitizers (two-component approach) that are capable of forming highly luminescent (overall sensitization yield, φse > 0.05) lanthanide complexes. The photophysical properties of these complexes together with the description of some emitting materials prepared in our group will be discussed in detail. In particular we focus the attention toward those complexes emitting in

Antenna-Lanthanide Complexes: A Growing Technology-Driven Research

43

the visible region that can be used in the lighting industry (e.g. for the preparation of photoand electro-luminescent materials) and for biological immunoassays.

Figure 1. Type of emission and related applications of lanthanides.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ELECTRONIC STRUCTURE AND COORDINATION PROPERTIES OF LANTHANIDE CATIONS Lanthanoides are the fifteen elements from lanthanum to lutetium (atomic number 5771) in the Mendeleiev’s periodic table, the electronic configuration being [Xe]f05d16s2 for La and [Xe]4f145d16s2 for Lu. The fourteen 4f electrons are added with some irregularities in the case of the atomic electronic configuration, while the progression is perfectly regular from 4f1 to 4f14 in the Ln3+ ions (Ln is the generic symbol of lanthanide elements). This particular electronic configuration is responsible for the constancy of lanthanide physical-chemical properties such as the oxidation state, the redox potentials and the ionic radii [30]. Indeed they show a very similar chemistry which makes their separation very complicate [31]. The lanthanide metals, and the lanthanum, are all relatively electropositive metals that strongly, although not exclusively, exhibit the +3 oxidation state. The potential for reduction Ln3+/Ln0 is around -2.3 V vs NHE (normal hydrogen electrode). Nevertheless other oxidation states have been observed namely +4 for Ce and, to a much lesser extent, Pr and Tb and +2 for Sm, Eu and Yb ions. These deviations from the generality of +3 oxidation state can be attributed to the special stability of empty, half-filled or completely filled f orbitals, i.e. Ce4+(4f0), Eu2+(4f7), Yb2+(4f14) although Pr4+(4f1) and Sm2+(4f6) do not fully satisfy this criterion [32]. The ionic radii of Ln3+ steadily decrease within the lanthanide series from La (1.17 Å) to Lu (1.00 Å) by an overall amount of 0.17 Å that is generally known as “lanthanide contraction” [33,34]. This is due in part to the increase of nuclear charge, that makes the electron cloud constantly shrinking as the 4f shell is filled and to relativistic effects that consequently become significant. The 4f electrons are shielded by the 5s2 and 5p6 orbitals and therefore they are not available for covalent interaction with the ligands. Hence interactions are largely of electrostatic nature and the geometry of the Ln3+ complexes is determined by steric factors

44

Silvio Quici, Lidia Armelao, Francesco Barigelletti et al.

rather than electronic ones [35]. As a consequence, the Ln3+ complexes of the same ligand are all isostructural. Due to their small size, Ln3+ ions have a high surface positive charge density so that they behave as hard Lewis acids. Accordingly, they strongly coordinate ligands having highly electronegative donor sites (hard Lewis bases), in the order: F- > HO- > H2O >NO3- > Cl[36]. The coordination numbers of Ln3+ are in the range 3-12 depending on the steric demand of the ligand, with 8 and 9 as the most frequently observed. Aqua ions found in crystalline compounds are generally 9 coordinate with the tricapped trigonal prism being the favoured structure [37]. The [Ln(H2O)n]3+ ions in aqueous solution are either 8 or 9 coordinate but this may vary with ionic strength and concentration [38]. The coordination number, in 1M perchlorate solution, appears to change from 9 for larger (earlier elements of the series) Ln3+ ions to 8 for smaller ones (at the end of the series). The elements in the middle such as Sm3+ show either 8 or 9 water molecules in the first coordination sphere [39,40]. Very low coordination numbers, e.g. 3 or 4, have been observed with bulk ligands such as bis(trimethylsilyl)amide or bis(isopropyl)amide [41,42]. Altough lanthanides are often referred to as rare earth elements they are not scarce. Lanthanum, cerium and neodymium are more abundant on earth’s crust than lead and over 100 minerals containing them have been described. The most common raw materials used for extraction of rare earths are bastnasite, monazite and xenotime. Europium is present at about 0.1% or less in these ores and is present in about 1 ppm in the earth’s crust. Promethium, which is radioactive, does not occur in nature, and was first made by man in 1945 [43].

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Spectroscopy and Photophysics in Antenna-Lanthanide Cations The electronic spectra of lanthanide-doped single crystals and lanthanide salts are usually interpreted according to the Russel-Saunders coupling scheme [32]. This in principle should be strictly valid for light atoms whereas the lanthanides actually are not, given that they display moderately large spin-orbit coupling constants, e.g. ζ = 556 and 1153 cm-1 for lantanum and lutetium respectively [44]. At any rate, the luminescence spectra show groups of narrow lines ascribed to transitions inside the 4f shell, that in solution can somewhat broaden in bands. Each small line within a group corresponds to a transition between two 2S+1LJ free ion levels (J-manifold); 2S+1 is the spin multiplicity, L is the total angular moment, and J = |L-S|. Within the Russel-Saunders coupling scheme, for L = 0, 1, 2, 3, 4, 5, and 6 the spectroscopic terms are labeled S, P, D, F, G, H, and I. In most cases, the lanthanide ions have ground states with a well-defined value of J, with the next J state not thermally accessible at room temperature. Table 1 lists properties of the lanthanide cations we are more interested here in view of the optical applications discussed later.

Antenna-Lanthanide Complexes: A Growing Technology-Driven Research

45

Table 1. Properties of selected lanthanide cationsa Symbol

Atomic number

Sm Eu Gd Tb Dy a

Electronic configuration (outside the [Xe] shell) Atom

M3+

62

4f66s2

4f5

63

7

4f 6s

64

7

4f 5d 6s

65

9

4f 6s

2

66

10

2

2

1

4f 6s

2

4f

6

4f

7

4f

8

4f

9

Radius Å

1.10

Electronic ground levelb 6

H5/2

7

1.09 1.08

F0

8

S7/2

7

1.06 1.05

F6

6

H15/2

b

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Ref [32]. Electronic transitions are intraconfigurational in character and are in principle only allowed between (2S+1)LJ levels for ΔL = ±1,0 or ΔJ = ±1,0.

Figure 2. The antenna effect for sensitization of the luminescence in some lanthanide cations (selected levels are displayed); blue arrows indicate non-radiative processes, red arrows indicate radiative processes. As exemplified here for the Tb(III), Eu(III), and Gd(III) centres, three cases are encountered with regard to energy transfer (en) as regulated by the energy gap, ΔE, between the triplet level (T) of the chromophore (L) and the emitting level of the cation: a) when ΔE ≤ 1500 cm-1 back energy transfer takes place and, as a consequence, O2-effects (particularly intense in solution) are observed, see text; b) when ΔE ≥ 1500 cm-1 energy transfer is complete and c) when energy transfer is exhothermic and does not take place.

A widespread approach to develop lanthanide complexes is based on the fact that the Ln(III) centers cannot efficiently absorb light, due to their forbidden intra-4f transitions (the extinction coefficient for such trivalent cations is ε ≈ 1-10 M–1 cm–1) [45]. Thus, the way of choice to sensitize the luminescence of these cations is to employ chromophores (L) as antennas for light absorption (for most organic chromophores in the UV region, ε ≈ 104-105 M–1 cm–1) [44,46].

46

Silvio Quici, Lidia Armelao, Francesco Barigelletti et al.

Accordingly, sequential sensitization steps include: (i) population of the lowest-lying singlet excited state (S) of the organic chromophore, (ii) subsequent intersystem crossing (ISC) to its triplet level (T) and (iii) energy transfer (en) to the Ln(III) centre. The sensitization process can also be performed by chromophores from the vast area of the transition metal complexes [4,47-53]. An outline for the sensitization steps is provided in Figure 2 [54-56]. The overall efficiency of Ln(III) sensitized emission (φse), consequent to the light absorption event, is therefore regulated by the intersystem-crossing efficiency (φISC), the energy-transfer efficiency (φen), and the intrinsic, metal centred (MC) luminescence quantum yield of the Ln(III) ion (φlumMC), eq. (1a).

φ se = φ ISC φ en φ lum MC

φ lum

MC

=

kr kr + knr

(1a)

(1b)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

In water, in order to inhibit non-radiative deactivation via interaction with OH oscillators, the 8-9 coordination positions of Ln(III) must be conveniently shielded against the intervention of solvent molecules [57-60], eq. (1b); kr and knr are radiative and non-radiative rate constants, respectively. In this regard, comparison of luminescence results as obtained in water and deuterated water solutions allows the assessment of water binding. For Eu(III) and Tb(III) centers, this can be done by using the following equations [58,61]

qEu = 1.2 (1/τ H2O – 1/τ D2O – 0.25)

(2a)

qTb = 4.2 (1/τ H2O – 1/τ D2O – 0.06)

(2b)

where q (uncertainty ± 0.5) is the number of coordinated water molecules and lifetimes (τ) are in ms. Thus, the ligand system selected for the building up of the complexes must meet a number of requirements. These are related both to the needed structural features (complex stability and saturation of the coordination positions) and to the fact that, for maximizing the emission intensity, each of the three steps involved in the sensitization event must be optimized, eq. 1a.

Design of Ligands for Luminescent Lanthanide Complexes The electronic, magnetic and photophysical properties of lanthanide ion complexes strongly depend on the control of the coordination sphere of the metal. Accordingly, particular care should be paid to the design of ligands in order to optimize the properties of complexes.

Antenna-Lanthanide Complexes: A Growing Technology-Driven Research

47

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

General Criteria An exhaustive review that describes the general principle and the basic knowledge for the design of organic ligands for the complexation of alkali and alkali-earth cations by JeanMarie Lehn dates back to 1973 [62]. In this paper all the parameters that should be considered in the design of ligands in order to achieve control over chemical, structural and thermodynamical properties of the complex, which in the end rule its functionality, have been clearly evidenced. Among these parameters the ligand topology (dimensionality, connectivity, shape, size, chirality etc.), the binding sites (nature, electronic properties, number, shape, arrangement), the layer properties (rigidity/flexibility and the lipophilicity/hydrophilicity ratios, thickness), the environment properties and counterions effect play a major role. These concepts are general and apply to any kind of metal complex but they become particularly important in the case of lanthanide cations that, because of their electronic configuration, show a coordination behaviour very simple and substantially based on Van der Waals electrostatic interactions. Furthemore the complexation properties of lanthanide cations are very similar to those of alkali-earth cations in particular to Ca2+. In general the complex formation is the result of an attractive interaction between a ligand and a metal cation and is associated with their partial or total desolvation. In other words the coordination sites of the ligand interact with the surface of the metal cation thus replacing partially or totally the first solvation sphere. From the thermodynamic point of view this process affords an increased entropy due to the desolvation step, while the enthalpy variation can be either positive or negative according to the difference between the energy of the bonds formed (ligand-cation) and the energy of the bonds broken (ligand-solvent; cationsolvent). In the case of lanthanide cations complexation in aqueous solution, dehydration is an endothermic process (ΔH is positive) and represents an unfavourable energy contribution to the variation in Gibbs free energy so that the overall process is entropy driven. Therefore it is convenient to use polydentate ligands, which, because of the chelate effect, can overcome this difficulty by formation of highly stable complexes even when water is the solvent. In order to increase the thermodynamic stability of the complex, the Ln3+-ligand interaction has to be optimized. Indeed, due to their hard character, lanthanide cations show preference for the formation of chemical bonds having large electrostatic components so that anionic ligands containing oxygen- or nitrogen-donor sites such as aminocarboxylates or βdiketonates will be preferred. As it has been already stated, a lanthanide luminescent complex is a multicomponent system in which the active components, namely the metal cation, the antenna and the coordination site are organized in a supramolecular structure. Consequently the choice of these components and their positioning in the overall structure are issues to be considered during the molecular design step in order to optimize the overall sensitization efficiency.

Choice of the Lanthanide The emission properties of the lanthanide cations are notable and cover an exceptionally wide spectral range that span from the UV (Gd3+) to the visible: orange (Sm3+), red (Eu3+),

48

Silvio Quici, Lidia Armelao, Francesco Barigelletti et al.

yellow (Dy3+), green (Tb3+) and blue (Tm3+) to the NIR (Yb3+, Nd3+ and Er3+). A simplified energy diagram of the emissive levels of some lanthanide cations is reported in Figure 3, together with the energies of singlet (blue) and triplet (green) excited states of some commonly used chromophores. The lanthanides usually possess relatively long-lived excited states, which can undergo energy transfer to high frequency vibrational oscillators such as OH, NH and to a lower extent CH. The presence of these oscillators in the proximity of the metal favours thermal dissipation of the energy (vibronic coupling) and consequently quenching of the luminescence. The vibronic coupling is one of the main deactivation pathways and is also related to the emitting properties of the metal. Indeed with lanthanides emitting in the NIR the emission is strongly reduced and in some circumstances it cannot even be observed when high frequency oscillators are present in the coordination site. As a rule, the lower is the excited state energy of the lanthanide ion the more efficient will be the deactivation by vibronic coupling, which has been termed “energy gap rule” [63,64].

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 3. Energy diagram of emissive levels of some lanthanide cations and chromophores: singlet states and triplet states of chromophores are evidenced in blue and green respectively, while principal emitting states of some lanthanides are drawn in red.

Choice of the Antenna The chromophore (antenna) that promotes the sensitization of the lanthanide light emission process plays an important role in determining the quantum yield and the emission lifetime. In general the antenna can be any aromatic or hetero-aromatic highly π-conjugated system characterized by high efficiency of light absorption (high extinction coefficient ε) and high efficiencies of intersystem crossing and energy transfer steps. The efficiency of a chromophore to behave as sensitizer is strictly related to the energy of its triplet excited state (Figure 2), which should be at least 1500 cm-1 higher than the lowest emitting levels of the lanthanide ion [65]. When this energy gap is higher than this value the

Antenna-Lanthanide Complexes: A Growing Technology-Driven Research

49

energy transferred from the triplet can flow through non-radiative excited states of the metal until it reaches the emissive levels and the metal centred emission occurs. On the contrary a lower energy gap strongly limits the emission quantum yield because of thermal deactivation due to back energy transfer and O2-quenching towards the chromophore triplet level (Figure 2). The triplet energies of commonly used chromophores are reported in Table 2. Table 2. Energy levels of commonly used chromophores.a Chromophore 2,2’-Bipyridineb 1,10-Phenanthroline 2,2’:6’,2’’-Terpyridineb 2,9-Dimethyl-4,7-diphenyl1,10-phenanthroline c Acetophenone 1,4-Naphthoquinone 8-Hydroxyquinolined 7-Amino-4-methyl-2hydroxyquinoline (Carbostyril 124) Tetraazatriphenylene b Naphthalene

Singlet (cm-1) 29200

Triplet (cm-1) 22100 21200

28200

λmax, ε (nm, M-1cm-1) 287, ~29200 264, 33900 305, ~22200 274, 68500

26000 20200

~27000

305, 23100

29000 32200

24000 21200

340, ~4000 275, ~56000

Ref. [66] [44] [67]

[44] [44] [68] [44]

[69] [44]

For the solvents employed see the original works. b The luminescence properties of 2,2’bipyridine (bpy) and 2,2’:6’,2’’-terpyridine (tpy) free ligands, whose structure is far from being substantially rigid and close to planarity, are difficult to ascertain; on the other hand, for the huge amount of known complexes of these ligands, inspection of their LC singlet and triplet levels is difficult because of very fast deactivation towards lower lying levels of MLCT, MC and LMCT nature. As an approximation, the listed absorption data are drawn from UV absorption profiles of [Ru(bpy)3]2+ [66] and [Ru(tpy)2]2+ [67] respectively and are to be considered with some care. c Results from our laboratories. d The spectroscopic properties of free 8hydroxyquinoline are heavily affected by photoinduced photoisomerization and solvation effects that result in a very weak or no emission [68], while for the complexes the same comments as in footnote (b) are likely to hold [70]. Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

a

On the basis of these considerations it is unlikely that the same chromophore can be used in an efficient way for the sensitization of different lanthanide ions. This is further complicated because the use of apparently suitable ligands might result in non luminescent complexes due to population of non radiative excited states such as Ligand-to-Metal Charge Transfer (LMCT). Another important point regards the excitation wavelength of the antenna that should be above 330 nm in order to avoid the use of expensive quartz optics, for instance in the immunoassay applications. According to the position and nature of the chromophore the energy transfer process can occur either by Förster [71] or Dexter [72] mechanisms. In order to ensure fast energy transfer a short distance between sensitizer and lanthanide ion is obviously advantageous. Direct coordination of the antenna chromophore to the lanthanide ion has been mainly exploited with aza-aromatic compounds (bipyridine, phenanthroline, azatriphenylene, terpyridine)

50

Silvio Quici, Lidia Armelao, Francesco Barigelletti et al.

(Table 2). In this last case, in order to satisfy the coordination properties of the lanthanide ion, beside the nature of the binding sites, their relative position in the overall structure of the ligand becomes particularly important.

Choice of the Coordination Site The coordination site is formed by a number of donor atoms or groups arranged in a covalently organized structure and capable of strongly bind the metal cation. According to the dimensionality, the coordination site can be: monodimensional as in the case of acyclic ligands which can be linear (podands) or branched (polypodands), bidimensional macrocyclic ligands (coronands) and tridimensional in macropolycyclic (cryptands) structures. As a consequence the complex formed can be of chelate (podates) or of inclusion (cryptates) types. The shielding of the cation depends on the length, flexibility and number of binding sites of the ligand in the case of podates and on the degree of preorganization in the case of cryptates. The positioning of the antenna chromophore within the coordination site together with the physical and chemical properties of the antenna is also very important for the preparation of highly luminescent lanthanide complexes. This point can be addressed mainly in two ways: with the antenna subunits (i) covalently attached through a suitable spacer or (ii) integrated into the coordination site structure. In this latter combination, which represents also the most suited one, the antenna coordinates the metal centre besides acting as sensitizer, Figure 4.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ACYCLIC LIGANDS There are many examples of acyclic coordination sites that have been used for the preparation of luminescent lanthanide complexes. Among them particularly interesting are branched polypodands bearing a number of binding sites [73]. In order to form stable complexes in aqueous solution, the donor atoms of the ligand should satisfy the demand of a hard polarizing lanthanide ion. Thus, among neutral donors, more polarizable amine nitrogen atoms are preferred to oxygen atoms of ether groups, and donors possessing large ground state dipole moment such as carboxamide and sulfoxides, interact better in ion-dipole and/or ion-induced dipole bonds than less polar substituents such as alcohols. Of course, hard anionic donors strongly coordinate to the highly positive metal cation so that carboxylates, phosphonates, phosphinates and β-diketonates are excellent binding groups. The structure of the ligand must be sufficiently flexible so that it can envelope the spherical lanthanide cation as tightly as possible giving rise to a cryptate-like complex. In this way the ligand operates through an “induced fit” way, which in the case of Ln(III) ions affords complexes showing very high thermodynamic and kinetic stabilities and a complete shielding of the metal centre. In other words in order to operate through an induced fit mechanism the ligand should feature a number of binding sites higher or equal to the coordination number of the Ln(III) cation arranged to match precisely the coordination properties of the metal (“predisposed ligand”). The stability of the complex increases with ligands having anionic binding groups that strongly interact with the cation thus favouring the coordination of nitrogen and/or oxygen donor atoms, contained into the ligand structure,

Antenna-Lanthanide Complexes: A Growing Technology-Driven Research

51

according to the well known “end group effect” that is a general concept in the design of highly efficient polypodands [73].

Figure 4. Possible way to position the antenna within the ligand.

Some of the acyclic ligands frequently used for the preparation of Ln(III) complexes are

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

reported in Figure 5. Ligand 1 features three chelating 2,2’-bipyridine (bipy) units covalently attached to a nitrogen atom through a methylene bridge and contains 7 binding sites while 2 shows four bipy units linked to an ethylenediamine group with 10 binding sites. In these ligands the bipyridine moieties work both as ligating sites and as antennas. The luminescence quantum yield for Eu3+⊂ 1 and Eu3+⊂ 2 in aqueous solution was 0.1 and 0.003 respectively, indicating that not all the bipy groups of 2 coordinate the metal centre probably because of negative steric effects. Furthermore the Eu3+ complexes of these ligands proved to be unstable in aqueous solution.

Figure 5.

52

Silvio Quici, Lidia Armelao, Francesco Barigelletti et al.

A number of examples of lanthanide complexes obtained with encapsulating acyclic, macrocyclic and macrobicyclic ligands based on bipyridine chelating group and their luminescence properties have been reported [45]. A very important class of acyclic ligands is based on polyaminopolycarboxylate systems such as 3 - 5 (Figure 6) which contain five and four carboxylic groups and are capable of forming anionic Ln(III) complexes highly stable and soluble in aqueous medium. Indeed diethylene triamine pentaacetic acid (DTPA) forms complexes with Ln(III) ions featuring very high thermodynamic stability (log K ~ 22) [74].

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 6.

For these reasons DTPA derivatives have been covalently linked to biologically active molecules and their Eu3+ and Tb3+ complexes have been largely used, by means of Dissociation-Enhanced Lanthanide Fluorescence Immunossay (DELFIA) technique, as label for biological assays [75]. The covalent insertion into the DTPA structure of appropriate chromophores that can promote the sensitized light emission gives rise to strongly luminescent and stable lanthanide complexes. Ligands as 5 with a carbostyril-124 antenna have been easily prepared starting from the commercially available DTPA bis-anhydride [76]. More symmetric luminescent ligands such as 4 have been prepared by multistep syntheses and the luminescent properties of a number of lanthanide complexes have been investigated [14] . Again Eu3+ and Tb3+ complexes of 4 and 5 showed high luminescence quantum yields and long emission lifetimes in aqueous solution being the lanthanide centre perfectly shielded from the environment. Indeed no water molecules have been found in the first coordination sphere of the complexed metal. A huge variety of ligands based on pyridine, bipyridine and terpyridine bearing a variable number of acetic groups capable of forming stable complexes with lanthanide cations have been reported by Mukkala research group [77], Figure 7. In

Antenna-Lanthanide Complexes: A Growing Technology-Driven Research

53

these complexes the polypyridine part of the ligand plays a manifold role, e.g. (i) structural element that determines the topology of the free ligands and of their complexes, (ii) coordination element that provides additional nitrogen binding sites for the lanthanide cation and (iii) sensitization element due to its photophysical properties it is an effective antenna for the visible light emitting Ln3+.

Figure 7.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Furthermore the polypyridine part of the molecule can be easily functionalized for binding to biologically active compounds thus opening a wide range of applications in bioanalytical assays [78].

MACROCYCLIC AND MACROPOLYCYCLIC LIGANDS A great variety of polyoxa-, polyaza-, and polyoxapolyaza-macrocyclic (coronands) and macrobicyclic (cryptands) compounds have been used as ligands for the preparation of Ln3+ complexes. Most of these compounds can be seen as the ring closed version of acyclic polypodands as it is clearly evidenced by structures 8a-b and 9a-d, Figure 8, where two or three bipyridines of the acyclic 1 are connected to a nitrogen atom through two or three methylene bridges respectively. Again bipyridine behaves both as binding subunit and as antenna so that 8a-b and 9a-d are suitable ligands for the preparation of luminescent lanthanide complexes showing an enhanced kinetic stability that derives from the increased rigidity of the system on going from ligand 1 to ligands 8a and 9a.

54

Silvio Quici, Lidia Armelao, Francesco Barigelletti et al.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 8.

It is well known that cryptands are the best candidates for the coordination of spherical alkali and alkali-earth cations [62]. Furthermore alkali criptates show picks of selectivity due to the complementarity of the size of the metal cation with the dimension of the coordinating cavity (“lock and key” criterion). However the control of the ligand topology based on the latter principle may fail in controlling the stability and selectivity of complexation with lanthanide ions due to the small variation of the cationic radii along the series. Nevertheless interesting macrobicyclic ligands have been synthesized which provide high stability and good energy transfer for the sensitization of Eu3+ and Tb3+. In fact Eu3+ and Tb3+ of the cryptands 9 have been used in Fluorescence Resonance Energy Transfer (FRET) immunoassays [79]. The design of ligands featuring an intermediate character between the highly preorganized “lock and key” based macropolycyclic and the preorganized “induced fit” systems proved to be successful with lanthanide ions. These ligands usually contain a polyaza-macrocycle with pendant additional anionic binding sites covalently linked to the secondary amine nitrogen atom of the ring. In this case the stability of the complexes can be tuned by adjusting the ring size and the number and nature of the lateral binding arms that envelope the lanthanide cation. Among all the ligands synthesized, 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA, 10) proved to be the best one for Ln(III) with the highest stability constants ever observed (log K in the range 22-29), Figure 9 [74,80]. The cyclen ring is predisposed for the complexation: the four nitrogen bind cooperatively to the face of the square antiprism, which

Antenna-Lanthanide Complexes: A Growing Technology-Driven Research

55

is the general crystal structure observed in DOTA-type complexes with lanthanide cations, where the macrocycle adopts the same quadrangular conformation either in the free or complexed state [81].

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 9.

Other aza-macrocycles such as 1,4,7-triazacyclononane (11) and 1,5,8,12tetraazacyclotetradecane (cyclam, 12) bearing appended acetic groups have been also used, but their lanthanide complexes featured several orders of magnitude lower thermodynamic and kinetic stability constants [76,80]. It has been recently reported that the introduction of an appropriate number of picolinate and acetic groups on the triazacyclononane ring affords ligands that form lanthanide complexes characterized by high thermodynamic stability and good water solubility, Figure 10. Depending on the Ln3+ cation these complexes are particularly interesting either as contrast agents or as luminescent labels for biological assays. Indeed the Tb3+ complex of an octadentate ligand bearing two picolinate and one acetic groups 11a showed φse = 0.43 in water with a luminescence lifetime τ =1.49 ms while only moderate values, φse = 0.05 and τ = 0.54 ms, have been found for the Eu3+ complex under the same conditions. These results indicate that the efficiency of the picolinate group as sensitizer is very high for Tb3+ compared to Eu3+ [82].

56

Silvio Quici, Lidia Armelao, Francesco Barigelletti et al.

It has been demonstrated that the size of the chelating ring also influences the stability of the complex. Indeed the substitution of one acetate group of DOTA or of DTPA with a propionate residue lowers the stability of Ln3+ complexes of at least two orders of magnitude [83]. The 1,4,7,10-tetraazacyclododecane-1,4,7-triacetic acid (DO3A, 13) bearing a chromophore covalently linked to the secondary amine nitrogen atom is a good model of ligand for the preparation of luminescent lanthanide complexes characterized by high stability and, depending on the nature of the chromophore and the lanthanide ion, high sensitization efficiency φse.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 10.

Of course the removal of one acetic binding site of DOTA plays a negative role on the stability of the complexes and this can be partly compensated by choosing antenna chromophores that contain additional binding sites, suitably positioned in order to complete the coordination sphere of the metal. With this idea in mind we designed the bipartite ligand 14, Figure 9, in which a phenanthroline derivative is covalently linked through one methylene bridge to a DO3A residue to afford a nonadentate system. Eu3+ and Tb3+ complexes of 14 proved to be highly stable in aqueous solution with the metal centre completely shielded (q = 0) from the environment. Indeed X-ray crystal structure of the Eu3+⊂ 14 and solution structural investigation by 1H NMR indicate that both the nitrogen atoms of the antenna bind the metal cation thus completing the coordination sphere. The good shielding ability was explained by the rigidity of phenanthroline chromophore and by its spatial arrangement within the complex thus accounting for the high overall sensitization efficiency (φse = 0.21) for Eu3+⊂ 14 in aqueous solution [65]. The overall sensitization efficiency of Tb3+⊂ 14 in air-equilibrated aqueous solution is somewhat lower (φse = 0.11) because of the small energy gap (~ 1400 cm-1) between the triplet state of the phenanthroline and the emitting state of Tb3+ and this accounts for the deactivation step through back energy transfer process (Figure 2). This is also demonstrated by the large increase of the sensitization efficiency (φse = 0.55) found in deareated conditions. Another interesting ligand reported by Beeby et al. [84] and conceptually similar to 14 is the compound 17, Figure 9, having an acetophenone residue as antenna. In this case also the

Antenna-Lanthanide Complexes: A Growing Technology-Driven Research

57

ketone group coordinates the metal centre and the coordination sphere of the metal is completed by one molecule of water (q = 1). The overall sensitization efficiencies found for Eu3+⊂ 17 and Tb3+⊂ 17 in aqueous solution are φse = 0.096 and 0.34 respectively. The presence of binding sites properly positioned on the chromophore is an important requisite for the luminescence efficiency of the lanthanide complexes. Thus, Eu3+ and Tb3+ complexes of ligands 15 and 16 (Figure 9) having a quinoline or a naphthyl group as antenna showed lower sensitization quantum yields. Indeed the quinoline group contains one nitrogen as a potential binding site that is however unsuitably positioned to effectively coordinate the metal. As a consequence water molecules (q = 1.5) are present within the coordination sphere of the metal [85].

The Two-Component Approach

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

We have demonstrated that conceptually simple two-component ligand systems can result in remarkable performances in air-equilibrated water solvent [14,15,65,85]. The systems are constituted by a single chromophoric unit flexibly linked via a methylene bridge to a hosting unit for the Ln(III) cation. Figure 11 displays representative cases. System 14 features the combination of a phenanthroline (phen) chromophore, the light antenna, and a tetraazacyclodedodecane-triacetic acid unit, DO3A, the hosting unit for the cation. DO3A provides 7 binding sites for coordination of the Ln(III) centre, and the 1:1 DO3A:Ln(III) association constant is remarkably high, K = 1023 M-1 [74,80]. The use of phen as a light-absorbing unit allows to set up an efficient antenna-lanthanide system because (i) the ISC step for phen takes place with φISC >> 0.65 (presumably, not far from unit) within the complex, (ii) the rate constant for the energy transfer step is ken ≈ 107 s-1, which compared to a deactivation rate constant kT ~ 3 ×104 s-1 for the phen T level (Figure 2) results in φen ~ 1, and (iii) the phen moiety cooperates to saturate the coordination sphere around the cation so that no water molecules are directly bound to the Ln(III) centre, q = 0, (eq. 2a). A similar approach led to the Eu3+⊂ 17 system where the single chromophore playing as an antenna was acetophenone, and q was found to be 1. Table 3 summarizes the luminescence results obtained in water solution with system 14.

Figure 11.

58

Silvio Quici, Lidia Armelao, Francesco Barigelletti et al. Table 3. Metal centred luminescence properties for Ln3+ complexes of systems 14 and 4 in H2O and D2O solutionsa H2O b

λmax (nm) Eu3+⊂ 14 Tb3+⊂ 14

D2O -1

-1

φse

τ (μs)

kr (s )

knr (s )

0.21

1240

169

637

0.11

310

355

3020

c

298c

c

1510

0.24

1250

0.15

780

0.55

Eu3+⊂ 4

616

Tb3+⊂ 4

544

Sm3+⊂ 4

598

2.5 x 10-3

Dy3+⊂ 4

478

5 x 10-4

0.45

c

c

364

φse

τ (μs) 1770 320

0.30

1880

0.14

820

13.0

2.5 x 10-3

34.0

1.2

5 x 10-4

1.1

2.3 x 103 c

a

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

At room temperature, excitation at 278/279 nm; for Sm(III) and Dy(III) complexes, results are for the Vis emission portion; no water molecules (q) are present in the coordination sphere of the Ln(III) centre, uncertainty on q is ± 0.5 [14,15,65]. b Highest-intensity line or band. c Degassed sample.

Excellent performances were likewise obtained with the water-soluble ligand 4. It is constituted by a single phen chromophore and a diethylenetriamine tetraacetic unit (DTTA) hosting site, expected to provide highly stable complexes. The coordination positions of the Ln(III) centre are protected against solvent access, as shown by the comparison of luminescence results obtained in water and deuterated water, Table 3. We provided evidence that for the 1:1 species obtained with ligand 4 (KA > 107 M-1), the luminescence sensitization process in air-equilibrated water is quite effective, resulting in φse = 0.24 and 0.15 for the visible emitters Eu3+⊂ 4 and Tb3+⊂ 4, respectively and φse = 2.5×10-3 and 5×10-4 for the visible portion of the Sm3+⊂ 4 and Dy3+⊂ 4 emitters, respectively [14]. In the following, for the sake of illustration we discuss some issues concerned with the spectroscopic properties of the complexes obtained with ligand 4. Absorption and emission properties of the complexes. Representative absorption spectra of 4 and Eu3+⊂ 4 in water solution are shown in Figure 12 together with results from luminescence titration of 4 upon EuCl3(H2O)6 addition. Results of the titration experiment depicted in insets (a) and (b) of Figure 12 indicate that a 1:1 association takes place, with an association constant KML ≥ 107 M-1 [65]. The spectral shapes of 4 and Eu3+⊂ 4 remarkably overlap and the two peaks in the UV region at 230 and 279 nm are ascribed to ligand-centred transition of the phenantroline group; the same results were obtained for other Ln3+⊂ 4 complexes. All of this amounts to be an uncommon observation because the presence of triply charged ions incorporated within a complex usually affects the electronic properties of the nearby chromophore, and is signaled by a change of its absorption profile. We concluded that coordination of the phen unit at the Ln(III) centre does not actually imply a strong electronic interaction between these two units;

Antenna-Lanthanide Complexes: A Growing Technology-Driven Research

59

1

H-NMR spectroscopy supported this view. Nevertheless, efficient sensitization of the MC luminescence takes place, as revealed both by the titration course in Figure 12 and by the final intense luminescence observed for Eu3+⊂ 4 (Table 3). Thus, the weak electronic interaction between the light absorbing phen and the light emitting Ln subunits is at any rate sufficient to promote the occurrence of ligand-to-metal energy transfer. A similar case from the literature was that of a Eu3+⊂ 17 system with appended a single acetophenone unit [84].

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 12. Titration experiment of water solution of 4 upon EuCl3(H2O)6 addition. Shown are (i) the absorption spectra of 4 (full line) and Eu3+⊂ 4 (taken at the end of the experiment, dotted line), (ii) the enhancement of the metal centred luminescence, λexc = 279 nm, upon addition of Eu3+, inset (a), (iii) the registered changes of the emission intensity at 616 nm (filled points) and the full line resulting from the fit of the data points according to a 1:1 stoichiometry, inset (b). Ref. [14].

Figure 13. Room temperature emission spectra in the Vis portion for the Ln3+⊂ 4 complexes of the indicated cations; in air-equilibrated D2O, λexc = 279 nm. For Sm3+⊂ 4 and Dy3+⊂ 4, in the NIR portion of the emission (not shown), the luminescence profile exhibits a maximum intensity peak at 948 and 994 nm, respectively.

60

Silvio Quici, Lidia Armelao, Francesco Barigelletti et al.

Ligand-centred luminescence. According to Figure 2 and to the results described above, light absorption via ligand-centred 1LC transitions is followed by an intersystem crossing step (ISC), leading to population of the lowest-lying triplet level (T) of phen origin. The observed lanthanide luminescence originates from subsequent ligand-to-metal intramolecular photoinduced energy transfer processes. Figure 13 shows the luminescence spectra of Eu3+⊂ 4, Tb3+⊂ 4, Sm3+⊂ 4 and Dy3+⊂ 4, as registered upon excitation at 279 nm. This reference scheme does not work for the case of Gd(III) complexes because its lowest-lying MC level is located at 32150 cm-1 (corresponding to the 6P7/2 → 8S7/2 transition), an energy content much higher than that of the T level of most organic chromophores. On this basis, the luminescence study of Gd3+⊂ 4 allows to evaluate the energy levels of the lowestlying singlet (S) and triplet (T) excited states of the chromophore, as affected by a nearby heavy and triply charged center. Figure 14 shows the luminescence results for 4 and Gd3+⊂ 4 as obtained at room temperature (fluorescence) and at 77 K (phosphorescence) and Table 4 lists luminescence data for the ligand-centred emission at both temperatures. Table 4. Ligand-centred luminescence properties for 4 and Gd3+⊂ 4 in H2O solution and rigid matrixa 298 K 4 Gd3+⊂4

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

a

λmax (nm) 370 380

φem

3.5 x 10-3 7.4 x 10-4

77 K τ (ns) 0.9 < 0.5

λmax (nm) 498 498

τ (s) 1.7 0.23

Fluorescence at room temperature and phosphorescence at 77 K (frozen medium) of 4; λexc = 279 nm.

Figure 14. Fluorescence (F, room temperature) and phosphorescence (P, 77 K, pulsed lamp, 10 ms delay after flash) spectra of 4 (solid line) and Gd3+⊂ 4 (dashed line) in water solution and rigid matrix; λexc = 279 nm.

Antenna-Lanthanide Complexes: A Growing Technology-Driven Research

61

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

With regard to the room temperature results as monitored at the fluorescence band for 4 (φem= 3.5 × 10-3 and τ = 0.9 ns) and for Gd3+⊂ 4 (φem = 7.4 × 10-4 and τ < 0.5 ns), one draws the indication that the metal centre causes an improved intersystem crossing within the complex. As for the 77 K results, from the highest energy peak of the phosphorescence profile, the T level of phenanthroline origin can be estimated to lay at 22100 cm-1 for both 4 and Gd3+⊂ 4, a result consistent with findings from previous investigations [44]. As seen above, Figure 2 compares typical energetic layouts for the cases of Tb(III), Eu(III), and Gd(III), with highest-energy transitions labeled 5D4→7F6, 5D1→7F0, and 6P7/2 → 8S7/2, respectively. Coordination features. Eu3+⊂ 4 and Tb3+⊂ 4 show a remarkably intense and long-lived luminescence in both H2O and D2O solutions (Table 3). Comparison of luminescence results for water and deuterated water provides an assessment of water binding at the Eu(III) and Tb(III) centres, (eqs 2). In particular, for Eu3+⊂ 4 and Tb3+⊂ 4 we found qEu = 0.02 and qTb = 0.01, respectively [14]. These results indicate that the coordination shell of system 4 effectively prevents water from binding at the metal centres. Of course, this is the basic reason for the high luminescence performances of the Ln3+⊂ 4 complexes of the series in water, in agreement with the fact that OH oscillators are kept far from the metal centres and cannot act as quenchers. Regarding the coordination features of Sm3+⊂ 4 and Dy3+⊂ 4, and based on the structural similarity of all Ln centres, we assumed that also in these cases no solvent molecules (q = 0) are present within the first coordination sphere. Oxygen effect. Comparison of luminescence intensity and lifetime data in air-equilibrated and degassed water solutions reveals an oxygen effect for the Tb3+⊂ 4 case (Table 3); by contrast, for the other cases no relevant changes are noticed. For Tb(III) complexes incorporating the phen chromophore, this effect is due to a small energy gap, ΔE, between ligand-centred T levels and MC levels (Figure 2). Because of thermal redistribution, both forward (ligand-to-metal) and backward (metal-to-ligand) energy transfer processes take place, with rate constants, kf and kb, respectively. This allows deactivation of the ligand T level through non-radiative quenching pathways by dissolved oxygen molecules, Figure 15.

Figure 15. Schematic energy levels for Tb3+⊂ 4 and the diffusional quenching effect by oxygen.

62

Silvio Quici, Lidia Armelao, Francesco Barigelletti et al.

We have modeled the room temperature equilibrium between the T level of phenanthroline origin and the MC luminescent level of Tb(III) by employing the following parameters: (i) the energy gap (ΔE) between the two levels was estimated from the profiles of Figure 14 (77 K case) and was varied within the 1400 and 1550 cm-1 range; (ii) the rate constant for the phen-like T decay was kP = 1/τP, with τP = 35μs and (iii) the rate constant for diffusional quenching by oxygen dissolved in water was kqO2 = 2.0 × 106 s-1, as evaluated

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

from kdiff × [O2], with kdiff = 7 × 109 M-1 s-1, and [O2] = 0.29 × 10-3 M. The three-fold changes of the luminescence features for Tb3+⊂ 4 could be reproduced for kf (the forward energy transfer rate constant, Figure 15) ≈ 5 × kqO2 [with the backward energy transfer kb = kf exp(-

ΔE/RT)]. Accordingly, the phen → Tb(III) energy transfer step is evaluated to be ken ≈ 1 × 107 s-1. For the all complexes but those of Tb(III), no changes of luminescence properties were observed for air-equilibrated and degassed samples (Table 3). A distinctive reason for the lack of oxygen effect is met for the case of Dy3+⊂ 4. The highest-energy level of Dy(III) is at ~20750 cm-1 (corresponding to the 4F9/2 → 6H15/2 transition) and is close-lying to that of Tb(III), ~20500 cm-1 (5D4→7F6). However, for Dy3+⊂ 4, the faster luminescence decay (1.2 μs, Table 3) in comparison to that of Tb3+⊂ 4 (780 μs) does not allow the establishment of an equilibrium between ligand (T) and Dy(III) centred levels. For the phen → Dy(III) energy transfer step, results of model calculations are consistent with a forward rate of kf ≈ 1 ×107 s-1, as for the case of Tb3+⊂ 4. Nature of the energy transfer step. As seen above, efficient phen → Ln(III) energy transfer takes place for all investigated complexes. On the basis of the observed or derived parameters, it can be concluded that φISC × φen ≈ 1 [from eq. 1, oxygen-free case for Tb(III) complexes]. This result is consistent with those from 1H-NMR spectroscopy, suggesting the involvement of the phen subunit in the coordination of the metal cation [15]. A dipole-dipole through-space (Förster-type) mechanism for energy transfer seems unlikely, given that the energy transfer step involves in all cases a triplet level for the donating chromophore, a phen subunit for many of the cases examined here. On the other hand, it is known that a very small electronic interaction, i.e. an interaction term H ~ 1 cm-1, is enough to permit the occurrence of Dexter-type through-bond energy transfer [86]. This interaction might originate via mediation by the intervening short sequence of bonds separating the phen chromophore and the hosting site for the Ln(III) centres. Metal centred luminescence. The Eu(III) and Tb(III) complexes are by far the most intense emitters among those of the lanthanide series, as shown for the representative cases collected in Table 3. It is interesting to examine the influence of radiative and non-radiative processes in these luminophores. In Table 3, values for the kr and knr rate constants, eq. 1b, are listed as obtained from the observed photophysical parameters and available literature sources. Estimates of the pure radiative lifetime (τr = 1/kr) can be obtained for Eu(III) complexes according to an approach which compares the intensity of the 5D0→7F1 band (593 nm, insensitive to the coordination environment) with the overall shape of the emission spectrum, (eq. 3).

k r = A(0,1)

I tot I (0,1)

(3)

Antenna-Lanthanide Complexes: A Growing Technology-Driven Research

63

In this equation, A(0,1) is the spontaneous emission probability of the 5D0→7F1 transition, that is evaluated 32.4 s-1 in water [84,87], and Itot/I(0,1) is the ratio of the total integrated intensity of the emission spectrum to the portion of the profile for the 5D0→7F1 band. Thus, for the Eu(III) complexes evaluated values for kr are in the range 160 to 240 s-1 and those for knr are even larger (Table 3). On the basis of eq. 1b, one draws the conclusion that even for the Eu(III) complexes, the most luminescent ones within the lanthanide series, the intrinsic efficiency φlumMC cannot reach unity. To notice that MC non-radiative processes can include both intrinsic and back energy transfer (kb, Figure 15 and concerned main text) contributions, k nr = k nr + kb . The former MC

are related to the presence of OH oscillators, and are therefore depressed in D2O as pointed out for the Eu(III) cases of Table 3. For Tb(III) complexes, the energy gap ΔE between the ligand T level and the emissive MC level is usually so small that Tb → L back energy transfer is by far the major contribution to the non-radiative deactivation, as noticed above for the cases of air-equilibrated water solution. In conclusion, also for the Tb(III) complexes, the intrinsic luminescence quantum yield is expected to be lower than unity.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Sol-Gel Techniques for the Preparation of Emissive Thin Films The peculiar luminescence properties of the lanthanide antenna complexes properly coupled with suitable inorganic or polymeric matrices can be exploited for a variety of modern applications [1,5,88] ranging from diagnostic tools in nanobiomedicine to optical and photonic materials for the development of lasers, displays and lighting devices, just to cite a few. The possibility to combine in one material the properties of inorganic or polymeric components with complex organic or organometallic units, such as chromophores, catalytically active groups or biomolecules, has witnessed important improvements in recent years with the development of low-temperature soft chemistry solution processes [89]. Among them, sol-gel and hydrothermal synthesis methods have allowed the preparation of original nanosystems and nanocomposite structures with sophisticated shapes and intriguing peculiarities. The notion is to create materials with new combinations of properties by integrating inorganic or organic components at nanoscale or molecular level. Nowadays, the incorporation of luminescent lanthanide complexes in solid matrices with controlled structural organization is of widespread interest in materials science as it affords functional materials with a variety of optical properties. In the areas of life science, biotechnology and clinical diagnostics, luminescent lanthanide chelates have been receiving increasing attention because of their applications as luminescent probes for highly sensitive Time-Resolved Fluoroimmunoassay (TR-FIA), DNA hybridization assay, fluorescence microscopy bioimaging, and other analytical techniques [78,90-93]. They display specific luminescence properties that conventional organic dyes do not show, such as sharp emission profiles, large Stokes shifts and long luminescence lifetimes (ms scale). Indeed a typical fluorescence bioassay drawback is that the fluorescence detection is easily affected by the background noises (emission lifetime ≈ ns to a few μs) caused by biological samples and analysis instruments [94,95]. By using luminescent lanthanide complexes non specific background luminescence from the specimen and cuvette materials

64

Silvio Quici, Lidia Armelao, Francesco Barigelletti et al.

and the scattering light can be effectively eliminated due to the very long fluorescence lifetime, usually over several hundreds microseconds. For practical uses, these bio-labelling reagents should be conveniently grafted on suitable substrates, such as inorganic or polymeric nanoparticles (NPs) or colloids [94]. Currently, silica-based NPs are used in many areas of bioanalysis and, compared with polymer-based NPs, they have a reduced tendency to aggregation and toward dye leakage [96,97]. In addition, the silica surface makes these NPs chemically inert and physically stable [98]. All these properties make silica NPs excellent labeling reagents for bioanalysis and bioimaging [99-101]. Through sol-gel synthesis, a single silica particle can be functionalized with a large number of dye molecules, up to tens of thousands, Figure 16.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 16. Schematic representation of a SiO2 nanoparticle functionalized with luminescent metal complexes.

Even though luminescence quenching phenomena within the NPs are not completely eliminated, thanks to the large amount of dye incorporated in a small volume, the goal of obtaining nanoparticles with brighter luminescence is achieved. Thus, the main advantage of luminophore-doped NPs concerns the capability of originate highly amplified optical signals compared with a single dye molecule. Moreover, as the dye is protected by the silica matrix, both photobleaching and photodegradation phenomena that often affect conventional dyes can be minimized [102]. Furthermore, thanks to the flexibility of silica chemistry, different types of functional groups can be easily introduced on the NPs surface and, besides single-dye doping, multiple-dye incorporation into the silica matrix is also possible. This can provide more information upon detection. In fact, by tuning the concentrations of the different luminophores within the NPs, excitation with a single wavelength leads to different emission, thus permitting simultaneous and sensitive detection of multiple targets. Concerning the field of optical materials, lanthanide antenna complexes are being intensively studied, among others, for the development of sensors [61], light-conversion molecular-devices, optical fibre lasers and amplifiers [26,103], and electroluminescent materials [104]. For most applications, such as tunable solid-state laser or phosphor devices,

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Antenna-Lanthanide Complexes: A Growing Technology-Driven Research

65

mechanical strength and chemical stability under variable temperature or moisture conditions are severe requisites. Recently, lanthanide hybrid materials obtained by incorporation of lanthanide complexes in the inorganic matrices have attracted considerable interest, and their luminescence properties have been extensively studied. The main goal of these studies was to understand how the photophysical properties of the emissive species can be affected by the interaction with the host structure [105]. Consequently, considerable research activity is being carried out to improve the chemical stability and to adapt the materials chemistry to the production technology of the concerned application. Depending on the chemical nature of the components, materials characterized only by weak interactions between organic and inorganic parts (such as hydrogen bonding, van der Waals forces, or electrostatic forces) can be obtained [105-107]. However inhomogeneous dispersion of two phases and leaching of the photoactive molecules frequently occur; in these cases the allowed concentration of complex is also severely restricted. Alternatively, an appealing possibility is the covalent bonding of rare-earth antenna complexes to the host network [108,109] through soft sol-gel chemistry. The tuning of the interactions between the luminescent species and the host medium represents a fundamental aspect [110]. Indeed, the properties that can be obtained for such materials certainly depend on the chemical nature of their components and on their molecular level structure. Sol–gel chemistry allows the combination at the nanoscale level of inorganic and organic components in a single hybrid composite [89], opening thus access to an expanding area of innovative materials and to new perspectives in materials science [89,110117]; moreover, a prominent advantage of the technique is that the microstructure and the external shape can be controlled by varying the sol-gel processing conditions. The use of the sol-gel technique for the synthesis of oxide-based materials [118,119] is thus attracting much attention. In particular, in the design of luminescent hybrid materials based on lanthanide antenna complexes, due to its chemical stability and transparency, silicon dioxide (SiO2, silica) is suitable as host material either for the development of optical materials or as a substrate in biotechnology and clinical diagnostic. The chemistry of the solgel process is mainly based on hydrolysis and polycondensation of metal alkoxides to form extended networks with an oxide skeleton. In order to prepare silica-based materials by this approach, tetramethylorthosilicate (TMOS, Si(OCH3)4) or tetraethylorthosilicate (TEOS, Si(OCH2CH3)4) are often used. In the first stage of the process, a sol of fine colloidal particles or polymers is formed by mixing the alkoxide precursor with water, a co-solvent, generally a low molecular weight alcohol, and an acid or base catalyst at room temperature, and further reactions advancement lead to gelation, i.e., wet gel formation. The basic sol–gel reaction starts when the metal alkoxide (Si–OR) is mixed with water (hydrolysis). The following three reactions are generally used to describe sol–gel processes [118]: The final intrinsic properties of the sol–gel matrix, i.e. cross-linking of the oxide network, porosity, surface area, polarity and rigidity are greatly related to the progress of hydrolysis and condensation reactions. Of course they are also affected by the choice of precursors, water to precursor molar ratios, solvent and co-solvent, temperature, aging, drying and curing conditions [120-122].

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

66

Silvio Quici, Lidia Armelao, Francesco Barigelletti et al.

Although the utility of sol–gel matrices as hosts for organic and organometallic components has been well known since long, it still attracts increased attention in basic research for designing appropriate sol compositions for development of host matrices for functional molecules such as lanthanide antenna complexes (guests). In some instances the sol–gel reactions run in highly acidic (or basic) conditions. Under such conditions biomolecules can be denaturated and functional (optical, magnetic…) complexes can lose their peculiar features. In order to prevent these undesired events the materials synthesis must be carefully designed so as to afford high quality and highly homogeneous host-guest nanosystems in which the functional properties of the guest molecules are preserved. To this aim, solvent, catalyst, and pH of the sol-gel solutions should be varied in order to obtain the desired results. For example, modified sol–gel procedures are reported [123] for proteins entrapment in a silica network starting from alcohol free solutions at pH between 5.0 and 8.0, to prevent protein denaturation. On the other hand, higher pH values speed up hydrolysis and condensation reactions which in turn accelerate the gelation process [124]. A shorter gelation time could affect the host matrix formation in monoliths and thin films thus leading to the formation of less homogeneous and transparent materials. Depending on processing conditions, sol-gel materials can be regarded as active substrates, being characterized by a porous structure in which a high amount of –OH groups are present, (a) in Figure 17. These groups can provide reaction sites for subsequent chemical functionalization both on the surface and sub-surface layers [125,126]. Alternatively, the luminophores (b) can be covalently anchored through –OH groups to the silica host during glass synthesis, thus yielding highly homogeneous layers with covalently linked complexes (Figure 17). Actually, the grafting of the guest molecules in/on the hosting phase is performed through an one-step process thanks to the molecular-level mixing of reactants in the starting solutions. This

Antenna-Lanthanide Complexes: A Growing Technology-Driven Research

67

allows, in principle, a fine control over luminophore dispersion and avoids undesired aggregation phenomena.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 17.

The introduction of further various functional groups such as amino, thiol, glycidoxy, epoxy, etc. [118,119], into the alkoxide monomers allows the grafting of many other different species to sol-gel materials. Moreover, it provides a versatile way to prepare tailor-made materials also allowing the synthesis of organically modified sol–gel glasses which present several attractive features [118,119]. Depending on the chemical conditions under which alkoxide precursors are polymerised, very different structures ranging from colloidal particles to randomly branched polymers can be obtained [118,119]. In the course of sol-to-gel conversion, which takes place at low temperatures, coating, fibre drawing and moulding into bulk shapes can be achieved. Among these, nanoparticles and films are the most important products of the method. Coating of glass, ceramic, metal and plastic substrates by the sol-gel route is very useful for modifying properties of substrates or providing substrates with active properties to develop new optical, electronic and chemical devices. Films can be prepared by dip, spin and spray coating techniques [118]. In case of dip-coating, a high degree of thickness uniformity is achievable and can be controlled via the withdrawal speed of the substrate. By comparison, thin film formation by spin coating causes a greater rate of solvent evaporation than dip coating and rapid changes in the physical properties of the sol–gel material [118]. The main factors that are important in development of thin films are the uniformity and thickness of film, its adhesion to the substrate and resistance to cracking, designing of stable internal environment and minimizing the leaching of entrapped species. The thickness of the sol–gel derived films is highly dependent on the gelation behavior of the sol, which in turn depends upon the viscosity of the casting solution. Surface characteristics as well as uniformity in monoliths/thin films are one of the desirable criteria for optical applications. A specific case study concerning the preparation of Eu/SiO2 and Tb/SiO2 materials is subsequently outlined. Starting from a unique dipartite ligand, consisting of DO3A covalently bonded to an acetophenone unit [108,109], silica-based thin films embedding Eu3+⊂ 18 and Tb3+⊂ 18 complexes were prepared by sol-gel using tetraethylorthosilicate Si(OC2H5)4 as silica source under acidic conditions. Before use for film deposition, the transparent and clear solutions were aged at room temperature. During this stage, hydrolysis and polycondensation

68

Silvio Quici, Lidia Armelao, Francesco Barigelletti et al.

reactions proceed, allowing the progressive formation of the silica network as well as the covalent grafting of the lanthanide complexes to the inorganic backbone through hydroxyl units. The layers were then obtained by dip-coating and subsequently used in the luminescence experiments both as-deposited and annealed at 100 and 200 °C. The thermal treatment was aimed to promote further densification of the glassy matrix while preserving the chemical integrity of the antenna units and their photophysical properties. A high homogeneous luminescence in the visible spectral window was detected in particular from the samples annealed at 200°C, Figure 18. To modulate the colour output, solutions containing variable Tb3+⊂ 18 / Eu3+⊂ 18 molar ratios were used. This approach allowed us to obtain Eu2-Tb1, Eu1-Tb1 and Eu1-Tb2 hybrids, featuring Tb/Eu molar ratio of 0.4, 1.0 and 2.2 respectively. The random distribution of the chromophores into the matrix resulted in the uniformity of the colour output and upon changing the ratio between metal complexes within the SiO2 host. Table 5 compares luminescence data for water solutions and SiO2 matrices at room temperature. Table 5. Luminescence data in H2O and SiO2 matrix at room temperature H2Oa

τ (ms)

φse

0.08

0.6

0.10

Eu2-Tb1

-

-

0.10

Eu1-Tb1

0.19

0.7c 1.7d

0.09

Eu1-Tb2

-

-

0.09

Tb3+⊂ 18

0.31

1.6

0.09

[Eu] = [Tb] = 10-5 M. b Molar ratio as mentioned in the text [108]. centred emission. d Tb(III)-centred emission.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

a

φse Eu3+⊂ 18

SiO2 matrixb τ (ms) 0.7 0.6c 1.1d 0.6c 1.0d 0.7c 1.0d 1.1 c

Eu(III)-

As can be seen from Table 5, each type of emitter maintained its luminescence properties within the silica matrix, an expected outcome owing to the dilution conditions employed, ca. 4.8 nm3 of film volume/luminophore. As a result, the produced colours arose from the statistical mixing of the green and red emissions and were shown to include degrees of pale yellow not far from white emission. These hybrid materials show attractive properties from both the mechanical and chemical viewpoints, and the doping of different inorganic matrices with lanthanide complexes emitting in the visible range represents a promising way to fabricate a variety of luminescent and electroluminescent devices.

Antenna-Lanthanide Complexes: A Growing Technology-Driven Research

69

Figure 18. Colors emitted by samples containing different ration of Eu3+⊂ 18 and Tb3+⊂ 18 complexes.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

CONCLUSIONS The research in luminescent lanthanide complexes is steadily growing mainly because of their unique optical properties that find application in many technologically interesting fields going from lasers and white light generation to optical fibre for telecommunication to biological Time Resolved FluoroImmunoAssays (TR-FIA). In order to correctly approach this research it is necessary to know: (i) the basic principle of rare earth elements e. g. their electronic structure and redox properties that are at the origin of their unique optical behaviour; (ii) the coordination features of the Ln3+ cations and the aspects that allow the design of ligands whose complexes show optimized luminescent properties; (iii) the basic concepts that rule the photophysical properties of luminescent lanthanide complexes; (iv) the basic concepts necessary to design highly stable and homogeneous materials that allow to transfer the results of the research to the daily life used macroscopic devices. Our effort in writing this chapter was to give, although not exhaustively, an answer to all these issues with the aim of contributing either to the enhancement of basic research and to the development of new and better performing materials and devices.

70

Silvio Quici, Lidia Armelao, Francesco Barigelletti et al.

REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17]

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

[18] [19]

[20] [21] [22] [23] [24] [25]

Yam, V. W. W.; Lo, K. K. W. Coord. Chem. Rev. 1999, 184, 157-240. Aime, S.; Crich, S. G.; Gianolio, E.; Giovenzana, G. B.; Tei, L.; Terreno, E. Coord. Chem. Rev. 2006, 250, (11-12), 1562-1579. Crich, S. G.; Biancone, L.; Cantaluppi, V.; Esposito, D. D. G.; Russo, S.; Camussi, G.; Aime, S. Magn. Reson. Med. 2004, 51, (5), 938-944. Benelli, C.; Gatteschi, D. Chem. Rev. 2002, 102, (6), 2369-2387. Faulkner, S.; Pope, S. J. A.; Burton-Pye, B. P. Appl. Spectrosc. Rev. 2005, 40, (1), 1-31. Parker, D.; Williams, J. A. J. Chem. Soc., Dalton Trans. 1996, (18), 3613-3628. Leonard, J. P.; Nolan, C. B.; Stomeo, F.; Gunnlaugsson, T. Top. Curr. Chem. 2007, 281, 1-43. Bünzli, J. C. G.; Piguet, C. Chem. Soc. Rev. 2005, 34, (12), 1048-1077. Bünzli, J. C. G. Acc. Chem. Res. 2006, 39, (1), 53-61. Beeby, A.; Botchway, S. W.; Clarkson, I. M.; Faulkner, S.; Parker, A. W.; Parker, D.; Williams, J. A. G. J. Photochem. Photobiol. B-Biol. 2000, 57, (2-3), 83-89. Klink, S. I.; Hebbink, G. A.; Grave, L.; Van Veggel, F. C. J. M.; Reinhoudt, D. N.; Slooff, L. H.; Polman, A.; Hofstraat, J. W. J. Appl. Phys. 1999, 86, (3), 1181-1185. Werts, M. H. V.; Woudenberg, R. H.; Emmerink, P. G.; van Gassel, R.; Hofstraat, J. W.; Verhoeven, J. W. Angew. Chem., Int. Ed. Engl. 2000, 39, ), 4542-4544. Foley, T. J.; Harrison, B. S.; Knefely, A. S.; Abboud, K. A.; Reynolds, J. R.; Schanze, K. S.; Boncella, J. M. Inorg. Chem. 2003, 42, (16), 5023-5032. Quici, S.; Cavazzini, M.; Marzanni, G.; Accorsi, G.; Armaroli, N.; Ventura, B.; Barigelletti, F. Inorg. Chem. 2005, 44, (3), 529-537. Quici, S.; Marzanni, G.; Forni, A.; Accorsi, G.; Barigelletti, F. Inorg. Chem. 2004, 43, (4), 1294-1301. Hasegawa, Y.; Wada, Y.; Yanagida, S. J. Photochem. Photobiol., C 2004, 5, (3), 183202. Hebbink, G. A.; Klink, S. I.; Grave, L.; Alink, P. G. B. O.; van Veggel, F. C. J. M. ChemPhysChem 2002, 3, (12), 1014-1018. Faulkner, S.; Pope, S. J. A. J. Am. Chem. Soc. 2003, 125, (35), 10526-10527. Dickins, R. S.; Aime, S.; Batsanov, A. S.; Beeby, A.; Botta, M.; Bruce, J.; Howard, J. A. K.; Love, C. S.; Parker, D.; Peacock, R. D.; Puschmann, H. J. Am. Chem. Soc. 2002, 124, (43), 12697-12705. Kamenskikh, I. A.; Guerassimova, N.; Dujardin, C.; Garnier, N.; Ledoux, G.; Pedrini, C.; Kirm, M.; Petrosyan, A.; Spassky, D. Opt. Mater. 2003, 24, (1-2), 267-274. Wong, W. K.; Liang, H. Z.; Wong, W. Y.; Cai, Z. W.; Li, K. F.; Cheah, K. W. New J. Chem. 2002, 26, (3), 275-278. Shavaleev, N. M.; Pope, S. J. A.; Bell, Z. R.; Faulkner, S.; Ward, M. D. J. Chem. Soc., Dalton Trans. 2003, (5), 808-814. Oshishi, Y.; Kanamori, T.; Kitagawa, T.; Takashashi, S.; Snitzer, E.; Sigel, G. H. Opt. Lett. 1991, 16, 1747. Slooff, L. H.; Polman, A.; Wolbers, M. P. O.; van Veggel, F. C. J. M.; Reinhoudt, D. N.; Hofstraat, J. W. J. Appl. Phys. 1998, 83, (1), 497-503. Heller, A. J. Am. Chem. Soc. 1967, 89, (1), 167 - 169.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Antenna-Lanthanide Complexes: A Growing Technology-Driven Research

71

[26] Kuriki, K.; Koike, Y.; Okamoto, Y. Chem. Rev. 2002, 102, (6), 2347 - 2356. [27] Nakamura, K.; Hasegawa, Y.; Kawai, H.; Yasuda, N.; Kanehisa, N.; Kai, Y.; Nagamura, T.; Yanagida, S.; Wada, Y. J. Phys. Chem. A 2007, 111, (16), 3029 - 3037. [28] Li, H. R.; Lin, J.; Zhang, H. J.; Fu, L. S.; Meng, Q. G.; Wang, S. B. Chem. Mater. 2002, 14, (9), 3651-3655. [29] Carlos, L. D.; Ferreira, R. A. S.; Bermudez, V. D.; Molina, C.; Bueno, L. A.; Ribeiro, S. J. L. Phys. Rev. B: Condens. Matter 1999, 60, (14), 10042-10053. [30] Moeller, T. J. Chem. Educ. 1970, 47, (6), 417-423. [31] Cotton, S. A. Lanthanide and Actinide Chemistry; Inorganic Chemistry: A Textbook Series; John Wiley & Sons: West Sussex, 2006, pp 280. [32] Cotton, F. A.; Wilkinson, G.; Murillo, C. A.; Bochmann, M. Advanced Inorganic Chemistry, 6th Edition; John Wiley & Sons LTD: New York, 1999, pp 1988. [33] Shannon, R. D. Acta Crystallogr., Sect. A: Found. Crystallogr. 1976, A32, 751-767. [34] Seitz, M.; Oliver, A. G.; Raymond, K. N. J. Am. Chem. Soc. 2007, 129, (36), 1115311160. [35] Rizkalla, E. N. Radiochim. Acta 1993, 61, (3-4), 181-189. [36] Peters, J. A.; Huskens, J.; Raber, D. J. Prog. Nucl. Magn. Reson. Spectrosc. 1996, 28, (3/4), 283-350. [37] Kurisaki, T.; Yamaguchi, T.; Wakita, H. J. Alloys Compd. 1993, 192, (1-2), 293-295. [38] Kanno, H.; Yokoyama, H. Polyhedron 1996, 15, (9), 1437-1441. [39] Kowall, T.; Foglia, F.; Helm, L.; Merbach, A. E. J. Am. Chem. Soc. 1995, 117, (13), 3790-3799. [40] Cossy, C.; Helm, L.; Powell, D. H.; Merbach, A. E. New J. Chem. 1995, 19, (1), 27-35. [41] Bradley, D. C.; Ghotra, J. S.; Hart, F. A. J. Chem. Soc., Dalton Trans. 1973, 10, 10211023. [42] Aspinall, H. C.; Tillotson, M. R. Polyhedron 1994, 13, (23), 3229-3234. [43] Werts, M. H. V. Science Progress 2005, 88, (2), 101-131. [44] Montalti, M.; Credi, A.; Prodi, L.; Gandolfi, M. T. Handbook of Photochemistry, 3rd Edition; CRC Press, Taylor & Francis: Boca Raton, 2006, pp [45] Sabbatini, N.; Guardigli, M.; Lehn, J. M. Coord. Chem. Rev. 1993, 123, (1-2), 201-228. [46] Sabbatini, N.; Guardigli, M.; Manet, I. Handbook Phys. Chem. Rare Earths; Elsevier: Amsterdam, 1996, 23, pp 69. [47] Lazarides, T.; Davies, G. M.; Adams, H.; Sabatini, C.; Barigelletti, F.; Barbieri, A.; Pope, S. J. A.; Faulkner, S.; Ward, M. D. Photochemical & Photobiological Sciences 2007, 6, 1152-1157. [48] Ward, M. D. Coord. Chem. Rev. 2007, 251, (13), 1663-1677. [49] Ronson, T. K.; Lazarides, T.; Adams, H.; Pope, S. J. A.; Sykes, D.; Faulkner, S.; Coles, S. J.; Hursthouse, M. B.; Clegg, W.; Harrington, R. W.; Ward, M. D. Chem. Eur. J. 2006, 12, (36), 9299-9313. [50] Xu, H. B.; Shi, L. X.; Ma, E.; Zhang, L. Y.; Wei, Q. H.; Chen, Z. N. Chem. Commun. 2006, (15), 1601-1603. [51] Herrera, J.-M.; Ward, M. D.; Adams, H.; Pope, S. J. A.; Faulkner, S. Chem. Commun. 2006, (17), 1851-1853. [52] Guo, D.; Duan, C. Y.; Lu, F.; Hasegawa, Y.; Meng, Q. J.; Yanagida, S. Chem. Commun. 2004, (13), 1486-1487.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

72

Silvio Quici, Lidia Armelao, Francesco Barigelletti et al.

[53] Klink, S. I.; Keizer, H.; Van Veggel, F. C. J. M. Angew. Chem., Int. Ed. Engl. 2000, 39, (23), 4319-4321. [54] Sato, S.; Wada, M. Bull. Chem. Soc. Jpn. 1970, 43, (7), 1955-1962. [55] Tanaka, M.; Yamaguchi, G.; Shiokawa, J. Bull. Chem. Soc. Jpn. 1970, 43, (2), 549-550. [56] Hayes, A. V.; Drickamer, H. G. J. Chem. Phys. 1982, 76, (1), 114-125. [57] Horrocks, W. D.; Sudnick, D. R. Acc. Chem. Res. 1981, 14, (12), 384-392. [58] Beeby, A.; Clarkson, I. M.; Dickins, R. S.; Faulkner, S.; Parker, D.; Royle, L.; De Sousa, A. S.; Williams, J. A. G.; Woods, M. J. Chem. Soc., Perkin Trans. 2 1999, 3, 493-504. [59] Weber, M. J. Phys. Rev. 1968, 171, (2), 283-291. [60] Werts, M. H. V.; Jukes, R. T. F.; Verhoeven, J. W. Phys. Chem. Chem. Phys. 2002, 4, (9), 1542-1548. [61] Parker, D. Coord. Chem. Rev. 2000, 205, 109-130. [62] Lehn, J.-M. Design of organic complexing agents Strategies towards properties; Structure & Bonding; Springer Berlin: Heidelberg, 1973, 16, pp 1-69. [63] Englman, R.; Jortner, J. Mol. Phys. 1970, 18, (2), 145-164. [64] Caspar, J. V.; Meyer, T. J. J. Am. Chem. Soc. 1983, 105, (17), 5583-5590. [65] Quici, S.; Marzanni, G.; Cavazzini, M.; Anelli, P. L.; Botta, M.; Gianolio, E.; Accorsi, G.; Armaroli, N.; Barigelletti, F. Inorg. Chem. 2002, 41, (10), 2777-2784. [66] Juris, A.; Balzani, V.; Barigelletti, F.; Campagna, S.; Belser, P.; Von Zelewsky, A. Coord. Chem. Rev. 1988, 84, 85-277. [67] Sauvage, J. P.; Collin, J. P.; Chambron, J. C.; Guillerez, S.; Coudret, C.; Balzani, V.; Barigelletti, F.; De Cola, L.; Flamigni, L. Chem. Rev. 1994, 94, (4), 993-1019. [68] Bardez, E.; Devol, I.; Larry, B.; Valeur, B. J. Phys. Chem. B 1997, 101, (39), 77867793. [69] Bakker, B. H.; Goes, M.; Hoebe, N.; Van Ramesdonk, H. J.; Verhoeven, J. W.; Werts, M. H. V.; Hofstraat, J. W. Coord. Chem. Rev. 2000, 208, 3-16. [70] Albrecht, M.; Fiege, M.; Osetska, O. Coord. Chem. Rev. 2008, 252, (8+9), 812-824. [71] Förster, T. Discuss. Faraday Soc. 1959, 27, 7-17. [72] [Dexter, D. L. J. Chem. Phys. 1952, 21, 836-850. [73] Vögtle, F.; Weber, E. Angew. Chem., Int. Ed. Engl. 1979, 18, (10), 753-776. [74] Reichert, D. E.; Lewis, J. S.; Anderson, C. J. Coord. Chem. Rev. 1999, 184, 3-66. [75] Sammes, P. G.; Yahioglu, G. Nat. Prod. Rep. 1996, 13, (1), 1-28. [76] Li, M.; Selvin, P. R. J. Am. Chem. Soc. 1995, 117, (31), 8132-8138. [77] Hakala, H.; Liitti, P.; Peuralahti, J.; Karvinen, J.; Mukkala, V.-M.; Hovinen, J. J. Lumin. 2005, 113, (1-2), 17-26. [78] Hemmila, I.; Mukkala, V.-M. Crit. Rev. Clin. Lab. Sci. 2001, 38, (6), 441-519. [79] Selvin, P. R. Annu. Rev. Biophys. Biomolec. Struct. 2002, 31, 275-302. [80] Izatt, R. M.; Pawlak, K.; Bradshaw, J. S.; Bruening, R. L. Chem. Rev. 1991, 91, (8), 1721-2085. [81] Aime, S.; Batsanov, A. S.; Botta, M.; Howard, J. A. K.; Parker, D.; Senanayake, K.; Williams, G. Inorg. Chem. 1994, 32, (21), 4696-4706. [82] Nonat, A.; Gateau, C.; Fries, P. H.; Mazzanti, M. Chem. Eur. J. 2006, 12, (27), 71337150. [83] Balogh, E.; Tripier, R.; Fouskova, P.; Reviriego, F.; Handel, H.; Toth, E. J. Chem. Soc., Dalton Trans. 2007, 32, 3572-3581.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Antenna-Lanthanide Complexes: A Growing Technology-Driven Research

73

[84] Beeby, A.; Bushby, L. M.; Maffeo, D.; Williams, J. A. G. J. Chem. Soc., Dalton Trans. 2002, 1, 48-54. [85] Quici, S.; Cavazzini, M.; Raffo, M. C.; Botta, M.; Giovenzana, G. B.; Ventura, B.; Accorsi, G.; Barigelletti, F. Inorg. Chim. Acta 2007, 360, (8), 2549-2557. [86] Barigelletti, F.; Flamigni, L. Chem. Soc. Rev. 2000, 29, (1), 1-12. [87] Kirby, A. F.; Richardson, F. S. J. Phys. Chem. 1983, 87, (14), 2544-2556. [88] Horrocks, W. D.; Sudnick, D. R. Acc. Chem. Res. 1981, 14, (12), 384-392. [89] Sanchez, C.; Soler-Illia, G. J. A. A.; Ribot, F.; Grosso, D. Compt. Rend. Chim. 2003, 6, (8-10), 1131-1151. [90] Nishioka, T.; Yuan, J.; Yamamoto, Y.; Sumitomo, K.; Wang, Z.; Hashino, K.; Hosoya, C.; Ikawa, K.; Wang, G.; Matsumoto, K. Inorg. Chem. 2006, 45, (10), 4088-4096. [91] Soini, E.; Lovgren, T. Crit. Rev. Anal. Chem. 1987, 18, (2), 105-154. [92] Dickson Gudgin, E. F.; Pollak, A.; Diamandis, E. P. Pharmacol. Ther. 1995, 66, (2), 207-235. [93] Matsumoto, K.; Yuan, J. In Metal Ions in Biological Systems; A. S. a. H. Sigel; Marcel Dekker: New York, 2003, 40, pp 191-232 [94] Yuan, J.; Wang, G. J. Fluoresc. 2005, 15, (4), 559-568. [95] Hemmilä, I.; Soini, E. Clin. Chem. 1979, 25, (3), 353-361. [96] Yan, J.; Estévez, M. C.; Smith, J. E.; Wang, K.; He, X.; Wang, L.; Tan, W. Nano Today 2007, 2, (3), 44-50. [97] Tan, W.; Wang, K.; He, X.; Zhao, X. J.; Drake, T.; Wang, L.; Bagwe, R. P. Med. Res. Rev. 2004, 24, (5), 621-638. [98] Smith, J. E.; Wang, L.; Tan, W. Trends Anal. Chem. 2006, 25, (9), 848-855. [99] Bagwe, R. P.; Yang, C.; Hilliard, L. R.; Tan, W. Langmuir 2004, 20, (19), 8336-8342. [100] Bagwe, R. P.; Hilliard, L. R.; Tan, W. Langmuir 2006, 22, (9), 4357-4362. [101] Zhao, X.; Bagwe, R. P.; Tan, W. Adv. Mater. 2004, 16, (2), 173-176. [102] Kim, H. K.; Kang, S.-J.; Choi, S.-K.; Min, Y.-H.; Yoon, C.-S. Chem. Mater. 1999, 11, (3), 779-788. [103] Sun, L. N.; Zhang, H. J.; Fu, L. S.; Liu, F. Y.; Meng, Q. G.; Peng, C. Y.; Yu, J. B. Adv. Funct. Mater. 2005, 15, (6), 1041-1048. [104] Kido, J.; Okamoto, Y. Chem. Rev. 2002, 102, (6), 2357-2368. [105] Li, Y.; Yan, B.; Yang, H. J. Phys. Chem. C 2008, 112, (10), 3959-3968. [106] Tanner, P. A.; Yan, B.; Zhang, H. J. Mater. Sci. 2000, 35, (17), 4325-4328. [107] Koslova, N. I.; Viana, B.; Sanchez, C. J. Mater. Chem. 1993, 3, (1), 111-112. [108] Armelao, L.; Bottaro, G.; Quici, S.; Cavazzini, M.; Raffo, M. C.; Barigelletti, F.; Accorsi, G. Chem. Commun. 2007, 28, 2911-2913. [109] Quici, S.; Cavazzini, M.; Raffo, M. C.; Armelao, L.; Bottaro, G.; Accorsi, G.; Sabatini, C.; Barigelletti, F. J. Mater. Chem. 2006, 16, (8), 741-747. [110] Sanchez, C.; Lebeau, B. MRS Bull. 2001, 26, (5), 377-387. [111] Ribot, F.; Sanchez, C. Comments Inorganic Chem. 1999, 20, (4-6), 327-371. [112] Sanchez, C.; Soler-Illia, G. J. A. A.; Ribot, F.; Lalot, T.; Mayer, C. R.; Cabuil, V. Chem. Mater. 2001, 13, (10), 3061-3083. [113] Soler-Illia, G. J. A. A.; Sanchez, C.; Lebeau, B.; Patarin, J. Chem. Rev. 2002, 102, (11), 4093-4138. [114] Boury, B.; Corriu, R. J. P. Chem. Commun. 2002, 8, 795-802. [115] Corriu, R. J. P. Angew. Chem., Int. Ed. Engl. 2000, 39, (8), 1376-1398.

74

Silvio Quici, Lidia Armelao, Francesco Barigelletti et al.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

[116] Corriu, R. J. P.; Leclercq, D. Angew. Chem., Int. Ed. Engl. 1996, 35, (13/14), 14211436. [117] Corriu, R. J. P.; Moreau, J. J. E.; Thepot, P.; Wong Chi Man, M.; Chorro, C.; LerePorte, J.-P.; Sauvajol, J.-L. Chem. Mater. 1994, 6, (5), 640-649. [118] Brinker, C. J.; Scherer, G. W. Sol-Gel Science: The Physics and Chemistry of Sol-Gel Processing; Academic Press Inc.: San Diego, 1990, pp 912. [119] Schubert, U.; Hüsing, N. Synthesis of Inorganic Materials; Wiley - VCH: Chichester, 2000, pp 396. [120] Gupta, R.; Chaudhury, N. K. Biosensors & Bioelectronics 2007, 22, (11), 2387-2399. [121] Winter, R.; Hua, D. W.; Song, X.; Mantulin, W.; Jonas, J. J. Phys. Chem. 1990, 94, (6), 2706-2713. [122] Lev, O.; Tsionsky, M.; Rabinovich, L.; Glezer, V.; Sampath, S.; Pankratov, I.; Gun, J. Anal. Chem. 1995, 67, (1), 22A-30A. [123] Ellerby, L. M.; Nishida, C. R.; Nishida, F.; Yamanaka, S. A.; Dunn, B.; Valentine, J. S.; Zink, J. I. Science 1992, 255, (5048), 1113-1115. [124] Keeling-Tucker, T.; Brennan, J. D. Chem. Mater. 2001, 13, (10), 3331-3350. [125] Armelao, L.; Barreca, D.; Bottaro, G.; Gasparotto, A.; Tondello, E.; Ferroni, M.; Polizzi, S. Chem. Vap. Dep. 2004, 10, (5), 257-264. [126] Armelao, L.; Barreca, D.; Bottaro, G.; Gasparotto, A.; Tondello, E.; Ferroni, M.; Polizzi, S. Chem. Mater. 2004, 16, (17), 3331-3338.

In: Antennas: Parameters, Models and Applications Editor: Albert I. Ferrero

ISBN978-1-60692-463-1 ©2009 Nova Science-Publishers, Inc.

Chapter 3

DESIGN AND OPTIMISATION OF ANTENNAS USING GENETIC ALGORITHMS FOR WIRELESS COMMUNICATIONS D. Zhou, R.A. Abd-Alhameed∗, C.H. See and P.S. Excell Mobile and Satellite Communications Research Centre Bradford University, Bradford, UK

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

1. INTRODUCTION The needs for global antenna design tools are always crucial due to the wide range of wireless applications appear nowadays for which the antennas should meet certain required performances. This poses numerous research challenges in this field to find the optimal solution and to overcome the limitations imposed by the design specifications. Thus a significant amount of a research work is needed to develop the scientific tools to state the art of the antenna analysis. Several types of optimizers are combined with these solutions tools such as the well known Genetic Algorithms (GA) that use binary based random search engine subject to various antenna design constraints. The work presented here, includes several designed and optimised antennas using GA. The Genetic algorithm driver, written in FORTRAN, was adopted in this work in conjunction with the industry-standard NEC-2 FORTRAN source code, which was used to evaluate the randomly generated antenna samples. Design examples of antennas were successfully demonstrated in this GA method and their results were verified through individual hardware realisation. Four types of antennas were proposed in this study for various applications and they include (1) design of quadrifilar helical antenna in the presence of small handset for mobile satellite wireless communication systems; (2) design of folded loop balanced antenna for mobile handsets; (3) design of microstrip patch antennas with circular polarisation; (4) design of antennas for wide harmonic suppression for active integrated antennas.

∗

Email: [email protected], Tel: +44 (0)1274234033, Fax: +44 (0)1274234525;

76

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

In this chapter, firstly, a summary on the sophisticated genetic algorithms and its design procedure in collaborating with NEC-2 for antenna designs and optimization were briefly described. Secondly, the first two examples on antennas using GA were demonstrated to prove the capability of GA as a quick optimization tool in antenna designs. The results for the proposed antennas from both the modelling and the prototype will be compared for validation. Thirdly, a novel program for adaptively meshing planar antennas using wire-grid structure is introduced, and then two examples on the design of microstrip patch antennas are presented. This program is written in FORTRAN by the present authors and added as a subroutine to the GA driver, with the primary objective of simulating air-dielectric planar microstrip patch antenna designs, using wire-grid models simulated with the NEC-2 code in cooperation with a GA. Finally, the antenna design of the two examples was investigated using adaptive meshing and genetic algorithms. Several simulated results including measurements were presented for comparison and discussion.

2. THE GENETIC ALGORITHM

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

2.1 Introduction to Genetic Algorithms Genetic algorithms are stochastic search procedures orchestrated by natural genetics, selection and evolution [1]. They are modelled on Darwinian concepts of natural evolution thus making them more inspiring during use [2]. After it’s first introduction in 1960’s by J. Holland, it has become an efficient tool for search, optimization and machine learning, but in the pre – GA era, concepts of it had been looming and applied in game playing and pattern recognition [3]. Over the recent years, it has proven to be a promising technique for different optimizations, designs and control applications. Basically, GA exerts pressure on a set or population of possible solutions managing them to evolve towards a global optimal point. This is achieved by a fitness weight selection process and severe exploration of the solution search space attained through recombination (crossover) and mutation of the characteristics present in the particular population considered. When the GA is used as an optimizer, it was found very effective and robust especially if the goal of the operation is to locate an approximate global maximum in complex combinatorial and search related problems. The powerful heuristics of the GA are essentially efficient to dynamically update the parameters applied to the input measurements, operates on them and produce near optimal solutions. Genetic algorithms show more promises because among other search algorithms, it examines all possible solutions in the search space of unknown parameters and eventually identifies the most suitable and fittest solution to the complex problem. Due to the unique ability to rigorously search the entire defined search space, it’s always been referred to as a robust and highly efficient technique with better performances; capable in solving complex problems in various engineering applications and fields.

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

77

2.2 Why Genetic Algorithms? At this point, it is highly necessary to point out the reasons why GA is mostly used nowadays in different optimization processes. Majority of the optimization methods are classified as either; • •

Global techniques with familiar examples such as Genetic Algorithms, random walk, simulated annealing and Monte Carlo. Local techniques with familiar examples such as conjugate gradient, quasi Newton and simplex methods.

A major difference between these two optimization techniques is that the local techniques tend to produce results extremely dependent upon initial start conditions and they couple tightly to the solution domain thereby converging relatively fast and producing local maximum results. On the other hand, global techniques are independent of these starting conditions and place certain constraints on them. Thus, this makes the global techniques to be robust and perform better even if there are discontinuities in the solution domain. Furthermore, it’s been observed that there are three main situations when the genetic algorithms tend to be more useful. These are; • • •

If the problem at hand requires quite a number of parameters. There are multiple local optima solutions present. A non – differentiable objective function.

From this point of view, a deeper look into these two groups of optimization process shows that the local techniques converge faster than the global techniques. However, in the electromagnetic design problems, the rate of convergence is relatively less important but optimal results are vital. Amongst the global techniques, the GA is more suitable for electromagnetic design issues. It is faster, reliable, robust and easily programmed and readily implemented [4].

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

2.3 Terminologies in GA To be able to appreciate and have a common understanding of this particular discussion about GA, we ought to dive more deeply into some fundamental definitions. The following summarizes the most important concepts many of which are similar and borrowed from the concepts of natural evolution. • • • •

Generation: A set of fit individuals which were successively created. Parent: These are members selected in a probabilistic manner from a particular initialized population. They are usually weighted relative to their fitness values [1]. Children: Usually generated by initially selected parents to form a new generation. They are products of the major genetic operators; crossover and mutation. Fitness: A value that dictates the measure of the goodness of each individual.

78

D. Zhou, R.A. Abd-Alhameed, C.H. See et al. • • •

•

Genes: Coded parameters required for optimization. Chromosomes: A couple of genes in string format. Objective function: Basically a numerical representation in simple equation formats, of the required goal in an optimization problem [5]. It defines if the complex problem is been maximized or minimized and could also be referred to as the cost function or fitness function. Search space: It’s the region that contains all possible solutions assumed by the design engineer [5]. Due to the chances of flexibility during optimization, the search space might contain solutions outside the feasible conditions. The intelligent GA tends to isolate them and select the appropriate optimal solutions from this pool. In addition to this, the format of the search space boundaries must be carefully taken to prevent early convergence on less optimal results.

From all described so far, the whole optimization process involves some basic tasks which are enumerated below as follows: • • • • • •

Solution parameters are encoded as genes. Chromosomes are formed from strings of genes. A random initiation to create a starting population. Individuals in the population are evaluated and assigned fitness values. Reproduction of the fit individuals selected. Genetic operators, crossover and mutation, to generate new set of individuals.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

2.4 GA Step by Step Implementations In general, a typical genetic algorithm optimizer consists of three main phases as described by Rahmat–Samii etc [1]. These stages include initiation, reproduction and new generation (i.e. generation replacement). The whole process of Genetic Algorithm is kicked–off by encoding the parameters into either a binary or real-valued format. The coding of the parameters (real values or binary) is highly required as this enables the GA to proceed in a style independent of the parameters themselves and thus independent of the solution space [5]. Genes are then allocated to represent these parameters which are used throughout the process to model the evolutionary algorithm. A set of population is selected randomly from the allocated genes and this is called the initiation stage. After the generation of the initial population, the fitness values of the individuals are evaluated. The evaluation of these values determines the survival of the individuals in this randomly generated generation to proceed to the selection stage. In simple terms, the fitness values describes or measures how good the individuals (i.e. antenna samples generated by GA in this study) are able to produce desired results to an extent after combination with themselves. This stage is usually referred to as the evaluation stage. The selection stage tends to identify the genes with the highest fitness values to enable them migrate into the mating phase in which a more ideal and better generation is produced by the algorithm. It is a stage whereby pressure is applied upon the population in an approach

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

79

similar to those of natural selection. This stage is usually executed with different techniques such as; tournament selection, population decimation and proportionate selection also known as roulette wheel. For the roulette wheel, the individuals are selected based on fitness proportional to a probability equation written in Eqn. 1, where f ( parent i ) is the fitness of the ith parent [1].

p selection =

f ( parent i ) ∑ f ( parent i )

1

i

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

It could be inferred from this equation after a close look that the probability of identifying and selecting an individual from the pool of population by the algorithm is solely a function of the relative fitness of the particular individual. The selection procedure is usually applied twice in an attempt to obtain a set of individuals suitable for the GA operators to act on appropriately [2]. The second popular technique is the tournament selection because it is relatively straight forward. A subpopulation of N individuals is chosen randomly from the selected population. These individuals in the subpopulation then compete with their fitness values and the one with the highest fitness value wins the tournament and is selected and isolated. The remaining less fit of the subpopulation is then replaced back to the former pool of individuals and the whole process repeated until all the members of the subpopulation are selected. It is a faster mode of selection and it is used in the GA optimizer implemented in this study. The preceding stage after selection is the reproduction phase in which a new generation is produced by crossovers and mutation operators. They are implemented in straightforward code segments. Simply put, crossover is the process by which the genes of a parent are combined with those of another parent to produce children with better genes. Several modes of crossover have been tried but the simplest of all is the single-point crossover demonstrated briefly below in Fig. 1 [6]. A random location is selected in the parent’s chromosomes and swapped to produce the children. Split Position …………………………………+ Fathers genes: 11101000111001100110 10101 Mothers genes: 10101110011101010111 00110 Left side genes from father + right side genes from mother Child A’s genes: 11101000111001100110 00110 Left side genes from mother + right side genes from father Child B’s genes: 10101110011101010111 10101 Figure 1. Single point crossover illustrations.

It is obvious that the individuals in the generation that are not selected for crossover operations are acted upon by mutation. It also changes part of the chromosomes string in order to maintain at least some traits in the new generation formed. All these processes are continued until a new generation is formed by replacing the old generation totally or partially and their fitness values evaluated intermittently. After some

80

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

certain runs and considering the input search space, GA tends to identify the near optimal value. Fig. 2 is the pictorial diagram illustrating the GA optimizer that summarizes all processes described earlier.

Figure 2. Design procedure of the GA optimizer.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

2.5 The Genetic Algorithm Driver and Implementation During the current research, it was noticed that various versions of the Genetic Algorithms driver are available for use as an optimizer, in which they are implemented in C, MATLAB and FORTRAN 77. From our experiences on home programmes, FORTRAN 77 seemed more friendly and easier to manipulate and thus was chosen for this optimization processes. The FORTRAN 77 version of the GA driver, written by David L. Carroll of the CU Aerospace USA [7], uses the randomized approach to initialize its start individuals and the tournament selection with shuffling techniques in choosing random pairs for mating. Binary coding is also enabled the uniform and non-uniform process of single point crossovers. Each one of the GA parameters presented, the GA driver can be controlled and adjusted through an associated file called GA input. Fig. 3 illustrates a sample of GA driver input file. As can be seen, some of the parameters in the input file have been highlighted and explained. These variables are the most important and influential to the GA driver in the antenna design and should be adaptively adjusted according to the various design types or objectives, in order

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

81

to maximize the GA driver performance in searching for solutions in antenna designs. It is notable that this GA driver is for maximize the design target objective.

2.6 Implementation of Antenna Designs Using GA Driver It is well known that, NEC-2 FORTRAN source code was adopted inside the GA fitness function to perform the required calculations answers for the cost functions. The source code was modified to process the input data file in which to support the calling function required. These modifications are found very helpful to reduce the execution processing time and manipulate the output data files.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 3. A sample of GA driver input file.

However, before the process of optimization is initiated, one required defining the target objectives and number of parameters required of the whole process in order to achieve the optimum desired goal. In simple terms, some of the most important antenna parameters which are directly targeted were selected for optimization and the desired objectives were those usually required by the end users. Sometimes a relationship was required to define a threshold for the GA which enables it to evaluate the designed antenna performance and terminate where necessary. Usually, this is a complex procedure to be applied; however, one can apply a certain constraints inside the cost function to support the data processing when nearly reaching the optimal design requirements. The cost function is usually included in the algorithm and it measures the fitness of the individuals produced in each generation of the algorithm [8].

82

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

A flow chart to represent the easiest way in which the GA optimizer coordinates its functions is represented in Fig. 4. The algorithm randomly initiates its population and converts the parameters of the initiated individuals into a file in a card format which can be called by NEC to determine the performance of these individuals. The results from NEC are fed again to the GA engine to evaluate individual fitness if the maximum value is obtained for convergence, if otherwise the whole process is repeated until optimal results are produced [9].

3. NEC-2 SOURCE CODE

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

NEC-2 was originated by the US Department of Defence and more specifically, it was developed at Lawrence Livermore National Laboratory in California under the sponsorship of the Naval Ocean Systems Centre and the Air Force Weapons Laboratory [10-11]. NEC is an advanced version of the antenna modelling program (AMP) developed in early 1970’s. NEC2 is the most popular and widespread electromagnetic code in the public domain. This is because it is free and easy to use with scripting programs. In addition to the full source code of NEC is available and can be modified without any restriction. Generally, NEC-2 is applied for thin wires (the basic modelling used as short straight segments) and closed conducting surface patch antenna structure (the basic modelling used as flat surface patches) and based on solving integral equations by the Method of Moments (MoM).

Figure 4. Flow chart of the genetic algorithm adopted in this study.

In most cases, the Method of Moments is used to form equations like the integral equations or the integro-differential ones. The essential idea in using MoM is the discretization of the problem into smaller linear bounded elements, which are treated as independent by using many functions, known as “basis functions”. Then the inner integral of the corresponding resulted functions is taken by applying the proper weighting functions. At

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

83

the end, the linear equations can take the form of a matrix and can be solved by a simple matrix inversion [12].

4. EXAMPLES ON ANTENNA DESIGNS USING GA 4.1 Design of Quadrifilar Helix Antenna using GA

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

In this section, two examples on antenna designs using the genetic algorithm driver in collaboration with NEC-2 source code [13] are presented. They are the quadrifilar helix antenna (QHA) for the use in mobile satellite communications and balanced folded loop antenna (FLA) for mobile handsets. The antennas firstly attempted to be designed and optimized using a genetic algorithm and then subsequently the performance of the optimal GA antennas was verified using a commercial EM simulator. By the end of optimization outcomes, a prototype of GA-optimized antenna was fabricated and tested in order to validate the GA solution. The quadrifilar helix antenna is a very attractive candidate antenna and has been widely used for satellite mobile handsets due to the symmetry of their geometry, properties of balanced feeding and their ability to provide circular polarization over a broad angular region [14]. The QHA is a circular polarised (CP) antenna consisting of four helices and fed with equal amplitude signals and with 90° phase difference between the feeding sources (i.e., 0°, 90°, 180°, and 270°). The presented results in this section are a case study in which GA are applied to design and optimise circular polarized QHA in the presence of a small size satellite mobile handset ( 2 x 5 x 10 cm). The QHA antenna can be defined by four parameters, including axial length (h), Pitch distance (Pd), Radius at bottom (Rb) and Radius at the top (Rt), as illustrated in Fig. 5.

Figure 5. QHA antenna configuration used by GA optimization

Real-valued GA chromosomes were used for this optimisation. Two most important antenna parameters such as VSWR and axial ratio (AR) are optimized at a single frequency (fo = 2.44 GHz). The antenna AR was calculated at fo at θ = 0° and φ = 0°. Each antenna

84

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

sample was computed using NEC-2 source code and its results were compared with desired fitness using the following cost function ‘F’:

F = W1 × (1 VSWR ) + W2 × A.R.

2

Where

VSWR = (1 + Γ ) (1 − Γ )

3

Γ = (Z in − 50) (Z in + 50)

4

Where F is the Fitness of the cost function, VSWR is the voltage standing wave ratio, A.R. is the axial ratio, Z in is the input impedance, Γ is the reflection coefficient and W1 and W2 are the weighting coefficients. The objective was to maximise F. Table 1. Comparative results of VSWR, AR and Fitness as the values of the weighting coefficients are varied

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Weighting (W1, W2) VSWR AR Fitness

0.2, 0.85 1.74531 0.92561 0.90064

0.3, 0.75 1.66912 0.92665 0.87472

0.4, 0.75 2.03473 0.98089 0.93266

0.5, 0.75 1.49462 0.92493 1.0282

Using Eqn. 2 the algorithm ensures that the maximum value of F is obtained through the combination of all the antenna parameters, although it should be noted that during all optimization designs, trade offs are usually expected. The parameter quantities of each helical antenna design were coded into chromosomes inside the source code of the algorithm. It has to be noted that the two weighting coefficients are optimally found to be 0.5 and 0.75 respectively for optimum design after a few tries, as illustrated in Table 1. It presents the comparative results of VSWR, AR and Fitness as the values of the weighting coefficients are varied. Within the maximum generation, the values of maximum fitness function for QHA design reached to be about 1.03. Moreover, a comparison of maximum fitness versus generations of different combination choice of the two weighting coefficients is shown in Fig. 6. The value of best fitness for the cost function tends to reach optimum solution after around 150 generations. Fig. 7 presents the progress of best fitness and average fitness against the number of generations for some selected values of weighting coefficients. Configurations for the GA-optimal QHA antenna, with Excellent VSWR and AR values were found within the maximum generation; the optimal values for each specified parameter are shown in Table 2. This Table also includes a summary of the GA input parameters and their constraints. The attained optimal antenna geometry for QHA antenna is presented in Fig. 8.

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

85

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 6. Maximum fitness versus number of generations (this example uses 4 populations in each generation).

Figure 7. The progress of best fitness and average fitness for w1 and w2 are 0.5 and 0.75, respectively.

86

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 8. The NEC-2 model of the QHA.

Figure 9. Prototype QHA antenna; internal view of the completed assembly (left) and overall complete assembly (right).

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

87

Table 2. Summary of GA input parameters, antenna variables and optimum values with the handset included GA parameters

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Number of population size = 4 Number of parameters = 4 Probability of mutation =0.02 Maximum generation =200 Number of possibilities=32768

QHA Design with the handset Parameters (m) Optimum (m) Pitch distance (Pd) 0.03026 (0.01-0.048) Axial length (h) 0.06294 (0.05-0.12) Radius at the bottom (Rb) 0.00721 (0.005-0.015) Radius at the top (Rt) 0.01194 (0.01-0.02) Radius of wires 0.00075 Distance above handset 0.005

For validation, a prototype of the GA-optimised QHA antenna was built up and tested. Photographs of prototype antenna, including configuration of the proposed antenna in NEC-2 model and an overall view of the complete assembly, are presented in Fig. 9. It is notable that the hybrid feeding network for the QHA was arranged to accommodate inside the handset box. The QHA arms were made of copper wires with radius of 0.75 mm. The relative bandwidth (for VSWR ≤ 2) at the input ports of the proposed circularly-polarised antenna was calculated and measured over the targeted frequency bands, i.e. 2.4 GHz band from 2400 MHz to 2485 MHz, as illustrated in Fig. 10. The bandwidth of the designed antenna was not considered in the GA cost function, the optimal antenna appears to have excellent impedance matching that covers the bandwidth requirements at 2.4 GHz band for satellite mobile communications. Both the simulated and measured VSWR are in good agreement. It is interesting to note that the resultant measured impedance bandwidth was found to be about 150 MHz (referring to 2:1 VSWR). It should be noted that the antenna bandwidth was not considered in the GA cost function, the resultant GA-optimized QHA antenna appear to have an excellent impedance matching. Fig. 11 illustrates the AR of the GA-optimized antenna against the elevation angle θ at 2350, 2400, 2420 and 2450 MHz for two vertical planes (i.e. zx and zy planes). As can be observed, the proposed antenna shows ±70° elevation angle variations for a working axial ratio less than 4 dB. The observations confirm the superior circular polarized characteristic of the proposed optimal design. Figs. 12 and 13 present the radiation pattern of the optimal antenna at 2400 and 2450 MHz for the same vertical cuts of the measured axial ratio. Radiation patterns at other frequencies across the intended band (not shown here) were also measured and it was found, that the symmetrical and identical variations were obtained for all the radiation patterns. In addition, the maximum gain of the antenna is found to be around 6 dBi over the 2.4 GHz band. In summary, the results confirm that an axial ratio of less than 4 dB over ±70° elevation angle can be achieved with acceptable 6 dBi power gain. The GA has proven its advantage for quickly finding solutions for antenna designs. The attained results in simulation and measurement indicate that the optimal antennas met design objectives under several certain constraints. Moreover, the capabilities of GA are shown as an efficient optimisation tool for

88

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

selecting globally optimal parameters to be used in simulations with an electromagnetic antenna design code, seeking convergence to designated specifications.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 10. Measured VSWR of the QHA.

Figure 11. Measured AR in dB against the elevation angles for two azimuth cuts φ = 0° and 90° and four operating frequencies.

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

89

Figure 12. Measured radiation pattern of the proposed antenna at 2400 MHz at zx plane (left) and zy plane (right); ‘o-o-o’: co-polar field component, ‘x-x-x’ cross-polar field component.

Figure 13. Measured radiation pattern of the proposed antenna at 2450 MHz at zx plane (left) and zy plane (right); ‘o-o-o’: co-polar field component, ‘x-x-x’ cross-polar field component.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

4.2 Design of a Balanced Folded Loop Antenna Using GA A FLA for mobile handsets with relatively wideband impedance was designed and optimized using GA. The geometry of proposed FLA was adopted from the Morishita’s work [15] (see Fig. 14) and applied here for this study. Parameters, used to define the FLA, were optimized and evaluated using GA in collaboration with NEC-2. Finally, GA optimal antenna structure was verified and compared using the commercial EM simulator CST and a good agreement in VSWR was observed. A prototype antenna was also fabricated and tested in order to validate the results obtained in the prediction. The FLA was optimized with GA using real-valued chromosomes. The intended antenna was designed for Global System for Mobile Communications (GSM) applications (1710-1860

90

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

MHz). Performance of the randomly generated antenna samples was computed using NEC-2 and its result was compared with desired fitness using a cost function, as follows, 3

F=

∑W

n

⋅ (1 VSWR ) fn

5

n =1

Where, Wn (n=1…3) are weighting coefficients for the cost function. All the weighted coefficients are set to be 1. Three pre-set specific frequencies fn (1710, 1785, and 1860 MHz) were applied for NEC-2 and GA for each erratically produced antenna structure, in order to ensure the optimal antenna covers the required impedance bandwidth.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 14. Antenna configuration studied. (a) balanced folded loop antenna with conducting plate; (b) front view of the antenna design; (c) side view of the antenna design.

GA input parameters, their constraints and the optimal values for each specified parameter of the design geometry are presented in Table 3. The GA generated antenna structures can be viewed using the NEC-Win Professional simulator. The basic geometry of the optimal FLA antenna with excellent VSWR covering entirely required GSM1800 frequencies bands was shown in Figs. 15 and 16. This was found within the maximum generations for which the antenna parameters of the best designs are shown in Table 3. CST simulator [16], based on the finite integration technique, was used to verify and validate the GA-optimized antenna structure. It should be noted that thin strip lines were employed in the CST model instead of thin wires used by GA optimization, due to the fact of the practical implementation (see Fig. 15).

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

91

Figure 15. Folded loop antenna model using CST simulator.

Table 3. Summary of GA input parameters, antenna variables and best solutions FLA for GSM1800 Parameters (m) Wire radius (a) Number of (0.0004-0.0008) population size = 6 Number of FLA length (b) (0.03-0.04) parameters = 7 Probability of FLA height (h) mutation =0.02 (0.003-0.012) Maximum FLA arm length (n) (0.002-0.015) generation =500 Parallel wires Number of possibilities=32768 distance (m) (0.005-0.015) Ground plane size FLA distance to GP edge (e) (0.0-0.002) (120 x 50 mm) Distance between FLA and GP (h0) (not shown in Fig. 15) (0.001-0.003)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

GA parameters

Optimal (m) 0.00074 0.03978 0.01173 0.008785 0.01489

0.0008643 0.001112

FLA for UMTS Parameters (m) Wire radius (a) (0.0004-0.0008) FLA length (b) (0.03-0.04) FLA height (h) (0.003-0.012) FLA arm length (n) (0.002-0.015) Parallel wires distance (m) (0.005-0.015) FLA distance to GP edge (e) (0.0-0.002) h0 (0.001-0.003)

Optimal (m) 0.0007905 0.03690 0.01179 0.009881 0.013599

0.001137 0.001146

In addition, the same design principle and antenna geometry was applied to design and optimize antennas for the third generation (3G) wireless mobile communication system application. The comparative antenna VSWR (see Fig. 17) shows a good impedance matching over the intended band (11.3% at f0 =2030 MHz). A prototype of the GA-optimised antenna for GSM 1800 was shown in Fig. 18. The conducting copper thickness of 0.15mm and 0.5mm was used for fabricating the balanced antenna and the ground plane, respectively. The measured return loss of the prototype antenna against simulated one is presented in Fig. 19. As can be seen, the resultant measured impedance bandwidth was found to be about 7.4% at f0 =1765 MHz (referring to -10 dB return loss), which is very encouraging, compared to the simulated result.

92

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

Figure 16. VSWR against frequency (a= 2.0705, b= 37.8802, h= 9.9991, h0=1, n= 13.884, e= 0.9409 and m= 12.8052, all dimensions in mm).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 17. VSWR against frequency (a= 1.95144, b= 37.0052, h= 9.9966, h0=1, n= 11.1638, e= 1.8835 and m= 12.9884, all dimensions in mm).

Figure 18. Photograph of the prototype antenna for GSM1800, excluding the feeding network (balun).

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

93

Figure 19. Measured and simulated return loss of the prototype antenna for GSM1800.

The radiation patterns in the zx plane and zy plane for the balanced folded antenna at 1780 MHz were calculated and plotted in Fig. 20, where the patterns of the proposed antenna are seen to be quite similar to each other at other frequencies inside the band. In addition, the zx plane presents a nearly omni-directional radiation pattern in all intended frequency bands. For the GSM1800 band, the calculated peak gain was found to be 4 dBi.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 20. Radiation patterns for the proposed antenna at 1780 MHz in dBi at zx plane (left) and zy plane (right) (‘+++’ Eθ and ‘o o o’ Eφ).

5. ADAPTIVE MESHING FOR NUMERICAL ANTENNA DESIGNS USING GA 5.1 Motivation on Development of Adaptive Meshing Program In the NEC-2 code, a conducting surface can be modelled using multiple, small flat surface patches similar to the segments used to the model wires. However, NEC-2 allows the use of surface patches for modelling a conducting surface that is incorporate with the magnetic field integral equation in which a closed surface can be implemented. In addition,

94

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

another alternative way for open and closed surfaces modelling can be implemented through the use of wire grids [17]. The wire-grid is an effective approach to employ thin wire to achieve the modelling of metallic planar structures using NEC for patch antenna designs. This technique, it requests a careful selection of parameters, such as segment length and the segment radius. A program to adaptively generate equivalent wire-grid structures for patch antennas for electromagnetic simulation of 2D structures has been developed and presented here. The main purpose of this program is to simulate planar microstrip patch antenna designs, using the NEC-2 code in collaboration with a genetic algorithm. In order to demonstrate how this program operates in meshing planar structures, two examples are illustrated, both involving design of circularpolarized coaxially-fed antennas. It has confirmed that the performances of both GAoptimized antennas were excellent and the presented examples show the capability of the proposed program in antenna design using GA.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

5.2 Principle of Adaptive Meshing Program This program is written in FORTRAN by the present authors and added as a subroutine to the GA driver, with the primary objective of simulating air-dielectric planar microstrip patch antenna designs, using wire-grid models simulated with the NEC-2 code in cooperation with a genetic algorithm. In addition to microstrip patch designs, the program can support the design of any 3D antenna geometry structure that does not contain large amounts of dielectric. Basically, the antenna under optimization needs to be defined by a number of parameters that can define the antenna configuration. Subsequently, the antenna geometry is adaptively divided into optimum numbers of trilateral and quadrilateral polygons by the code user. Each polygon can be represented using either three or four nodes. Each node is specified by its x, y and z co-ordinates subject to the defined antenna parameters. Then, the fictitious wire boundaries of these polygons can be optimally segmented to a pre-set segment length and connected to each other using a designated algorithm, this also creating a mesh of wires within the polygon to make it approximate to the behaviour of sheet metal. The method avoids closely separated wire segments, in which the minimum separation distance considered is four times the wire radius. It also provides the equivalent surface areas between the wire grid model and actual antenna geometry. Assuming that the surface area of the wire grid should approximate the surface area of the polygons plane being modelled, the segmented element radii can be decided using a relationship that the grid wire radius should equal the segment length divided by 2π. Obviously, the more wires in a grid of a certain set of polygon plane dimensions, the smaller the segment length becomes and hence, the smaller the wire radius needs to be. It is notable that segmentations on the adjoining lateral of neighbouring polygons are expected to be overlapped. If a model contains duplicate elements it will not be apparent from the graphical display, but it may significantly corrupt the accuracy of the NEC analysis. Another algorithm for checking and removing duplicate elements of these overlapped segments is consequently applied. Adaptive meshing code creates the required wire grid models of antennas and structures that might be used by other antenna modeling programs. Once the antenna model has been

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

95

created, its wire grid geometry is saved as an ASCII text file that can be read by NEC-2 source code. In addition, this FORTRAN code has also the capability to calculate the total number of segments in the discretised structure and allowing the user to determine whether the size of the model is still within the limits of maximum number of wires used by NEC (it should be noted that no duplicate segments of the overlapping laterals are counted). Since the NEC-2 source code adopted in this work is restricted to the maximum number of segments of 2000. The presentation of the adaptive meshing to any randomly generated antenna configurations using GA can be viewed using graphic support available from NEC-Win Professional Package for checking [18].

5.3 Design Examples on Implementation of Adaptive Meshing

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

In order to demonstrate how this program operates in meshing planar structures and how it is subsequently applied in antenna designs using GA, two examples are illustrated, both involving design of air-dielectric circular-polarized (CP) coaxially-fed patch antennas. Both antennas assumed to operate around 2.48 GHz WLAN band. The first antenna design, having two cutoffs at the diagonal corners, is shown in Fig. 21. In a first pass, the antenna is subdivided into three quadrilaterals, requiring six parameters to define it (including the height h). Fig. 22 demonstrates the adaptive segmentation results as delta (minimum segment length considered on the edge of polygons) was chosen to be 3 mm and was viewed using NEC-Win Professional package. The other example considered a square-slot CP antenna (see Fig. 23). This design was first adaptively segmented into four trilaterals and four quadrilaterals and requires eight parameters. The meshed surface structure can be seen Fig. 24, where delta was also set to be 3 mm. The automatically-generated meshes are generally of a good shape (high area-to-perimeter ratio).

Figure 21. Geometry applied for adaptive meshing using GA.

96

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

Figure 22. Mesh used for Fig. 21 using GA (the dot indicates the optimal feeding point).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 23. A square slot CP antenna geometry applied for adaptive meshing using GA.

Figure 24. Mesh used for Fig. 23 using GA (the dot indicates the optimal feeding point position).

For this optimisation, real-valued GA chromosomes were used and antenna parameters of VSWR and AR are optimized at a single frequency of 2.48 GHz (fo). The antenna AR was calculated at fo with θ= 0° and φ=0°. The antenna performance of each antenna was computed using NEC-2 source code and its input impedance (Zin) and AR were evaluated for desired fitness using the same cost function presented for QHA antenna design. The optimal structure configurations for both optimal circular polarized microstrip antennas, with excellent VSWR and AR, were found within the maximum generations. The

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

97

attained optimal antenna geometries for each antenna using wire-grid mesh are presented in Figs. 22 and 24. Tables 4 and 5 present the GA input parameters, their constraints and the optimal values for each specified parameter of the design geometry. It is notable that the two weighting coefficients were found to be 0.4 and 0.8 correspondingly for both designs after a few attempts. Within the maximum generation, the values of maximum fitness function for the two designs proposed were found around 1.08 and 1.03 for Figs. 22 and 24, respectively. Table 4. Summary of GA input parameters, antenna variables and best solutions GA parameters

No. of population size = 4 No. of parameters = 6 Probability of mutation =0.02 Maximum generation =100 No. of possibilities=32768

Air-dielectric CP antenna with two corners chopped Parameters (m) Optimal (m) Antenna length (L) (0.03-0.07) 0.05835 Antenna width (W) (0.03-0.07) 0.04972 Truncated length (d) (0.002-0.02) 0.00858 Antenna height (h) (0.003-0.01) 0.00643 Feeding position at x-axis (xf) (0.0-0.024) 0.01191 Feeding position at y-axis (yf) (0.0-0.024) 0.01256

Table 5. Summary of GA input parameters, antenna variables and best solutions GA parameters

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

No. of population size = 4 No. of parameters = 8 Probability of mutation =0.02 Maximum generation =200 No. of possibilities=32768

Air-dielectric CP patch antenna with a square slot Parameters (m) Optimal (m) Antenna length (L) (0.04-0.07) 0.05239 Antenna width (W) (0.04-0.07) 0.04571 Antenna height (h) (0.003-0.01) 0.00647 Slot centre position at x-axis (x1) (0.02-0.03) 0.02777 Slot centre position at y-axis (y1) (0.02-0.03) 0.02863 Distance from slot centre to side (r) (0.005-0.01) 0.00804 Feeding position at x-axis (xf) (0.0-0.02) 0.01000 Feeding position at y-axis (yf) (0.0-0.02) 0.01763

The GA-optimized antenna structure for each design achieved by the algorithm was validated using NEC-Win simulator and the results were compared with an electromagnetic source code based on Method of Moment (MOM) [19]. The return loss bandwidth achieved at |Γ| < 10dB are around 400 MHz and 300 MHz for antenna structures shown in Figs. 25 and 26 respectively. It can be clearly noted the operated frequency band at axial ratio ≤ 3dB (for acceptable operated circular polarization) for both optimal antenna structures shown in Figs. 27 and 28 were calculated around 180 MHz and 150 MHz respectively. As also can be seen, the performance of the optimal antenna structures were excellent, in which the presented results are in close agreement as expected to the constraints applied at GA through the cost function. It is worth to mention that antenna bandwidth was not be the main target considered in the GA cost function, but the GA-optimized antennas appear to have excellent impedance matching. The presented results of the GA-optimized method were agreed well with that results obtained from MoM code [19]. This proves the use of grid wire adaptive meshing applied for planner antenna structure using GA.

98

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

Figure 25. Comparison of simulated return loss of GA-optimized antenna with two-corner cutoffs.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 26. Comparison of simulated return loss of GA-optimized antenna with a square slot.

Figure 27. Simulated GA-optimized antenna axial ratio for the CP antenna with two corner cutoffs.

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

99

Figure 28. Simulated GA-optimized antenna axial ratio for the CP antenna with a square slot.

6. DESIGN OF HARMONIC SUPPRESSION ANTENNAS WITH ADAPTIVE MESHING USING GA

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

6.1 Design Objectives for Harmonic Suppression Antennas Active transmitting antennas normally contain significant non-linearity and are always integrated compact design. Therefore, the transistor drain (or collector) will be producing harmonic currents directly into the radiator, and these would be expected to be radiated fairly unwanted power [21]. In active antenna design, the unwanted harmonic contents can be terminated (or eliminated) using the radiating element. In this way, active circuit does not request any additional circuitry for harmonic tuning, and thus, can simplify the circuit design and ended with small compact design. In [22], the modified rectangular patch antenna with a series of shorting pins added to the patch centre line was applied to shape the radiated second harmonic from the active amplifying-type antenna, in order to increase the transmitter efficiency. Unfortunately, the proposed design does not provide the termination for the third harmonic. A circular sector patch antenna with 120° cut out in [22] was investigated and proved to provide additional harmonic termination for the third harmonic, claiming a further enhancement in the transmitter efficiency. Moreover, an H-shaped patch antenna for harmonic suppression was designed and applied in oscillator-type active integrated antennas for the purpose of eliminating the unwanted harmonic radiation [23-24]. In the scenario of designing the active integrated antenna, the microstrip patch antenna not only acts as a radiator, but also provides some circuit functionalities such as matching circuit and band pass filter as in active amplifier-type antenna. In this case, without proper designs to suppress the harmonic radiation from the radiator, some possible unwanted harmonic power can be radiated, which could cause detrimental electromagnetic interference (EMI) to the system [25]. In order to overcome this problem, several techniques have been proposed and demonstrated to control such harmonics for patch antennas, for example: using shorting pins, slots, Photonic Bandgap (PBG) structures, or matching stubs on the antenna feeding line [25-38]. Above these proposed techniques, one interesting fact can be found in

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

100

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

away that most of the modified patch antennas for harmonic suppression were achieved based on a specific reference antenna. It implies that the proposed techniques for successfully rejecting harmonic radiation has certain constrains applied on them, such as the type of feeding used (microstrip line or slot feed) for the patch. In [35], a microstrip-line fed slot antenna was developed for harmonic suppression without using a reference antenna and this was achieved with a rather complex geometry for 5 GHz operation. Assuming the operating frequency was changed, and then the whole antenna structure has to be redesigned, which is believed to be long and complicate in process. Therefore, there is a motivation to develop a novel technique to design harmonic suppression microstrip patch antennas for active integrated applications. It has no much constrains on feeding types or antenna geometries. Moreover, it has to be easily manufactured and redesigned if there is a request on the antenna design specification for different applications (i.e. different operating frequency). In this study, a technique in designing microstrip patch antennas for harmonic suppression is presented and implemented using a genetic algorithm. An antenna that presenting a good impedance matching at the fundamental design frequency (fo) and an ideally maximum reflection at harmonic frequencies (mainly by considering the first two harmonics (2fo and 3fo)), is said to be a harmonic suppression antenna (HSA). Strictly speaking, the response of the HSA in terms of antenna return loss (S11) mostly is a band pass filter having a perfect rejection outside the interested frequency bands. It has to be noted that in some HSA designs [27, 31 and 38], the proposed antenna can still have resonances at frequencies other than the targeted frequencies (i.e. fo, 2fo and 3fo); but they are still claimed as antennas for harmonic suppression as long as the designs present good termination at the intended frequencies. In addition, another constrain to the HSA is that in theory, the input impedance of any HSA design has to be purely reactive at the harmonic frequencies [22]. This is because originally HSA was developed for harmonic termination in order to achieve a Class F operation for the amplifying-type active antennas [22]. In this way, antenna will not radiate any power at the harmonic frequencies and the unwanted power will be reflected back to the active device. The design objectives of antennas for harmonic suppression are necessary to satisfy the two aspects in terms of return loss and input impedance. In the following, four designs of coaxially-fed air-dielectric microstrip patch antenna for rejecting harmonics using a genetic algorithm were presented, including patch antenna with a fully shorted wall, or partially shorted wall, and with a folded patch [39]. They are all designed to operate at 2.4 GHz. The presentation of the antenna geometry and adaptive meshing for the optimal antenna configuration using GA for each one of the designs were presented. For the design cases, prototype antennas of optimal antenna configurations using GA for each one of the designs were tested. The return loss was validated and the results were compared using measurement.

6.2 Microstrip Patch Antenna with a Fully Shorted Wall A simple coaxial-fed air-dielectric patch antenna with a fully shorted wall, operating at 2.4 GHz, was firstly attempted for this study as a simple technique to provide the acceptable harmonic rejection [21]. A FORTRAN adaptive meshing subroutine, used to adaptively generate equivalent wire-grid structures to support the structure design of this antenna was

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

101

added to the GA program. This subroutine provides the suitable link of the GA cost function to the NEC-2. The proposed antenna design is shown in Figs. 29 to 32. The proposed design antenna is subdivided into four trilaterals and two quadrilaterals (including the conducting shorted wall). This model was requiring six parameters (including the patch height h) to be defined. Figs. 30 and 32 demonstrate the top view and 3D of the adaptive wire grid segmentation results.

Figure 29. Top view subdivision of the antenna geometry used for adaptive meshing using GA.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 30. Top view of resulted wire mesh used for Fig. 28.

Figure 31. Side view of the antenna geometry of Fig. 28.

102

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

Figure 32. 3D view of the resulted wire mesh using GA.

Table 6 presents the GA input parameters in which the possible range of parameters magnitudes were shown. For this optimisation, real-valued GA chromosomes were used. In this work, the fundamental, first and second harmonic frequencies were considered inside the GA cost function. The randomly generated antenna configurations were evaluated for maximum fitness using the following cost function:

F = w1

n 1 + ∑ wi Γ(if o ) VSWR( f o ) i = 2

6

Where F is the fitness of the cost function; n = 3; W1, W2 and W3 are the weight coefficients of the cost function and they were optimally found to be 0.6, 0.4 and 0.4 after a few attempts. The geometry configuration of the optimal antenna was found within the maximum generations and it is presented in Fig. 32, in which this was placed on infinite ground plane (the antenna can also be mounted on finite ground plane using the present grid wire method). The computation time consumed for each of the erratically generated antenna samples varied between 60 to 70 seconds, according to the different combination of length, width and height of the patch antenna selected for comprising the antenna configuration. This was achieved by using a PC: 2.8 Pentium IV of 1 GB RAM.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 6. Summary of GA input parameters, antenna variables and best solutions GA parameters

No. of population size = 4 No. of parameters = 6 Probability of mutation =0.02 Maximum generation =200 No. of possibilities=32768

Air-dielectric patch antenna with a shorting wall Parameters (m) Optimal (m) Antenna length (L) (0.03-0.06) 0.04093 Antenna width (W) (0.02-0.06) 0.03305 Shorting wall position (d) (0.002-0.03) 0.00972 Antenna height (h) (0.003-0.01) 0.0079 Feeding point at x-axis (fx) (0.004-0.02) 0.00723 Feeding point at y-axis (fy) (0.004-0.02) 0.01752

For validation, a prototype of the GA-optimised harmonic suppression antenna with a fully shorted wall (see Fig. 33) was subsequently designed and tested using Network

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

103

Analyzer HP 8510C. A copper sheet with thickness of 0.5 mm was used for the production of patch antenna, the shorting wall and the ground plane. The ground plane size was 140 mm x 140 mm and this relatively large size for the purpose of eliminating effect of the finite ground plane. The return loss was validated and measured and compared to the calculated one, as shown in Fig. 34. As can be seen, the results for the rejection of 2nd and 3rd harmonics were quite encouraging and no other resonance or ripples were found at the harmonic frequency bands of interest. Moreover, the performance of the measured return loss of the proposed harmonic suppression antenna at the fundamental and first two harmonic frequencies was also investigated. It was found that the prototype antenna is resonant at 2.47 GHz and presents a quite wide bandwidth around 500 MHz as observed. The reflection coefficient levels at the first and second harmonic frequencies were found to be 1.71 dB and 2.45 dB, respectively, and these results are quite acceptable, as compared with harmonic suppression antennas published in the open literature [27]. It is notable that the measured resonance frequency of the prototype antenna has good agreement with the prediction. Taking the manufacture error into the account and the errors due to the the adaptively meshing wire-grid structures the proposed antenna design for harmonic suppression presents a fairly good accuracy and minimum computation time to evaluate the antenna performance using NEC-2 source code.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 33. Photograph of the fabricated harmonic rejection antenna with a shorting wall.

Figure 34. The measured and simulated return loss of the patch antenna with a shorting wall.

104

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

In addition, input impedance of the prototype antenna was also measured over a wide frequency band, as shown in Fig. 35. A different-scaled input impedance plot is also presented in Fig. 36, in order to precisely observe the variations of the input impedance over the frequencies of interest. As can be shown in Fig. 36, the real part of the input impedance of the proposed antenna is almost constant (less than 10 Ω) at harmonic frequency bands. This is very promising characteristic of the designed procedure of such harmonic suppression. The antenna can also achieve the harmonics rejection from the nonlinear active devices, even when the operating frequency is slightly varied. This is because the proposed antenna provides the required reactive termination around the harmonic frequencies. The measured input impedance of the harmonic rejection antenna with a fully shorted wall for fundamental frequency and its first two harmonics is summarised in Table 7. It shows that almost a perfect matching to 50 Ω was attained at fundamental frequency and fairly small resistive impedance at harmonics was found. This clearly meets the design objectives set up by the proposed model.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 35. The overall measured input impedance of the patch antenna with a fully shorted wall.

Figure 36. Measured input impedance of the harmonic suppression antenna (expanded scale of Fig. 34).

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

105

Table 7. Performance of antenna input impedance of the harmonic rejection antenna with a fully shorted wall at the fundamental and first two harmonics

Frequency (GHz) fo : 2.47 2fo : 4.94 3fo : 7.41

Antenna input impedance (Ω) Real Imaginary 48.92 6.4834 7.9395

-0.5879 -28.917 17.64

Measurements of the far field radiation patterns of the prototype were carried out in a farfield anechoic chamber. The fixed antenna (reference antenna) was a broadband horn (EMCO type 3115) and the spacing between the test antenna and the horn was kept 4 m. Two pattern cuts were taken for four selected operating frequencies that cover the designated whole bandwidth in this study. The radiation patterns in the zx plane and zy plane for the GAoptimized HSA with fully shorted wall at fundamental, second and third harmonic frequencies were measured. These results were presented in Fig. 37, in which the second and third harmonic radiations of the proposed HSA with fully shorted wall are less than 13 dB and 18 dB for the zx plane and 10 dB and 9 dB for the zy plane, subject to the normalised accepted power of the fundamental frequency, respectively. The measured maximum gain of the GA-optimised antenna was found to be 4.14 dB.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

6.3 Microstrip Patch Antenna with a Partially Shorted Wall Following a successfully development and demonstration of harmonic suppression antenna design with a fully shorted wall in the previous section, a follow-up work on investigating the possibility to control the harmonics with the variation of the width of the shorting wall is presented in this section. The antenna design was continued to operate at 2.4 GHz. The antenna geometry for harmonic suppression, having a partially shorted wall, is shown in Figs. 38 - 40. Initially, the antenna is subdivided into six trilaterals and two quadrilaterals (including the partially shorted wall), this is requiring seven parameters to predefine, including the width of the shorting wall. Fig. 39 demonstrates the top view of adaptive segmentation results. Table 8 presents the GA input parameters for each required variable. Again real-valued GA chromosomes were used. The frequency band that covers fundamental, first and second harmonic frequencies were considered. The randomly generated antenna configurations were evaluated for maximum fitness using the same cost function stated in the previous section. The weighting coefficients (W1, W2 and W3) for the cost function were still optimally found to be 0.6, 0.4 and 0.4. The geometry configuration of optimal antenna was found within the maximum generations as shown in Fig. 41, in which infinite ground plane is also considered.

106

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

Figure 38. Top view of the geometry applied for adaptive meshing using GA.

Table 8. Summary of GA input parameters, antenna variables and best solutions GA parameters

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

No. of population size = 4 No. of parameters = 7 Probability of mutation =0.02 Maximum generation =200 No. of possibilities=32768

Patch antenna design with a partially shorted wall Parameters (m) Optimal (m) Antenna length (L) (0.03-0.05) 0.03486 Antenna width (W) (0.03-0.05) 0.03820 Shorting wall position (d) (0.005-0.015) 0.00986 Antenna height (h) (0.003-0.01) 0.00336 Variable shorting wall width (Ws) (0.001-0.03) 0.02474 Feeding point at x-axis (fx) (0.004-0.02) 0.01685 Feeding point at y-axis (fy) (0.004-0.025) 0.01923

For validation, a prototype of the GA-optimised harmonic suppression antenna with a partially shorted wall was shown in Fig. 42. A copper sheet of thickness of 0.5 mm was used for fabrication of the patch antenna, the shorting wall and the ground plane. The ground plane size is 140 x 140 mm. The return loss was validated and measured results were compared with the prediction, as shown in Fig. 43. As can be seen, the rejection levels of 2nd and 3rd harmonics were quite encouraging. Moreover, it can be easily noticed that the rejection level of the antenna at the second harmonic is superior to the previous design and is almost the same at the third harmonic. In addition, other resonance was found at the third harmonic frequency bands. The prototype antenna is resonant at 2.48 GHz and presents a less wide bandwidth (around 150 MHz), compared to the first design. This is mainly because the height of this antenna is much lower (about 3.36 mm) than the first design (7.9 mm), in which the antenna height has the most important influence on bandwidth enhancement of the microstrip patch antenna.

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

107

2.47 GHz

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

4.94 GHz

7.41 GHz Figure 37. Measured radiation patterns of the proposed GA-optimized HSA with fully shorted wall for 2.47 GHz, 4.94 GHz and 7.41 GHz at: (left) zx plane; (right) zy plane; (‘───’measured Eθ, and ‘- - - -’ measured Eφ).

108

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

Figure 39. 2D Mesh used for Fig. 38 using GA.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 40. Side view of the antenna geometry of Fig. 38.

Figure 41. 3D Mesh used for Fig. 40 using GA.

The measured input impedance over the frequency band of 1 GHz to 9 GHz is shown in Fig. 44. Extended scale plot of the impedance is presented in Fig. 45. Again it can be seen that the real part of the input impedance of the proposed antenna is close to zero for a wide frequency band around the second and third harmonic frequencies. This is also indicating the

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

109

influence of the reactive effects to harmonic termination at harmonic frequencies are realised. A summarised detail of the input impedances for fundamental and two first harmonics are presented in Table 9. Similar to previous section the input impedance shows a fairly good matching to 50 Ω load was attained at fundamental frequency and relative small resistive impedances at harmonics were observed.

Figure 42. Prototype HSA with partially shorted, side view (left) and top view (right).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 43. Comparison of the measured and calculated return loss of the patch antenna with partially shorted wall.

Table 9. Performance of measured harmonic rejection antenna with partially shorted wall at the fundamental and first two harmonics

Frequency (GHz)

Antenna input impedance (Ω) Real Imaginary

fo : 2.48

53.852

7.8281

2fo : 4.96

5.0547

-73.672

3fo : 7.44

7.002

5.5452

110

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

Figure 44. The overall measured input impedance of the patch antenna with partially shorted wall.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 45. The measured input impedance of the patch antenna with partially shorted wall (expanded scale of Fig. 44).

Measurements of the far field radiation patterns of the prototype were carried out in a farfield anechoic chamber. Two pattern cuts were taken for four selected operating frequencies that cover the designated whole bandwidth in this study. The radiation patterns in the zx plane and zy plane for the GA-optimized HSA with partially shorted wall at 2.47 GHz, 4.94 GHz and 7.41 GHz were measured. The measured results were presented in Fig. 46, in which the second and third harmonic radiations of the proposed HSA with fully shorted wall are less than 12 dB and 9 dB for the zx plane and 11 dB and 8 dB for the zy plane, for the normalised accepted power of the fundamental frequency, respectively. The measured maximum gain of the GA-optimised antenna was found to be 3.56 dB.

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

111

2.47GHz

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

4.94GHz

7.41 GHz Figure 46. Measured radiation patterns of the proposed GA-optimized HSA with partially shorted wall for 2.47 GHz, 4.94 GHz and 7.41 GHz at: (left) zx plane; (right) zy plane; (‘───’ measured Eθ, and ‘- - - -’ measured Eφ).

112

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

6.4 Microstrip Patch Antenna with a Folded Patch In this section, a new technique in designing microstrip patch antennas for harmonic suppression using GA is extended here to develop other harmonic suppression antennas. A novel coaxial-fed air-dielectric microstrip patch antenna for suppressing harmonics with a folded patch, resonating at the same frequency is studied. The antenna geometry for harmonic suppression, having a folded patch extended underneath the main patch, is shown in Figs. 47 to 50. The antenna is subdivided into four trilaterals and three quadrilaterals (including the folded patch). Thus the structure combination requires eight parameters to be defined. Table 10 presents the GA input parameters for each variable used. Again the weighting coefficients W1, W2 and W3 were found to be 0.4, 0.4 and 0.6 respectively.

6.4.1 Results and Discussions

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The computed results were validated using the hardware realisation. Fig. 51 presents the computed return loss for microstrip patch with a folded patch against the measured one. The results were quite encouraging especially on the required harmonic suppression. The results of both packages for each antenna were agreed very well. However, the first and second harmonic levels for the folded patch for both harmonics were around 2 dB. The measured resonant frequency of the GA-optimised antenna with folded patch configuration was found 2.426 GHz. It should be noted that the operating bandwidth were around 500 MHz , in which this results will totally support the design procedure of using this type of antenna for wide range of wireless communication application.

Figure 47. Top view of the antenna geometry applied for adaptive meshing using GA.

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

Figure 48. 2D Mesh used for Fig. 47 using GA.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 49. Side view of the antenna geometry applied for adaptive meshing using GA for Fig. 47.

Figure 50. 3D Mesh used for Fig. 49 using GA.

113

114

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

Table 10. Summary of GA input parameters, antenna variables and best solutions. GA parameters

No. of population size = 4 No. of parameters = 8 Probability of mutation =0.02 Maximum generation =500 No. of possibilities=32768

Air-dielectric folded patch antenna design Parameters (m) Optimal (m) Antenna length (L) (0.03-0.06) 0.04316 Antenna width (W) (0.02-0.06) 0.03006 folded wall position (d) (0.005-0.015) 0.00748 Antenna height (h) (0.004-0.01) 0.00989 Extend folded wall length (Lf) (0.005-0.015) 0.01327 Extend folded wall height (hf) (0.001-0.0035) 0.00159 Feeding point at x-axis (fx) (0.004-0.015) 0.00571 Feeding point at y-axis (fy) (0.004-0.025) 0.01392

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 51. Comparison of the measured and simulated return loss of the patch antenna with a shorting wall.

The measured input impedance over the frequency band of 0.5 GHz to 12 GHz is shown in Fig. 52. Extended scale plot of Fig. 52 is presented in Fig. 53. Again it can be seen that the real part of the input impedance of the proposed antenna is close to zero for a wide frequency band around the second and third harmonic frequencies. This is also indicating the influence of the reactive effects to harmonic termination at harmonic frequencies is achieved. A summarised detail of the input impedances for fundamental and two first harmonics are presented in Table 11. Similar to previous two sections it is a fairly good matched to load of 50 Ω attained at fundamental frequency and relative small resistive impedances at harmonics were observed.

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

115

Figure 52. The overall measured input impedance of the patch antenna folded patch.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 53. The measured input impedance of the patch antenna with folded patch (expanded scale of Fig. 52).

Table 11. Performance of measured harmonic rejection antenna with folded patch at the fundamental and first two harmonics

Frequency (GHz)

Antenna input impedance (Ω) Real Imaginary

fo : 2.426

47.12

-12.59

2fo : 4.852

5.021

-24.81

3fo : 7.278

4.606

20.07

116

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

2.45 GHz

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

4.90 GHz

7.35 GHz Figure 54. Measured radiation patterns of the proposed GA-optimized HSA with a folded patch for 2.45 GHz, 4.90 GHz and 7.35 GHz at: (left) zx plane; (right) zy plane; (‘───’measured Eθ, and ‘- - - -’ measured Eφ).

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

117

Measurements of the far field radiation patterns of the prototype antenna were also carried out. Two pattern cuts were taken for four selected operating frequencies at fundamental and harmonic frequencies that cover the designated whole bandwidth in this study. The radiation patterns in the zx plane and zy plane for the GA-optimized HSA with folded patch at fundamental, second and third harmonic frequencies were measured. The measured results were presented in Fig. 54, in which the second and third harmonic radiations of the proposed HSA with fully shorted wall are found less than 19 dB and 13 dB for the zx plane and 10 dB and 9 dB for the zy plane respectively. These gain patterns were normalised to the accepted power of the fundamental frequency. The measured maximum gain of the GA-optimised antenna was found to be 5.01 dB.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

7. SUMMARY The numerical solution technique for antenna designs applied for several wireless communication system applications using adaptive meshing and genetic algorithm has been presented in this chapter. A FORTRAN code genetic algorithm driver was adopted in this work in conjunction with the industry-standard NEC-2 FORTRAN source code, which was used to evaluate the randomly generated antenna samples. Several antenna designs were considered and investigated using GA. These include QHA antenna, balanced folded loop antenna, two circular polarized antennas, and three different microstrip patch antennas for harmonic suppressions. For these designs the GA was successfully proved as an efficient optimizer tool that can be adopted and used to search and find the quicker solutions for complex antenna design geometries. A novel FORTRAN program for adaptively meshing planar antennas in terms of wiregrid structure has been implemented and embedded inside the GA source code. The programme was fully tested and applied for complex antenna structures. It was shown that the results of several examples modelled by the adaptive meshing illustrate good stability and accuracy. In addition, three novel microstrip patch antennas, for which the first and the second harmonics were mostly suppressed, have been successfully developed using GA and the adaptive meshing technique. The results of the optimum designs of the proposed antennas exhibit an excellent harmonic suppression. The measured results in terms the impedance response and radiation power gains of 2nd and 3rd harmonics were found reasonable and agree well with the computed one. In general, the presented examples in this chapter show the capability of the proposed program in antenna design using GA, however, this work can be further extended to include complex antenna structures mounted next or attached or inside a lossy/lossless dielectric materials. Moreover, this will also permit to use the proposed method with or embedded inside a hybrid solution domain that will strongly add another potential capability to the available numerical tools.

118

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

ACKNOWLEDGEMENTS The authors would like to gratefully acknowledge the facility services provided by Mobile and Satellite Communciations Research Centre at Bradford University and the support by the UK Engineering and Physical Sciences Research Council under grant EP/E022936/1.

REFERENCES [1] [2] [3]

[4] [5] [6] [7] [8] [9]

[10]

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

[11] [12]

[13] [14]

Y. Rahmat–Samii and E. Michielssen, “Electromagnetic optimization by Genetic Algorithms”, John Wiley & Sons, Canada, 1999. D.A. Coley, “An introduction to Genetic Algorithms for scientists and engineers”, World Scientific, Singapore, 1999. H.H. Ammar and Y. Tao, “Fingerprint registration using Genetic Algorithms”, IEEE Symposium on applications specific systems and software engineering technology, pp. 148–154, March 2000. D.E. Goldberg, “Genetic Algorithms in search, optimization and machine learning”, Addison – Wesley Publishing Company, Canada, 1989. R.H. Dinger, “Engineering design and optimization with Genetic algorithms”, Northcon/98 conference proceedings, pp.114–119, October 1998. http://ai.eller.arizona.edu/%7Emramsey/ga.html. D.L.Carroll, Fortran Genetic Algorithm Driver, Version 1.7, Download from: http://www.staff.uiuc.edu/~carroll/ga.html, 12/11/98. J. Johnson and Y. Rahmat–Samii, “Genetic Algorithms in engineering electromagnetic”, IEEE Antenna & Prop. Magazine, Vol. 39, pp.7 – 21, Aug. 1997. D. Zhou, R.A. Abd-Alhameed, C.H. See, P.S. Excell, and E.A. Amushan, “Design of quadrifilar helical and spiral antennas in the presence of mobile handsets using genetic algorithms”, The first European Conference on Antennas and Propagation (EuCAP2006), Session 3PA1, Paper no.122, Nice, France, 6-10 November 2006. G.T. Burge and A.J. Poggio, “Numerical electromagnetic code (NEC): method of moments”, US Naval Ocean Systems Center, San Diego, Rep. No. TD116, pp. 1-37, 1981. B.J. Strait, “Applications of the Method of Moments to electromagnetic fields”, Ed, SCEEE Press, pp. 450-457, 1980. M.A. Mangoud, R.A. Abd-Alhameed and P.S. Excell, “Simulation of human interaction with mobile telephones using hybrid techniques over coupled domains”, IEEE Transactions on microwave theory and techniques, vol. 48, no. 11, pp. 20142021, November 2000. G.L. Burke and A.J. Poggio, “Numerical Electromagnetics Code (NEC)-Method of Moments”, Lawrence Livermore Laboratory, Livermore, CA, 1981. R.A. Abd-Alhameed, M. Mangoud, P.S. Excell, and K. Khalil, “Investigations of polarization purity and specific absorption rate for two dual-band antennas for satellite-mobile handsets”, IEEE Transactions on antennas and propagation, vol. 53, no. 6, pp. 2108-2110, June 2005.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Design and Optimisation of Antennas Using Genetic Algorithms for Wireless …

119

[15] H. Morishita, Y. Kim, Y. Koyanagi and K. Fujimoto, “A folded loop antenna system for handsets”, IEEE AP-S Proc., vol. 3, pp. 440-443, July 2001. [16] Computer Simulation Technology Corporation, CST Microwave Studio, Version 5.0, Germany. [17] WireGrid for Windows: A graphic user interface for NEC, version 3.20. [18] NEC-Win Professional, version 1.l, Nittany Scientific, Inc., Hollister, CA. [19] R.A. Abd-Alhameed, P.S. Excell and M.A. Mangoud, “Broadband antenna response using hybrid technique of frequency domain of MoM and FDTD”, ACES Journal on Computational of Electromagentics, vol. 20, No. 1, pp. 70-77, March 2005. [20] E. Elkhazmi, N.J. McEwan and J. Moustafa, “Control of harmonic radiation from an active microstrip patch anrtenna”, Journees Intern. De Nice Sur Les Antennas, pp. 313-316, November 1996. [21] V. Radisic, S.T. Chew, Y. Qian, and Tatsuo Itoh, “High efficiency power amplifier integrated with antenna”, IEEE Microwave guided wave letter, vol. 7, no. 2, pp. 39-41, February 1997. [22] V. Radisic, Y. Qian, and T. Itoh, “Class F power amplifier integrated with circular sector microstrip antenna,” IEEE MTT-S Digest, pp. 687-690, 1997. [23] A.F. Sheta, “A novel H-shaped patch antenna”, Microwave and optical technology letters, vol. 29, no. 1, pp. 62-66, April 2001. [24] Q.-X. Chu and M. Hou, “An H-shaped harmonic suppression active integrated atnenna”, International Journal of RF and microwave computer-aided engineering, vol. 16, Issue 3, pp. 245-249, May 2006. [25] F.-R. Hsiao, T.-W. Chiou and K.-L. Wong, “Harmonic control of a square microstrip antenna operated at the 1.8 GHz band”, Proceedings of APMC2001, pp. 1052-1055, Taipei, Taiwan, R.O.C., 2001. [26] B.M. Lee, S.W. Kwon and Y.J. Yoon, “Dual feeding active integrated antenna”, Electronics Letters, vol. 38, no. 19, pp. 1073-1075, September 2002. [27] S. Kwon, B.M. Lee, Y.J. Yoon, W.Y. Song and J.-G. Yook, “A harmonic suppression antenna for an active integrated antenna”, IEEE antennas and wireless propagation letters, vol. 13, no. 2, pp. 54-56, February 2003. [28] Y.J. Sung, M. Kim, and Y.-S. Kim, “Harmonics reduction with defected ground structure for a microstrip patch antenna”, IEEE antennas and wireless propagation letters, vol. 2, pp. 111-113, 2003. [29] X.-C. Lin and L.-T. Wang, “A broadband CPW-fed loop slot antenna with harmonic control”, IEEE antennas and wireless propagation letters, vol. 2, pp. 323-325, 2003. [30] H. Kim, K.S. Hwang, K. Chang and Y.J. Yoon, “Novel slot antennas for harmonic suppression”, IEEE antennas and wireless components letters, vol. 14, no. 6, pp. 286288, 2004. [31] S.K. Lee, Y. Qin and E. Korolkiewicz, “Reduction of the second and third harmonics for a rectangular microstrip patch antenna”, Microwave and optical technology letters, vol. 40, no. 6, pp. 455-460, March 2004. [32] X.-C. Lin, L.-T. Wang and J.-S. Sun, “Harmonic suppression by photonic bandgap on CPW-fed loop-slot antenna”, Microwave and optical technology letters, vol. 41, no. 2, pp. 154-156, April 2004.

120

D. Zhou, R.A. Abd-Alhameed, C.H. See et al.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

[33] H. Liu, Z. Li, X. Sun and J. Mao, “Harmonic suppression with photonic band gap and defected ground structure for a microstrip patch antenna”, IEEE antennas and wireless components letters, vol. 15, no. 2, pp. 55-56, February 2005. [34] Y.J. Sung and Y.-S. Kim, “An improved design of microstrip patch antennas using photonic bandgap structure”, IEEE Transactions on antennas and propagation, vol. 53, no. 5, pp. 1799-1804, May 2005. [35] H. Kim and Y.J. Yoon, “Microstrip-fed slot antennas with suppressed harmonics”, IEEE Transactions on antennas and propagation, vol. 53, no. 9, pp. 2809-2817, September 2005. [36] S.-Y. Lin, K.-C. Huang and J.-S. Chen, “Harmonic control for an integrated microstrip antenna with loaded transmission line”, Microwave and optical technology letters, vol. 44, no. 4, pp. 379-383, February 2005. [37] K.S. Hwang, H. Kim and Y.J. Yoon, “Bandwidth-enhancement approach for a smallsized harmonic-suppressed antenna”, Microwave and optical technology letters, vol. 44, no. 6, pp. 585-587, March 2005. [38] S.-Y. Lin, “Modes-controlled slot antennas with frequency selective surface”, Microwave and optical technology letters, vol. 48, no. 1, pp.47-49, January 2006. [39] D Zhou, RA Abd-Alhameed and PS Excell, “Design of antenna for wide harmonic suppression using adaptive meshing and genetic algorithms”, Seventh international conference on computation in electromagnetics (CEM2008), pp. 187-188, Brighton, UK, 7-10 April 2008.

In: Antennas: Parameters, Models and Applications Editor: Albert I. Ferrero

ISBN 978-1-60692-463-1 © 2009 Nova Publishers, Inc.

Chapter 4

APPLICATION OF SMA WIRE ACTUATORS IN FLATNESS CONTROL OF MEMBRANE SAR ANTENNAE Fujun Peng1, Yan-Ru Hu 2, Xin-Xiang Jiang3 and Alfred Ng4 Canadian Space Agency, St.-Hubert, Canada

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ABSTRACT A membrane SAR(Synthetic Aperture Radar) antenna will be subjected to flatness problems during its lifetime in orbit due to the thermal variations in space. A pure passive control method may not be sufficient to maintain the membrane’s flatness since the thermal loads are changing. Therefore, an active control system, which is used to adjust the tensions according to the thermal variations, is developed. SMA wire actuators are selected to exert required tension, due to their unique properties such as high force, long stroke, small size, light weight, and silent operation, etc. First, the selected SMA wire actuators are tested in the ambient air environment as well as in a vacuum chamber to characterize their stability and controllability properties. Tests are conducted under a constant load and variant input currents are used to activate the actuator. The results indicate that, in the ambient air environment, the displacement outputs of the SMA actuator are not stable if the full phase transformation is not achieved. This clearly reflects the influence of the natural air convection on the stability of the actuation. As a contrast, the displacement outputs of the SMA in the vacuum condition are much more stable at all given input currents. This implies that SMA actuators probably can be used directly in space without any assistance of control strategy. However, it is also noticed that, in vacuum environment, it mostly takes much 1

Directorate of Spacecraft Engineering, Canadian Space Agency, St.-Hubert, QC, J3Y 8Y9, Canada, [email protected] 2 Directorate of Spacecraft Engineering, Canadian Space Agency, St.-Hubert, QC, J3Y 8Y9, Canada, [email protected] 3 Directorate of Spacecraft Engineering, Canadian Space Agency, St.-Hubert, QC, J3Y 8Y9, Canada, [email protected] 4 Directorate of Spacecraft Engineering, Canadian Space Agency, St.-Hubert, QC, J3Y 8Y9, Canada, [email protected]

122

Fujun Peng, Yan-Ru Hu, Xin-Xiang Jiang et al. more time to achieve a stable displacement output, and in some cases even more than 20 minutes are needed. This may limit SMA's application in space if no other means is utilized to increase its response speed. To overcome SMA’s poor stability and increase its response speed, an actuator control strategy is developed based on the idea of adjusting the SMA wire temperature as fast as possible. This strategy is simple, stable, and no hysteresis model or thermal model is required. Tests are performed and the results demonstrated that the proposed control strategy is very effective in controlling SMA actuators. Under this control strategy, SMA wire actuator can track square wave, ramp, and sinusoidal signal with very high accuracy ⎯ small steady error and overshoot. By increasing the input current, SMA response speed can be increased remarkably. In spite of an increase of rising overshoot and possibly the RMS error, the absolute values of them are still very small. It is also indicated that shorter sampling interval is helpful in getting higher tracking accuracy. After getting the above testing results with satisfactory accuracy, the actuator control strategy is integrated into a control system based on genetic algorithm, and used to control the flatness of a membrane SAR antenna model. Tests are performed 20 times at room temperature. Each time all tension combinations generated by the genetic algorithm are very well realized by the SMA actuators and the standard deviation of the membrane flatness goes down quickly from around 0.22mm to less than 0.05mm. Another 20 tests are then performed with local thermal load applied. Again optimal tension combinations are realized very well. The membrane flatness goes down quickly from around 0.33mm to less than 0.05mm.

Key words: shape memory alloy, precision control, inflatable structures, SAR antenna, shape control

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

INTRODUCTION There has been an increasing interest in the application of inflatable structures in space programs. This kind of structures has unique advantages in achieving low mass and high packaging efficiency. Their ultra-lightweight and small-volume properties in turn can potentially reduce the overall space mission cost by reducing the launch vehicle size requirement. Inflatable structures can also reduce total system mass and deployment system complexity, thereby increasing system reliability. This type of structures has been envisioned for many space applications such as large telescopes, antennas, solar sails, sun shields, solar arrays, etc. [1-4]. We are currently working on an in-house R&D project in the development of a large surface area to mass ratio inflatable space structure with possible applications as a Synthetic Aperture Radar (SAR) antenna. The key components of this inflatable structure are inflatable tubes, membrane and the links installed in-between stretching the membrane (Fig. 1). It can be rolled into a small volume and fixed on a satellite bus for launching. When the satellite arrives into the orbit, the inflatable tubes are filled with gas and roll out, and the Kapton membrane will be deployed accordingly. It is expected that the membrane will be subjected to flatness problems during its lifetime in orbit due to the thermal variation in space. A pure passive control method may not be sufficient to maintain the membrane flatness since the inflatable structures will experience

Application of SMA Wire Actuators in Flatness Control…

123

changing space environment. Hence we are developing an active control scheme to adjust the stretching tensions in real time so that the membrane could be maintained flat.

Inflatable tubes Membrane Links and actuators

Figure 1. Sketch of the inflatable structure

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The control system involves a vision system for membrane flatness measurement, the genetic algorithm (GA) for tensions optimization. For exerting required tensions to the membrane, SMA wire actuators are installed in-between the membrane and the frame, considering that SMA actuators have the unique properties such as high force, long stroke, small size, light weight, and silent operation, etc. However, SMA wires poor stability and controllability are obstacles on the way of using them to exert tensions to the real SAR antenna. It is hard to establish a stable and reliable relationship between its displacement output and the electrical input current, i.e., the actual displacement output is not only affected by the input current, but also by some other factors such as cooling conditions, load conditions. It is not possible to achieve the tensions we need by simply input a fixed electrical current to SMA actuators. Therefore, a special controller must developed first which can regulate the SMA wire actuators to achieve tensions with enough accuracy. Then involves the controller as a sub-controller into the whole active shape control system. This paper presents some results in the study of trying to use SMA wire actuators in the active flatness control of membrane SAR antennae, including preliminary characterization of SMA wires in air and in vacuum, SMA controller development and testing, active flatness control system development and testing.

SMA PRELIMINARY TESTING RESULTS Preliminary testings are performed to characterize the stability and controllability properties of selected SMA wire actuators. The obtained results will also be used for comparison with that to be obtained under control of the proposed strategy. Fig. 2 shows the test bench, which includes a mechanical fixture, a high precision linear voltage displacement transducer, data acquisition devices, a PC, and a power amplifier. The actuator tested is a 170mm long, 0.152mm diameter Ni-Ti SMA wire. Its technical data are listed in Table 1. Tests are conducted under a constant load of 1.57N, and variant input currents are used to activate the actuator. This process is then repeated in a thermal vacuum chamber (Fig. 3, vacuum level at 10-5 Torr. temperature at 25°C), but different input currents are used for activation. Fig. 4 shows the time histories of the tests.

124

Fujun Peng, Yan-Ru Hu, Xin-Xiang Jiang et al.

+ SMA Wire

LDVT Displacement Transducer

Power Amplifier

¯

DAQ & Computer

Load

Figure 2. Test bench for testing SMA Wires

Figure 3. Thermal vacuum chamber for space environment simulation

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 1. Technical data of 0.152mm diameter Nitinol wire Austenite finishing temperature (ºC) Electrical resistance (Ohms/mm)

90 0.051

Maximum pull force (N)

3.2

Input current* (mA)

400

Contract time (s)

1.0

Off time (s)

1.2

* Current that activates wire in 1 s and can be left on without over heating

Application of SMA Wire Actuators in Flatness Control…

Time (s) (a) In air

125

Time (s) (b) In vacuum chamber

Figure 4. Testing results in air under different input electrical currents

Current (mA)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 5. Ultimate displacement outputs under different constant input currents

Fig. 4(a) indicates that, in the ambient air environment, the displacement outputs of the SMA actuator are not stable if the full phase transformation is not achieved. This clearly reflects the influence of the natural air convection on the stability of the actuation. As a contrast, the displacement outputs of the SMA in the vacuum condition are much more stable at all given input currents. This implies that SMA actuators probably can be used directly in space without any assistance of control strategy. However, we also notice that it mostly takes much more time to achieve a stable displacement output, and in some cases even more than 20 minutes are needed. This may limit SMA's application in space if no other means is utilized to increase its response speed. Also, Fig. 4 implies the poor controllability of SMA in the sense that it is hard to find a precise constant electrical current which can effectively activate SMA actuator to offer a desired actuation output. For the case in Fig. 4(a), when the input current changes from 220mA to 230mA, the SMA output increases from 1.6mm to 6.6mm. This means, in this region, an increase of 1mA current may lead to an output change of over 0.5mm. Similarly, in Fig. 4(b), a 1mA variation in input current may cause an output change of more than 0.74mm. Therefore, in these cases, SMA is too sensitive to the variation of input currents to offer an actuation with good accuracy. The sharp rises of the curves in Fig. 5 give a clearer and more intuitive explanation of the poor controllability of SMA actuators. It can also be seen in Fig. 5 that, to achieve the same actuation output, much less current is required in vacuum condition than in air. For the case in Fig.5, to achieve a 4mm

126

Fujun Peng, Yan-Ru Hu, Xin-Xiang Jiang et al.

actuation, 50mA is required in vacuum chamber as compared to 220mA in ambient air condition. This means a roughly 95% energy saving in space environment.

SMA CONTROL STRATEGY DEVELOPMENT AND TESTING SMA’s poor stability and controllability imply that a control to SMA actuators becomes essential to achieve precise actuations. Some attempts have been made to solve this type of problems by adjusting the heating electrical current flowing in the SMA wire. Several control algorithms, such as Proportional Integral Derivative (PID) control, Pulse Width Modulation (PWM) control, optimal control, have been proposed, and Preisach model and neural networks are developed to describe hysteresis property of SMA actuators [5–7]. However, it is hard to model the SMA hysteresis precisely, and also the stability of the control system is not guaranteed. This may become an obstacle to use SMA actuators in space missions. Here we put emphasis on control strategy's effectiveness, reliability and practicability to space applications, instead of addressing complex theory analysis and modeling of SMA. A simple control strategy is proposed, which is based on the idea of adjusting the SMA wire temperature as fast as possible. This control strategy is simple, stable, and requires no hysteresis model or thermal model. Consider a SMA wire actuator subject to electrical current activation, its response can be expressed as:

S (t ) = f (C (t ), I (t ))

(1)

where, S(t) is the output actuation, which could be displacement or force, C(t) and I(t) are cooling conditions and input current. If we don't use active cooling, the SMA wire actuator response can only be actively adjusted by changing the input current. Consider a simple feedback control, at time t, the input current could be designed as:

I (t ) = I (t − 1) + ΔI (t )

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

= I (t − 1) + K (t )e(t )

(2)

where, K(t) is a feedback gain and e(t) is the error response denoted by

e(t ) = S * (t ) − S (t )

(3)

where S*(t) and S(t) are the desired response and actual response, respectively. Now the question becomes how to determine K(t) of Eq.(2) such that SMA wire temperature changes rapidly and consequently the error response reduces quickly. Consider e(t ) > 0 , the input current should be given a positive increment, and K(t) should also be a positive value. If K(t) is small, the obtained I(t) may not be large enough to heat the SMA wire very fast, or many more steps are needed to have a proper SMA temperature. So it is desirable to have a relatively large value of K(t). But at the same time,

Application of SMA Wire Actuators in Flatness Control…

127

the input current should not exceed an upper limit Ic, which is set to avoid burning the SMA wire. To have the fastest response, for e(t ) > 0 , the input current is directly designated as:

I (t ) = I c . For e(t ) < 0 , input current increment should be negative so that SMA wire temperature may decrease and consequently SMA wire actuation becomes smaller. To have the fastest response, input current is set to 0. For e(t ) = 0 , I(t) could be either I c or 0. The control law can be written as:

⎧I , I (t ) = ⎨ c ⎩0,

e(t ) > 0

(4)

e(t ) ≤ 0

The value of Ic can be determined experimentally. It should be large enough to activate full transformation of the SMA wire, but should not burn the SMA wire within an update interval. This control strategy is then tested. The test setup, SMA wire actuator and load are the same as those for the preliminary testing. This control strategy is implemented in Labview. The control system monitors the actual displacement output of the SMA wire in real time with a high sampling rate. When the actual displacement is less than the desired value, the input current is set Ic. When the actual displacement output becomes equal or higher than the desired value, the input current is set to zero. Tests are first performed on tracking square waves in ambient air condition. Different activation input currents Ic and different sampling intervals are tested. Fig. 6 shows the time history of the displacement response and control signal under 280mA input current and 10ms sampling interval. The SMA actuator reacts very fast and tracks the square wave very well. More detailed results under different input currents are given in Table 2 and Figs. 7–9.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 2. Test results under different input current

3.417

Descending time (s) 0.927

Mean error (mm)* -0.0113

Steady RMS error (mm) 0.0182

280

0.967

1.077

-0.0020

0.0209

320

0.547

1.033

0.0077

0.0287

360

0.373

0.983

0.0243

0.0475

400

0.233

0.940

0.0416

0.0690

Input current(mA)

Rising time (s)

240

* The values listed here are averages of -1×e(t), where e(t) is expressed as Eq.(3)

From Table 2 and Fig. 7(a) one can find that an increase of input current shortens rising time effectively. When the input current is greater than 280mA, the rising time is less than 1.0s. This is much faster than the results obtained without control shown in Fig. 4(a). The descending time is not affect much by the changing of input current since the cooling

128

Fujun Peng, Yan-Ru Hu, Xin-Xiang Jiang et al.

condition remains unchanged. But local airflow usually exists which induces fluctuation of descending time.

Time (s) Figure 6. Time history of SMA displacement response under 280mA current and 10ms updating interval 3.5 3

Time (s)

2.5

Rising time Descending time

2 1.5 1 0.5 0

240

280

320

360

400

440

Current (mA) (a) Rising time and descending time

Current (mA) (b) RMS error and mean error

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 7. Test results under different input currents

Regarding the steady tracking error, one can find from Table 2 and Fig. 7(b) it becomes larger along with the increase of input current. In the mean time, larger input current pushes the mean tracking error changing from negative values to zero and positive values. This is mainly due to the time delay between the control action and the actual phase transformation because of the diffusion nature of heat transfer. When the actuation crosses the desired value from below, this time delay will allow the actuation continue to go up for a short period of time, although the power is removed. When a larger input current is used, the SMA temperature changes faster and the time delay becomes larger, thus causes larger positive error and pushes up the mean steady tracking error. The similar phenomenon also exists when the actual actuation crosses the desired value from above. At this time the time delay is mainly affected by the cooling speed. Fig. 8 shows clearly how steady tracking errors are

Application of SMA Wire Actuators in Flatness Control…

129

pushed up when a larger input current is applied. However, the steady tracking error is quite small, less than 1.5% in our cases. The performance is much better than that shown in Fig. 4(a), where no control is involved.

Time (s) (a) Under input current of 240mA

Time (s) (b) Under input current of 360mA

Figure 8. Mean steady errors under different input currents

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The effect of sampling interval is given in Fig. 9. It is seen that basically larger sampling interval causes larger steady tracking error. This is easy to understand since larger sampling interval will more likely lead to the delay of identifying if the response reaches the desired level, while during the delay the practical SMA response may go farther away from the desired value. Theoretically, changing of sampling interval should not affect rising time and descending time. But the time delay introduced by a larger sampling interval will definitely reduce the measurement accuracy of rising time and descending time. We also performed tests tracking sinusoidal signal. The input current and sampling interval are 320mA and 8ms, respectively. The results are shown in Fig. 10. SMA wire responses catch up to the desired signal at around 1.2 second, and then track it very well.

Sampling interval (ms) Figure 9. RMS errors under different sampling intervals (input current is 320mA)

130

Fujun Peng, Yan-Ru Hu, Xin-Xiang Jiang et al.

Time (s)

signal, (320mA, 8ms)

Figure 10. Test results on tracking sinusoidal

Similar tests are also performed in vacuum chamber under different input currents. The test setup, SMA wire actuator and load are the same as those in ambient air condition. The results are shown in Table 3 and Figs. 11–13. Basically, the trends of the results are quite similar to those obtained in ambient air environment, i.e., along with the increase of input current, the RMS error become larger, while the rising time is reduced. One unique characteristics we can find from Table 3 and Figs. 11–12 is that the RMS errors are much less than those obtained in ambient air condition (roughly 20% of the latter). This is probably because the required input current is much less, and the time delay in the process of heat diffusion and phase transformation becomes smaller at the time the power is removed. This phenomenon implies that the proposed control strategy will have better accuracy on SMA actuator control in space than on ground.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 3.Test results in vacuum chamber under different input current Input current(mA) 100 120 140 160 180 200 220 240 260

Rising time (s) 12.20 7.30 4.80 3.70 2.70 2.20 1.70 1.20 0.88

Mean error (mm)* 0.0033 0.0030 0.0031 0.0037 0.0039 0.0051 0.0063 0.0069 0.0104

Steady RMS error (mm) 0.0039 0.0036 0.0037 0.0045 0.0047 0.0062 0.0078 0.0087 0.0136

* The values listed here are averages of -1×e(t), where e(t) is expressed as Eq.(3)

Another difference from the air condition case is that the mean errors are always positive values. This is because the cooling speed is much slower in vacuum condition due to lack of air convection, and consequently the time delay is much smaller than the heating process at the time of crossing over the desired value. This phenomenon is clearly shown in Fig.13. Considering this characteristics, we can further improve the control accuracy by adding a bias to the desired value. For example, if we use 200mA input current and the desired value is

Application of SMA Wire Actuators in Flatness Control…

131

2.5mm, we can reduce the steady tracking error from 0.0062mm to 0.0035mm by setting 2.4995mm as the desired value with bias 0.0005 in control system. The error is reduced by 43.6%

Figure 11. Time histories of SMA displacement under different input currents (10ms updating interval) 14

x 10

-3

Amplitude (mm)

12 Mean error RMS error

10 8 6 4 2 80

120

160

200

240

280

Input current (mA)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 12. Mean error and RMS error under different input currents (in vacuum, 10ms)

Time (s) Figure 13. Desired value and steady actuations under different input current (in vacuum)

132

Fujun Peng, Yan-Ru Hu, Xin-Xiang Jiang et al.

ACTIVE FLATNESS CONTROL SYSTEM INTEGRATION

Tension Measurement

Flatness Measurement

(Strain Gages)

(Vision System)

Hardware Software Control Parameters

Display and save

GA Parameters

GA

SMA controller and data acquisition

Vision system

Loop 3 Loop 2 Loop 1

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 14. Block diagram of the whole control system implementation

The control system is integrated using LabView, Matlab and Automation Manager. LabView code realizes the SMA actuator sub-controller, tension measurement, as well as controls control parameters input, results display and saving and data I/O. Matlab code implements the genetic algorithm. Automation Manager realizes the membrane shape measurement. LabView code is the master, which coordinates the whole system functioning by using ActiveX technique. The whole control system is implemented using three layers of loops, which is illustrated in Fig. 14. Loop 1 receives parameters from GUI (Graphical User Interface), executes GA (genetic algorithm) and then transfers control parameters and a group of candidate tension combinations to Loop 2. Loop 2 realizes the required tension combinations and measures the membrane flatness (standard deviation) corresponding to every tension combination. First, Loop 2 transfers one tension combination and control parameters to Loop 3, which executes the actuator controller and data acquisition. When the desired values of the tension combination are achieved, Loop 3 triggers the vision system to record the corresponding membrane flatness. To obtain steady membrane states, Loop 3 dose not triggers the vision system at once. Instead, it delays a number of iterations after the desired tension are obtained, so that the influence of the tension overshoots can be avoided. This number could be set from GUI. In order for the vision system to have enough time to finish its work, Loop 3 is still running to hold the desired tensions until the vision system competes the measurement. When the vision system and Loop 3 complete an execution, Loop

Application of SMA Wire Actuators in Flatness Control…

133

2 runs again to realize another tension combination and the obtain corresponding membrane flatness. This process is repeated multiple times until all the candidate tension combinations are realized and the associated membrane flatness values are recorded. Loop 1 gets back the data obtained by Loop 2, displays, saves to files, and then goes to the next generation. To complete the test of one generation of tension candidates, Loop 1 runs only one circle. The circle number of Loop 2 running is the total number of new individuals in every population. While for Loop 3, the running circles is much more than that of Loop 2, the specific number is always changing depending on the largest difference of the two adjacent tensions. If the difference is large, the actuator then needs more time to change its status from one to the other, and thus more running circles are needed.

TESTING OF MEMBRANE SAR FLATNESS CONTROL The structure to be tested is shown in Fig. 15. Basically, it is a 200mm×300mm rectangular Kapton Membrane, stressed by 12 discrete links installed between the membrane boundaries and aluminum frame. A local thermal load source is placed under the membrane (not visible in Fig.15). The flatness of the membrane is dependent on local thermal loads and the tension combinations. To realize active control, six shape memory alloy wire actuators are installed along the edge as a part of tension links. To monitor the values of tensions, strain gages are glued onto small and thin aluminum strips, which are also parts of tension links. The arrangement of SMA actuator, strain gage and links is sketched in Fig. 16. These tension measurement elements are calibrated using a load cell before tests are performed. A vision system is used to measure the flatness of the membrane. A projector shines lines on the membrane surface, and the vision system gives 3D coordinates of points selected on the line edges. In order for the vision system camera to see these lines clearly, a very thin coating is put on one side of the membrane. The whole setup including vision system is shown in Fig. 17.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

X

Y

Z [ \ ]

Figure 15. Picture of Membrane Structure Used for Flatness control testing

134

Fujun Peng, Yan-Ru Hu, Xin-Xiang Jiang et al.

Membrane

Aluminum link

Kaption tape 300mm× 200 mm

Strain gage

SMA Actuator

Strain Amplifier

Power Amplifier

Control System

Figure 16. Arrangement of Actuators and Sensors

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 17. Picture of the Whole Active Flatness Control System and Setup

Before performing structure shape control, the six SMA actuators (0.25mm in diameter and 100mm in length) installed on the membrane structure are tested. Set the six desired tensions as 3.75N, 3.53N, 3.31N, 3.09N, 2.87N and 2.65N, each corresponding to one actuator, and see how well these tensions can be achieved. The obtained time histories of the six tensions are shown in Fig. 18. One can see the obtained tensions tracks the desired tensions very well. The RMS errors (after 30 seconds) are 0.0072N, 0.0076N, 0.0079N, 0.0057N, 0.0068N, 0.0068N, respectively. The response speed is not fast due to the low input currents. The activation voltage is 1.7V for all SMA actuators, and their different electrical resistances lead to different input currents as well as different response speeds.

Application of SMA Wire Actuators in Flatness Control…

135

Figure 18. Time Histories of Obtained Tensions

These actuators are also tested using 20 tension combinations. During the test, the actuators transfer from one desired tension combination to another directly, without cooling down to the room temperature. The desired tensions, measured tensions and the error tensions are shown in Fig. 19. The values of error tensions are given in Table 4.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 4. Tension errors (N) Actuator 1

Actuator 2

Actuator 3

Actuator 4

Actuator 5

Actuator 6

0.0030

0.0096

-0.0019

-0.0058

-0.0127

-0.0051

0.0028

0.0039

-0.0063

-0.0001

-0.0138

-0.0033

-0.0132

0.0028

-0.0019

0.0014

-0.0151

0.0074

0.0035

-0.0084

0.0047

0.0111

0.0017

0.0073

0.0015

-0.0016

-0.0063

-0.0086

-0.0042

0.0025

-0.0032

0.0048

0.0004

0.0003

-0.0117

0.0006

-0.0123

-0.0075

-0.0018

-0.0103

-0.0148

0.0054

-0.0051

0.0052

0.0047

0.0053

0.0112

-0.0019

-0.0072

0.0211

-0.0096

0.0001

0.0005

0.0114

0.0003

0.0029

-0.0030

-0.0028

-0.0069

-0.0054

-0.0049

0.0017

0.0036

-0.0089

0.0036

0.0027

0.0009

0.0091

0.0026

0.0033

-0.0060

0.0070

-0.0007

0.0037

0.0048

0.0014

0.0024

0.0009

0.0002

0.0028

-0.0007

0.0028

0.0055

0.0056

-0.0001

0.0035

-0.0007

-0.0090

0.0028

-0.0062

0.0021

0.0065

0.0014

-0.0011

0.0072

-0.0006

0.0135

0.0002

0.0015

-0.0059

0.0025

-0.0012

-0.0029

0.0001

-0.0018

0.0007

0.0026

-0.0012

0.0065

-0.0031

-0.0030

-0.0003

0.0015

0.0146

-0.0053

0.0060

-0.0085

-0.0032

-0.0005

-0.0007

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

136

Fujun Peng, Yan-Ru Hu, Xin-Xiang Jiang et al.

Number of tension combinations Figure 19. Test Results of 20 Tension Combinations

Application of SMA Wire Actuators in Flatness Control…

137

We can see that actual tensions achieved are very close to the desired values, and the maximal tension error of these obtained tensions is 0.021N. All the relative errors are less than 0.65%. After getting the above test results with satisfactory accuracy, we performed tests on flatness control of the membrane of our setup. First, tests are performed 20 times at room temperature. Each time the tension combinations converge very fast (with different speeds) to the optimal values, and all tension combinations generated by the genetic algorithm are very well realized by the SMA actuators. The average membrane flatness of the 20 tests is shown in Fig. 20. The average standard deviation goes down quickly from around 0.22mm to less than 0.05mm. Another 20 tests are then performed with local thermal load applied, which is realized by heating the membrane locally using the heater placed under it. Again optimal tension combinations are found and realized very well. The average membrane flatness of the 20 tests is shown in Fig. 21. The average standard deviation goes down quickly from around 0.33mm to less than 0.05mm.

Time (s)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 20. The Average Membrane Flatness at room temperature

Time (s) Figure 21. The Average Membrane Flatness with Local Thermal Load

138

Fujun Peng, Yan-Ru Hu, Xin-Xiang Jiang et al.

CONCLUDING REMARKS SMA wire actuators are used for active flatness control of membrane SAR antennae. To overcome SMA's poor stability and controllability, a simple control strategy is proposed based on the idea of adjusting the SMA wire temperature as fast as possible. Testing Results show that the proposed control strategy is very effective in controlling SMA wire actuators. Their output displacements or tensions can track desired values with very good accuracy. The SMA wire actuators together with the proposed SMA control strategy are then integrated into the active flatness control system. Testing results show that the desired control tensions are realized with satisfactory accuracy, and membrane flatness is improved remarkably.

REFERENCES [1]

[2]

[3]

[4]

[5]

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

[6]

[7]

Jenkins, C. H. M. (ed.), Gossamer Spacecraft: Membrane and Inflatable Structures Technology for Space Applications, Progress in Astronautics and Aeronautics, Vol.191, 2001 Cadogan, D., and Grahne M., "Inflatable Space Structures: A New Paradagm for Space Structure Design", Proceedings of the 49th International Astronautical Congress, Sept. 28-Oct 2, 1998, Melbourne, Australia, IAF-98-I.1.02 Lin, J. K. H., and Cadogan, D. P., "An Inflatable Microstrip Reflectarray Concept for Ka-Band Applications", Proceedings of the 41st AIAA/ASME/ASCE/ AHS/ASC Structures, Structural Dynamics and Materials Conference & Exhibit, April 3-6, 2000, Atlanta, AIAA2000-1831 Karooka, D. K., Jensen, and D. W., "Advanced Space Structure Concepts And Their Development", Proceedings of the 42nd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference and Exhibit, April 16-19, 2001, Seattle, AIAA-2001-1257 Song, G., Chaudhry, V., and Batur, V., "Precision tracking control of shape memory alloy actuators using neural networks and sliding-mode based robust controller," Smart Materials and Structures, Vol.12 , pp.223-231, 2003 Webb, G., et al., "Control of SMA actuators in dynamic environments," Proceedings of SPIE Conference on Mathematics and Control in Smart Structures, Vol. 3667, pp.278289, 1999 Heintze, O., Seelecke, S., and Buskens, C., "Modeling and optimal control of microscale SMA actuators," Proceedings of SPIE Conference on Smart Structures and Materials, pp.495-505, 2003

In: Antennas: Parameters, Models and Applications Editor: Albert I. Ferrero

ISBN 978-1-60692-463-1 © 2009 Nova Publishers, Inc.

Chapter 5

MULTIPLE ANTENNA CODING & SIGNAL PROCESSING: SPACE-TIME CODING FOR WIRELESS COMMUNICATIONS S. Manioudakis1 and A.M. Maras2 1

2

Siemens Enterprise Network, Athens, Greece Department of Communication Sciences and Technology, School of Applied Sciences and Technology, University of Peloponnese, Tripolis, Greece

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

1. INTRODUCTION Wireless communication systems are experiencing a rapid growth in the number of subscribers and the range of services and hence, limitations on the availability of the radio frequency spectrum are growing as well. As a result, a substantial amount of research effort has focused on transmission techniques, along with sophisticated coding and signal processing algorithms in order to overcome the various detrimental effects of wireless propagation phenomena. These techniques must also allow for efficient use of the spectrum and provide high-performance, reliable and cost-effective wireless communication. In order to provide improved performance in wireless systems and to cope with various propagation impairments it is possible to design various channel codes or modulation schemes jointly or independent of each other [1]. In a joint manner, modulation and coding can be used in schemes like trellis-coded modulation (TCM) [2] resulting in increased performance due to the increased Euclidean distance. In an independent manner, channel codes can be used to boost system performance at the expense of increasing resource requirements such as transmission power or bandwidth. This results in decreasing the spectral efficiency of a system. Alternatively, various modulation schemes can be used for different applications giving different trade-offs between performance and implementation complexity

1

Siemens Enterprise Networks, Software Development, 15 Andrea Metaxa str, Room 2.01, Athens, Greece, email:[email protected] 2 Department of Communication Sciences and Technology, School of Applied Sciences and Technology, University of Peloponnese, Tripolis, Greece, email:[email protected]

140

S. Manioudakis and A.M. Maras

[1]. Another popular alternative is to use diversity techniques. This usually (but not always) results in employing multiple transmit and multiple receive antenna systems.

2. DIVERSITY TECHNIQUES Diversity can occur in time, frequency or space domains [3-9]. The main idea is that multiple replicas of the transmitted signals arrive at the receiver, all carrying the same information but with small correlation in channel impairments affecting any pair of received replicas. In such situations the probability that all received replicas of a given signal are below an acceptable level of integrity is much smaller than the probability of any individual signal being below that level. This increases the reliability of the receiver (i.e., decreases the probability of error). The philosophy behind diversity is to transmit signal replicas by using as few system resources (e.g., bandwidth, power, complexity) as possible whilst using these replicas at the receiver in a manner that gives the maximum reduction in error rates. Diversity gain can be defined as the slope of the curve of the error probability versus receiver signal-tonoise ratio (SNR), expressed in a logarithmic scale as shown in (1). Diversity Gain = − lim

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

SNR →∞

log( Pe ) log(SNR )

(1)

Time diversity can be achieved by transmitting identical messages in different time slots provided that the channel is treating each transmission slot in an uncorrelated manner. If retransmissions occur within the coherence time of the channel then no performance gains are expected. Frequency diversity can be achieved by transmitting the same information in different frequencies provided that these frequencies have enough separation so that the channel treats each transmission frequency in an uncorrelated manner. If retransmissions occur within the coherence bandwidth of the channel then no performance gains are expected. Both timefrequency diversity techniques induce a loss in bandwidth efficiency of a system due to retransmissions of the same information. Space diversity at the transmitter also referred to as transmit diversity is implemented by transmitting the same information from multiple antennas. Given that there is sufficient spacing between the antennas, uncorrelated versions of the transmitted signals are expected at the receiver. The antenna separation that is necessary varies with antenna height, propagation environment and frequency. In indoor environments, half a wavelength ( λ w / 2 ) is adequate for uncorrelated signals. In outdoor to indoor environments two to three wavelengths are considered to be adequate and in rural outdoor environments tens of wavelengths of antenna separation may be required for uncorrelated signals. By employing multiple receive antennas, space diversity at the receiver, also referred to as receive diversity, may be achieved. Uncorrelated versions of the transmitted signals are expected at the receiver provided that there is sufficient antenna spacing. In cases when both the transmitter and the receiver employ multiple antennas and are surrounded by a large number of scatterers, the effect of antenna spacing is small provided that it is more than half a wavelength. A very attractive feature of space diversity is that it does not induce any loss of bandwidth efficiency. Communication

Multiple Antenna Coding and Signal Processing

141

systems that employ both transmit and receive diversity are referred to as MultipleInput/Multiple-Output (MIMO) systems. Such systems are well known for taking advantage of multipath diversity (a traditional pitfall of wireless transmission) and resulting in performance gains. In works like [10-12] it is shown that MIMO systems take advantage of random fading and if available, they also take advantage of multipath delay spread [13-14] for increasing transfer rates. The first approach on the capacity gains achieved by MIMO systems can be found in [15].

2.1 Transmit Diversity Transmit diversity, has been studied from the early 1990’s whilst receive diversity was in a more mature state since it had already been studied extensively. Transmit diversity was initially explored as a method of combating detrimental effects in wireless fading channels [16-19]. Using multiple transmit antennas for diversity provides better performance without increasing the bandwidth or transmission power. The first bandwidth-efficient transmit diversity scheme was proposed in [20]. It is proved that the diversity advantage of this scheme is optimal (i.e., equal to the number of transmit antennas). A multilayered space-time (ST) architecture was introduced in [11]. This scheme uses spatial multiplexing to increase the data rate and may not provide transmit diversity. Simple iterative decoding algorithms that have been proposed in conjunction with spatial multiplexing can achieve spatial diversity, mostly receive diversity.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

3. SPACE-TIME BLOCK CODES & SPACE-TIME TRELLIS CODES ST coding is a technique that has been employed for a wide range of applications, from cellular networks, radio communications and wireless transmission in general [1-90], to underwater acoustic communications [91-93]. The information theoretical results in [10] and [12] showed that there is a huge advantage in utilizing spatial diversity. This led to the first approach towards ST coding in [18]. Taking into account the research efforts of the early 1990’s a unification theory was introduced by Tarokh et al. [21-23], proposing ST coding. A complete study of design criteria for maximum diversity and coding gains was given. Delay diversity [24] was included as a special case. In the search for jointly designing channel coding, modulation and transmit diversity with no bandwidth expansion in MIMO systems, the proposed solution is to carefully generate correlation into the transmitted signals in both spatial and temporal domains. From a mathematical perspective, Tarokh et al. used combinatorics [25] to devise a tool that can incorporate channel coding (i.e., redundancy) in space and time. The introduction of ST block coding provided a theoretical framework that started an increasing interest on the subject. As a result, a great amount of advanced scientific books has been published that are merely or solely dedicated to MIMO systems and ST coding [3-4], [7-8], [19], [26-46]. Apart from the academic interest, ST coding and MIMO technology play a significant role in real life applications. For example, recent standardisation efforts such as IEEE 802.11n WLAN draft standard [47] and the WIMAX standard in [48] have adopted MIMO technology. MIMO

142

S. Manioudakis and A.M. Maras

technologies will be a part of wireless standards including the cellular systems EV-DO Rev C, UMTS LTE, and the IEEE 802.xx family of standards 802.16e, 802.16j, 802.16m, and 802.11n. Future enhancements to these standards, such as those proposed for the IMT Advanced concept, will use MIMO to achieve required data rates in the order of hundreds of Mbps and spectral efficiencies in the order of tens of bps/Hz. ST coding can be divided into space-time trellis codes (STTCs) and space-time block codes (STBCs). STTCs extend the Ungerboeck approach to the spatial domain. Instead of set partitioning large modulation constellations, STTCs set partition various spatial transmission constellations. STBCs can provide maximum diversity gains at low decoding complexity while STTCs can provide maximum diversity gains together with high coding gains at the expense of a decoding complexity that can be prohibitive for practical implementation. This is due to the fact that given a fixed number of antennas the number of trellis states at the decoder is an exponentially increasing function of the diversity level and the transmission rate. Much insight about ST coded systems can be gained by classifying them from a coding theory as well as a signal processing perspective. From a coding theory perspective ST codes can be considered as channel codes, as shown in Fig. 1, since they introduce spatio-temporal redundancy.

Coding Theory

Source Coding

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Compression

Channel Coding

Error Correction via Data Translation

Error Correction via Redundancy

Cryptography

Security/Privacy

Space-Time Coding Figure 1. ST coding from a coding theory perspective

From a signal-processing point of view, conventional processing of signals is usually done either in the time domain or in the frequency domain. By introducing an antenna array at the transmitter and/or the receiver, the spatial domain is an extra domain that is made available for processing. Assuming time-domain processing and multiple antennas at the transmitter, ST codes have multiple spatial dimensions a fact that can justify their name. If frequency-domain processing is performed, by introducing an antenna array, the spatial

Multiple Antenna Coding and Signal Processing

143

domain made available leads to efficient processing/coding in space and frequency. This, in turn, can justify the names “space-frequency processing” and “space-frequency coding” as shown in Fig. 2.

Signal Processing

Time Domain Processing

Frequency Domain Processing

Antenna Array

Space-Time Processing

Space-Time Coding

Space-Frequency Processing

Space-Frequency Coding

Figure 2. ST coding from a signal-processing perspective

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Similar to conventional channel codes, the rate of an ST code can be defined as N RSTC = c where N c is the number of input snapshots and N is the number of output N snapshots per transmit antenna for a given data frame. Some popular STBCs include the full rate-two transmit antenna STBC

⎡ s (1) − s(2)* ⎤ ⎢ * ⎥ ⎢⎣ s(2) s (1) ⎥⎦

(2)

also known as the Alamouti STBC [49]. This simple construction is part of both the WCDMA and CDMA 200 standards. In all STBCs considered herein, columns encapsulate the spatial dimension while rows encapsulate the temporal dimension. A transmitter using the Alamouti STBC is shown in Fig. 3. The source (user) signals are modulated. The modulated symbols are then input to the STBC and the output is transmitted. Two transmission intervals are required for the Alamouti STBC. In the first transmission interval, transmit antenna Tx1

transmits symbol s (1) and transmit antenna Tx2 transmits symbol s (2 ) . In the second

144

S. Manioudakis and A.M. Maras

transmission interval transmit antenna Tx1 transmits symbol − s (2 ) and transmit antenna *

Tx2 transmits symbol s (1) . *

Constellation Mapper

Space-Time Block Code

Information Source

⎡ s(1) − s (2 )* ⎤ * ⎥ ⎣⎢ s (2 ) s(1) ⎦⎥

[s(1) s(2)]

Tx1

→ ⎢

(N = 2)

( Nc = 2 )

T x2

Figure 3. A transmitter using the Alamouti STBC

The works in [21-23] included the Alamouti STBC in (2) as a special case. Some popular orthogonal STBCs include the rate ½ and rate ¾, four transmit antenna STBCs

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

⎡ s (1) − s (2 ) − s (3) − s (4 ) ⎢ s (4 ) − s (3) ⎢ s (2 ) s (1) ⎢ s (3) − s (4) s (1) s (2 ) ⎢ ⎢ s (4 ) s (3) − s (2) s (1) ⎣

⎡ ⎢ s (1) − s (2 ) ⎢ ⎢ ⎢ s (2) s (1) ⎢ ⎢ ⎢ s (3) s (3) ⎢ 2 2 ⎢ ⎢ s (3) − s (3) ⎢⎣ 2 2

s (1)

*

s (2)

*

s (3)

*

s (4)

*

− s (2 )

− s (3)

*

*

*

s (1)

*

s (4 )

− s (4)

s (1)

*

s (3)

*

− s (2)

*

*

s (3)*

* − s (4 ) ⎤ *⎥ − s (3) ⎥ * s (2) ⎥ ⎥ * s (1) ⎥⎦

⎤ ⎥ 2 ⎥ * ⎥ s ⎥ − 3 ⎥ 2 * * ⎥ s (2 ) + s (2 ) + s (1) − s (1) ⎥ ⎥ 2 * * ⎥ − s (1) − s (1) − s (2 ) + s (2 ) ⎥ ⎥⎦ 2

(3)

s 3*

2 * s (3) 2 * * − s (1) − s (1) + s (2 ) − s (2 ) 2 * * − s (2 ) − s (2 ) + s (1) − s (1) 2

(4)

and the rate ½ and rate ¾, three transmit antenna STBCs

⎡ s(1) − s(2) − s (3) − s(4) s(1)* ⎢ * s(4) − s (3) s(2) ⎢ s(2) s(1) * ⎢ s(3) − s(4) s(1) s(2) s(3) ⎣

− s(2)

− s(3)

*

*

*

s(1)

− s(4)

*

*

s(4 ) s(1)

*

* − s (4 ) ⎤ *⎥ − s(3) ⎥ * s(2 ) ⎥ ⎦

(5)

Multiple Antenna Coding and Signal Processing

⎡ ⎢ s(1) − s(2)* ⎢ ⎢ ⎢ s(2) s(1)* ⎢ ⎢ s(3) ⎢ s(3) ⎢ 2 2 ⎣

s(3)

*

2 * s(3) 2 * * − s(1) − s(1) + s(2) − s(2) 2

145

s(3)

⎤ ⎥ 2 ⎥ * ⎥ s(3) ⎥ − ⎥ 2 * *⎥ s(2) + s(2) + s(1) − s(1) ⎥ ⎥ 2 ⎦ *

(6)

In the case of complex signal constellations and for more than two transmit antennas, 3 there exist no complex orthogonal designs with RSTC > [23]. Although some findings in 4 [23] have been further generalised in [50] and orthogonal STBCs of higher rates have been found in works like [51] and [52], the STBCs produced from these works involve code matrices that are too long for the channel to be considered constant. In the case where real signal constellations are used there exist full rate STBCs for two, three and four antennas as shown in (7), (8) and (9) respectively.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

⎡ s(1) − s(2)⎤ ⎢ s(2) s (1) ⎥ ⎣ ⎦

(7)

⎡ s(1) − s (2 ) − s (3) − s(4 )⎤ ⎢ s (2) s (1) s (4) − s (3)⎥⎥ ⎢ ⎢⎣ s(3) − s(4 ) s(1) s(2) ⎥⎦

(8)

⎡ s(1) − s(2) − s(3) − s(4)⎤ ⎢ s (2 ) s(1) s(4) − s(3)⎥⎥ ⎢ ⎢ s(3) − s(4) s(1) s(2 ) ⎥ ⎥ ⎢ ⎣ s (4 ) s (3) − s (2) s (1) ⎦

(9)

The STBCs considered in (2) through (9) admit the desirable property ⎛ Tx ⎞ 2 S S* = c ⎜ s (m ) ⎟ I ⎜ ⎟ ⎝ m =1 ⎠

∑

(10)

where c is a constant. Assuming perfect channel estimates at the receiver it can be shown that ~ the possibility of transmitting S and deciding in favour of an error matrix S can be upper bounded as [9]

146

S. Manioudakis and A.M. Maras ~ ⎛ PS→S ≤⎜ ⎜ ⎝

(

) ∏ r

i =1

⎞ βi ⎟ ⎟ ⎠

− Rx

(E s / 4 N 0 )−rRx

(11)

where P(.) denotes the probability operator, Es is the symbol energy, N0 is the noise spectral density, r is the rank of the error matrix defined in [9], Rx is the number of receive antennas ⎛ and βι are the non-zero eigenvalues of the error matrix. The term ⎜ ⎜ ⎝

r

∏ i =1

⎞ βi ⎟ ⎟ ⎠

− Rx

represents the

coding gain of the ST code and the (E s / 4 N 0 )− rRx term is the diversity gain of rRx. It can be concluded from (11) that when designing STTCs the rank of the error matrix should first be maximised. As a second step, the coding gain should also be maximised. This means that, from a probability of error viewpoint, diversity gain is more important than coding gain for STTCs and if one should make a choice between diversity gain and coding gain, the latter should be sacrificed.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

4. NON-ORTHOGONAL ST CODES There are many advantages in employing orthogonal ST codes. The orthogonality of the ST codes allows simple maximum likelihood (ML) decoding by decoupling the decoding of different symbols. In a similar manner, pairwise decoding can be applied in quasi-orthogonal STBCs. Orthogonal ST codes can be easily incorporated into a communication system due their structure, a fact that makes them tractable for application. They have a straightforward parallelisation to multiple single-input/single-output (SISO) systems (the transmission of an orthogonal STBC in a MIMO channel can be easily translated into N parallel SISO channels). At high transmission rates and for more than two transmit antennas there exists a relatively limited number of ST codes with small block lengths and hence codes that can be adopted in rapidly varying channels. Furthermore, the aforementioned simple decoding algorithms may restrict the cardinality of the transmitted symbols and/or reduce the error performance of the ST code. Alternatives to orthogonal ST codes include the linear dispersion codes [53], the Bell Labs Layered ST (BLAST) architectures [11], the codes based on number theory [54], the threaded algebraic codes in [55], the codes from the Cayley transform [56], the codes from division algebras [57], the Khatri-Rao ST codes [58], the codes based on frame theory [59], the differential ST codes [60], [61], the unitary codes in [62], the complex-field ST codes in [63] and others. Although a complete list of all kinds of ST codes available in the literature is beyond our scope, the existence of a large number of alternative ST code designs is evident. Each design gives a different trade-off between diversity gain, coding gain, spatial multiplexing gain, complexity, and transmission rate. Hence, there are many candidate ST codes that can be employed in practice and a decision for the most appropriate one should take into account the application scenario under consideration.

Multiple Antenna Coding and Signal Processing

147

5. MATHEMATICAL MODELS Consider the single user case ( K = 1 ) where there exist m antennas at the transmitter for m = 1, L , Tx and m R antennas at the receiver for m R = 1, K , R x . There are two alternative mathematical formulations to a STBC system. The first approach is to impose an ST structure on the coded information matrix (source matrix) as shown in (12). Uppercase bold letters represent matrices (i.e. S is the matrix of the transmitted signals), lowercase bold letters represent vectors (i.e. h stands for the vector of channel coefficients) and small letters stand for scalars. The R x × N receive array matrix X (observation matrix) can be expressed as follows

⎡ x1 ⎤ ⎡ h1 ⎤ ⎡ n1 ⎤ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ X= ⎢ M ⎥ = ⎢ M ⎥S+⎢ M ⎥ =HS+N ⎢x R ⎥ ⎢h R ⎥ ⎢n R ⎥ ⎣ x⎦ ⎣ x⎦ ⎣ x⎦

(12)

The receive array matrix is constructed by stacking the 1 × Tx received vectors x mR for all receive antennas. The R x × Tx channel matrix H is constructed by stacking the 1 × Tx channel vectors h mR for all receive antennas. In (12), S enjoys the structure of an STBC while h is a vector of independent and identically distributed (i.i.d.) random channel coefficients. The assumption is that the noise matrix N consists of i.i.d. and circularly symmetric complex Gaussian random variables with zero mean and variance σ2/2 per dimension. In the case of K users, the model in (12) can be extended as

X = [H1

⎡ S1 ⎤ L H K ] ⎢⎢ M ⎥⎥ + N = H S + N ⎢⎣S K ⎥⎦

(13)

The R x × KTx channel matrix H is constructed by stacking the channel matrices of all Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

users and the KTx × N source matrix S is constructed by stacking their corresponding source matrices. The second equivalent approach is to impose an ST structure on the channel matrix. In the single user case we arrive at

⎡ n1 ⎤ ⎡ x1 ⎤ ⎡ H 1 ⎤ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ x = ⎢ M ⎥ = ⎢ M ⎥ s+⎢ M ⎥ = H s+n ⎢n R ⎥ ⎢x R ⎥ ⎢H R ⎥ ⎣ x⎦ ⎣ x⎦ ⎣ x⎦

(14)

The NR x × 1 receive array vector x is constructed by stacking the N × 1 vectors x mR for all receive antennas. In this case the NR x × Tx channel matrix H is constructed by stacking

148

S. Manioudakis and A.M. Maras

the N × Tx matrices H mR for all receive antennas. Each H mR enjoys the structure of an STBC while s is a Tx × 1 source vector of constellation symbols. The noise vector n models the i.i.d. noise process. In the case of K users the model in (14) can be extended as

x = [H 1

⎡ s1 ⎤ L H K ] ⎢⎢ M ⎥⎥ + n = H s + n ⎢⎣s K ⎥⎦

(15)

In this case the NR x × KTx channel matrix H is constructed by stacking the channel matrices of all users and the KTx × 1 source vector s is constructed by stacking their corresponding source vectors.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

6. CHANNEL CHARACTERISTICS The best use of multiple transmit antennas depends on the amount of channel state information that is available to the transceiver. In a flat fading channel, the channel has a constant gain and linear phase response over a bandwidth which is greater than the bandwidth of the transmitted signal. Assuming that the channel does not change rapidly during the transmission of a frame of data, this leads to the so-called quasi-static flat fading channel model. Most of the early MIMO transmission schemes (e.g., STTC, Alamouti, orthogonal STBC) were proposed for frequency-flat fading channels. However, if there are multipath signals with rather large propagation delays, the assumption of a frequency flat fading channel might not be valid, depending on the symbol duration used. In such cases an ST equalizer is required at the receiver in order to counteract the intersymbol interference (ISI). Alternatively, multi-carrier approaches may be pursued using multiple frequency-flat subbands to circumvent the problem of ISI. In general, the ST equalizer has to account for the specific structure of the employed ST coding scheme. This means that standard equalizer algorithms already available for single antenna transceivers are not suitable. This necessitates MIMO equalisation algorithms. For example, a trellis-based equalizer/detector may be used that is based on the Viterbi algorithm [64] or the Bahl-Cocke-Jelinek-Raviv (BCJR) algorithm [65]. As opposed to the Viterbi algorithm, which provides “hard” outputs (e.g., ±1 in the binary case), the BCJR algorithm calculates “soft” outputs that are optimal in the sense of the maximum a posteriori (MAP) criterion.

7. CONCATENATED ST CODING Since the invention of turbo codes in [66-67], the BCJR algorithm became popular. This is because soft outputs are of particular interest in systems employing concatenated codes in conjunction with an iterative (“turbo”) detection scheme at the receiver. Such turbo schemes are known to yield excellent system performances. Turbo codes had a rapid success since they manage to cope with conflicting tasks from information and coding theory. Although

Multiple Antenna Coding and Signal Processing

149

coding theory suggests that codes chosen “at random” should perform well if the block size is large enough, it is mainly the high complexity of conventional ML decoding that introduces a decoding challenge. Since block and convolutional codes are highly structured, encoders and decoders with reasonable implementation complexity are possible. On the other hand, due to their high structure, these codes have inferior performance as compared with the random coding bounds predicted by Shannon [68]. Taking these facts into consideration, turbo codes contain enough structure to allow practical encoding and decoding algorithms while they also possess random like properties due to the use of interleavers. They are considered as a milestone in modern channel coding theory. The turbo principle may, for example, be applied in a MIMO transmission scheme consisting of an inner ST code and an outer channel code, used in order to further improve performance. It is considered to be common practice that when ST codes with low coding gains are employed they are also concatenated with channel codes of high coding gains (e.g., turbo codes, LDPC codes etc.). This gives an overall system that achieves high coding and diversity gains. ST coded code concatenations exist in many forms. In [69], a turbo code is serially concatenated with a short STBC (Fig. 4).

Turbo Code

Space-Time Block Code

M

Figure 4. The concatenation structure in [69]

In [70], STTCs are modified to be recursive and two encoder structures are proposed. The first is a serially concatenated encoder that uses a convolutional code as an outer code and a recursive STTC as an inner code (Fig. 5). The second is a parallel concatenation (Fig. 6) and it is actually a self concatenated [71] recursive STTC. The structures of codes in [69-70] guarantee full space diversity but the code rate cannot achieve the full rate.

Convolutional Code

Interleaver

Recursive Space-Time TCM Code

M

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 5. The first structure in [70]

Interleaver

Recursive Space-Time TCM Code

M

Figure 6. The second structure in [70]

In [72], the outputs of a turbo code are bit-interleaved, mapped to QPSK symbols and transmitted using multiple antennas (Fig. 7). Full code rate is achieved but the code is not guaranteed to achieve full space diversity.

150

S. Manioudakis and A.M. Maras

Turbo Code

Bit Interleaving & Symbol Mapping

Figure 7. The structure in [72]

In [73], a BPSK parallel concatenated turbo code is used for a multiple transmit antenna system. The encoder structure is shown in Fig. 8.

Recursive Convolutional Encoder Recursive Convolutional Encoder

Interleaver

De-interleaver

Figure 8. The structure in [73]

In [74] full rate and full diversity QPSK ST turbo codes are designed. The design shown in Fig. 9 can be generalised to higher-order constellations.

Puncture 1

Interleaver 1

Recursive Encoder 2

Puncture 2

Interleaver 2

Multiplex

Recursive Encoder 1

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Interleaver

Figure 9. The structure in [74]

Other possible turbo ST coded concatenations can be found in [75]. Under various channel conditions, ST codes have also been concatenated with other channel codes such as LDPC codes [76] and hybrid combinations [77], giving a variety of performance and complexity characteristics.

8. CHANNEL KNOWLEDGE In most practical cases, the system estimates the channel at the receiver by transmitting some known pilot signals. The receiver utilizes the corresponding received signals to estimate

Multiple Antenna Coding and Signal Processing

151

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

the channel. In such a system, a coherent detection is utilized in which the decoder uses the value of the path gains estimated at the receiver [78]. A popular training based method is to correlate the known pilot sequence with the corresponding received samples. The quality of the resulting channel estimates in single antenna transceivers is determined by the autocorrelation properties of the employed pilot sequence, the number of pilot symbols used, and the noise variance. If the channel model is frequency selective with an effective channel memory length L, good auto-correlation properties are required for any cyclic shift of the pilot sequence by up to L symbols. By contrast, in MIMO systems a pilot sequence is required for each transmit antenna and hence the quality of the channel estimates is also significantly influenced by the mutual (cyclic) cross-correlation properties of the employed pilot sequences. Defining good pilot sequences for multiple transmit antenna sequences can be a challenging task. It must be clarified that the assumption of perfect channel knowledge at the transceiver is ideal and unrealistic. Although ST coding was initially developed under the assumption of perfect channel estimates at the receiver (via training data), there exist wireless links where channel tracking is undesirable or infeasible, either because of rapid changes in the channel characteristics or because of limited system resources. The main drawback in using training data is that the bandwidth efficiency of the system will decrease. In the case of GSM systems, for example, around 22% of the bandwidth is devoted to training sequences. The bandwidth efficiency issue is even more pronounced in ST coded systems since ST decoding usually requires multi-channel information at the receiver and hence the achievable diversity gain comes at the expense of a proportional increase in the amount of training that is necessary. To cope with this issue, non-coherent reception such as, differential, blind, semi-blind or hybrid combinations may be sought. Although a differential detection scheme does not require channel knowledge at the receiver, it incurs a performance penalty of around 3 dB with respect to coherent detection. Fortunately, there exist techniques like multiple-symbol DSTC [79] that can recover some of this loss. Blind schemes assume very little on the environmental characteristics and/or the transmitted signals. Based in minimal assumptions and no training data they perform signal recovery at the receiver under certain ambiguities [80]. While the blind case represents the worst case to be developed since no training is used at all, improved trade-offs between performance and complexity may be achieved by the semi-blind alternative. Semi-blind schemes employ minimal training and can be generated from blind schemes. Blind and semi-blind techniques for ST coded systems can be found in works like [36], [81-84].

9. DIVERSITY GAIN VERSUS SPATIAL MULTIPLEXING GAIN Although our discussion focused on the diversity and coding gains achieved by using multiple transmit and receive antennas, there are also other gains to be achieved. Higher capacity and hence higher transmission rates may be achieved by increasing the number of transmit antennas. In a MIMO channel with, for example, an equal number of transmit and receive antennas existing in a rich scattering environment it can be shown that the capacity increases linearly with the number of antennas. This is achieved without increasing the transmission power and as a result, higher transmission rates may be achieved due to spatial

152

S. Manioudakis and A.M. Maras

multiplexing gain. In the general case of T x transmit and R x receive antennas, T x symbols can be transmitted in order to achieve a diversity gain of R x - T x +1 (if T x = R x then the diversity gain equals to 1). The maximum spatial diversity when transmitting one symbol per time slot equals to the product T x R x . It is intuitively satisfying to view spatial multiplexing gain as the increase in transmission rate that can be achieved by employing multiple transmit antennas with respect to employing a single transmit antenna. The capacity of a MIMO channel increases by increasing the SNR and since transmission rate can be related with capacity, an increase in the transmission rate is also expected by increasing the SNR. This motivated the authors in [85] to define spatial multiplexing gain as follows Spatial

Multiplexing

Gain = lim

SNR →∞

STC rate log(SNR )

(16)

where STCrate is the rate of the ST code in bits per channel use and it is a function of the SNR. The authors in [85] derive an optimal trade-off between spatial multiplexing gain and diversity gain. This may give great insight to the trade-off between transmission rate and probability of error.

10. MAXIMUM LIKELIHOOD STBC DECODING Consider the single-user case where a user is equipped with T x = 2 transmit antennas and uses the Alamouti STBC. A receiver with two antennas is shown in Fig. 10. A single receive antenna is assumed in this Section. The following analysis is similar for all STBCs considered herein and is applicable for more than one receive antenna. The Alamouti STBC and a single receive antenna are chosen due to mathematical tractability.

Channel Estimator

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

H1

H2

Linear Combiner

ML Decoder

sˆ(1)

sˆ(2)

Figure 10. A two antenna receiver for the Alamouti STBC

Adopting the model in (14) for R x = 1 the receive array vector can be expressed as

h( 2) ⎤ ⎡ s (1) ⎤ ⎡ n(1) ⎤ ⎡ x(1) ⎤ ⎡ h(1) x=⎢ * ⎥=⎢ * ⎥⎢ ⎥+⎢ * ⎥ * ⎣ x (2)⎦ ⎣h (2) − h (1)⎦ ⎣ s( 2)⎦ ⎣n (2)⎦

(17)

Multiple Antenna Coding and Signal Processing

153

where x(1) , x(2) are the received symbols during the first and second symbol period, respectively, that can be expressed as

x(1) = h(1) s (1) + h(2) s (2) + n(1) x (2) = h * (2) s (1) − h * (1) s (2) + n * (2)

(18)

*

Let the output from the linear combiner at the receiver be expressed as

⎡ h * (1) h(2) ⎤ ⎡ x(1) ⎤ ( x = H* x = ⎢ * ⎥⎢ * ⎥ ⎣h (2) − h(1)⎦ ⎣ x (2)⎦

(19)

( ( ⎡ x (1) ⎤ where x = ⎢ ( ⎥ . From (19) we get ⎣ x ( 2) ⎦

( (

) )

2 2 ( x (1) = h(1) + h(2) s (1) + h * (1) n(1) + h(2) n * (2) 2 2 ( x (2) = h(1) + h(2) s( 2) + h * (2) n(1) − h(1) n * (2)

(20)

Hence, due to the orthogonality property of the channel matrix the received symbol during the first symbol period depends on the corresponding transmitted symbol only. Equivalently, the received symbol during the second symbol period depends only on the corresponding transmitted symbol. In vector form, (20) can be expressed as

(

)

( 2 2 ( x = h(1) + h(2) s + n

(21)

( where n = H * n . Assuming that the channel is known at the receiver and that all transmitted

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

symbols are equiprobable, an ML ST decoder chooses a vector of estimates (pair of symbols), ⎡ sˆ(1) ⎤ in the form of sˆ = ⎢ ⎥ , that minimise the distance metric ⎣ sˆ(2)⎦

(

)

2

2 2 ( sˆ = arg min x − h(1) + h(2) sˆ sˆ∈S

(22)

F

The above ST decoder can be simplified to two separate decoding rules that are more straightforward and can be expressed as

(

) + h (1) n(1) + h(2) n (2) + h(2) ) + h (2) n(1) − h(1) n (2)

sˆ(1) = arg min (s (1) − sˆ(1) ) h(1) + h(2) sˆ (1)∈S

2

(

sˆ( 2) = arg min (s (2) − sˆ(2) ) h(1) sˆ ( 2 )∈S

2

2

*

2

*

*

*

(23)

154

S. Manioudakis and A.M. Maras

11. SIMULATION RESULTS Generally, the error performance of a multiple transmit and/or multiple receive antenna system is expected to be superior to that of a single transmit and single receive antenna system due to diversity. In the following figures, a frequency flat and quasi-static Rayleigh fading channel is assumed. Perfect channel estimates exist at the receiver. The performance of various systems is investigated using 10000 Monte Carlo trials and 500 snapshots in each trial. In Fig. 11, a 3 bits/sec/Hz scheme is evaluated for one, two, three and four transmit antennas. A single antenna exists at the receiver. The Tx = 1 (single transmit antenna uncoded) scheme employs 8-PSK modulation. 3 bits/sec/Hz

0

10

4 antennas 3 antennas 2 antennas uncoded

-1

10

-2

Bit Error Rate

10

-3

10

-4

10

-5

10

-6

10

5

10

15

20 25 SNR (dB)

30

35

Figure 11. Performance of various STBCs and a single transmit antenna system for 3 bits/sec/Hz. One receive antenna is used in all cases

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The Tx = 2 (two transmit antenna) scheme employs the Alamouti STBC and 8-PSK

3 STBC in (6) 4 3 and 16-QAM modulation. The Tx = 4 (four transmit antenna) scheme employs the RSTC = 4 STBC in (4) and 16-QAM modulation. At a bit error rate (BER) of 10-5 the four transmit antenna scheme outperforms the three transmit antenna and the two transmit antenna schemes by around 2.6 and 7 dB respectively. At an SNR more than around 8 dB, the Tx = 1 scheme has the worst performance from all other simulated schemes. As the SNR increases the performance gap also increases due to diversity. However, at an SNR lower than 8dB, the Tx = 3 and Tx = 4 schemes perform worse than all other simulated schemes due to the use of modulation. The Tx = 3 (three transmit antenna) scheme employs the RSTC =

Multiple Antenna Coding and Signal Processing

155

16-QAM modulation. At low SNR, diversity can not compensate the effect of increasing the signal constellation size. 2 bits/sec/Hz

0

10

4 antennas 3 antennas 2 antennas uncoded

-1

10

-2

Bit Error Rate

10

-3

10

-4

10

-5

10

-6

10

5

10

15

20 SNR (dB)

25

30

35

Figure 12. Performance of various STBCs and a single transmit antenna system for 2 bits/sec/Hz. One receive antenna is used in all cases

In Fig. 12, a 2 bits/sec/Hz scheme is evaluated for one, two, three and four transmit antennas. A single antenna exists at the receiver. The Tx = 1 scheme employs QPSK modulation. The Tx = 2 scheme employs the Alamouti code and QPSK modulation. The

Tx = 3 scheme employs the RSTC =

1 STBC in (5) and 16QAM modulation. The Tx = 4 2

1 STBC in (3) and 16QAM modulation. At a BER of 10-5 the 2 Tx = 4 scheme outperforms the Tx = 2 scheme by around 5 dB. The phenomenon of diversity not compensating for the increased signal constellation is evident again and hence the 16QAM schemes perform worse from all other simulated schemes at low SNR. At high SNR, the Tx = 1 scheme is again of inferior performance. In Fig. 13, a 1 bit/sec/Hz scheme is evaluated for one, two, three and four transmit antennas. A single antenna exists at the receiver. The Tx = 1 scheme uses BPSK modulation.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

scheme employs the RSTC =

The Tx = 2 scheme uses the Alamouti STBC and BPSK modulation. The Tx = 3 scheme uses the RSTC =

RSTC =

1 STBC in (5) and QPSK modulation while the Tx = 4 scheme uses the 2

1 STBC in (3) and QPSK modulation. 2

156

S. Manioudakis and A.M. Maras 1 bit/sec/Hz

0

10

4 antennas 3 antennas 2 antennas uncoded

-1

10

-2

Bit Error Rate

10

-3

10

-4

10

-5

10

-6

10

5

10

15

20

25

30

SNR (dB)

Figure 13. Performance of various STBCs and a single transmit antenna system for 1 bit/sec/Hz. One receive antenna is used in all cases

The Tx = 4 scheme outperforms the Tx = 3 scheme and the Tx = 2 scheme by around 2.6 and 7.5 dB respectively at a BER of 10-5. The Tx = 1 scheme has the worst error performance compared with the other simulated schemes. In Fig. 14, the four schemes from Fig. 13 are evaluated when two antennas exist at the receiver. Note that in this case the performance gap of the simulated schemes is decreased. The Tx = 4 scheme outperforms the Tx = 3 scheme and the Tx = 2 scheme by around 3.5 and 1 dB respectively at a BER of 10-5. This is due to the fact that much of the diversity gain is already achieved using two transmit and two receive antennas. 1 bit/sec/Hz

0

10

4 antennas 3 antennas 2 antennas uncoded

-1

10

-2

Bit Error Rate

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

10

-3

10

-4

10

-5

10

-6

10

0

2

4

6

8 SNR (dB)

10

12

14

16

Figure 14. Performance of various STBCs and a single transmit antenna system for 1 bit/sec/Hz. Two receive antennas are used in all cases

Multiple Antenna Coding and Signal Processing

157

12. CONCLUSIONS AND FUTURE WORK Significant performance gains may be achieved by designing and optimising wireless networks with a cross-layer perspective [86-87]. MIMO systems designed for the physical layer may be more effective when the appropriate higher layer techniques are employed. Joint optimisation of physical layer and higher layer techniques is an issue of current research interest and development. Collaborative diversity [88] is another hot topic of investigation. Instead of placing more antennas in a handset the idea in collaborative diversity systems is to make handsets less egoistic and collaborate with each other. When a user is in a conversation his mobile could use the mobile of other users who are not in a conversation. Cooperative relays have recently been proposed for mobile and ad hoc networks. Transmit beamforming is favoured instead of ST coding when perfect channel state information is available at the transmitter. Since this assumption is ideal and unrealistic, transmitters employing hybrid combinations of beamforming and ST coding are promising. Opportunistic beamforming can be employed in slowly varying channels [89] since multiuser diversity gain diminishes when the channels vary rapidly. This is another area of ongoing research. As a final concluding remark we mention that the theoretical gains promised by MIMO systems are in some cases questionable in practice. Some causes and effects have been given in works like [90] and the references therein. Tackling such causes and effects, and finding the solutions that produce in practice as much as possible of the aforementioned theoretical gains is another area of ongoing research effort.

REFERENCES [1] [2] [3] [4]

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

[5]

[6] [7] [8] [9]

J. G. Proakis, Digital Communications, 4th edition, McGraw-Hill, 2001. G. Ungerboeck, “Channel coding with multilevel/phase signals”, IEEE Trans. Inform. Theory, vol. 28, pp. 55-67, Jan. 1982. D. Shiu, Wireless Communication Using Dual Antenna Arrays, Kluwer Academic Publishers, 2000. E. G. Larsson, P. Stoica, G. Ganesan, Space-time block coding for wireless communications, Cambridge University Press, 2003. Naguib, N. Seshadri, A. R. Calderbank, “Increasing Data Rate over Wireless Channels: Space Time Coding and Signal Processing for High Data Rate Wireless Communications”, IEEE Signal Processing Mag., vol. 17, pp. 76-92, May 2000. Paulraj, C. Papadias, "Space-time processing for wireless communications", IEEE Signal Processing Mag., vol. 46, pp. 49-83, Nov. 1997. Paulraj, D. Gore, R. Nabar, Introduction to Space-Time Wireless Communications, Cambridge University Press, 2003. G. Ganesan, P. Stoica, Signal processing advances in wireless and mobile communications, Volume II, pp. 59-88, Prentice-Hall, 2001. D. Gesbert, M. Shafi, D. Shiu, P. J. Smith, A. Naguib, “From theory to practice: An overview of MIMO space-time coded wireless systems”, IEEE J. Select. Areas Commun., vol. 21, pp. 281-302, April 2003.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

158

S. Manioudakis and A.M. Maras

[10] G. J. Foschini, M. J. Gans, “On the limits of wireless communications in a fading environment when using multiple antennas”, Wireless Pers. Commun., vol. 6, pp. 311– 335, Mar. 1998. [11] G. J. Foschini, “Layered space-time architecture for wireless communication in a fading environment when using multi-element antennas”, Bell Labs Tech. J., Autumn, pp. 41– 59, 1996. [12] E. Telatar, “Capacity of multi-antenna Gaussian channels”, Euro. Trans. Commun., vol. 10, pp. 585–595, Dec. 1999. [13] H. Bolcskei, D. Gesbert, and A. J. Paulraj, “On the capacity of OFDM-based spatial multiplexing systems”, IEEE Trans. Commun., vol. 50, pp. 225-234, Feb. 2002. [14] G. Raleigh, J. M. Cioffi, “Spatial-temporal coding for wireless communications", IEEE Trans. Commun., vol. 46, pp. 357-366, 1998. [15] J. H. Winters, “On the capacity of radio communication systems with diversity in a Rayleigh fading environment”, IEEE J. Select. Areas Commun., vol. 5, pp. 871-878, June 1987. [16] J.-C. Guey, M. P. Fitz, M. R. Bell, W. Y. Kuo, “Signal design for transmitter diversity wireless communication systems over Rayleigh fading channels”, IEEE Trans. Commun., vol. 47, pp. 527- 537, Apr. 1999. [17] Narula, M. Trott, G. Wornell, “Performance limits of coded diversity methods for transmitter antenna arrays”, IEEE Trans. Inform. Theory, vol. 45, pp. 2418-2433, Nov. 1999. [18] N. Seshadri, J. H. Winters, “Two signaling schemes for improving the error performance of frequency-division-duplex (FDD) transmission systems using transmitter antenna diversity”, in Proc. Vehicular Technology Conference, pp. 508-511, May, 1993. [19] L. Hanzo, C. H. Wong, M. S. Yee, Adaptive Wireless Transceivers, IEEE Press, Wiley, 2002. [20] Wittneben, “Base station modulation diversity for digital SIMULCAST”, in Proc. Vehicular Technology Conference, pp. 848-853, May 1991. [21] V. Tarokh, N. Seshadri, A. R. Calderbank, “Space-time codes for high data rate wireless communication: Performance criterion and code construction”, IEEE Trans. Inform. Theory, vol. 44, pp. 744-765, March 1998. [22] V. Tarokh, H. Jafarkhani, A. R. Calderbank, “Space-time block coding for wireless communications: Performance Results”, IEEE Sel. Areas Commun., vol.17, pp. 451460, March 1999. [23] V. Tarokh, H. Jafarkhani, A. R. Calderbank, “Space-time block codes from orthogonal designs”, IEEE Trans. Inform. Theory, vol. 45, pp. 1456-1467, July 1999. [24] Wittneben, “A New bandwidth efficient transmit antenna modulation diversity scheme for linear digital modulation”, in Proc. International Conference on Communications vol. 3, pp. 1630-1634, Geneva, 1993. [25] V. Geramita, J. Seberry, Orthogonal Designs, Quadratic Forms and Hadamard Matrices. Lecture Notes in Pure and Applied Mathematics, 1979. [26] M. A. Karim, Space-Time Coding: An adaptive approach with iterative detection and decoding, Verlag, Germany, 2007. [27] T. M. Duman, A. Ghrayeb, Coding for MIMO Communication Systems, Wiley, 2007. [28] G. Tsoulos, MIMO System Technology for Wireless Communications, CRC Press, 2006.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Multiple Antenna Coding and Signal Processing

159

[29] E. Biglieri, R. Calderbank, A. Constantinides, A. Goldsmith, A. Paulraj, H. V. Poor, MIMO Wireless Communications, Cambridge University Press, 2007. [30] L. C. Tran, T. A. Wysocki, A. Mertins, J. Seberry, Complex Orthogonal Space-Time Processing in Wireless Communications, Springer, 2006. [31] P. van Rooyen, M. P. Lötter, D. van Wyk, Space-Time Processing for CDMA Mobile Communications, Kluwer, 2000. [32] Oestges, B. Clerckx, MIMO Wireless Communications, Elsevier, 2007. [33] V. Kuhn, Wireless Communications over MIMO Channels, Wiley, 2006. [34] H. Bölcskei, D. Gesbert, C. B. Papadias, and A.-J. van der Veen, Space-Time Wireless Systems: From Array Processing to MIMO Communications, 2006. [35] G. Durgin, Space-Time Wireless Channels, Pearson Higher Education, 2002. [36] G. B. Giannakis, Z. Liu, Space-Time Coding for Broadband Wireless Communications, Wiley, 2007. [37] L. Hanzo, T. H. Liew, B. L. Yeap, Turbo Coding, Turbo Equalization and Space-Time Coding, IEEE Press, Wiley, 2002. [38] Hottinen, O. Tirkkonen, R. Wichman, Multi-antenna Transceiver Techniques for 3G and Beyond, Wiley, 2003. [39] M. Jankiraman, Space-Time Codes and MIMO Systems, Artech House, 2004. [40] Paulraj, D. Gore, R. Nabar, Introduction to Space-Time Wireless Communications, Cambridge University Press, 2003. [41] Vucetic, J. Yuan, Space-Time Coding, Wiley, 2003. [42] X. Wang and H. V. Poor, Wireless Communication Systems, Wiley, 2003. [43] H. Jafarkhani, Space-time coding, Cambridge University Press, 2005. [44] B. Gershman, N. D. Sidiropoulos, Space-time processing for MIMO communications, Wiley, 2005. [45] Goldsmith, Wireless communications, Cambridge University Press, 2005. [46] Tse, P. Viswanath, Fundamentals of Wireless Communication, Cambridge University Press, 2005. [47] IEEE 802.11n. IEEE Standard for Enhancements for Higher Throughput. Draft Standard. [48] IEEE 802.16. IEEE Standard for Local and metropolitan area networks – Part 16: Air Interface for Fixed Broadband Wireless Access Systems http://standards.ieee.org/ getieee802/download/802.16-2004.pdf. [49] S. M. Alamouti, "A simple transmit diversity technique for wireless communications", IEEE J. Sel. Areas Commun., vol. 16, pp. 1451-1458, Oct. 1998. [50] M. Zafar Ali Khan, B. Sundar Rajan, “A generalization of some existence results on orthogonal designs for STBCs”, IEEE Trans. Inform. Theory, vol. 50, pp. 218-219, Jan. 2004. [51] H. Wang, X. Xia, "Upper bounds of rates of complex orthogonal space-time block codes", IEEE Trans. Inform. Theory, vol. 49, pp. 2788-2796, Oct. 2003. [52] X. Liang, "Orthogonal designs with maximal rates", IEEE Trans. Inform. Theory, vol. 49, pp. 2468-2503, Oct. 2003. [53] Hassibi, B. M. Hochwald, "High-rate codes that are linear in space and time", IEEE Trans. Inform. Theory, vol. 48, pp. 1804-1824, July 2002. [54] M. O. Damen, A. Tewfik, J. C. Belfiore, “A construction of a space-time code based on number theory”, IEEE Trans. Inform. Theory, vol. 48, pp. 753–760, Mar. 2002.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

160

S. Manioudakis and A.M. Maras

[55] M. O. Damen, H. El-Gamal, N. C. Beaulieu, “Linear threaded algebraic space-time constellations”, IEEE Trans. Inform. Theory, vol. 49, pp. 2372–2388, Oct. 2003. [56] Y. Jing, B. Hassibi, "Unitary space-time modulation via Cayley transform", IEEE Trans. Signal Processing, vol. 51, pp. 2891–2904, Nov. 2003. [57] Sethuraman, B. S. Rajan, V. Shashidhar, "Full-diversity, high-rate space-time block codes from division algebras", IEEE Trans. Inform. Theory, vol. 49, pp. 2596-2616, Oct. 2003. [58] N. D. Sidiropoulos, R. S. Budampati, “Khatri-Rao space-time codes”, IEEE Trans. Signal Processing, vol. 50, pp. 2396-2407, Oct. 2002. [59] R. W. Heath Jr., A. Paulraj, "Linear dispersion codes for MIMO systems based on frame theory", IEEE Trans. Signal Processing, vol. 50, pp. 2429-2441, Oct. 2002. [60] H. Jafarkhani and V. Tarokh, “Multiple transmit antenna differential detection from generalized orthogonal designs”, IEEE Trans. Inform. Theory, vol. 47, pp. 2626-2631, Sept. 2001. [61] M. Hochwald, W. Sweldens, "Differential unitary space-time modulation", IEEE Trans. Commun., vol. 48, pp. 2041-2052, Dec. 2000. [62] M. Hochwald, T. L. Marzetta, T. J. Richardson, W. Sweldens, “Systematic design of unitary space-time constellation”, IEEE Trans. Inform. Theory, vol. 46, pp. 1962-1973, Sept. 2000. [63] X. Ma, G. B. Giannakis, “Full-diversity full-rate complex-field space-time coding”, IEEE Trans. Signal Processing, vol. 51, pp. 2917-2930, Nov. 2003. [64] J. Viterbi, “Error bounds for convolutional codes and an asymptotically optimum decoding algorithm”, IEEE Trans. Inform. Theory, vol. 13, pp. 260-269, April 1967. [65] L. Bahl, J. Cocke, F. Jeinek and J. Raviv, “Optimal decoding of linear codes for minimizing symbol error rate”, in Proc. International Symposium on Information Theory, p. 90, 1972. [66] Berrou, A. Glavieux, P. Thitimajshima, "Near Shannon limit error correcting coding and decoding: Turbo-codes", in Proc. International Conference on Communications, pp. 1064-1070, Geneva, 1993. [67] Berrou and A. Glavieux, “Near optimum error correcting coding and decoding: Turbocodes”, IEEE Trans. Commun., vol. 44, pp. 1261-1271, Oct. 1996. [68] E. Shannon, “A mathematical theory of communication”, Bell Systems Technical Journal, vol. 27, pp. 379-423, July 1948. [69] G. Bauch, “Concatenation of space-time block codes and turbo-TCM”, in Proc. International Conference on Communications, vol. 2, pp. 1202–1206, June 1999. [70] K. R. Narayanan, “Turbo decoding of concatenated space-time codes”, in Proc. Allerton Conf. on Commun, Control and Comput, Monticello, Illinois, USA, Sept. 1999. [71] S. Benedetto, D. Divsalar, G. Montorsi, and F. Pollara, “Self-concatenated codes with self-iterative decoding for power and bandwidth efficiency”, in Proc. International Symposium on Information Theory, p. 177, 1998. [72] Stefanov and T. M. Duman, “Turbo coded modulation for wireless communications with antenna diversity”, in Proc. Vehicular Technology Conference, Amsterdam, Netherlands, September 1999. [73] H. Su and E. Geraniotis, “Spectrally efficient turbo codes with full antenna diversity”, in Proc. Multiacc. Mob. and Teletraffic for Wireless Commun, Italy, Oct. 1999.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Multiple Antenna Coding and Signal Processing

161

[74] Y. Liu, M. P. Fitz, O. Y. Takeshita, “Full rate space-time turbo codes”, IEEE J. Sel. Areas Commun., vol. 19, pp. 969–980, May 2001. [75] Cui, “Turbo space-time coded modulation: Principles and performance analysis”, Ph.D. dissertation, New Jersey Institute of Technology, New Jersey USA, May 2001. [76] Lu, X. Wang, K. Narayanan, “LDPC-based space-time coded OFDM systems over correlated fading channels: Performance analysis and receiver design”, IEEE Trans. Commun., vol. 50, pp. 74-88, Jan. 2002. [77] S. Manioudakis, W. L. Woo, B. Sharif, “Code concatenations for blind space-time coded systems”, in Proc. Vehicular Technology Conference, vol. 1, pp. 682-686, Orlando, USA, Oct. 2003. [78] V. Tarokh, A. Naguib, N. Seshadri, A. R. Calderbank, “Space-time codes for high data rates wireless communications: performance criteria in the presence of channel estimation errors, mobility and multiple paths”, IEEE Trans. Commun., vol. 47, pp. 199–207, Feb. 1999. [79] Gao, A. M. Haimovich, D. Lao, “Multiple-symbol differential detection for space-time block codes”, in Proc. Conf. Info. Sciences and Sys., Princeton, March 2002. [80] S. Manioudakis, A. M. Maras, Progress in wireless communications research, Chapter 11: pp. 259-280, Editor: A. P. Martinhoff, NOVA publishers, 2007. [81] S. Manioudakis, “Blind estimation of space-time coded systems using the finite alphabet-constant modulus algorithm”, IEE proc. Vision, Image & Signal Processing, vol. 153, pp. 549-556, Oct. 2006. [82] S. Manioudakis, Blind estimation for space-time coded systems, PhD dissertation, University of Newcastle Upon Tyne (UK), 2004. [83] L. Swindlehurst, G. Leus, “Blind and semi-blind equalization for generalized spacetime block codes”, IEEE Trans. Signal Processing, vol. 50, pp. 2489-2498, Oct. 2002. [84] Reynolds, X. Wang, H. V. Poor, "Blind adaptive space-time multiuser detection with multiple transmitter and receiver antennas", IEEE Trans. Signal Processing, vol. 50, pp. 1261-1276, June 2002. [85] L. Zheng, D. N. C. Tse, “Diversity and multiplexing: a fundamental trade-off in multiple-antenna channels”, IEEE Trans. Inform. Theory, vol. 49, pp. 1073–1096, May 2003. [86] C. Comanciu, N. Mandayam, Wireless networks: Multiuser detection in cross-layer design, Kluwer, 2004. [87] R. Jurdak, Wireless ad hoc and sensor networks: A cross-layer design perspective, Springer, 2007. [88] J. C. Laneman, Cooperative diversity in wireless networks: Algorithms and architectures, PhD Thesis, MIT, 2002. [89] M. Kobayashi, G. Caire, D. Gesbert, “Transmit diversity vs. opportunistic beamforming in data packet mobile downlink transmission”, IEEE Trans. Commun., vol. 55, pp. 151157, Jan. 2007. [90] J. Mietzner, P. A. Hoeher, “Boosting the performance of wireless communication systems: Theory and practice of multiple-antenna techniques”, IEEE Comms. Mag., pp. 40-47, Oct. 2004. [91] S. Roy, T. M. Duman, V. McDonald, J. G. Proakis, “High-rate communications for underwater acoustic channels using multiple transmitters and space-time coding:

162

S. Manioudakis and A.M. Maras

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Receiver structures and experimental results”, IEEE J. Oceanic Eng., vol. 32, pp. 663– 688, Jul. 2007. [92] B. Kilfoyle, J. C. Preisig, A. B. Baggeroer, “Spatial modulation experiments in the underwater acoustic channel”, IEEE J. Oceanic Eng., vol. 30, pp. 406–415, Apr. 2005. [93] C. Pelekanakis, A. B. Baggeroer, “Achieving maximum space, time and frequency diversity in shallow water acoustic channels: System design and experimental results”, in preparation

In: Antennas: Parameters, Models and Applications Editor: Albert I. Ferrero

ISBN 978-1-60692-463-1 © 2009 Nova Publishers, Inc.

Chapter 6

RESISTIVE RECTANGULAR PATCH ANTENNA WITH UNIAXIAL SUBSTRATE Amel Boufrioua Electronics Department, University of Constantine, 25000 Constantine, Algeria

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ABSTRACT The moment method technique, based on Galerkin’s procedure, is developed to examine the scattering properties of a rectangular microstrip patch antenna with non zero surface resistance and containing anisotropic substrate. The electric field integral equation for a current element on a grounded dielectric slab of infinite extent is developed by different basis functions, and the asymptotic forms of these basis functions are also given in this study. The electric field integral equation, which enforces the boundary condition, must vanish on the patch surface, and can then be discritized into a matrix form. The necessary terms for representing the surface resistance on the patch are derived and are included in the equation in the form of a resistance matrix. Once the impedance matrix and the resistance matrix are calculated, the results form a system of simultaneous equations. The resulting system of equations is then solved for the unknown current modes on the patch. The complex resonant frequency, the radiation and the scattering radar cross section of a microstrip antenna—including the effect of the surface resistance and the effect of uniaxial anisotropy in the substrate—are analyzed. Also, a theoretical analysis of a rectangular patch antenna excited by a microstrip line or an electromagnetic coupled feeding with a perfectly or an imperfectly conducting patch and isotropic or uniaxial anisotropic substrate are presented in this chapter. Note that the currents on the feed line and the patch are expanded in terms of three types of modes which will be given in detail in this study. Numerical results show that the surface resistance significantly affects the radar cross section and radiation of the rectangular microstrip patches. Also, it is worth noting that our calculated frequencies do not depend on the surface resistance. Moreover, the results indicate that the resonant frequency is slightly increased due to the positive uniaxial anisotropy; on the other hand, the resonant frequency is decreased due to the negative uniaxial anisotropy. However, the resonant frequency, the radiation and the radar cross section are slightly shifted due to the ε x permittivity change and drastically change due to the ε z permittivity change.

164

Amel Broufrioua

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

1. INTRODUCTION In recent years, microstrip patch antennas became one of the most popular antenna types for use in aerospace vehicles, telemetry and satellite communication. In general these structures are poor radiators, but by proper design the radiation performance can be improved and these structures can be used as antenna elements [1, 2]. For the analysis and the design of microstrip antennas, several techniques have been developed, such as the cavity model and the transmission line model [3, 4]; however, the accuracy of these approximate models (simple analytical methods) is limited and only suitable for analysing simple, regularlyshaped antennas or thin substrates. The most important criteria in selecting methods are the rapidity and effort to derive the algorithm and write the program, storage requirement and computer run time necessary to obtain desired accuracy [5]. The full-wave moment method in the spectral domain achieves greater computational efficiency [5], has been applied extensively, and is now a standard approach for analysis of microstrip geometry. In such an approach, the spectral dyadic Green’s function, which relates the tangential electric fields and currents, is developed. As a rigorous solution to the problem of a rectangular microstrip antenna, which is the most widely-used configuration because its shape readily allows theoretical analysis, Galerkin’s method is employed in the spectral domain where various sets of patch current expansions are used. Clearly the choice of the basis functions used to calculate the current of a rectangular patch is crucial for obtaining numerical convergence with faster times. Fast numerical convergence is obtained using subdomain roof top basis functions to expand the current on the patch with a simplified algebraic formulation. Another type of basis functions is based on the complete set of orthogonal modes of the magnetic cavity, and the other employs Chebyshev polynomials with the proper edge condition for the patch currents [6]. The boundary condition for the electric field on a thin resistive sheet has been examined by Senior and is valid as long as the sheet is electrically thin. It is found that the substrate permittivity is a very important factor to be determined in microstrip antenna designs; many substrate materials used for printed circuit antenna exhibit dielectric anisotropy, especially uniaxial anisotropy [7]. However, the study of anisotropic substrate materials is interesting since it has been found that the use of such materials may have a beneficial effect on a circuit or antenna [1]. The designers should carefully check for the anisotropic effects in the substrate materials with which they will work. Moreover, in this chapter we will describe spectral domain analysis of imperfectly conducting microstrip patch antennas by using entire domain sinusoid basis functions to model the patch current density. Several authors have examined the scattering response of resistive strips and tapered resistive strips. This approach has also been used in order to study frequency selective surfaces [8]. One important aspect is that there is a variety of feeding techniques that can be applied to microstrip patch antennas. Generally, the most well-known feeding methods of a microstrip antenna are microstrip feeding, probe feeding and electromagnetic coupling feeding. Microstrip feeding is easily fabricated by connecting the microstrip line to the edge of the patch directly [9]. The actual flow of current from the feed line to the patch is modeled by expanding the traveling wave currents on the feed line, as in [9], and using entire domain modes for the patch currents; continuity of current from the feed line to the patch is provided by several piecewise sinusoidal modes that overlap the feed line and the patch.

Resistive Rectangular Patch Antenna with Uniaxial Substrate

165

2. THEORY Consider a perfectly conducting rectangular patch of dimensions a × b on a grounded dielectric substrate of a uniform thickness h, shown in Figure 1. y Radiating conductor

b

x a

(2) (1)

ε x ,ε z

h

Ground plane

Figure 1. Geometry of a rectangular microstrip antenna.

The substrate material is taken to be isotropic or uniaxially anisotropic with the optical axis normal to the patch. Mathematically, the permittivity of a uniaxially anisotropic substrate can be represented by a tensor or dyadic of this form [1]:

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ε = ε 0 .diag[ε x ,ε x ,ε z ].

(1)

ε0 is the free-space permittivity; all the dielectric materials are assumed to be nonmagnetic with permeability μ0. ε z is the relative permittivity in the direction of the optical axis.

ε x is the relative permittivity in the direction perpendicular to the optical axis. For all our computations the mode that we will be studying is the TM01 mode with the dominant component of the current in the y direction. In this section, to simplify the analysis, the antenna feed will not be considered. The boundary condition on the patch is given by [1]:

E scat + Einc = 0 Einc Tangential components of incident electric field.

(2)

166

Amel Broufrioua

E scat Tangential components of scattered electric field. From Maxwell’s equations in the Fourier transform domain, knowing the general form of

~

~

the longitudinal components Ez and H z [1] the transverse fields in the (TM, TE) representation can be written in terms of these components as [10]:

~ e ⎡E (k , z )⎤ ~ E s (k s , z ) = ⎢ ~ s h s ⎥ = e i k z z A(k s ) + e −i k z z B(k s ) ⎣E s (k s , z )⎦

(3)

~ e ⎡H (k , z )⎤ ~ H s (k s , z ) = ⎢ ~ s h s ⎥ = g (k s ) e i k z z A(k s ) − e −i k z z B(k s ) ⎣H s (k s , z )⎦

[

]

(4)

The superscripts e and h denote the TM and the TE waves, respectively. A and B are two unknowns vectors to be determined [10], ks is the transverse wave

vector; k s = xˆ k x + yˆ k y , k s = k s and g (k s ) is determined by:

⎡ωε 0 ε x ⎢ ke ⎢ z ⎢ 0 ⎢ ⎣

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

⎡k e kz = ⎢ z ⎣0

⎤ 0 ⎥ ⎥ k zh ⎥ ω μ 0 ⎥⎦

(5)

1

1 0⎤ e ⎛ εx 2 ⎞2 h 2 2 2 2 ⎟ ⎜ ε , k k k , k = ( ε k − k ) , k0 = ω ε 0 μ 0 = − ⎥ z ⎜ x 0 z x 0 s s ε z ⎟⎠ k zh ⎦ ⎝

k ze and k zh are respectively propagation constants for TM and TE waves in the uniaxial dielectric. ω is the angular frequency,

μ 0 and ε 0 are the permeability and permittivity of free

space, respectively. By eliminating the unknowns A and B, in the equations (3) and (4) we obtain the following matrix which combine the tangential field components on both sides z1 and z2 of the considered layer as input and output quantities:

( (

) )

( (

) )

~ ⎡E k s , z 2− ⎤ ⎡ I cos k z h =⎢ ⎢ ~ e(h ) e(h ) − ⎥ ⎣⎢H k s , z 2 ⎦⎥ ⎢⎣i g sin k z h e(h )

e(h )

(

)

( (

) )

e(h ) ~ i g −1 sin k z h ⎤ ⎡ Ee(h ) k s , z1+ ⎤ ⎡ 0 ⎤ ⎥ ×⎢~ e(h ) ⎥ − ⎢~ e ( h ) ⎥ (6) I cos k z h ⎦⎥ ⎣H e(h ) k s , z1+ ⎦ ⎣J (k s )⎦

(

)

Resistive Rectangular Patch Antenna with Uniaxial Substrate

167

I is the unit matrix. In the spectral domain the relationship between the patch current and the electric field on the patch is given by:

( ) = G (k )⋅ ~J (k )

~ E k s

s

s

(7)

s

Where G is the spectral dyadic Green’s function which is efficiently determined by

~

equation (8) and J (k s ) is the current on the patch which relates to the vector Fourier transform of J(rs) [11].

⎡G e G=⎢ ⎣ 0

0 ⎤ ⎥ Gh ⎦

(8)

Ge =

− k ze k z sin (k z1 h ) iωε 0 ik ze sin (k z1 h ) + ε x k z cos(k z1 h )

(8a)

Gh =

− k 02 sin (k z1 h ) iωε 0 ik z sin (k z1 h ) + k zh cos(k z1 h )

(8b)

1

1

In the case of the isotropic substrate:

⎡ μ0 cos(kz1 h) ⎢ ε (1− i ε r kz cot(kz1 h) kz1 ) G=⎢ 0 ⎢ 0 ⎢ ⎢⎣

⎤ ⎥ ⎥ ⎥ μ0 1 ⎥ ε 0 cos(kz1 h) (1− i kz1 cot(kz1 h) kz )⎥⎦ 0

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Where:

k z1 = k0 cos(k z h ) k z = k 02 − k s2

The surface current J on the patch can be expanded into a series of known basis functions Jxn and Jym: N ⎡ 0 ⎤ ⎡ J (r )⎤ M J (rs ) = ∑ a n ⎢ xn s ⎥ + ∑ bm ⎢ ⎥ n =1 ⎣ 0 ⎦ m =1 ⎣ J ym (rs )⎦

(9)

168

Amel Broufrioua

Where an and bm are the unknown coefficients to be determined in the x and y direction respectively. The vector Fourier transforms for (9) are expressed by

~ J (k s ) =

∞

∫ ∫ d r F (k s

s

,−rs ) J (rs )

(10)

−∞

F (k s ,−rs ) is the kernel of the vector Fourier transforms [7]. F (k s , rs ) =

1 ks

⎡k x ⎢k ⎣ y

k y ⎤ ik s ⋅rs , rs = xxˆ + yyˆ . e − k x ⎥⎦

(11)

kx and ky are the spectral variables corresponding to x and y respectivelly. Expression (9) is substituted into equation (10), the results equation is given by:

1 ⎡k x ⎤ N 1 ⎡k ⎤M ~ ~ ~ J (k s ) = ⎢ ⎥ ∑ a n J xn (k s ) + ⎢ y ⎥ ∑ bm J ym (k s ) k s ⎣k y ⎦ n =1 k s ⎣− k x ⎦ m =1

(12)

~ ~ J xn (k s ) , J ym (k s ) are the Fourier transforms of J xn (rs ) , J ym (rs ) respectively. The choice of the basis functions is very important for a rapid convergence to the values; we have chosen four types of basis functions on the patch: • • • •

Entire domain sinusoidal basis functions without edge condition. Entire domain sinusoidal basis functions with edge condition. A combination of Chebyshev polynomials with edge condition. Roof top subdomain basis functions.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

A comparative study between these four basis functions with their asymptotic forms has been developed in this chapter.

2. 1. Exact forms of the currents The entire-domain sinusoidal basis functions with and without edge condition are used to expand the unknown currents on the patch. Many subsequent analyses involve entire-domain basis functions that are limited to canonical shapes such as rectangles, circles and ellipses. However there is little theoretical analysis concerning the use of Chebyshev basis functions for modelling the current on the patch antenna. Recently, much work has been published regarding the scattering properties of microstrip antennas on various types of substrate geometries. Virtually all this work has been done with entire domain basis functions for the current on the patch.

Resistive Rectangular Patch Antenna with Uniaxial Substrate

169

2. 1. 1. Sinusoid basis function without edge condition For the resonant patch, entire domain expansion currents lead to fast convergence and can be related to a cavity model type of interpretation [9]. The currents can be defined using a sinusoidal basis function defined on the whole domain, without the edge condition [12, 13]; there currents are associated with the complete orthogonal modes of the magnetic cavity. Both x and y directed currents were used, with the following forms [7]:

( )

( )

( )

( )

⎤ ⎡ n1π ⎤ ⎡ n2π y+ b ⎥ x+ a ⎥cos ⎢ J xn(rs )=sin ⎢ b a 2 2 ⎦ ⎣ ⎦ ⎣

(13a)

⎤ ⎡ m1π ⎤ ⎡ m2π y+ b ⎥ x+ a ⎥sin ⎢ J ym(rs )=cos ⎢ b a 2 2 ⎦ ⎣ ⎦ ⎣

(13b)

The transforms of these basis functions (13.a) and (13.b) can be written as: ∞ ~ J xn (k s ) = ∫ ∫ d rs e −i k s rs J xn (rs ) −∞

⎡ a/2 ⎛nπ ⎛ ~ J xn (k s ) = ⎢ ∫ dx e −ik x x sin ⎜⎜ 1 ⎜ x + ⎝ a ⎝ ⎣− a / 2

⎛n π ⎛ b ⎞ ⎞⎤ a ⎞ ⎞⎤ ⎡ −ik y ⎟ ⎟⎟⎥.⎢ ∫ dy e y cos⎜⎜ 2 ⎜ y + ⎟ ⎟⎟⎥ (14a) 2 ⎠ ⎠ ⎦ ⎣ −b / 2 2 ⎠ ⎠⎦ ⎝ b ⎝ b/2

∞ ~ J ym (k s ) = ∫ ∫ d rs e −i k s rs J ym (rs ) −∞

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

b/2 ⎡ a/2 ⎛ m1π ⎛ ⎛m π ⎛ a ⎞ ⎞⎤ ⎡ b ⎞ ⎞⎤ ~ − ik y − ik x x J ym (k s ) = ⎢ ∫ dx e cos⎜⎜ ⎜ x + ⎟ ⎟⎟⎥.⎢ ∫ dy e y sin ⎜⎜ 2 ⎜ y + ⎟ ⎟⎟⎥ (14b) 2 ⎠ ⎠⎦ ⎣ −b / 2 2 ⎠ ⎠⎦ ⎝ a ⎝ ⎝ b ⎝ ⎣−a / 2

2. 1. 2. Sinusoid basis function with edge condition The second set of basis functions includes the edge condition for the patch currents. Computations can be carried out very efficiently if the edge condition is satisfied [12]. The basis functions are:

J xn (rs ) =

J ym (rs ) =

1 2 1 − (2 y / b )

⎛n π sin ⎜⎜ 1 ⎝ a

1 1 − (2 x / a )

2

⎛n π a ⎞⎞ ⎛ ⎜ x + ⎟ ⎟⎟ cos ⎜⎜ 2 2 ⎠⎠ ⎝ ⎝ b

b ⎞⎞ ⎛ ⎜ y + ⎟ ⎟⎟ 2 ⎠⎠ ⎝

⎛m π ⎛ b ⎞⎞ a ⎞⎞ ⎛ m π ⎛ cos ⎜⎜ 1 ⎜ x + ⎟ ⎟⎟ sin ⎜⎜ 2 ⎜ y + ⎟ ⎟⎟ 2 ⎠⎠ ⎝ b ⎝ 2 ⎠⎠ ⎝ a ⎝

(15a)

(15b)

Using equation (10), the Fourier transforms of (15.a) and (15.b) are expressed by:

170

Amel Broufrioua b/ 2 ⎡ a/ 2 ⎛ n1π ⎛ a ⎞⎞⎤ ⎡ ⎛ n π ⎛ b ⎞⎞ ~ −ik y −ikx x J xn (ks ) = ⎢ ∫ dx e sin⎜⎜ ⎜ x + ⎟⎟⎟⎥ ⎢ ∫ dy e y cos⎜⎜ 2 ⎜ y + ⎟⎟⎟ ⎝ a ⎝ 2 ⎠⎠⎦ ⎣−b / 2 ⎝ b ⎝ 2 ⎠⎠ ⎣−a / 2

2⎤ 1− (2y / b) ⎥ ⎦

(16a)

⎡ a/ 2 ⎛ m π ⎛ a ⎞⎞ ~ J ym(ks ) = ⎢ ∫ dx e−ikx x cos⎜⎜ 1 ⎜ x + ⎟⎟⎟ ⎝ a ⎝ 2 ⎠⎠ ⎣−a / 2

b/ 2 ⎛ m π ⎛ b ⎞⎞⎤ −ik y 2⎤⎡ 1− (2x / a) ⎥ ⎢ ∫ dy e y sin⎜⎜ 2 ⎜ y + ⎟⎟⎟⎥ ⎝ b ⎝ 2 ⎠⎠⎦ ⎦ ⎣−b / 2

(16b)

2. 1. 3. Chebyshev polynomials with edge condition Entire-domain basis functions are useful for analysing rectangular or circular patches, but become cumbersome for other shapes. Some work has been published concerning the use of Chebyshev basis functions for modelling the current on the patch antenna. The current on the patch is approximated in x and y direction by combination of Chebyshev polynomials of first and second kind Tn(x), Un(x) and an additional factor chosen to incorporate the edge condition [14]:

J xn (rs ) =

J ym (rs ) =

1 − (2 x a )

2

1 − (2 y b )

2

1 − (2 y b ) 1 − (2 x a )

2

2

U n1 (2 x a ) Tn 2 (2 y b )

(17a)

Tm1 (2 x a ) U m 2 (2 y b )

(17b)

The Fourier transforms of these basis functions (17.a) and (17.b) are given by:

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

~ −i k y 2 ⎡ a ⎤⎡ b J xn (k s ) = ⎢ ∫− a2 d x e −i k s x 1 − (2 x a ) U n1 (2 x a ) ⎥ ⎢ ∫−b2 d y e y Tn2 (2 y b ) ⎣ 2 ⎦⎣ 2

~ ⎡ a J ym (k s ) = ⎢ ∫− a2 d x e −i k s x Tm1 (2 x a ) ⎣ 2

2 ⎤ 1 − (2 y b ) ⎥ ⎦ (18a)

b −i k y 2 ⎤ ⎡ 2 ⎤ 1 − (2 x a ) ⎥ ⎢ ∫−b2 d y e y 1 − (2 y b ) U m2 (2 y b )⎥ ⎦⎣ 2 ⎦ (18b)

After some straigt forward algebraic manipulation, the results are:

π ~ n +n J xn (k s ) = ab(− i ) 1 2 4 2

⎤ ⎡ ⎢ ⎛ a ⎞⎥ ⎛ b ⎞ n +1 J n2 ⎜ k y ⎟ ⎢ 1 J n1 +1 ⎜ k x ⎟⎥ ⎝ 2 ⎠⎥ ⎝ 2 ⎠⎢ k a ⎥⎦ ⎢⎣ x 2

(19a)

Resistive Rectangular Patch Antenna with Uniaxial Substrate

⎤ ⎡ ⎢ m 1 + a b ~ ⎛ ⎞ ⎛ ⎞⎥ m +m J ym (k s ) = ab(− i ) 1 2 J m1 ⎜ k x ⎟ ⎢ 2 J m2 +1 ⎜ k y ⎟⎥ 4 ⎝ 2 ⎠⎢ k b ⎝ 2 ⎠⎥ ⎥⎦ ⎢⎣ y 2

π2

171

(19b)

2. 1. 4. Roof top sub-domain basis functions Roof top functions are very versatile in application and provide a clear and systematic structure of the computer program, whereas the entire-domain basis functions have the useful property that many combinations of modes on the same antenna element are uncoupled and yield a smaller number of unknowns in the solution [15]. A comparative study between these two basis functions will be developed in this chapter. In the next step, a set of roof top sub-domain basis functions are employed to model the current density distribution on the conductor. Roof top functions are characterized by their triangular shape along the direction of current flow and rectangular cross section in the orthogonal direction [16]. Mathematically, the sub-domain basis functions for the components of the current are described as: M N +1

J x (rs ) = ∑∑ I xmn Λ m ( x )  n ( y )

(20a)

m =1 n =1

J y (rs ) =

M +1 N

∑∑ I m =1 n =1

mn y

Λ n ( y )  m (x )

(20b)

Where, the functions Λ and  are "triangle" and "pulse" functions, respectively. The transforms of the current densities of (20.a) and (20.b) can be written as [14] M N +1 ~ ~ J x (k s ) = ∑∑ I xmn K xmn (k s )

(21a)

M +1 N ~ ~ J y (k s ) = ∑∑ I ymn K ymn (k s )

(21b)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

m =1 n =1

m =1 n =1

Where

⎛ Δx ⎞ ⎛ Δy ⎞ sin 2 ⎜ k x ⎟ ⎟ sin ⎜ k y 8 ~ mn 2 ⎠ ⎝ 2 ⎠ ⎝ exp(− i k x xm − i k y yn + i k y (Δy / 2)) K x (k s ) = 2 ky Δx kx (22a)

172

Amel Broufrioua

⎛ Δx ⎞ 2 ⎛ Δy ⎞ sin ⎜ k x ⎟ ⎟ sin ⎜ k y 8 ~ mn 2 ⎠ 2 ⎠ ⎝ ⎝ K y (k s ) = exp(− i k x xm − i k y yn + i k y (Δy / 2)) kx k2y Δy (22b)

(xm , yn ) coordinates current mode. Note that the rectangular patch is divided into (M+1) × (N+1) cells along the x and y directions, with each cell having the dimensions of Δ x and Δ y .

Δx = a (M + 1) and Δy = b ( N + 1)

2. 2. Asymptotic forms of the currents For high ks arguments the asymptotic Fourier transforms of the four previous basis functions are calculated.

2. 2. 1. Sinusoid basis function without edge condition The respective asymptotic forms of the basis functions (14.a), (14.b) for large ks are given by: a b −ik x ⎤ ⎡ ik y ⎡ ik x a2 2 2 n1π ⎢ e ~ n1 e n ⎥ ⋅ ⎢e ⎯ ⎯ ⎯ → − − ( ) − (− 1) 2 J xn (k s ) ⎯⎯ ⎯ i 1 k x , k y >> 2 2 ⎥ ⎢ ⎢ a k k kx ⎢⎣ x ⎥⎦ ⎢⎣ y

−ik y

e

b 2

ky

⎤ ⎥ ⎥ ⎥⎦

(23a)

⎡ i k x a2 m π ~ m J ym (k s ) ⎯⎯⎯ ⎯⎯⎯→ i 2 ⎢ e − (− 1) 1 k x , k y >> ⎢ b kx ⎣⎢

−i k x

e

kx

a 2

b −i k y ⎤ ⎤ ⎡ i k y b2 2 m ⎥ ⋅ ⎢e − (− 1) 2 e 2 ⎥ 2 ⎥ ⎢k ky ⎥ y ⎦⎥ ⎣⎢ ⎦⎥

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

(23b)

2. 2. 2. Sinusoid basis function with edge condition For large ks the asymptotic forms of (16.a) and (16.b) are expressed by: a b −ikx ⎤ ⎡ iky ⎡ ikx a2 2 2 n1π ( ) 1 i − ~ (− i)n2 J xn (k s ) ⎯⎯ π b ⎢e 2 − (−1)n1 e 2 ⎥ ⋅ ⎢e + i(−1)n2 ⎯ ⎯→ − kx , ky >> ⎢k 2a kx ⎥ ⎢ k y 2 x ⎣⎢ ⎦⎥ ⎣⎢

⎤ ⎥ ky ⎥ ⎦⎥

−iky

e

b 2

(24a)

Resistive Rectangular Patch Antenna with Uniaxial Substrate

173

a b b −ikx ⎤ ⎡ iky −iky ⎤ ⎡ ikx a2 2 2 m2π ~ m2 (1 − i ) m m ⎢e + i(−1) 1 e ⎥ ⋅ ⎢e + i(−1) 2 e 2 ⎥ ( ) J ym (k s ) ⎯k⎯ i a π ⎯ ⎯ → − 2 2 x , ky >> ⎢ k 2b ky ⎥ kx ⎥ ⎢ k y 2 ⎢⎣ x ⎥⎦ ⎢⎣ ⎥⎦

(24b)

2. 2. 3. Chebyshev polynomials with edge condition For large ks the asymptotic forms of (19.a) and (19.b) are expressed by:

π ~ J xn (k s ) ⎯k⎯ ⎯ ⎯→ x , ky >> 2

a b −ikx ⎤ ⎡ iky ⎡ ikx a2 2 2 b n n1 +n2 ⎢ e n1 e e ⎥ ⎢ (n1 +1)(−1) ⎢ 3 − i(−1) ⋅ + i(−1) 2 3 ⎥ ⎢ a k kx ⎥ ⎢ k y ⎣⎢ x ⎦ ⎣

⎤ ⎥ ky ⎥ ⎦⎥

−iky

e

b 2

(25a)

π ~ J ym (k s ) ⎯⎯ ⎯ ⎯→ kx , ky >> 2

a b −ikx ⎤ ⎡ iky ⎡ ikx a2 2 2 a m m1 e m1 +m2 ⎢ e e ⎥⋅⎢ (m2 +1)(−1) ⎢ + i(−1) − i(−1) 2 3 ⎢ ⎥ b kx kx k ⎦⎥ ⎣⎢ y ⎣⎢

⎤ ⎥ 3 ⎥ ky ⎥ ⎦

−iky

e

b 2

(25b)

2. 2. 4. Roof top sub-domain basis functions Using the Taylor development, for large ks the asymptotic forms of (21.a) and (21.b) are expressed by:

~ ⎛ 9 ⎞ ⎛ 81 ⎞ 2 ⎛ 3159 ⎞ 4 J x (k s ) ⎯⎯ k + ⎜ 31 ⎟ k x ⎯ ⎯→ ⎜ k x , k y >> 10 ⎟ ⎜ 20 ⎟ x ⎝ 10 ⎠ ⎝ 2.10 ⎠ ⎝ 2.10 ⎠

(26a)

~ ⎛ 9 ⎞ ⎛ 81 ⎞ 2 ⎛ 3159 ⎞ 4 J y (k s ) ⎯⎯ k + ⎜ 31 ⎟ k y ⎯ ⎯→ ⎜ k x , k y >> 10 ⎟ ⎜ 20 ⎟ y ⎝ 10 ⎠ ⎝ 2.10 ⎠ ⎝ 2.10 ⎠

(26b)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The tangential electric field due to the surface current J can be expressed as:

Es (rs ) =

1 4π

∞

∫ ∫ dk F (k

2

−∞

s

s

~ , rs ) G (k s ) J (k s )

(27)

Enforcement of the boundary condition to vanish on the patch yields: ∞

∫ ∫ dk F (k −∞

s

s

~ , rs ) G (k s ) J (k s ) = 0

(28)

Using Galerkin’s method the integral equation describing the field E in the rectangular patch can be discretized into the following matrix:

174

Amel Broufrioua

⎡(Z 1 )N × N ⎢ ⎣⎢(Z 3 )M × N

(Z ) (Z )

⎤ ⎡(a ) N ×1 ⎤ ⎥⋅⎢ ⎥=0 ⎥ ⎣(b )M ×1 ⎦ M ×M ⎦

2 N ×M 4

(29)

Where: ∞

(Z )

∫ ∫ dk

=

1 kn

[

−∞

∞

(Z )

=

2 km

kxky

∫ ∫ dk s

k

−∞

(Z )

3 ln

∞

=

∫ ∫ dk

k

∞

=

4 lm

2 s

kxky s

−∞

(Z )

∫ ∫ dk

]

1 2 e ~ ~ k x G + k y2 G h J xk (− k s ) J xn (k s ) . 2 ks

s

2 s

[G

[G

e

e

(29a)

]

~ ~ − G h J xk (− k s ) J ym (k s ) .

(29b)

]

~ ~ − G h J yl (− k s ) J xn (k s ) .

[

(29c)

]

1 2 e ~ ~ k y G + k x2 G h J yl (− k s ) J ym (k s ) . 2 ks

s

−∞

(29d)

The numerical integration of the matrix elements in equations (29) can be facilitated by conversion from the k x , k y coordinates to the polar coordinates k ρ , α i.e.

(

∞

∫ ∫ dk

−∞

s

=∫

∞

∫

∞

−∞ −∞

)

(

∞

2π

0

0

)

dk x dk y = ∫ dk ρ k ρ ∫ dα

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Since the resonant frequencies are defined to be the frequencies at which the field and the current can sustain themselves without a driving source. Therefore, for the existence of nontrivial solutions, the determinant of the [Z] matrix must be zero. This condition is satisfied by a complex frequency f = f r + if i that gives the resonant frequency f r and the other antenna characteristics.

2. 3. Effects of a resistive patch In this section the integral equation includes a resistive boundary condition on the surface of the patch is developed. In this case the boundary condition at the surface of the patch is given by [17, 18]:

Resistive Rectangular Patch Antenna with Uniaxial Substrate

E scat + E inc = R s ⋅ J

175 (30)

R s Surface resistance on microstrip patch antenna. E inc Tangential components of incident electric field. E scat Tangential components of scattered electric field. J Surface current on the patch. The right side of equation (30) represents the field dissipated on the patch. The surface resistance R is in general, a function of x and y and is equal to zero for a perfectly s

conducting patch. As previously, an integral equation can be formulated by using the Green’s function on a thick dielectric substrate to determine the electric field at any point. The solution of such an integral equation is finally obtained by the moment method with the given set of boundary conditions. From this analysis, the current distribution on the patch is determined [1], in this case the choice of entire domain defined in the field of the patch was illustrated to develop the unknown currents on the patch. The electric field integral equation which enforces the boundary condition must vanish on the patch surface, can then be discritized into a matrix form as

⎡(Z1 )N × N ⎢ ⎢⎣(Z3 )M × N

(Z ) (Z )

⎤ ⎡(a )N ×1 ⎤ ⎡(R1 )N × N 0 ⎤ ⎡(a )N ×1 ⎤ + ⎥⋅⎢ ⎥⋅⎢ ⎢ ⎥=0 ⎥ ( ) ( ) b b ( ) R 0 ⎥ ⎢ M × 1 M × 1 ⎦ ⎣ ⎦ ⎣ 4 M ×M ⎥ M ×M ⎦ ⎣ ⎦

2 N ×M 4

(31)

(Z ) The impedance matrix terms, i = 1, 2, 3, 4 given in equation (29).

(R ) Resistance matrix terms representing surface resistance on the patch j = 1, 4. i

j

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Once the impedance matrix and the resistance matrix have been calculated, the results form a system of simultaneous equations. The resulting system of equations is then solved for the unknown current modes on the patch. Some examples of the moment method technique have been developed here to study the resonant frequency and the scattering properties of a rectangular patch antenna with non zero surface resistance. The boundary condition for the electric field was used to derive an integral equation for the electric current; the necessary terms for representing the surface resistance on the patch were derived and were included in the equation in the form of a resistance matrix. It is important to note that the terms of the resistance matrix do not depend on frequency. Computer programs have been written to evaluate the elements of the impedance and resistance and then to solve the matrix equation. In Table 1 comparaison of the real and immaginary resonant frequency with measured and calculated data of Pozar in the case of isotropic and anisotropic substrate is presented and show a very good agreement.

176

Amel Broufrioua Table 1. Comparaison of the calculated resonant frequency with measured and calculated data

h (cm)

0.127 0.127 0.254

a (cm)

3.00 1.50 3.00

b (cm)

Freq (Ghz) Measured [1]

2.00 0.95 1.90

2.264 4.495 2.242

ε x = ε z =10.2 Calculated [1] 2.285 4.580 2.290

Our results 2.264 4.613 2.306

ε x =13.0, ε z =10.2 Calculated [1] 2.268 4.520 2.260

Our results 2.254 4.558 2.279

In Figures 2 and 3 the real and imaginary frequency shifts versus the substrate thickness are studied. It is found that the real frequency is shift to higher frequencies for the positive uniaxial case, in the other hand, is shifted to lower frequencies for the negative uniaxial case. As for the imaginary frequency, it is increased due to the positive uniaxial anisotropy and decreased due to the negative uniaxial anisotropy. 9

9.8

x 10

9.6 9.4 9.2 9 8.8 8.6 8.4

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

8.2

0

0.02 0.04 0.06 0.08 0.1

0.12 0.14 0.16 0.18 0.2

h(cm) Figure 2. Real part of frequency shifts versus the substrate thickness for the isotropic (

ε x = ε z =2.35), positive uniaxial (

ε x =1.88, ε z =2.35) and negative uniaxial

(

ε x =2.82, ε z =2.35) substrates, a=1.5cm, b=1.0cm, Rs=0 Ω

Resistive Rectangular Patch Antenna with Uniaxial Substrate

177

8

5

x 10

4.5 4 3.5 3 2.5 2 1.5 1 0.5 0

0

0.02 0.04 0.06 0.08 0.1

0.12 0.14 0.16 0.18 0.2

h(cm)

Figure 3. Imaginary part of frequency shifts versus the substrate thickness for the isotropic ( (

ε x = ε z =2.35), positive uniaxial ( ε x =1.88, ε z =2.35) and negative uniaxial ε x =2.82, ε z =2.35) substrates, a=1.5cm, b=1.0cm, Rs=0 Ω

Figure 4 shows the results for the half power band width of the patch antenna. The positive uniaxial anisotropy slightly increases the bandwidth, while the negative uniaxial anisotropy slightly decreases the bandwidth, These behaviors agree very well with those reported by King-Lu Wong et al [7] for these three figures. The influence of uniaxial anisotropy in the substrate on the resonant frequency of a rectangular microstrip patch antenna for different pairs of relative permittivity ( ε x , ε ) is z

shown in Table 2. The obtained results show that when the permittivity ε is shanged and Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

z

εx

remains constant, the real and imaginary part of resonant frequency change drastically, on the other hand, we found a slight shift in the real and imaginary part of the resonant when the permittivity ε x is shanged and ε remains constant. These results are given in detail in our z

paper [18]. The variations of the radar cross section and the radiation of a rectangular patch antenna due to the uniaxial anisotropy shown by Figure 5 and Figure 6, can be seen to be the same as discussed previously for the case of real and imaginary part of the resonant frequency in Table 2.

178

Amel Broufrioua BW% 12 10 8 6 4 2 0

0

0.02 0.04 0.06 0.08 0.1

0.12 0.14 0.16 0.18 0.2

ε x = ε z =2.35),

Figure 4. Bandwidth shifts versus the substrate thickness for the isotropic (

ε

ε x =2.82, ε z =2.35)

ε z =2.35) and negative uniaxial (

x =1.88, positive uniaxial ( substrates, a=1.5cm, b=1.0cm, Rs=0 Ω

h(cm) -27 -28 -29 -30 -31

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

-32 -33

0

10

20

30

θ

Figure 5. Radar cross section versus angle

40

ε

z

ε

z

= 2.32,

=2.32

εx

=2.32,

ε

z

60

70

80

90

for the isotropic, negative and positive uniaxial

substrates, a=1.9cm, b=2.29cm, h=0.159cm, Rs(Ω)=0, 1.16,

50

= 4.64

εx ε z

φ =0°

=

εx

= 2.32,

ε

z

εx

= 2.32,

= 1.16,

εx

=

= 4.64,

Resistive Rectangular Patch Antenna with Uniaxial Substrate 90

2.5

120

2

60

1.5

150

179

30

1 0.5

180

0

330

210

300

240 270 Figure 6. Radiation pattern versus angle

θ

for the isotropic, negative and positive uniaxial substrates,

a=1.9cm, b=2.29cm, h=0.159cm, Rs(Ω)=0,

ε ε

z z

= 2.32,

ε x =2.32, ε z = 4.64

φ =0°

ε x = ε z = 2.32, ε x = 2.32, ε z = 1.16,

ε x = 1.16, ε x = 4.64,

=2.32

Table 2. Frequency and radar cross section for different pairs of relative permittivities ( ε x , ε ), a =1.5cm, b=1.0cm, h=0.2cm, Rs(Ω)=0, θ =60°, φ =0° z

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

εx

εz

Freq (Ghz)

RCS (dBsm)

5.0 5.0 5.953+i 0.178 -30.29 5.0 3.6 6.753+i 0.272 -30.00 5.0 6.4 5.356+i 0.121 -30.55 3.6 5.0 6.078+i 0.184 -30.40 6.4 5.0 5.842+i 0.172 -30.45 Table 3 shows the scattering radar cross section RCS for an imperfectly conducting patch with the surface resistance Rs=30 Ω compared to a perfectly one and printed on a uniaxial anisotropic substrate of thickness h=0.2 cm, where isotropic, positive and negative uniaxial anisotropic substrates are considered, the patch dimensions a= 1.5cm, b=1.0 cm. It can be seen clearly that the permittivity ε have a stronger effect on the scattening radar cross z

section than the permittivity ε x for both cases. Also we observe that when the surface resistance is increased, the level of the radar cross section decreases. Consequently the addition of a resistance on the surface of a microstrip patch antenna has been shown to decrease the scattered energy from the antenna.

180

Amel Broufrioua

Table 3. Effects of the surface resistance on the radar cross section for isotropic, negative and positive uniaxial substrates, a=1.5cm, b=1.0cm,h =0.2cm, θ =60°, φ =0°

εx

εz

2.32 4.64 2.32 1.16 2.32

2.32 2.32 1.16 2.32 4.64

RCS (dBsm) Rs(Ω)=30 -29.27 -29.39 -29.12 -29.25 -29.64

Rs (Ω)=0 -28.57 -28.82 -28.05 -28.68 -29.50

In Table 4, we compare our results obtained from the entire domain and roof top basis functions with those of Nelson et al [19] for anisotropic substrate. Comparisons agree very well with those of Nelson et al, for the two sets of currents with slight shifts in frequency between subdomain and entire domain data are noted. Computations show that the roof top subdomain basis functions provides a significant improvement in the computations time with less iterations in the evaluation of the resonant frequency of a microstrip patch compared to the entire domain sinusoid basis functions. It is estimated that the computation time for the case of roof top basis functions is about 50% of that for the case of entire domain. Table 4. Real and imaginary part of resonant frequency for two sets of basis functions ε x =9.4, ε =2.35, h= 0.158cm, Rs=0 Ω z

Structure ( a × b ) cm

[19]

1.0 × 1.5 1.0 × 0.2

7.773+i 0.233 8.112+i 0.112

Freq (Ghz) Our results Entire domain Roof top 7.722+i 0.316 7.604+i 0.194 8.125+i 0.121 7.909+i 0.088

Figure 7 and 8 show the scattering properties for the Eθ component of the electric field at φ = 0 ° plane and the scattering properties for the E at φ = 90 plane displayed as a Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

φ

function of the angle θ and as a function of surface resistance at the frequency 5.95 Ghz for

a rectangular patch with dimensions a × b = (1.5cm × 1cm ) printed on an isotropic substrate with height h= 0.2cm which has a permittivity of

ε r = 5 , it is clear that when the surface

resistance on the patch is increased, the level of the components Eθ decreases consequently. However it is important to note that our results for the Eφ component do not change with the surface resistance at φ = 0 ° . Also we have a slight shift in the variation of the E

φ

component at φ = 90 We conclude that the addition of a resistance on the surface of a

Resistive Rectangular Patch Antenna with Uniaxial Substrate

181

microstrip patch has been shown to decrease the Eθ component at φ = 0 ° and slightly increases the E component at φ = 90 . φ

90 6 120

90 60

1

120

60 0.8

4

0.6

150

150

30

2

30

0.4 0.2

0

180

180

330

210

240

0

300

240

270

Figure 7.

θ

Eθ component at φ = 0 versus angle

for various values of surface resistance

330

210

300 270

Figure 8.

θ

E φ component at φ = 90 versus angle

for various values of surface resistance.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The numerical results of the resonant frequency using different asymptotic basis functions are given in detail in our paper [20], when we have compared the use of asymptotic sinusoid basis functions with and without edge condition with asymptotic Chebyshev polynomials. Computations show that the asymptotic forms provide results with minimal computation effort. Asymptotic Chebyshev polynomials provide a significant improvement in the computation time in the evaluation of the resonant frequency of a microstrip patch compared to the asymptotic sinusoid basis function with and without edge condition. However the asymptotic sinusoid without edge condition presents fast convergence compared to the sinusoid with edge condition. The results obtained by these forms of currents provide a significant convergence with less computationally complex [20] compared to the exact forms of currents.

2. 4. Rectangular patch antenna with excitation In this section a microstrip patch antenna excited by an electromagnetic proximity or by a microstrip transmission line as shown in Figure 9. For the case of the proximity coupled, the microstrip feed line is on a substrate of thickness h covered with a superstrate of thickness d, The feed line is inset a distance s, it is this overlap of expansion modes that provides the continuity of current flow from the feed line to the patch, however for the microstrip transmission line feeding method consists of a rectangular patch and a uniform microstrip feed line. Both the patch and the feed line are located on a dielectric substrate which has a uniform thickness of h and a relative permittivity ε r .

182

Amel Broufrioua

z

y Radiating conductor

Feed line

wf s

b

0

x

a d

ε r2

h

ε r1

Ground plane

Figure 9. Geometry of a rectangular patch with excitation

The integral equation for the unknown currents on the antenna and the feed is solved by applying the Galerkin method of moment in the Fourier transform domain. The currents on the feed line and the patch are expanded in terms of three types of modes [9]: a. Traveling wave currents on the feed line: These currents are used to describe the current on the feed line and are extended to infinity [9]. It can be given by this equation:

J f x ( x, y ) = g f ( y ) [(1 − R ) f c ( x ) − j (1 + R ) f s ( x )]

(32)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Where R is the reflection coefficient along the line to be determined g f ( y ) is the transverse distribution of the current which may include the edge effect or is assumed as constant for a narrow feedline [9, 21]: Constant transverse distribution of the current is given by:

⎧ 1 ⎪ g f (y) = ⎨ w f ⎪⎩ 0

for

y ≤

wf

2 Otherwise

The Fourier transform for this equation is:

(33)

Resistive Rectangular Patch Antenna with Uniaxial Substrate

g~ f (k y ) =

∞

∫ dy e

−i k y y

183

g f (y)

−∞

−w f 2

∫ dy e

=

−i k y y

−w f 2

1 wf

(34)

⎛ wf ⎞ ⎟⎟ = sin c⎜⎜ k y 2 ⎠ ⎝ The transverse distribution of the current including the edge condition is given by:

2 ⎧ ⎪ 2 ⎛2 y⎞ ⎪ g f ( y ) = ⎨π w f 1 − ⎜ ⎟ ⎜w ⎟ ⎪ f ⎝ ⎠ ⎪ 0 ⎩

for

y ≤

wf 2

(35)

Otherwise

The Fourier transform of this expression is:

g~ f (k y ) =

∞

∫ dy e

−i k y y

g f (y)

−∞

−w f 2

∫ dy e

=

−w f 2

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

=

1

π

−w f 2

2

−i k y y

∫ dy e

πw f

⎛ 2y ⎞ ⎟ 1− ⎜ ⎜w ⎟ f ⎠ ⎝

(36)

2

−i k y y

⎛ wf ⎜⎜ ⎝ 2

−w f 2

2

2

⎞ ⎟⎟ − y 2 ⎠

⎛ wf ⎞ ⎟ = J 0 ⎜⎜ k y 2 ⎟⎠ ⎝ wf width of the feed line. Expressions f c ( x ) , f s ( x ) in the equation (32) are:

f c ( x ) = cos k e x, f s ( x ) = sin k e x,

for − ∞ < x < − π 2 k e for − ∞ < x < 0

(37) (38)

184

Amel Broufrioua

ke =

2π

λg

is the propagation constant of the microstrip line

Their Fourier transforms are:

~ f c (k x ) =

∞

∫ dx e

−i k x x

f c (x )

−∞

=

(39)

−π 2 k e

∫ dx e

−i k x x

cos k e x

−∞

⎡ ⎛ iπk x 2k e exp⎜⎜ ⎢ ~ π ⎝ 2k e f c (k x ) = ⎢δ (k x − k e ) + δ (k x + k e ) + 2⎢ π k x2 − k e2 ⎢ ⎢⎣

⎞⎤ ⎟⎟ ⎥ ⎠⎥ ⎥ ⎥ ⎥⎦

(40)

⎤ 2k e ~ π⎡ f s (k x ) = −i ⎢δ (k e − k x ) − δ (k e + k x ) − 2 2 ⎥ 2⎣ π i (k x − k e )⎦

(41)

(

~ f s (k x ) =

∞

∫ dx e

−i k x x

)

f x (x )

−∞

0

==

∫ dx e

−i k x x

sin k e x

−∞

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

b. Overlap currents: Which is used to expand the current on the patch and a portion of the feed line, the current will be non-uniform because of the discontinuity, piecewise sinusoidal PWS modes were used to model this non-uniformity [9].

J

f n

Nf

(x, y ) = g f ( y ) ∑ I nf n =1

f nf (x )

(42)

Where:

f nf (x ) =

(

sin k e h f − x + nh f

)

sin k e h f

With Lf is the finite lenght of the feed line

for

x + nh f < h f

(43)

Resistive Rectangular Patch Antenna with Uniaxial Substrate

185

Nf : l'ensemble des modes PWS sur la ligne d'alimentation.

I nf : are coefficients of subsectional basis functions. hf : is the length of the sinusoidal rooftop basis function. Equation f n ( x ) can be written as: f

⎧ sin k e (x + nh f + h f ) ⎪ sin k e h f ⎪ ⎪ sin k e (x + nh f − h f f n f ( x ) = ⎨− sin k e h f ⎪ ⎪0 ⎪ ⎩

for

)

− h f < x + nh f < 0

for 0 < x + nh f < h f

(44)

Otherwise

The Fourier transform of the previous equation is:

⎧ ⎪ ⎪⎪ ~f f n (k x ) = ⎨ ⎪ ⎪ ⎪⎩

i n ke h f

if

k x = ke

−i n k e h f

if

k x = −k e

if

k x ≠ ke

hf e i n kx hf

− 2e h f sin c(k e h f

)

hf e cos(k x h f ) − cos(k e h f k x2 − k e2

)

(45)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

c. Patch currents: The sinusoid basic function without edge condition was used in this case to model the patch current. After the different calculation of the transforms of these three types of currents using the Fourier transform domain with the Galerkins method of moments, the integral equation for the unknown currents on the patch and the feed line is solved by the electric field integral equation which enforces the boundary condition. The integral equation for the unknown currents on the patch is given for z=h+d and the integral equation for the unknown currents on the feed line is given for z=h, the results can be discitized into a matrix form as:

⎡ Z 111 ⎢ 11 ⎢Z3 ⎢ Z 121 ⎢ 21 ⎢⎣ Z 3

Z 211

Z 11 2

Z 411 Z 221

Z 31 2 Z 12 2

Z 421

Z 32 2

Z 21 2 ⎤ ⎥ Z 41 2 ⎥ Z 22 2 ⎥ ⎥ Z 42 2 ⎥⎦

⎡ R ⎤ ⎡V11 ⎤ ⎢I f ⎥ ⎢ 1 ⎥ ⎢ n ⎥ = ⎢V2 ⎥ ⎢ a n ⎥ ⎢V12 ⎥ ⎢ ⎥ ⎢ 2⎥ ⎣bm ⎦ ⎢⎣V2 ⎥⎦

The impedance matrix elements are given by:

(46)

186

Amel Broufrioua

Z

11 1

2

⎡k ⎤ ~ = ∫ ∫ BB dk s ⎢ x ⎥ J xj (k s ) G11e −∞ ⎣ ks ⎦ ∞

Nf

Z = ∑∫ 11 2

2

∞

−∞

n=1

(47)

∫k

s

hf

2

⎡k ⎤ ~ ⎛k −k ⎞ ⎛ k + ks ⎞ ink h dks ⎢ x ⎥ J xj (k s ) G11e e x f sin c⎜ x s h f ⎟ sin c⎜ x hf ⎟ k 2 2 ⎝ ⎠ ⎝ ⎠ s ⎣ ⎦ (48)

Z

11 3

2

⎡ky ⎤ ~ = ∫ ∫ BBdk s ⎢ ⎥ J xj (k s ) G11h −∞ ⎣ ks ⎦ ∞

Nf

Z = ∑∫ 11 4

2

∞

−∞

n=1

(49)

∫k

s

hf

2

⎡ky ⎤ ~ ⎛k −k ⎞ ⎛k +k ⎞ ink h dks ⎢ ⎥ J xj (k s ) G11h e x f sin c⎜ x s h f ⎟ sin c⎜ x s h f ⎟ ⎝ 2 ⎠ ⎠ ⎝ 2 ⎣ ks ⎦ (50)

Z

22 1

⎡k = ∑ ∫ ∫ dk s ⎢ x −∞ n =1 ⎢⎣ k s N

M

Z 222 = −∑ ∫

−∞

m =1

Z

22 3

m =1

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

∫ dk

∞

M

Z

(51)

s

⎡kxky ⎤ ~ e ~ ⎢ 2 ⎥ J xj (k s ) G22 J ym (k s ) ⎢⎣ k s ⎥⎦

(52)

⎡ ky = ∑ ∫ ∫ dk s ⎢ −∞ n =1 ⎢⎣ k s N

2

⎤ ~ h ~ ⎥ J xj (k s ) G22 J xn (k s ) ⎥⎦

⎡kxky ⎤ ~ ~ h dk ⎢ 2 ⎥ J xj (k s ) J ym (k s ) G22 s −∞ ∫ ⎢⎣ k s ⎥⎦

Z 422 = −∑ ∫

12 1

∞

2

⎤ ~ e ~ ⎥ J xj (k s ) G22 J xn (k s ) ⎥⎦

∞

∞

M

∞

Z 212 = ∑ ∫ m =1

∞

−∞

(54)

2

⎡k ⎤ ~ ~ = ∑ ∫ ∫ dk s ⎢ x ⎥ J xj (k s ) G12e J xn (k s ) −∞ n =1 ⎣ ks ⎦ N

(53)

∫ dk

s

⎡kxk y ⎤ ~ e ~ ⎢ 2 ⎥ J xj (k s ) G12 J ym (k s ) ⎢⎣ k s ⎥⎦

(55)

(56)

Resistive Rectangular Patch Antenna with Uniaxial Substrate

Z

12 3

2

⎡ ky ⎤ ~ ~ = ∑ ∫ ∫ dk s ⎢ ⎥ J xj (k s ) G12h J xn (k s ) −∞ n =1 ⎣⎢ k s ⎦⎥ N

∞

M

Z 412 = −∑ ∫

Z

Z

21 2

∞

−∞

m =1

12 1

187

∫ dk

s

⎡ kx k y ⎤ ~∗ ~ h ⎢ 2 ⎥ J xj (k s ) J ym (k s )G12 ⎢⎣ k s ⎥⎦

Nf

∞

= ∑∫ n =1

∞

−∞

(58)

2

⎡k ⎤ ~ ~ = ∑ ∫ ∫ dk s ⎢ x ⎥ J xj (k s ) G12e J xn (k s ) −∞ n =1 ⎣ ks ⎦ N

(57)

(59)

2

∫k

s

⎡k ⎤ ~ ⎛ k + ks ⎞ ⎛ k − ks ⎞ ink h e e x f sin c⎜ x h f ⎟ sin c⎜ x hf ⎟ dks ⎢ x ⎥ J xj (k s ) G21 ⎠ ⎠ ⎝ 2 ⎝ 2 ⎣ ks ⎦

2

hf

(60)

Z

21 3

2

⎡ky ⎤ ~ e = ∫ ∫ BBdk s ⎢ ⎥ J xj (k s ) G21 −∞ k ⎣ s⎦ ∞

Nf

∞

Z = ∑∫ 21 4

n=1

−∞

(61)

2

∫k

s

hf

2

⎡ky ⎤ ~ ⎞ ⎛k −k ⎞ ⎛k +k ink h h dks ⎢ ⎥ J xj (k s ) G21 e x f sin c⎜ x s h f ⎟ sin c⎜ x s h f ⎟ ⎠ ⎝ 2 ⎠ ⎝ 2 ⎣ ks ⎦ (62)

The voltage vector elements are given by: ∞

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

2

⎛k V = − ∫ dk s ⎜⎜ x ⎝ ks −∞

⎞ ~ ⎟⎟ J xj (k s )G11e AA ⎠

∞

⎛ ky V = − ∫ dk s ⎜⎜ ⎝ ks −∞

⎞ ~ ⎟⎟ J xj (k s )G11h AA ⎠

∞

⎛k V = − ∫ dk s ⎜⎜ x ⎝ ks −∞

⎞ ~ e ⎟⎟ J xj (k s )G21 AA ⎠

∞

⎞ ~ ⎟⎟ J xj (k s )G21h AA ⎠

1 1

1 2

2 1

⎛ ky V = − ∫ dk s ⎜⎜ ⎝ ks −∞ 2 2

(63)

2

(64)

2

(65)

2

(66)

188

Amel Broufrioua

Where:

⎡ π π π ⎤ 2nπ − i sin(kx + ks ) cos(kx − ks ) sinc(kx + ks ) ⎥ ⎢4nsinc(kx − ks ) 4ks 4ks 4ks 4ks ⎦ ⎣ π π −i ( 2n+1) ⎛ π ⎞ ⎛ π ⎞ −i(2n+1) 4 4 ⎜ ⎟ ⎜ ⎟ ) + cos⎜ (kx − ks )2n ⎟ sinc⎜ (kx − ks ) ⎟ e e 4ks ⎠ 4ks ⎠ ⎝ ⎝

N ⎛ AA= −∑⎜⎜ n=1 ⎝

N ⎛ nπ π π π BB = ∑ ⎜⎜ 4n sin c(k x + k s ) cos(k x + k s ) sin c(k x + k s ) + cos(k x + k s ) + 4k s 4k s 2k s 4k s n =1 ⎝

⎛ π i sin⎜⎜ (k x − k s )2n 4 ks ⎝

⎞ ⎛ π ⎟⎟ sin c⎜⎜ (k x − k s ) 4 ks ⎠ ⎝

⎞ −i (2 n +1)π4 ⎟⎟ e ⎠

⎞ ⎟⎟ ⎠

G11e , G11h Represent the Green’s functions correspond to the first layer which contain the feed line i.e. for z=h. e G22 , G22h Represent the Green’s functions correspond to the second layer which contain

the patch i.e. for z=h+d. e G12e , G12h , G21 , G21h Represent a two dimensional dyadic Green’s functions [10], which

describes the coupling between the two layers, the first containing the feed line and the second containing the patch. Note that the effect of the superstrate on k e must be included, in the case of the proximity coupled feeding. Moreover, if the substrate and the superstrate have different permittivities, it is straightforward matter to modify the Green’s functions [10]. Also it is worth noting that for the case of the microstrip transmission line feeding, if the substrate is anisotropic and the patch is resistive the Green’s functions is taken to be as given in the previous sections.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

3. CONCLUSION A computationally efficient method including a resistive boundary condition on the surface of the patch and the effect of anisotropic substrate has been presented in this chapter. The boundary condition for the electric field is used to derive an integral equation for the electric current. The necessary terms for representing the surface resistance on the patch are derived and are included in the equation in the form of a resistance matrix. The formulation is carried out in the spectral domain. Entire domain sinusoid basis functions with and without edge condition, Chebychev polynomial with edge condition and roof top sub-domain basis functions were introduced to expand the unknown current on the metal patches. In this chapter, the numerical convergence is investigated for the case of the entire sinusoidal basis functions without considering the edge singularity condition. The effects of uniaxial anisotropy in the substrate on the resonant frequency, the radiation and the radar cross section of the microstrip patch antenna are also investigated.

Resistive Rectangular Patch Antenna with Uniaxial Substrate

189

The numerical results show that the use of uniaxial substrates significantly affects the characterization of the microstrip antennas. Our obtained results indicate that the resonant frequencies are increased due to the positive uniaxial anisotropy, and on the other hand, they are decreased due to the negative uniaxial anisotropy. The ε permittivity has a stronger z

effect on the radar cross section, the radiation and the resonant frequency than the permittivity ε . Also, it is worth noting that our calculated frequencies do not depend on the x

surface resistance; the results show also that the addition of a resistance on the surface of a microstrip patch antenna decreases the antenna dispersed energy as well as the Eθ component of the electric field at φ = 0 ° plane and slightly increase the E component at φ

φ = 90 . Also, computation results show that the roof top sub-domain basis functions provide a significant improvement in the computation time with fewer iterations in the evaluation of the resonant frequency of a microstrip patch compared to the entire domain sinusoid basis functions. It is worth noting that the main advantage of the roof top basis functions is the ease with which the spectral Green’s function is obtained with a minimum analytical effort. Moreover, the currents on the feed line are developed and the impedance matrix is calculated. Comparative study between our results and those available in the literature shows a very good agreement.

REFERENCES

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

[1]

Pozar, D. M. (1987). Radiation and scattering from a microstrip patch on a uniaxial substrate: IEEE Trans. Antennas and propagation, 35, 613-621 [2] Lafond, O., Himdi, M., & Daniel, J. P. (1999). Aperture coupled microstrip patch antenna with thick ground plane in millimetre waves. Electron. Lett, 35, 1394-1395. [3] Damiano, J. P., & Papiernik A. (1994). Survey of analytical and numerical models for probe-fed microstrip antennas: IEE Proc.-Microw. Antennas Propag, 141, 15-22 [4] Mirshekar-Syahkal, (1990). Spectral domain method for microwave integrated circuits: Wiley, New York [5] Booton, Jr. R. C. (1994). Computational methods for electromagnetics and microwaves: Wiley-Interscience Publication, New York, Wiley. [6] Tulintsef A. N., Ali S. M., & Kong J. A. (1991). Input impedance of a probe-fed stacked circular microstrip antenna: IEEE Transactions on antennas and propagation, 39, 381-390. [7] Wong. K-L.,& Row. J-S. (1993). Resonance of a rectangular microstrip patch on a uniaxial substrate: IEEE Trans on Microwave Theory and Tech, 41, 698-701. [8] Mittra, R., Hall, R. C., & Tsao, C. H. (1984). Spectral domain analysis of circular patch frequency selective surfaces: IEEE Trans. Antennas and propagation. 32, 533-536 [9] Pozar, D. M. & Voda, S. M. (1987). A rigorous analysis of a microstripline fed patch antenna. IEEE Trans. Antennas Propagat, 35, 1343-1350. [10] Bouttout, F., Benabdelaziz, F., Fortaki, T., & Khedrouche, D. (2000). Resonant frequency and bandwidth of a superstrate-loaded rectangular patch on a uniaxial

190

Amel Broufrioua

anisotropic substrate: Communications in numerical methods in engineering, 16, 459473. [11] Chew, W. C., & Liu, Q. (1988). Resonance frequency of a rectangular microstrip patch: IEEE Trans. Antennas Propagate. 36, 1045-1056

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

[12] Row, J. S., & Wong K. L. (1993). Resonance in a superstrate-loaded rectangular microstrip structure: IEEE Transactions on antennas and propagation, 41, 1349-1355. [13] Newman, E. H., & Forrai, D. (1987). Scattering from a microstrip patch: IEEE Trans. Antennas Propagat. 35, 245-251. [14] Abramowitz, M., & Stegun I. A. (1965). Handbook of mathematical functions: Dover, New York. [15] Pozar, D. M. (1982). Input impedance and mutual coupling of rectangular microstrip antennas: IEEE Trans. Antennas Propagat, 30, 1191-1196. [16] Park, S. O., Balanis, C. A., & Birtcher, C. R. (1998). Analytical evaluation of the asymptotic impedance matrix of a grounded dielectric slab with roof top functions: IEEE Trans. Antennas Propagat. 46, 251-259. [17] Shively. D., (1994). Scattering from Perfectly Conducting and Resistive Strips on a Grounded Dielectric Slab: IEEE Trans. Antennas Propagation, 42, 552-556. [18] Boufrioua, A., & Benghalia, A. (2006). Effects of the resistive patch and the uniaxial anisotropic substrate on the resonant frequency and the scattering radar cross section of a rectangular microstrip antenna: AST, Aerospace Science and Technology. 10, 217221. [19] Nelson, R. M., Rogers, D. A., & D’assunçao, A. G. 1990. Resonant frequency of a rectangular microstrip patch on several uniaxial substrates: IEEE Trans. Antennas and propagation, 38, 973-981. [20] Boufrioua, A., Benghalia, A., & Bouttout, F. (2008). Resonant frequency of a rectangular patch antenna using asymptotic basis functions: COMPEL, The International Journal for Computation and Mathematics in Electrical and Electronic Engineering, 27, 638-650. [21] Yang, X. H., & Shafai, L. 1995. Characteristics of aperture coupled microstrip antennas with various radiating patches and coupling apertures: IEEE Trans. Antennas Propagat. 43, 72-78.

In: Antennas: Parameters, Models and Applications Editor: Albert I. Ferrero

ISBN:978-1-60692-463 ©2009 Nova Science Publishers, Inc.

Chapter 7

ANTENNAL SENSILLAR MORPHOLOGY, STRUCTURE AND FUNCTION IN PARASITIC WASPS Yan Gao a, Li-Zhi Luo b,* and Abner Hammond c a

Jining Vocational Technological College, Shangdong, China State Key Laboratory for Biology of Plant Diseases and Insect Pests, Institute of Plant Protection, Chinese Academy of Agricultural Sciences, Beijing 100094, China c Department of Entomology, Louisiana Agricultural Experimental Station, Baton Rouge, LA 70803, USA ∗ b

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ABSTRACT Antennae of insects play an important role in their adult life and vary greatly in their morphology, structure, and function. This is particularly true in the parasitic wasps, greater understanding is still lacking in this area. Understanding antennal sensillar type and structures will, therefore, add to our knowledge of the variations in parasitic wasp behavior. In this paper, the morphology, structure, and function of antennae in parasitic wasps were reviewed and summarized based on our research results and published literature. Antennal length, diameter and shape between male and female in the major species/groups of parasitic wasps were described and compared. Function and structure and distribution of antennal sensilla elucidated by scanning and transmission microscopy had been presented separately as trichoid sensilla (SW), chaetica sensilla (TP), basiconic sensilla (SW), coeloconic sensilla (DW), and placoid sensilla (SW). Variations in type, number and size of antennal sensilla in the different species as well as sex’s differences, were also discussed in light of behavior. The significance of antennal patterns and ontogeny of sensilla as phylogenetic tool is discussed.

Keywords: antennal sensilla, parasitic wasps, morphological, structure, function, ontogeny

∗

Corresponding author. Tel.: +86 10 62815620; fax: +86 10 62895365. E-mail address: [email protected] (L.-Z. Luo).

192

Yan Gao, Li-Zhi Luo and Abner Hammond

INTRODUCTION Insect antennae are mobile, segmented, paired head-appendages and are found in nearly all insect groups (Gullan and Cranston, 1994; Schneider, 1964; Chiappini and Mazzoni, 2000). Antennae are often considered to be the nostrils of insects, and also referred to as their feelers, and possess different kinds of sensilla (Zacharuk, 1985). Numerous sensory organs, or sensilla (singular: sensillum) occur on antennae in the form of hairs, pegs, pits or cones. Functions of antennal sensilla include chemoreceptivity (gustatory and olfactory), mechanoreceptivity, thermoreceptivity, hygroreceptivity and CO2 receptivity (Quicke, 1997; Keil, 1999). Many parasitic Hymenoptera commonly find their hosts using chemical stimuli produced by the host or by the plant. Antennal sensory receptors of female parasitic Hymenoptera are involved in habitat searching, host location, host examination, host detection, host acceptance, oviposition and host discrimination (Weseloh 1972; Dahms 1984; Vinson et al. 1986; Bin et al. 1989; Isidoro et al. 2001). On the other hand, the antennal sensory receptors of male parasitic Hymenoptera are involved in female recognition (Bin et al. 1999; Battaglia et al. 2002). In this paper, we will first present the nomenclature of the different types of antennal sensilla, then we will illustrate the morphological variations and distribution and function of sensilla in different species of hymenopteran parasitoids, and ontogeny of insect sensilla, and finally, we will examine the antennal differences between the sexes and species in relation to their functions in host finding and evaluation.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

1. Nomenclature of Sensilla Antenna1 sensilla from several species of parasitic Hymenoptera, adults of Braconidae and Chalcidoidea, have been described (Steinberg et al., 1993; Alborn et al., 1995; Parè and Tumlinson, 1999; Ochieng et al., 2000; Fukushima et al., 2002; Hoballah et al., 2002; Voegelé et al., 1975; Olson and Andow, 1993; Isidoro et al., 1996; Amornsak et al. 1998; van Baaren et al., 1996, 1999; Bleeker et al., 2004)(fig.1). Different names have been assigned to sensilla types despite similarity in form and distribution. For example, basiconic sensilla found on Braconidae have been called fluted basiconic sensilla (Norton and Vinson, 1974a), S. basiconica A (Navasero and Elzen, 1991) and Sensilla trichodea TP (Bleeker et al., 2004). Schenk (1903) named these sensilla after the structure of their cuticular parts. The difficulty has been in achieving an unambiguous nomenclature. To circumvent this problem, Altner (1977) proposed a system in which the presence and structure of pores was used as the prime characteristic. The sensilla were divided into three groups: poreless (NP), wall pore sensilla with two subgroups [single-walled (SW) and double-walled (DW)], and terminal pore (TP) sensilla. Nevertheless, the traditional nomenclature generally follows that of Snodgrass (1935), Schneider (1964), Zacharuk (1985) and Keil (1999). Ultrastructural nomenclature of the sensilla has been based on that of Norton and Vinson (1974a), Navasero and Elzen (1991), Amornsak et al. (1998) and Ochieng et al. (2000). Sensilla have been divided into five groups: trichoid sensilla (SW), chaetica sensilla (TP), basiconic sensilla (SW), coeloconic sensilla (DW), and placoid sensilla (SW).

Antennal Sensillar Morphology, Structure And Function In Parasitic Wasps

193

Figure1. A–G Diagrams of the seven different types of sensilla found on the antennae of female A. victus and A. listronoti. (A)Trichoid sensillum. (B) Sensillum chaetica type 1. (C) Sensillum chaetica type 2. (D) Sensillum chaetica type 3. (E) Sensillum chaetica type 4. (F) Basiconic sensillum. (G) Placoid sensillum(van Baaren et al., 1999)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

2. Functional Morphology of Sensilla Most authors consider that sensilla with pores serve as olfactory receptors, contact chemoreceptors or mechanoreceptors (Voegelé et al., 1975; Bin et al., 1989; Olson and Andow, 1993; van Baaren et al., 1996,1999; Isidoro et al., 1996; Bleeker et al., 2004; Gao et al., 2007). In addition to chemosensory sensilla, insect sensilla reportedly respond to humidity, temperature, carbon dioxide, mechanical stimuli (either hair sensilla or campaniform organs), and even IR radiation. Humidity receptors are probably depending on humidity monitored mechanosensory function: hygrometer or evaporimeter theories (Steinbrecht, 1998). The sensilla on the scape and pedicel are primarily mechanoreceptors, which may function as proprioceptors to monitor movements, deflection and vibration of antenna1 segments (Keil, 1997). The single-walled sensilla have a variable outer form, and can be shaped as hairs, pegs, or plates. Steinbrecht (1997) described the pore system in great detail; these sensilla consist of cuticular canals through the hair wall, and the inner side is connected to pore-tubules. The pore-tubules, which are suspended in sensillum lymph, are believed to serve as the route for odour molecules to reach the inside of the sensillum. The number of sensory cells in these sensilla is usually two or three cells, but can be more numerous as in the pore plate sensilla of Hymenopterans (Keil, 1999).

194

Yan Gao, Li-Zhi Luo and Abner Hammond

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The double-walled sensilla are usually rather short, and are apically provided with slitlike pores that penetrate the cuticle. Pore tubules are lacking (Hunger and Steinbrecht, 1998). The number of sensory cells is commonly around five. The dendritic segments are unbranched and are tightly kept together within the hair (Keil, 1999). Terminal pore sensilla are bimodal sensilla responding to taste and tactile stimuli. These sensilla are usually innervated by sensory cells. The single one has a tubular body, indicative of a mechanosensory function (Keil, 1998). The remaining cells reach to the terminal pore, and function as contact chemoreceptors (Keil, 1999).

2.1. Trichoid Sensilla Trichoid Sensilla (TS) have been reported as the most abundant sensilla type present on all antennal segments (van Baaren et al. 1996; Gao et al., 2007). TS are classified according to shape, cuticular attachment and their distribution. These socketless pointed sensilla arise smoothly from the cuticle and exists as one example before the groove on the pedicel at the pedicel hook (Fig. 1a). The major distinction between chaetica sensilla and trichoid sensilla is by the respective presence and absence of a cuticular socket at the point of insertion into the antennal surface (Schneider, 1964). Olson and Andow (1993) classify these “hairlike” structures with a diameter to length ratio less than 0.30µm as s. trichodea. The nonporous s. trichodea of two Braconidae wasps had a grooved surface and a small bulbous structure at the tip. This type was inserted in a socket and had a length of approximately 30 µm with a diameter of approximately 2 µm at the base (Bleeker et al., 2004). These of sensilla had a thick solid wall without pores and the sensilla lymph was not innervated by dendrites of sensory neurons (Fig. 2a, b). They were the most abundant hair-type present on all antennomeres, and they were predominantly situated in between the sensilla placodea. These sensilla were slightly curved towards the apex of the antennomere and usually bent over the sensilla placodea. The second type of nonporous s. trichodea, has a grooved surface and a double pore on the apex. It was described as more perpendicular on the antenna than the other s. trichodea and, had a socket. A single nonporous wall surrounded the inner lumen, which was usually innervated by 5 dendrites. Several s. trichodea TP were present on each antennomere, predominantly at the apical side. Tip pore s. trichodea have been described as s. basiconica type I (Ochieng et al., 2000), s. basiconica A (Navasaro and Elzen, 1991), fluted basiconic sensilla (Norton and Vinson, 1974), and as having curved trichoid formations with an apical pore (Barbarossa et al., 1998). The third type of TS described is the wall pores s. trichodeum. This sensillum type has a smooth surface with numerous pores of about 20 nm in diameter. Numerous dendritic branches innervated the lumen. Wall pores s. trichodea have been described as s. basiconica type 2 (Ochieng et al., 2000), s. basiconica B (Navasaro and Elzen, 1991) and as curved nonfluted basiconic sensilla (Norton and Vinson, 1974).These sensilla are usually present in a circular arrangement along the distal end of each antennomere.

Antennal Sensillar Morphology, Structure And Function In Parasitic Wasps

195

Figure 2. Electromicrograph of Sensillum trichodeum on the male antenna of M. pallidipes. (a) SEM S. trichodeum (T). (b) TEM photograph of transverse section of S. trichodeum. Note the thick and grooved sensillum wall (W), which is not innervated by dendrites.( Gao et al., 2007)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The nonporous TS are the most abundant type of sensillum occurring on all the antennomers of both sexes of parasitic wasps (van Baaren et al., 1996, 1999; Gao et al., 2007). We did not find any pore systems and sensory neurons on the TS dendrites of either sexes of Microplitis pallidipes under TEM examination (Gao et al., 2007), as reported in Microplitis croceipes (Ochieng et al., 2000) and therefore, no olfactory function can be attributed to this sensillar type at this time. TS may serve as mechanoreceptors due to their socket-like insertion into the antennal cuticle and their spatial arrangement (Keil, 1999; Bleeker et al., 2004). Keil (1999) cites that TS may be olfactory, but sensilla found on the pedicel are usually mainly mechanoreceptive. The position of TS across the join of the pedicel hook suggests a proprioceptive function, but it does not possess standard proprioceptor morphology (it lacks a flexible socket) (Fig. 2). Schneider (1964) suggests that TS may be dye-permeable and so may possess chemoreceptivity. Further study by close examination of the surface structure of this sensillum is required to clarify its function.

2.2. Chaetica Sensilla Chaetica sensilla (CS) are commonly seen in braconids (Navasero and Elzen, 1991; Ochieng et al., 2000；Gao et al., 2007) and Chalcidoidea (Amornsak et al. 1998) and mymarids (van Baaren et al., 1999), but they are absent in some Cotesia species (Bleeker et al., 2004; Roux et al., 2005) (Fig. 3). CS are refered to as mechanoreceptive bristles on the antennae of Rhodniu (McIver and Siemicki, 1984). Five types of chaetica sensilla (ChS types l-5) occur on Trichogramma australicum antennae. These were classified on the basis of their cuticular surface, structure and basal socket. ChS l-3 occurs on the female antenna (Amornsak et al. 1998). The location and number of this sensillum type is similarl to that reported in M. croceipes and M. pallidipes (Navasero and Elzen, 1991; Ochieng et al., 2000). The sizes of

196

Yan Gao, Li-Zhi Luo and Abner Hammond

CS in M. pallidipes varied as those reported in M. croceipes. Whether the size variations in CS are suggestive of different types of chaetica as described for mymarids (van Baaren et al., 1999) needs further study. Based on location and structure, CS in other parasitic wasps have usually been considered to have a mechanistic function (Ochieng et al., 2000) or perhaps mixed functions (van Baaren et al., 1999). The function of CS in M. pallidipes and M. croceipes is similarly considered to be mechanistic (Ochieng et al., 2000).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 3. Electromicrograph of Sensilla chaetica distribution on the proximal part of the pedicellus of M. pallidipes. The numbers refer to their variation in size. (Gao et al., 2007)

2.3. Basiconic Sensilla Basiconic sensilla (BS) have been divided into types 1 and 2. Type 1 is practically the same as type 2 but can be differentiated on the basis of the basal depression and stalk (Amornsak et al. 1998). The basiconica types1 and 2 are present in both sexes of M. croceipes and M. pallidipes (Ochieng et al., 2000; Gao et al., 2007) (fig 4a-d, fig 5). The basiconica type 1 has been previously described as fluted basiconic sensilla (Norton and Vinson, 1974a), S. basiconica A (Navasero and Elzen, 1991) and s. trichodea TP (Bleeker et al., 2004). The basiconica type II was formerly described as S. basiconica B (Navasero and Elzen, 1991) and s. trichodea WP (Bleeker et al., 2004). A structure similar to BS 2 has been reported on the flagellum of the eulophid Sympiesis sericeicornis Nees, but it has also been called “peg-like sensillum” or “sensillum coeloconicum” (Meyhofer et al., 1997). BS 2 in T. australicum resembles sensillum described and termed “ampullacea” by Voegelé et al. (1975) for T. brasiliensis, T. maltbyi and T. evanescens, and these were named “multiporous grooved basiconica (MPG) C” in T. nubdale (Olson and Andow, 1993). Similar structures to BS type 1 and type 2 have been reported on the pteromalid Nasonia vitripennis (Walker) and the

Antennal Sensillar Morphology, Structure And Function In Parasitic Wasps

197

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

eulophid Tetrastichus hagenowii (Ratz.) and referred to as “basiconic capitate pegs” (Miller, 1972) and “multiporous pegs” (Barlin et al., 1981), respectively.

Figure 4. Electromicrograph of the distal part of the 10th antennomere of a female M. pallidipes showing the types of S. basiconica. (A) SEM photograph of S. basiconicaN type I (1) and S. basiconica II (2) (Gao et al., 2007); (B) Transverse section of a S. basiconica I. Note the thick and groove nonporous sensillum wall (W) and the sensillum lymph (L), which is innervated by dendrites (D) (TEM) (Gao et al., 2007); (C) Transverse section of S. basiconica type II. Note the numerous dendritic branches (D) and the pores (arrow) in the sensillum wall (W) (TEM) (Gao et al., 2007);(D) Transverse section of S. basiconica type II, showing the smooth sensillum wall (W) and dendritic branches (D) in the base (TEM) (Gao et al., 2007);(E) The S. basiconica I (1) and S. basiconica type II (2) on the distal portion of the last antennomere (SEM). (Gao et al., 2007).

198

Yan Gao, Li-Zhi Luo and Abner Hammond

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 5. close-up view of a flagellum sub-segment showing S. basiconica I (arrow), S. basiconica 2 (arrowhead), SEM, scale bar =1µm (Ochieng et al., 2000).

These two types of sensilla described in the report by Gao et al. (2007) contained on the tip or the walls. The differences in basiconica I and II of M. pallidipes are not only in shape but also in the dendritic innervations. The basiconica type I had grooves on the surface and the inner lumen was surrounded by a thick nonporous wall (Gao et al., 2007; Ochieng et al., 2000). The BS type II was encircled by a thick-wall with multiple pores, similar to those reported by Ochieng et al. (2000). The wall of BS type II of M. pallidipes has more pores than that of M. croceipes (Gao et al., 2007). BS type I is reasoned to be a gustatort sensillum, since it had four to six dendrites inside the wall, although the terminal pore is not evident in SEM photos. The function is based on the behaviour of Habrobracon hebetor since BS are involved in host location, especially in the detection of short-range cues emitted from the host (Dweck and Gadallah, 2007). Darwish et al. (2003) in their study on the behaviour of H. hebetor, found that once the host has been detected the antennae are lowered and the female begins to drum alternately with each antennal tip on the body surface of the host. Finally, she bent her abdomen forward, the ovipositor is unsheathed and probing of the host begins. The female often makes contact with the stung host using the tips of her antennae during probing. This behavior was shown to be in agreement with Weseloh (1972), who suggested that the antennal tips are the principle location for organs involved in initiating host acceptance behaviour in the encyrtid Cheiloneurus noxius. These two reports were as explained by with Roux et al. (2005) who postulated that a gustatory stimulus following antennal contact is probably the key stimulus inducing oviposition behaviour in the braconid Cotesia plutellae.

Antennal Sensillar Morphology, Structure And Function In Parasitic Wasps

199

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

BS has numerous dendritic branches within the sensillum lymph innervating the lumen and numerous pores in the wall; therefore, it is considered to have an olfactory function, or to have hygro-, thermo, or mechanoreceptor function (Wibel et al., 1984; Dahms, 1984; Steinbrecht, 1998; van Baaren et al., 1996; van Baaren et al., 1999; Ochieng et al., 2000).

2.4. Coeloconic Sensilla By definition, coeloconic sensilla (CoS) are pit organs or have a peg protruding from a pit (Snodgrass, 1935; Schneider, 1964: Zacharuk, 1985); for example as in N. uitripennis (termed “campaniform”, Miller, 1972) in the scelionid Gryon pennsylvanicum (Ashmead) (termed “campaniform sensillurn”, Villa and Mineo, 1990), in the cynipid Tryb- Ziographa rapae (Westw.) (termed “coeloconic sensillum”, Butterfield and Anderson, 1994) and in the scelionid T. basalis (termed “campaniform sensillum”, Isidoro et al., 1996). Similar structures have been reported as “campaniform sensilla” or “papilla” (Voegelé et al., 1975) and “campaniform-like structures” in T. nuhilale (Olson and Andow, 1993). CoS have been reported in N. vitripennis and identified as 2 types of campaniform sensilla (Wibel et al., 1984). Only one type CoS was reported to be on the lateral surface of the second anellus and second funicle in the female and on the club in the male of T. australicum, which is similar to an unperforated domed cupola. CoS occur on the distal surface of the pedicel of both sexes in N. vitripennis and T. australicum, but have not been reported on the pedicel of other Trichogramma species (Voegelé et al., 1975; Olson and Andow, 1993). Two types of CoS have been described in Cotesia species of braconids and they were referred to as CoS type I and II (Bleeker et al., 2004; Roux et al., 2005)(fig 6D,E). However, there was only CoS type I found in both sexes and on each flagellomere of M. pallidipes was found (Gao et al., 2007) (fig 6A-C). The shape and cuticular insertion of CoS in M. pallidipes was similar to those in M. croceipes (Navasero and Elzen, 1991; Ochieng et al., 2000) and similar to type I of Cotesia species (Steinbrecht, 1997; Bleeker et al., 2004). The CoS type II, which has been previously described in M. croceipes (Navasero and Elzen, 1991; Ochieng et al., 2000), was not found in M. pallidipes (Gao et al., 2007). TEM investigations show that fingerlike projections containing sensillum lymph that were usually innervated by one dendrite, thus CoS type I is thought to have olfactory function, as reported in previous studies (Altner et al., 1983; Steinbrecht, 1997; Keil, 1999; Ochieng et al., 2000; Roux et al., 2005). Hunger and Steinbrecht (1998) described spoke channels connecting the central lumen with the sensilla grooves through which odour molecules may be transported. These spoke channels seemingly are not always present when examining ultrathin sections of CoS in Habrobracon hebetor and Locusta migratoria (Schneider and Steinbrecht, 1968; Dweck and Gadallah, 2007). In other insects, the CoS may function either as olfactory and/or thermo/hygroreceptors (Kaissling 1971; Altner et al. 1983; Steinbrecht 1997). Thermo- and hygrosensitive receptors have been distinguished from chemoreceptors electrophysiologically or by the presence of a lamellated dendrite beneath unperforated cuticle (Steinbrecht 1999).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

200

Yan Gao, Li-Zhi Luo and Abner Hammond

Figure 6. Electromicrograph of Sensilla coeloconicum of M. pallidipes. (A) Distribution of S. coeloconicum (arrow) on the antennomere (SEM) (Gao et al., 2007);(B) The characteristic peg like stalk of S. coeloconicum (SEM), showing the peg surrounded by a donut-shaped ring (asterisk) (Gao et al., 2007); (C) Transverse section in the middle region of S. coeloconicum showing fingerlike projections (F) and dendritic branches (D) within it (TEM) (Gao et al., 2007);(D) Detail of sensillum coeloconicum type I, showing the characteristic peg. Note the deep grooves between the fingerlike projections. SEM (Bleeker et al., 2004); (E) S. coeloconica type II, with bulbous structure (arrow) and donut-shaped ring (asterisk). SEM. (Bleeker et al., 2004).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Antennal Sensillar Morphology, Structure And Function In Parasitic Wasps

201

2.5. Placoid sensilla Placoid sensilla (PS) have been described in various sizes and shapes on the antennae of nearly all parasitic Hymenoptera (Barlin and Vinson, 1981a, b; Navasero and Elzen, 1991; Ochieng et al., 2000; Bleeker et al., 2004; Roux et al., 2005) (fig 7). PS in Chalcidoidea have been described as 2 morphological types of multiporous plate sensilla which vary in the number of pores as follows: (1) with a relatively thin porous plate and many pores (10 pores perµm2) occurring both in female and male antennae; and (2) with a thicker porous plate and fewer pores (4 pores perµm2) occurring in the female antenna only (Barlin and Vinson, 1981a). Barlin and Vinson (1981a) and Barlin et al. (1981) used external morphology to distinguish sensilla types (width and attachment to the antenna surface), for example: (1) the sensilla are broad and attach to the antenna1 cuticle for almost their entire length, except for the apices that are free; (2) sensilla that are free from the antenna1 cuticle for one half of their length and then taper near the apex. Two morphological types, PS 1 and 2 were found in Tetrastichus australicum (Barlin et al., 1981). But, these two types of PS have not been reported in other Trichogramma spp. (Voegelé et al., 1975; Barlin and Vinson, 1981a; Olson and Andow, 1993). Two types of multiporous plate sensilla have also been found in the eulophids T. hagenowii and C. pulvinariae and the pteromalid Muscidljiirax zaraptor Kogan and Legner (Barlin and Vinson, 1981a). Both types 1 and 2 generally alternate in a ring around the distal segment (Barlin and Vinson, 1981a). PS 1 in T. australicum is longer than PS 2 whereas in M. zaruptor PS 1 is slightly shorter than PS 2. PS of M. pallidipes are elongated and sausageshaped, and commonly occur in both sexes. In comparison to M. croceipes (Ochieng et al., 2000), the size of S. placodea in M. pallidipes is larger and the total number is smaller. The ultra-structures of the PS are also varied to some extent in these two Microplitis species. There were significant differences found in the number, size and density of PS between the sexes of M. pallidipes (Gao et al., 2007). This difference is consistent with that reported for M. croceipes, where males have longer PS, longer antennomeres, and longer antennae (Navasero and Elzen, 1991; Ochieng et al., 2000). The length of PS correlates with the length of antennomeres (Borden et al., 1978b; Navasero and Elzen, 1991; Olson and Andow, 1993; Ochieng et al., 2000). In Bleeker et al. (2004) they posited that the length of antennomere is correlated with the length of PS in C. glomerata. The function of S. placodea is assumed to be olfactory because they posses a multiple cuticular pore system (van Baaren et al. 1996; Barlin and Vinson, 1981a). Single-sensillum research shows that S. placodea in M. croceipes are indeed olfactory receptors that responded in a dose-dependent manner to plant volatiles (Ochieng et al., 2000). The PS in M. pallidipes is considered to have the same function as reported for M. croceipes, since their structures are similar. Their specific function in M. pallidipes however, has yet to be confirmed elctrophysiologically.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

202

Yan Gao, Li-Zhi Luo and Abner Hammond

Figure 7. Electromicrograph of the elongated Sensilla placodea on the base of the13th antennomere of a male M. pallidipes. (A) SEM photograph of the elongated S. placodea (P) distribution on the antennomere in parallel with the antennal longitudinal axis(Gao et al., 2007); (B) Transverse section in the middle region of a placoid sensillum showing elevated grooves (G) that surround sensillum and numerous dendritic branches (D) within the median channel (TEM); note the porous sensillum wall (arrow) (Gao et al., 2007); (C) The septum (S) under the dendritic branches (D) and the porous surface sensillum wall (arrow) in the distal region, and (G) is the elevated grooves (TEM) (Gao et al., 2007); (D) Detail of s. placodeum showing multiple pores arranged in rows (arrows). SEM. Inset: Longitudinal TEM section of s. placodeum showing the cuticular pores (arrows) and numerous

Antennal Sensillar Morphology, Structure And Function In Parasitic Wasps

203

dendritic processes in the sensillum lymph (SL). Bar= 0.5µm (Bleeker et al., 2004); (E) LM micrograph of longitudinal section through a s. placodeum. Note the antennal nerve (AN) and cell bodies of sensory neurons (SN) from which the dendrites (DE) run through the aperture in the septum (SE) into the lumen (LU) under the multiporous plate (MP). Bar= 0.5µm (Bleeker et al., 2004).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

3. Ontogeny of Sensilla According to Walther (1983), an evaluation of the antennal patterns of sensilla in Hymenoptera can be used as a phylogenetic tool. The patterns include (1) density of sensilla and their arrangements; and (2) morphology and ultrastructure of the sensilla. Short and cylindrical basiconic sensilla are considered to be more primitive than long, slender ones. trichodea curvata of ants are considered to be homologous to the sensilla placodea of other Hymenoptera (Walther, 1983). This sensillar type is quite similar to the ascoid sensilla of the Micropterigidae, a similarity examplifying convergent evolution. The number of enveloping cells is variable in Hymenopterans, as additional enveloping cells are often present; and (3) sexual dimorphism exists in patterns of sensilla. Males often possess longer sensilla in larger numbers than females. According to Walther (1983), this is a primitive trait expressed in the males, but the theoretical basis for this assumption is not explained (Hallberg and Hansson, 1999). Hallberg and Hansson (1999) considered that the developmental relation between scales and sensilla can be considered as established. Evolutions of the different sensillar types are independent. But, Cuperus (1983) suggested a relation between scales and different sensillar types. He envisaged an evolution from plate-like to hair-like sensilla, so PS should be more wide-spread among the insect orders, especially in those considered as being ancestral, but they are for instance lacking in Thysanura (Adel, 1984; Larink, 1976). Plate-like sensilla are present in Hymenoptera, Coleoptera, and in certain Trichoptera and Lepidoptera. However, the sensilla placodea are possibly quite heterogenous among different insect groups. Furthermore, it is questionable if it is possible to derive one sensillar type from another, since it is difficult to imagine a shift in perception of adequate stimuli of a sensillum. In some cases the ontogentic development can provide clues regarding the evolution of an organ system. The sensillar ontogeny has been studied in the ametabolous Lepisma and in the hemimetabolous Gryllus, Oncopeltus, and Locusta (Schmidt and Berg, 1994), as well as in a number of holometabolous species: Lepidoptera (Keil, 1997; Keil and Steiner, 1991; Sanes and Hildebrand 1976; Waku, 1991), Coleoptera (Ernst, 1972), Hymenoptera (Martini and Schmidt, 1983, 1984; Schmidt and Kuhbander, 1983; Stepper et al., 1983), and Diptera (Kuhbander,1984, 1985). During the ontogeny of the insect sensillum, supernumerary enveloping cells are present in some cases. These cells may or may not degenerate during the subsequent developmental processes. Often an extra tormogen or trichogen cell can be present. The outer enveloping cells secrete different parts of the cuticular hair, and withdraw when the hair is fully formed. The innermost enveloping cell will give rise to the dendritic sheath. The sensory cells form axons that grow to the brain and the dendrite forms a cilium, which invades the hair after the enveloping cells, have withdrawn. The cilia are formed in the future dendrite in connection with the centriole pair, beneath a membrane vesicle, and the central microtubules of the axoneme are lacking (Keil, 1997; Seidl, 1991; Hallberg and Hansson, 1999).

204

Yan Gao, Li-Zhi Luo and Abner Hammond

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

4. Differences between the Sexes and Species 4.1 Differences Between the Sexes The length of the antennae differs between males and females. In M. croceipes (Navasaro and Elzen, 1991; Ochieng et al., 2000), in C. glomerata and C. rubecula (Bleeker’s et al., 2004), in Microplitis pallidipes (Gao et al., 2007), two mymarid species (van Baaren et al., 1999) and a pteromalid (Pettersson et al., 2001) antennae of males are longer than those of females. This sexual dimorphism may be correlated with the difference in length of s. placodea between males and females. The length of antennomeres is correlated with the length of s. placodea. Males have longer s. placodea, which might result in longer antennomeres and, therefore, longer antennae. (Navasaro and Elzen, 1991; Ochieng et al., 2000; Bleeker’s et al., 2004; Gao et al., 2007). Some studies have also found that male s. placodea are shorter than female s. placodea (Amornsak et al., 1998; Borden et al., 1978a). And, there is also a difference in the number of s. placodea between the sexes. Males have a larger number of s. placodea due to a higher density of these sensilla. Hymenopteran males often have a higher number of s. placodea (van Baaren et al., 1999; Borden et al., 1978b; Navasaro and Elzen, 1991; Ochieng et al., 2000; Bleeker’s et al., 2004; Gao et al., 2007). Romani et al. (2002) found two types of sensilla, multiporous gustatory sensilla of type 1 and 2 (MGS1 and MGS2) on the female antennal functional area of the pupal parasitoid Trichopria. drosophilae, and these sensilla are found only on females. Only three of these five type chaetica sensilla occur on the female antennae of T. australicum (Amornsak et al. 1998). Two types of multiporous plate sensilla may be found in Chalcidoidea, were the female has two types and male only one type (Barlin and Vinson, 1981a). Because of their appearance only on the female’s antennae, PS 2 may have a female-specific function, such as receiving olfactory cues during oviposition behaviour (Barlin and Vinson, 1981a; Barlin et al., 1981). Male insects in general often have a higher number of olfactory sensilla compared to females (Chapman, 1982). In most of these species, males are attracted to females by sex pheromones (Chapman, 1982; Field and Keller, 1993; Tagawa, 1977; Tagawa and Kitano, 1981). Higher numbers of sensilla may indicate an increase in sensitivity (Chapman, 1982; Ignell et al., 1999). This could indicate a mate locating function of these sensilla in males and, more specifically, the detection of sex pheromones. In the scarabid beetle Anomala cuprea, males have a higher number of sensilla placodea than females, and these s. placodea are sensitive to sex pheromones (Larsson et al., 2001; Leal and Mochizuki, 1993). The males of C. glomerata and C. rubecula may use the s. placodea to detect sex pheromones, possibly in conjunction with host plant odors. In a moth species, host plant volatiles have been discovered that increase the response of a sex pheromone specific olfactory receptor neuron (Ochieng et al., 2002). In both species, no differences in type and topographical arrangement were found between males and females. In another Ichneumonoid, M. croceipes, this difference was also absent (Ochieng et al., 2000). Such differences in types and location of sensilla are present in other parasitoids that belong to the Chalcidoidea and Platygasteroidea, and these differences are thought to be associated with sex-specific differences in behavior, e.g., courtship and host recognition (Amornsak et al., 1998; van Baaren et al., 1999; Cave and Gaylor, 1987).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Antennal Sensillar Morphology, Structure And Function In Parasitic Wasps

205

4.2. Differences Between the Species In all these families of Hymenoptera, they are always the most abundant type of sensillum covering most of the antennomeres. Sensilla numbers differ among parasitic Hymenoptera depending on the number of flagellar segments. In N. vitripennis and T. hagenowii 22 sensilla and 15 sensilla are present, respectively (Miller, 1972; Barlin et al., 1981).van Baaren et al. (1996) reported that basiconic sensilla are found on all flagellum segments in E. lopezi but only on segments 2-10 in L. dactylopii. Meyhofer et al. (1997) found only 8 peg-like sensilla on the 6th flagellomere and about 30-60 sensilla on other flagellomeres in S. sericeicornis. The presence of PS on the antennae is typical for Hymenoptera, as well as the presence of multi-innervated s. trichodea. In most other groups, this sensillar type is innervated by 2–3 sensory cells, but in some Hymenoptera there are around 10 sensory cells (Hallberg, 1979). There is a significant difference in quantity of s. placodea between the species. Bleeker’s et al.(2004) study shows that C. rubecula females possess a larger number of s. placodea compared to C. glomerata females and C. rubecula males have a larger amount of s. placodea compared to C. glomerata males. The large number of s. placodea might indicate a higher olfactory sensitivity for C. rubecula compared to C. glomerata. However, the number of sensilla placodea is positively correlated with body size in bees (Johnson and Howard, 1987) and such a correlation may also explain the higher number of s. placodea in the slightly larger C. rubecula. These studies showed that the olfactory receptive range (Smid et al., 2002) as well as the antennal lobe structure is similar (Smid et al., 2003). The study of three very similar species of the genus Anaphes victus and A. listronoti and redescribed A. sordidatus showed that most differences were in the length of wings, ovipositors, heads or antennal antennomeres, but most measurements presented overlaps (Huber et al. 1997). For example, the antennae of females from the Quebec populations of A. listronoti presented two sensory ridges (placoid sensilla) on the second antennal antennomere, while the Texas populations of A. victus had no, or occasionally one, sensory ridge on the second antennal antennomere. And important size overlap is present for all characters measured and both species have two placoid sensilla on the second antennal antennomere (van Baaren et al., 1999). van Baaren et al. (1999) reported that females of Quebec populations of A. victus and A. listronoti can be morphologically distinguished on the base of the number of chaetica type 4. However, male antennae present no differences between the two species. When comparing Coptera occidentalis Muesebeck to Trichopria drosophilae, the touch and taste sensilla areas show remarkable differences, since there is only one type of gustatory sensilla, i.e., MGS1 on Coptera occidentalis Muesebeck, and these are arranged in patches on the ventral side of the apical six antennomeres, while the MGS2 are not present at all (Romani et al., 2002 ). This could be related to differences between the host puparia (Romani et al., 2002).

CONCLUSION It is significant to note that there are certain similarities in the structure of the chemosensory sensilla of Chalcidoidea, and Mymaridae and Braconidae. These similarities

206

Yan Gao, Li-Zhi Luo and Abner Hammond

demonstrate that adaptation to a certain environment will likely result in similar morphology. It is also worth noting that the difference between sexes of the same species, is closely connected with the different behaviors of females and males. The abundance of contact chemoreceptors on male antenna as well as on female antennae, should stimulate research for differences in function should be detected between the sexes. Numerous morphological studies further emphasize the value of this model system as a comparative approach to studying the neurobiological mechanisms underlying differences in insect associative learning. Additional knowledge of sensilla function will dependent on future electrophysiological research. New research is needed to explain the relationship among these many different types of olfactory sensilla within parasitoid species. Parasitoid larvae have little opportunity to switch to another host once a decision has been made by the ovipositing female, indicating that insects cannot determine the suitability of potential hosts. Host preferences of ovipositing females are sometimes much narrower than the range of plants on which their can survive, but utilize indirect cues that have proved reliable in the past (Schoonhoven et al., 1998). It is necessary that clarify the relation between these sensilla function and the environment of different parasitoid based on life-history characteristics of both parasitoids and hosts.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

REFERENCES Adel T. (1984). Sensilleninventar und Sensillenmuster auf den Antenne vom Thermobia domestica und Lepisma saccharina (Insecta: Zygentoma). Braunschw Naturk Schr, 2, 191–217. Alborn, H.T., Lewis, W.J., Tumlinson, J.H., (1995). Host-specific recognition kairomone for the parasitoid Microplitis croceipes (Cresson). J. Chem. Ecol., 21, 1697–1708 Altner H. 1977. Insektensensillen: Bau- und Funktionsprinzipien. Verh Dtsch Zool Ges, 1977:139–153. Altner I, Hatt H, Altner H. (1983). Structural properties of bimodal chemo- and mechanosensitive setae on the perciopod ?? of the crayfish, Austro-potemobius torrentium. Cell Tiss Res 228, 351–374. Amornsak W, Cribb B, Gordh G. (1998). External morphology of antennal sensilla of Trichogramma australicum Girault (Hymenoptera : Trichogrammatidae). Int J Insect Morphol Embryol 27, 67–82. Barbarossa IT, Muroni P, Dardani M, Casula P, Angioy AM. (1998). New insight into the antennal chemosensory function of Opius concolor (Hymenoptera, Braconidae). It J Zool 65,367–370. Barlin MR, Vinson SB, Piper GL. (1981). Ultrastructure of the antennal sensilla of the cockroach egg parasitoid, Tetrastichus hagenowii (Hymenoptera: Eulophidae). J Morphol 168, 97–108 Barlin, M. R., Vinson, S. B. (1981a). The multiporous plate sensillum and its potential use in braconid systematics (Hymenoptera: Braconidae). Canadian Entomologist, 113 , 931938.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Antennal Sensillar Morphology, Structure And Function In Parasitic Wasps

207

Barlin, M. R., Vinson, S. B. (1981b). Multiporous plate sensilla in antennae of the Chalcidoidea (Hymenoptera). International Journal of Insect Morphology and Embryology, 10, 29-42. Battaglia D, Isidoro N, Romani R, Bin F, Pennacchio F. (2002). Mating behaviour of Aphidius ervi (Hymenoptera: Braconidae): the role of antennae. Eur J Entomol, 99, 451– 456 Bin F, Colazza S, Isidoro N, Vinson SB. (1989). Antennal chemosensilla and glands and their possible meaning in the reproductive behaviour of Trissolcus basalis (Woll.) (Hymenoptera: Scelionidae). Entomologica, 24, 33–97 Bin F, Wäkers F, Romani R, Isidoro N. (1999). Tyloids in Pimpla turionellae (L.) are release structures of male antennal glands involved in courtship behaviour (Hymenoptera: Ichneumonidae). Int J Insect Morphol Embryol, 28, 61–68 Bleeker MAK, Smid HM, van Aelst AC, van Loon JJA, Vet LEM. (2004). Antennal sensilla of two parasitoid wasps: a comparative scanning electron microscopy study. Microsc Res Tech, 63, 266–273 Borden, J. H., Chong, L., Rose, A. (1978a). Morphology of the elongate placoid sensillum on the antennae of Itoplectis conquisitor. Annals of the Entomological Society of America, 71, 223-227. Borden, J. H., Rose, A., Chorney, R. J. (1978b). Morphology of the elongate sensillum placodeum on the antennae of Aphidius smithi (Hymenoptera: aphidiidae). Canadian Journal of Zoology, 56, 519-525. Butterfield A, Anderson M. (1994). Morphology and ultrastructure of antennal sensilla of the parasitoid Trybliographa rapae (Westw.) (Hymenoptera: Cynipidae). Int J Insect Morphol Embryol, 23, 11–20 Cave RD, Gaylor MJ. (1987). Antennal sensilla of male and female Telonomus reynoldsi Gord and Coker (Hymenoptera: Scelionidae). Int J Insect Morphol Embryol, 16, 27–39. Chapman RF. (1982). Chemoreception: the significance of receptor number. Adv Insect Physiol, 16, 247–357 Chiappini Elisabetta, Mazzoni Emanuele. (2000). DiOEering morphology and ultrastructure of the male copulatory apparatus in species-groups of Anagrus Haliday (Hymenoptera: Mymaridae). Journal of Natural History, 34, 1661–1676 Cuperus PL. (1983). Distribution of antennal sense organs in male and female ermine moths, Yponomeuta vigintipunctatus (Retzius) (Lepidoptera: Yponomeutidae). Int J Insect Morphol Embryol, 12, 59–66. Dahms, E. C. (1984). An interpretation of the structure and function of the antennal sense organs of Melittobia australica (Hymenoptera: Eulophidae) with the discovery of a large dermal grand in the male scape. Memoirs of the Queensland Museum, 21, 361-385. Darwish E, El-shazly M, El-Sherif H. (2003). The choice of probing sites by Bracon hebetor Say (Hymenoptera: Braconidae) foraging for Ephestia kuehniella Zeller (Lepidoptera: Pyralidae). J Stored Prod Res, 39, 265–276 Dweck K. M H., Gadallah S. N. (2007). Description of the antennal sensilla of Habrobracon Hebetor, BioControl (In printing) Ernst K-D. (1972). Die Ontogenie der basiconischen Riechsensillen auf der Antenne von Necrophorus (Coleoptera). Z Zellforsch, 129, 217–236 Field SA, Keller MA. (1993). Courtship and intersexual signaling in the parasitic wasp Cotesia rubecula (Hymenoptera: Braconidae). J Insect Behav, 6, 737–750.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

208

Yan Gao, Li-Zhi Luo and Abner Hammond

Fukushima, J., Kainoh, Y., Honda, H., Takabayashi, J. (2002). Learning of herbivore-induced and nonspecific plant volatiles by a parasitoid, Cotesia kariyai. J. Chem. Ecol. 28, 3, 579–586. Gao Y, Luo L-Z, Hammond A, (2007) Antennal morphology, structure and sensilla distribution in Microplitis pallidipes (Hymenoptera: Braconidae). Micron 38, 6, 684–693 Gullan PJ, Cranston PS. (1994). The insects: an outline of entomology. Chapman & Hall, London. Hallberg E. (1979). The fine structure of the antennal sensilla of the pine saw fly Neodiprion sertifer (Insecta: Hymenoptera). Protoplasma, 101, 111–126. Hallberg Eric, Hansson Bill. (1999). Arthropod Sensilla: Morphology and Phylogenetic Considerations. Microscopy Research and Technique, 47, 428–439. Hoballah, M.E., Tamo, C., Turlings, T.C.J. (2002). Differential attractiveness of induced odors emitted by eight maize varieties for the parasitoid Cotesia marginiventris: is quality or quantity important? J. Chem. Ecol. 28, 951– 968. Huber JT, Côté S, Boivin G. (1997). Description of three new Anaphes species (Hymenoptera: Mymaridae), egg parasitoids of the carrot weevil, Listronotus oregonensis (Leconte) (Coleoptera: Curculionidae), and redescription of Anaphes sordidatus Girault. Can Entomol 129, 959–977. Hunger T, Steinbrecht A. (1998) Functional morphology of a double-walled multiporous olfactory sensillum: the sensillum coeloconicum of Bombyx mori (Insect, Lepidoptera). Tissue Cell 30,1, 14–29 Ignell R, Anton S, Hansson BS. (1999). Integration of behaviourally relevant odours at the central nervous level in solitary and gregarious third instar locusts, Schistocerca gregaria. J Insect Physiol, 45, 993–1000. Isidoro, N., F. Bin, S. Colazza, S.B. Vinson. (1996). Morphology of antennal gustatory sensilla and glands in some parasitoid Hymenoptera with hypothesis on their role in sex and host recognition. Journal of Hymenoptera Research, 5, 206–239. Isidoro N, Romani R, Bin F. (2001). Antennal multiporous sensilla: their gustatory features for host recognition in female parasitic wasps (Insect, Hymenoptera: Platygastroidea). Microsc Res Tech, 55, 350–358 Johnson LK, Howard JJ. (1987). Olfactory disc number in bees of different sizes and ways of life (Apidae: Meliponinae). J Kansas Entomol Soc, 60, 380–388. Kaissling K-E. (1971). Insect olfaction. In: Beidler LM (ed) Handbook of sensory physiology, vol 4, issue (1). Springer Verlag, Berlin, pp 351–431 Keil TA, Steiner C. (1991). Morphogenesis of the antenna of the male silkmoth, Antheraea polyphemus. III. Development of olfactory sensilla and the properties of hair-forming cells. Tissue Cell, 23, 821–851. Keil T.A. (1997). Comparative morphogenesis of sensilla: A review. Int J Insect Morphol Embryol, 26, 151–180 Keil TA. (1998). The structure of integumental mechanoreceptors. Microsc Anat Invertebrates, 11B, 385–404. Keil TA. (1999). Morphology and development of the peripheral olfactory organs. In: Hansson BS, editor. Insect olfaction. Heidelberg: Springer. p 5–47. Kuhbandner B. (1984). Ultrastructure and ontogeny of the hair sensilla on the funicle of Calliphora erythrocephala (Insecta, Diptera). I. The basiconic sensilla. Zoomorphology, 104, 373–385.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Antennal Sensillar Morphology, Structure And Function In Parasitic Wasps

209

Kuhbandner, B. (1985). Ultrastructure and ontogeny of the doublewalled sensilla on the funicle of Calliphora erythrocephala Meigen (Dipter: Calliphoridae). Int J Insect Morphol Embryol, 14, 227–242. Larink O. 1976. Entwicklung und Feinstruktur der Schuppen bei Lepismatiden und Machiliden (Insecta, Zygentoma und Archaeognatha). Zool Jb Anat, 95, 252–293. Larsson MC, Leal WS, Hansson BS. (2001). Olfactory receptor neurons detecting plant odours and male volatiles in Anomala cuprea beetles (Coleoptera: Scarabaeidae). J Insect Physiol, 47, 1065–1076 Leal WS, Mochizuki F. 1993. Sex pheromone reception in the scarab beetle Anomala cuprea. Enantiomeric discrimination by sensilla placodea. Naturwissenschaften, 80, 278–281. Martini R, Schmidt K. (1983). Cell degeneration during early development of hymenopteran olfactory sensilla. Tissue Cell,15, 823–827. Martini R, Schmidt K. (1984). Ultrastructure and early development of the pore plate sensilla of Gymnomerus laevipes (Schukard)(Vespoidea, Eumenidae). Protoplasma, 119,197– 211. McIver, S.B., and Siemicki, R. (1984). Fine structure of antennal mechanosensilla of adult Rhodnius prolixus Stal (Hemiptera: Reduviidae).J. Morphol., 180, 19–28. Meyhofer, R., Casas, J. and Dom, S. (1997). Mechano- and chemoreceptors and their possible role in host location behavior of Sympiesis sericeicornis (Hymenoptera: Eulophidae). Ann. Entomol. Sot. Amer., 90, 208-2 19. Miller, M. C. (1972). Scanning electron microscope studies of the flagellar sense receptors of Peridesmia discus and Nasonia vitripennis (Hymenoptera: Pteromalidae). Ann. Entomol. Sot. Amer. 65, 1119-1124. Navasero, R.C., Elzen, G.W. (1991). Sensilla on the antennae, fortarsi and palpi of Microplitis croceipes (Cresson) (Hymenoptera: Braconidae). Proc. Entomol. Soc. Wash. 93, 737–747. Norton, W. N., Vinson, S. B. (1974). A comparative ultrastructural and behavioral study of the antennal sensory sensilla of the parasitoid Cardiochiles nigriceps (Hymenoptera: Braconidae). Journal of Morphology, 142, 329-350. Ochieng SA, Park KC, Baker TC. (2002). Host plant volatiles synergize responses of sex pheromone-specific olfactory receptor neurons in male Helicoverpa zea. J Comp Physiol A, 188, 325–333. Ochieng SA, Park KC, Zhu JW, Baker TC. (2000). Functional morphology of antennal chemoreceptors of the parasitoid Microplitis croceipes (Hymenoptera : Braconidae) Arthropod Structure and Development, 29, 231-240. Olson DM and Andow DA. (1993). Antennal sensilla of female Trichogramma nubilale (Ertle and Davis) (Hymenoptera : Trichogrammatidae) and comparisons with other parasitic Hymenoptera. International Journal of Insect Morphology and Embryology, 22, 505-520. Parè, P.W., Tumlinson, J.H. (1999). Plant volatiles as a defense against insect herbivores. Plant Physiol. 121, 325–331. Pettersson EM, Hallberg E, Bigersson G. (2001). Evidence for the importance of odour perception in the parasitoid Rhopalicus tutela (Walker) (Hym., Pteromalidae). J Appl Entomol, 125, 293–301. Quicke, D.L.J. (1997). Parasitic Wasps. Chapman and Hall, London, 470 p.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

210

Yan Gao, Li-Zhi Luo and Abner Hammond

Romani R., Isodoro N., Bin F. and Vinson S.B. (2002). Entomologia Experimentaalis et Applicata, 105, 119. Roux O, van Baaren J, Gers C, Arvanitakis L, Legal L. (2005). Antennal structure and oviposition behavior of the Plutella xylostella specialist parasitoid: Cotesia plutellae. Microsc Res Tech, 68, 36–44. Sanes JR, Hildebrand JG. (1976). Origin and morphogenesis of sensory neurons in an insect antenna. Dev Biol, 51, 300–319. Schenk O. (1903). Die antennalen Hautsinnesorgane einiger Lepidopteren und Hymenopteren. Zool Jb Anat, 17, 573–618. Schmidt K, Kuhbandner B. (1983). Ontogeny of the sensilla placodea on the antennae of Aulacus striatus Jurine (Hymenoptera: Aulacidae). Int J Insect Morphol Embryol, 12, 43– 57 Schmidt K, Berg J. (1994). Morphology and ontogeny of single-walled multiporous sensilla of hemimetabolous insects. Tissue Cell, 26, 239–247 Schneider D. (1964). Insect Antennae. Annual Review of Entomology, 9, 103-122. Schneider, D., Steinbrecht, R. A. (1968). Checklist of insect olfactory sensilla. Symposia of the Zoological Society of London, 23, 279-297. Schoonhoven LM, Jermy T, van Loon JJA. (1998). Insect–Plant Biology. Chapman & Hall,London. Seidl S. (1991). Structure and differentiation of the sensilla of the ventral sensory field on the maxillary palps of Periplaneta Americana (Insecta, Blattodea), paying special attention to the ciliogenesis of the sensory cells. Zoomorphology, 111, 35–47. Smid HM, Bleeker MAK, Loon JJA, Vet LEM. (2003). Three-dimensional organization of the glomeruli in the antennal lobe of the parasitoid wasps Cotesia glomerata and C. rubecula. Cell Tiss Res, 312, 237–248. Smid HM, Loon JJA, Posthumus MA, Vet LEM. (2002). GC-EAGanalysis of volatiles from Brussels sprouts plants damaged by two species of Pieris caterpillars: olfactory receptive range of a specialist and a generalist parasitoid wasp species. Chemoecology, 12, 169– 176. Snodgrass RE. (1935). Principles of Insect Morphology. McGraw-Hill, New York; London. Steinbrecht RA. (1997). Pore structures in insect olfactory sensilla. A review of data and concepts. Int J Insect Morphol Embryol, 26, 229–245 Steinbrecht RA. (1998). Bimodal thermo- and hygrosensitive sensilla. Microsc Anat Invertebrates, 11B, 405–422. Steinbrecht RA. (1999). Bimodal thermo- and hygrosensitive sensilla. In: Harrison FW, Locke M (eds) Microscopic anatomy of invertebrates, vol 11b. Wiley-Liss, New York, pp 405–422. Stepper J, Becker C, Schmidt K. (1983). Feinbau und Ontogenese der Porenplatten auf den Antennen von Pimpla turionellae (Hymenoptera, Ichneumonidae). Zoomorphology, 102, 11–32. Tagawa J. (1977). Localization and histology of the female sex pheromone producing gland in the parasitic wasp Apanteles glomeratus.J Insect Physiol, 23, 49–56. Tagawa J, Kitano H. 1981. Mating behaviour of the braconid wasp, Apanteles glomeratus L. (Hymenoptera: Braconidae) in the field. Appl Entomol Zool, 16, 345–350.

Antennal Sensillar Morphology, Structure And Function In Parasitic Wasps

211

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

van Baaren J, Barbier R, Ne´non JP. (1996). Female antennal sensilla of Epidinocarsis lopezi and Leptomastix dactylopii (Hymenoptera: Encyrtidae), parasitoids of pseudococcid mealybugs. Can J Zool, 74, 710–720. van Baaren J, Boivin G, Lelannic J, Ne´non JP (1999) Comparison of antennal sensilla of Anaphes victus and A. listronoti (Hymenoptera: Mymaridae), egg parasitoids of Curculionidae. Zoomorphology, 199, 1–8 Villa, L. and Mineo, G. (1990). Mapping of the antenna] sensilla structures of Gryon pennsylvanicum (Ashmead): A SEM study (Hym., Scelionidae). Frustula Entomol. XIII, 225-235. Vinson SB, Bin F, Strand MR. (1986). The role of the antennae and host factors in host selection behavior of Trissolcus basalis (Woll.) (Hymenoptera: Scelionidae). Colloq INRA, 43, 267–273. Vinson, S.B. (1991). Chemical signals used by parasitoids. In: F. Bin (ed.), Proceedings of the 4th European Workshop, Insect Parasitoids (Perugia, April 3–5, 1991). Redia, Perugia, pp. 43–93 Voegelé J, Cals-Usciati J, Pihan JP, Daumal J. (1975). Structure de l’antenne femelle des trichogrammes. Entomophaga, 20, 161–169 Waku Y. (1991). Developmental changes of the antenna and its neurons in the silkworm, Bombyx mori, with special regard to larval-pupal transformation. J Morphol, 207, 253– 271. Walther JR. (1983). Antennal patterns of sensilla of the Hymenoptera - A complex character of phylogenetic reconstruction. Verh Naturwiss Ver Hamburg, 26, 373–392. Weseloh, R. M. (1972). Sense organs of the hyperparasite Cheiloneurus noxius (Hymenoptera: Encyrtidae) important in host selection processes. Annals of the Entomological Society of America, 61, 41-46. Wibel, R. G., Cassidy, J. D., Buhse, H. E., Cummings, M. R., Bindokas, V. P., Charlesworth, J. and Baumgartner, D. L. (1984) Scanning electron microscopy of antenna1 sense organs of Nasonia vitripennis (Hymenoptera: Pteromalidae). Trans. Amer. Microsc. Sot. 103, 329-340. Zacharuk RY. (1985). Antennae and Sensilla. In Comprehensive Insect Physiology Biochemistry and Physiology. Nervous System: Sensory, pp 1-69,

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

In: Antennas: Parameters, Models and Applications Editor: Albert I. Ferrero

ISBN 978-1-60692-463-1 © 2009 Nova Publishers, Inc.

Chapter 8

MEASUREMENT OF PARAMETERS OF THE ACOUSTIC ANTENNA ARISING AT BRAKING AND STOPPING OF THE PROTON BEAM IN WATER AND RESEARCH OF CHARACTERISTICS OF CREATED FIELD V. B. Bychkov1, V. S. Demidov2 and E. V. Demidova2 1

All-Russia Scientific Research Institute of Physicotechnical and Radio Engineering Measurements, State Scientific Center, Mendeleevo, Moscow Region, Russian Federation 2 Institute of Theoretical and Experimental Physics, State Science Center, Moscow, Russian Federation

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ABSTRACT The purpose of the present paper is the experimental research of properties of the acoustic antenna arising at the braking of an intensive beam of accelerated protons in the water environment. Research was conducted in a near-field zone that had allowed to allocate signals from separate elements of the antenna and to carry out the analysis of parameters of signals, such as amplitude, width and time of their propagation. As a source of protons, the external beam of the accelerator at the Institute of Theoretical and Experimental Physics (ITEP, Moscow), with energy of 200 MeV and a time duration of 70 ns, was used. The beam intensity was supported at the level of 4⋅1010 protons per pulse and supervised by the current transformer. An experiment was carried out in the parallelepiped plexiglass basin of a square section 95 cm in length and with a volume of 250 liters filled to 85% with water. Input of the proton beam inside the volume was realized through a pipe with a diameter of 59 mm, 46 cm length and wall thickness of 1.5 mm inserted into a lateral side of the basin and closed by a plug made from organic glass with thickness of 2mm. The average ionizing range of protons in water was 25.2 cm. So, the sizes of the basin and the applied equipment have allowed study of the nondeformed structure of a hydroacoustic field induced by the proton beam. Measurements of an acoustic field were made by means of a relocatable hydrophone in two mutual-perpendicular directions. Along the beam axis hydrophone movement was carried out with a step of 8.9 mm at a distance of 3.5 cm from the beam axis. In the cross-

214

V. B. Bychkov, V. S. Demidov and E. V. Demidova cut direction the trace passed in the horizontal plane, passing through the beam axis at a distance of 35.6 cm from the point of the entrance of the proton beam in water. In this case the scanning step was equal to 4.45 mm. According to thermoacoustic model in the area of beam action for time, comparable with the action time, an acoustic antenna arises. In the present work the problem of reconstruction of the form of the antenna using the experimental results is being solved. The technique of calculation of the hydrophone response to the radiation of separate elements of the acoustic antenna has been developed. The dependences of amplitude of the signals and their time parameters on the relative position of the antenna and the hydrophone have been obtained. The angular distribution of the field created by the terminal area of the radiation zone has been obtained. This characteristic, generally speaking, is similar to the directional diagram of an audio antenna. To test the experimental results, the full-scale simulation of set-up geometry and the physical processes accompanying the propagation of protons in water was carried out using GEANT-3.21 package. The simulation of the process of generation of an acoustic signal was performed as a first approximation in the assumption of proportionality of the signal intensity to the energy that is generated at the ionisation of atoms of water by a proton without taking into account heat conductivity and the elastic properties of the environment, leading to relaxation. The model calculations confirm the qualitative conclusions and the results obtained at the processing of experimental data.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

1. INTRODUCTION Experimental and theoretical research on radiation acoustics have been carried out during several of the most recent decades [1-2]. It was established that the intensive fluxes of ionising radiation create in a substance lengthy acoustic antenna (АА), the sizes and form of which are defined by distribution in the medium of the thermal field induced by radiation. General characteristics of the radiation acoustic waves arising in the process of propagation of ionising particles through a substance [3-9] are well studied: the proportional dependence of the response of acoustic receiving systems on the proton beam intensity, the dependence of the signal duration on the beam diameter, temperature dependence of intensity of the acoustic signal arising in liquids, etc. These characteristics are well described by a thermoacoustic model of the occurrence of mechanical oscillations during the propagation of ionising radiation through the substance. However, many problems of radiation acoustics remain unsolved. The properties of a radiation acoustic antenna were studied mainly with reference to hadron-electromagnetic showers (HES), created by cosmic rays of ultrahigh energy in the water environment, with hope to measure their energy. It was shown theoretically and by the simulation that acoustic antenna as well as showers have the form described by smooth function with a maximum in the middle of a shower. On the other hand, to specify the mechanisms of ultrasonic generation experiments on the study of braking of intensive monochromatic beams of low-energy protons to their stoppage can be interesting. In this case the distribution of energy loss has another shape: an almost uniform distribution along the entire range of protons comes to the end with a sharp maximum in a zone of so-called Bragg peak in the area of ionisation range. In papers [10,11] the space-time picture of a field arising in water during the stoppage of a proton beam has been registered for the first time. Measurements have been fulfilled in the points located on trajectories, parallel to beam

Measurement of Parameters of the Acoustic Antenna Arising at Braking …

215

direction. Three sources of radiation АА—from the area nearest to the receiver, from Bragg peak and from the point of beam entrance in water—have been identified. The purpose of the present work is the restoration of the form of acoustic antenna arising at the braking of an intensive beam of accelerated protons in the water environment as well as the research of properties of radiation—dependences of pressure and frequency of radiation on co-ordinates of a point of reception of a signal and on the observation angle.

2. METHODICAL ASPECTS 2.1. The Arrangement of Experiment

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The experiment was carried out at the external proton beam of ITEP accelerator similar to that which was used in papers [7, 9-11]. The proton beam energy was equal to (200±0.4) MeV, duration of a bunch—70 ns. The average beam intensity was supported at the level of about 4·1010 protons per pulse and was supervised by the current transformer. The spatial form of beam in the cross-section direction was quasi-gauss with root-mean square deviation equal to 1.5 cm. The beam was limited by the lead collimator with the diameter of 6 cm and a length of 5 cm. The acoustic oscillations were generated in water basin (see Figure 1). The box of basin 1 made of organic glass has a length of 94.5 cm and a cross-section of 50.8×52.3cm (in vertical direction). The input of the proton beam into the centre of measuring volume was carried out through a duralumin pipe 2 with a diameter of 59 mm, length 46 cm and thickness of a wall of 1.5 mm, inserted into the lateral side of the basin and closed by the Teflon end-cap 3 with a thickness of 2 mm.

Figure 1. Photo of the experimental basin.

216

V. B. Bychkov, V. S. Demidov and E. V. Demidova

The basin was filled with salty water to 85% of its volume. Concentration of sea salt was about 3%. The water temperature was equal to 18.5°С and did not change during the experiment. The research was carried out by the method of a scanning hydrophone [10-11]. The electromechanical scanner 4 with manual remote control allowed to place the hydrophone 5 discretely with step s1 =8.9 mm or s2 = 4.45 mm within the linear aperture with the length 40 cm.

2.2. The Run of Experiment

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The measurements of an acoustic field were made in two mutual-perpendicular directions. The scheme of the arrangement of points in which the measurements in coordinates Z (along beam) –Х (in horizontal direction) were carried out is shown in Figure 2. The cross-section of the area of beam action is represented as a contour of twodimensional distribution of average loss of energy by protons in water in which 67% of the energy deposited by protons concentrate approximately. Scanning traces are marked by symbols I, II. Trace I passed in parallel with beam axis at Х=3.4 cm, trace II transversely to the beam direction at Z=35.6 cm

Figure 2. The schematic picture of beam (simulation) and the arrangement of points in which measurements were made.

Measurement of Parameters of the Acoustic Antenna Arising at Braking …

217

Along trace I measurements were carried out in 74 points, located from each other by step size s2, along trace II—in 39 points with step s1. In the experiments, two different hydrophones and various operating modes of reading devices were applied.

2.2. Hydrophones, Amplification Equipment and Readout Technique Transducing of acoustic signals to the electric ones was carried out by means of specially designed hydrophones based on piezoceramics ZnTiPb of spherical (trace I) and cylindrical shapes (trace II). In the latter case piezoceramics has been tangentially polarised. The cylindrical hydrophone had the built-in preamplifier, at the trace I the charge amplifier 2635 produced by Bruel&Kjaer was applied for signal amplification. In the table the parameters of hydrophones, amplifiers, readout equipment and their operating modes are presented. The quantization of signals from the exit of preamplifiers was made by 2-beam digital oscilloscope TEKTRONIX TDS 3032 connected with the personal computer by means of interface GPIB. The information in volume of 104 points was recorded to the computer disk in the format *.sht. Except for the hydrophone response, the signal from the current transformer (CT) measuring the proton beam intensity was recorded. Electronics of the CT worked in integrating mode, so the amplitude of its signal was proportional to the proton beam current. In Figure 3 the example of the oscillogram obtained on trace I at the point № 32 is presented. The exit of the acoustic signal amplifier was connected to the first channel of the oscilloscope. The signal is shown for time evolvement of 200 μs per a square at the scale of 200 mV per a cell. The signal of current transformer is presented on the scale of 20 μs х 500 mV. Parameters

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The shape and size of a piezoelement Average sensitivity of a piezoelement, mcv/Pa Range of uniformity the amplitudefrequency characteristics in the limits ± 5 dB, Hz Signal / noise The amplifier, the amplification coefficient Frequency of numbering of signal, MHz Own noise of the amplifier, nV/SQRT (Hz)

A series of measurements I Sphere ∅ = 1.5 cm

A series of measurements II Cylinder ∅= 4 mm, h=6 mm

2500

1500

2.102 ÷5.104

102 ÷105

> 10 dB Bruel&Kjaer – 2635, k=400

> 12 dB

5

2. 5

-