Understanding Fiber Optics [5 ed.] 1511445653, 9781511445658

Understanding Fiber Optics is the fifth edition of an intuitive introduction to fiber optics widely used as a textbook,

2,440 454 74MB

English Pages 802 [801] Year 2015

Report DMCA / Copyright


Polecaj historie

Understanding Fiber Optics [5 ed.]
 1511445653, 9781511445658

Table of contents :
Preface to the Laser Light Press Edition
Chapter 1 An Introduction to Fiber Optics
Chapter 2 Fundamentals of Fiber-Optic Components
Chapter 3 Fundamentals of Communications
Chapter 4
Types of Optical Fibers
Chapter 5 Properties of Optical Fibers
Chapter 6 Fiber Materials, Structure, and Manufacture
Chapter 7 Specialty Fibers
Chapter 8 Cabling
Chapter 9 Light Sources
Chapter 10 Transmitters
Chapter 11 Receivers
Chaper 12 Amplification, Regeneration, and Wavelength Conversion
Chapter 13 Connectors and Splices
Chapter 14 Couplers and Other Passive Components
Chapter 15 Wavelength-Division Multiplexing Optics
Chapter 16 Optical Switches, Modulators, and Other Active Components
Chapter 17 Fiber-Optic Measurements
Chapter 18 Troubleshooting and Test Equipment
Chapter 19 System and Optical Networking Concepts
Chapter 20 Fiber System Standards
Chapter 21 Single-Channel System Design
Chapter 22 Optical Networking System Design
Chapter 23 Global Telecommunications Applications
Chapter 24 Regional and Metro Telecommunications
Chapter 25 Local Telephone or "Access" Networks
Chapter 26 Internet Access and Local-Area Networks
Chapter 27 Video Transmission
Chapter 28 Mobile Fiber-Optic Communications
Chapter 29 Fiber-Optic Sensors
Chapter 30 Imaging and Illuminating Fiber Optics
APPENDIX A Important Constants, Units, Conversion Factors, and Equations
APPENDIX B Decibels and Equivalents
APPENDIX C Standard Time-Division Multiplexing Rates
APPENDIX D ITU Frequencies and Wavelengths for L- and C-bands, 100-GHz Spacing, 100 Channels
APPENDIX E Laser and Fiber Safety
APPENDIX F Fiber-Optic Resources

Citation preview

Understanding Fiber Optics Fifth edition, revised

Jeff Hecht LaserLight Press Auburndale, Massachusetts

Library of Congress Cataloging-in-Publication Data Hecht, Jeff Understanding fiber optics I [Jeff Hecht].-5th ed. p. cm. Includes index. 1. Fiber optics.

I. Title.

TA1800.H43 2006 621.36'92-dc22

Copyright ©2015, 2006, 2002, 1999, 1993, 1987 by Jeff Hecht Published by Laser Light Press, 525 Auburn St., Auburndale, Massachusetts 02466 USA Previously published by Pearson Education, Inc.

All rights reserved. Printed in the United States of America. This publication protected by Copyright and permission should be obtained from the publisher prior to any prohibited reproduction, storage in a retrieval system, or transmission in any form or by any means, electronic, mechanical, photocopying, recording, or likewise. For information regarding permissions, write to Rights and Permissions Department, Laser Light Press, 525 Auburn St, Auburndale, MA 02466, USA


Preface to the Laser Light Press Edition About This Edition Except for this preface, the front matter, and the errata that follows, this Laser Light Press edition reprints the fifth edition of Understanding Fiber Optics published in 2006 by Pearson Education, Inc. I am planning a sixth edition, but because that will take a while to prepare and with the Pearson edition is no longer available, I am reprinting the fifth through Laser Light Press. It may not cover the cutting edge of fiber optics, but it does cover the fundamentals you need to understand the field. This edition also is an experiment. I want to see how reducing the book's price will affect sales and make Understanding Fiber Optics more accessible to students. Thus Laser Light Press offers a low-cost PDF electronic version and a relatively inexpensive print-on-demand paperback. The many diagrams make an e-reader version more difficult. Whether you are an instructor, a student or a general reader, I would appreciate your comments and suggestions. If you are teaching a course based on the book, please contact me at [email protected] for an instructor's manual. You can find more information on the book's status and on associated material at through http://www.understandingfiberoptics.com or through http://www.jeffhecht.com.

About Fiber Optics Fiber optics has come a long way since I wrote the first edition of Understanding Fiber Optics in 1987. Optical-fiber communications was a radical new technology then, used mostly for high-capacity, long-distance transmission of telephone signals. I used a 1200-baud modem to send text messages from my computer through proprietary networks. Today a fiber-optic cable to my home provides a broadband connection to the Internet. A global network of fiber-optic cables links my phone and my computer to every continent except Antarctica, and a new cable is being laid through the Arctic Ocean. Fiber optics has revolutionized telecommunications in the same way the railroads revolutionized land transportation in the years my great-great-grandfather worked for one. Like the railroad business, the fiber-optic business has had its spectacular booms and busts. The telecommunications bubble brought dreams of riches, but the bust that followed left nightmares of ruin and grim jokes about the stocks of once high-flying companies. Yet the bubble and its aftermath are reminders that fiber optics is a technology that may be too good for its own good. Like the railroads and the Internet, fiber optics was something so good that the stock market wildly overvalued it; and like the Internet, fiber optics will be part of our future. I wrote the first edition of this book mainly for self-study, but it is now used widely in classroom settings. My goal is to explain principles rather than to detail procedures. When you finish, you should indeed understand fiber optics. You should be able to understand what the field is all about, comprehend what you read in trade journals such as Lightwave or Laser Focus World, make sense of what people in the field are saying, and explain fiber optics to your Aunt Millie or your niece. You won't be ready to design a brand new system, but you will be literate in the field.


  Think of it as Fiber Optics 101, a foundation for your understanding of a growing technology. To explain the fundamentals of fiber optics, I start with ideas that may seem basic to some readers; the details will follow. To make concepts accessible, I include drawings to show how things work, limit math to simple algebra, and step through some simple calculation to show how they work. I compare fiber optics with other common technologies and highlight similarities and differences. I have organized the book to facilitate cross-referencing and review of concepts, and made a point of adding a thorough index to make its contents accessible. I also include some information on the business side of the technology, and boxes that talk about key issues that the fiber-optics community needs to think about. The book introduces basic concepts first, then digs deeper into hardware and applications. The chapters are organized as follows: • The first three chapters are an introduction and overview. Chapter 1 tells how fiber optics are used and how the technology developed. Chapter 2 introduces optics, light, and the concept of light guiding. Chapter 3 introduces other basic concepts of communications and fiber-optic systems. They assume no background in optics or telecommunications. • Chapters 4 through 8 cover optical fibers, their properties, and how they are assembled into cables. The material is divided into five chapters to make it easier to digest. Chapters 4 through 6 explain the fiber concepts used in the rest of the book. Chapter 7 covers special-purpose fibers used in optical amplifiers and fiber gratings, photonic-crystal or microstructured fibers, and planar waveguides. Chapter 8 is an overview of cabling. • Chapters 9 to 12 cover laser and LED light sources including diode and fiber lasers, optical transmitters, optical detectors, receivers, optical amplifiers, and electro-optic regenerators. Chapter 12 compares and contrasts the operation of optical amplifiers and electro-optic regenerators. • Chapters 13 to 16 cover other components. Chapter 13 covers connectors and splices that join fibers. Chapter 14 covers optical couplers and other passive components in simple fiber systems and describes integrated optics. Chapter 15 covers optics that send signals at many separate wavelengths through the same fibers. Chapter 16 covers optical modulation and switching for optical networking. • Chapter 17 covers fundamentals of optical and fiber-optic measurements and explains the quirks of optical measurements. Chapter 18 describes fiber-optic testing. • Chapters 19 to 22 cover general principles of fiber communication. Chapter 19 describes fundamental concepts of fiber-optic systems and optical networking and how they work in practice. Chapter 20 describes communication standards. Chapter 21 outlines design of point-to-point single-wavelength systems, with sample calculations, so you can understand their operation. Chapter 22 describes the design of optical networks. • Chapters 23 to 27 explain how fiber optics fit into networks used for global and regional telephone and Internet transmission, cable television, and data networks. These chapters focus on different levels and aspects of the global network to keep concepts manageable. Chapter 28 covers special systems that don't fit elsewhere, such as networks in cars, military systems, and aircraft. • The final two chapters describe non-communication applications. Chapter 29 explains the principles and operation of fiber-optic sensors. Chapter 30 covers imaging and illumination with fiber optics. The glossary at the back of the book gives you quick translations of specialized terms and acronyms.

  v   Appendices tabulate useful information, including values of important physical constants, conversion factors, and a few key formulas. They're all in one place to make them easier to find. They also include an annotated list of resources, in addition to the suggestions for further reading in each chapter. So many resources are available on the Internet that I can't hope to compile a thorough list; I encourage you to use search engines creatively. I welcome your comments, questions, and suggestions at [email protected].

Acknowledgments Over the years many members of the fiber-optics community have given generously of their time to patiently answer my questions. I owe special thanks to John Jay, Shane Nipple, Craig Kegerise, Jerry Jackson, Eric Udd, Dana McEntire, and Joel Orban for feedback on draft chapters of this edition. Thanks to Kevin Able, Bill Chang, David Charlton, Marc Duchesne, Erich Dzakler, Robert Gallawa, Jim Hayes, Dennis Horwitz, Larry Johnson, Jim Masi, Nick Massa, Mike Pepper, Jim Refi, John Schlager, and Wayne Siddal for help on earlier editions and other material. Thanks to Jeffrey Rankinen, Pennsylvania College of Technology; Richard Windley, FCPI College of Technology; and Dave Whitmore, Champlain College for their helpful reviews. Any errors that remain are my own. This book draws on a series of articles on optical networking that I wrote for Laser Focus World. I think Steve Anderson for commissioning and editing them, Carol Settino for ably steering them into print, and the magazine's readers for feedback. I thank the Optical Society of America and SPIE - The International Society for Optical Engineering for inviting me to reach short courses based on Understanding Fiber Optics. I owe special thanks to the editorial and production staff at Pearson Education for their excellent work and their assistance in making this book possible. Thanks also to Lisa Cohen for updating me on the changing world of book publishing. Jeff Hecht, Auburndale, MA March 2015


Errata What have you learned? item 6 on page 35 should read: Refractive index (n) of a material is the speed of light in a vacuum divided by the speed of light in the material. It is always greater than 1.0 at optical wavelengths. Figure 14.12 on page 356 does not correctly show the operators performed on light of different polarizations in an optical circulator. Figure 15.1 on page 365 should have an * on 𝜆4 on the right side of the drawing to show that wavelength comes from the local transmitter at the bottom. Table A.3 on page 764 should give the value of Planck's constant in J-s (jouleseconds) or eV-s (electronvolt-seconds), not J/s or eV/s. The numerical values are correct, but the units are not.

Contents Chapter 1 An Introduction to Fiber Optics, 1 A Personal View: Ups and Downs, 1 * The Roots o f Fiber Optics, 2 • The Very Basics of Communications, 8 • Fiber Terms: Terminology and Units, 12

Chapter 2 Fundamentals of Fiber-Optic Components, 17 Basics o f Optics, 17 • Light G uiding, 26 • Fiber Transmission, 28 • Electro-Optics and O ther Components, 33 • Fiber-Optic Applications, 34

Chapter 3 Fundamentals of Communications, 39 Communications Concepts, 39 • Signals and Formats, 46 • Connectivity, 50 • Communications Services, 54 • The Business o f Telecommunications, 58

Chapter 4 Types of Optical Fibers, 65 Light G uiding, 65 • Step-lndex M ultim ode Fiber, 68 • Modes and Their Effects, 71 • G raded-lndex M ultim ode Fiber, 75 • Single-Mode Fiber, 7 7 • Dispersion-Shifted Single-Mode Fiber, 80 • Polarization in Single-Mode Fiber, 85 • O ther Fiber Types, 87

Chapter 5 Properties of Optical Fibers, 93 Fiber Attenuation, 93 • Light Collection and Propagation, 99 • Dispersion, 103 • N onlinear Effects, 1 1 5 * Mechanical Properties, 119

Chapter 6 Fiber Materials, Structure, and Manufacture, 127 Requirements for M aking O ptical Fibers, 1 27 • Glass Fibers, 128 • Fused-Silica Fibers, 130

• Plastic Fibers, 137 • Exotic Fibers and Light Guides, 140

Chapter 7 Specialty Fibers, 151 W h a t A re "Specialty" Fibers?, 151 • Dispersion-Compensating Fibers, 152 • Polarization-M aintaining Fibers, 153 • Bend-Insensitive and Coupling Fibers, 153 • Reduced-Cladding Fibers, 155 • Doped Fibers for Am plifiers and Lasers, 156 • Fiber G ratings and Photosensitive Fibers, 159 • Photonic o r "H oley" Fibers, 165 • Special Noncommunications Fibers, 166

Chapter 8 Cabling, 173 Cabling Basics, 173 • Reasons for Cabling, 174 • Types o f Cable, 178 • Elements of Cable Structure, 183 • Cable Installation, 190 • Cable Changes and Failure, 191

Chapter 9 Light Sources, 197 Light Source Considerations, 197 • LED Sources, 200 • The Laser Principle, 203 • Simple Semiconductor Lasers, 2 07 ’ Laser W avelength, 213 • Fiber Lasers, 2 19 • O ther Solid-State Laser Sources, 221

Chapter 10 Transmitters, 227 Transmitter Terminology, 2 2 7 • O perational Considerations, 228 • M ultiplexing, 232 • M odulation, 234 • Single-Channel Transmitter Design, 238 • Sample Transmitters, 241

Chapter 11 Receivers, 249 Defining Receivers, 249 • Performance Considerations, 258 • Electronic Functions, 265 • Sample Receiver Circuits, 267


Chapter 12 Amplification, Regeneration, and Wavelength Conversion, 275 Am plification and Regeneration, 275 • System Requirements, 279 • Repeaters and Regenerators, 280 • Optical Amplifiers, 281 • Erbium-Doped Fiber Amplifiers, 284 • Other Doped Fiber Amplifiers, 291 • Raman Am plification in Fibers, 292 • Semiconductor Optical Amplifiers, 295 • Optical Regeneration, 299 • Wavelength Conversion, 299

Chapter 13 Connectors and Splices, 307 Applications of Connectors and Splices, 307 • Fiber-to-Fiber Attenuation, 309 • Internal Reflections, 314 • M echanical Considerations in Connectors, 315 • Connector Structures, 3 17 • Standard Connector Types, 320 • Splicing and Its Applications, 326 • Splicing Issues and Performance, 3 27 • Types of Splicing, 328

Chapter 14 Couplers and Other Passive Components, 339 Coupler Concepts and Applications, 339 • Coupler Characteristics, 343 • Coupler Types and Technologies, 347 • Attenuators, 353 • Optical Isolators, 354 • Optical Circulators, 355

Chapter 15 Wavelength-Division Multiplexing Optics, 363 W D M Requirements, 363 • W D M Systems, 364 • Optical Filters and W D M , 370 • W D M Technologies, 375 • Building Multiplexers and Demultiplexers, 385

Chapter 16 Optical Switches, Modulators, and Other Active Components, 391 Defining Active Components, 391 • M odulators and M odulation, 392 • Switching in O ptical

Networks, 3 97 • O ptical Switching Technologies, 4 0 3 • Wavelength Switching and Conversion, 4 0 9 • Integrated Optics, 4 10

Chapter 17 Fiber-Optic Measurements, 417 Basics o f O ptical Power Measurement, 4 1 7 • Wavelength and Frequency Measurements, 4 25 • Phase and Interference Measurements, 428 • Polarization Measurements, 4 30 • Time and Bandwidth Measurements, 4 30 • Signal Q uality Measurements, 4 34 • Fiber-Specific Measurements, 4 36

Chapter 18 Troubleshooting and Test Equipment, 447 Fiber-Optic Troubleshooting, 4 4 7 • Test and Measurement Instruments, 4 50 • Troubleshooting Procedures, 462

Chapter 19 System and Optical Networking Concepts, 471 An Evolving Network, 471 • Telecommunication N etwork Structure, 4 73 • Transmission Topologies, 4 7 5 • Directing Signals, 481 • Signal Formats, 4 83 • Transmission Capacity, 4 8 7

Chapter 20 Fiber System Standards, 499 W h y Standards A re Needed, 4 9 9 • Families of Standards, 501 • Layers o f Standards, 502 • Interchange Standards, 5 0 7 • Fiber Transmission Standards, 5 0 9 • Current Standards Issues, 513

Chapter 21 Single-Channel System Design, 521 Variables, 521 • Power Budgeting, 523 • Examples o f Loss Budgeting, 528 • Transmission C apacity Budget, 534 • Cost/Performance Trade-offs, 541


Chapter 22 Optical Networking System Design, 549 O ptical Networking Concepts, 5 4 9 • O ptical Channel Density, 5 50 • O perating Ranges of W D M Systems, 5 55 • Factors in W D M Design, 5 5 7 • O ptical Am plification and W D M Design, 5 62 • Switching and O ptical N etworking, 563 • Design Examples, 566

Chapter 23 Global Telecommunications Applications, 573 Defining Telecommunications, 5 7 4 • The G lobal Telecommunications N etwork, 5 7 7 • Internet Transmission, 5 82 • Submarine Cables, 585 • Long-Haul Terrestrial Systems, 5 94 • Types o f Long-Distance Services, 598

Chapter 24 Regional and Metro Telecommunications, 605 Defining Regional and Metro Telecommunications, 605 • Regional Distribution, 6 06 • Regional Telecommunications Networks, 6 10 • Metro Networks, 612 • R egional/M etro Services and Equipment, 614

Chapter 25 Local Telephone or "Access" Networks, 623 Structure o f the Local Phone Network, 623 • Subscriber and Access Services, 6 30 • Emerging Services and Competing Technologies, 6 32 • Fiber to the Home or Premises, 636

Chapter 26 Internet Access and Local-Area Networks, 651 Data and Voice Transmission, 651 • The Internet and Its Structure, 654 • Data Transmission Technologies, 6 6 0 • Fiber Data-Link Design, 665 • Fiber in Standard Data Networks, 665

Chapter 27 Video Transmission, 677 Video Basics, 6 7 7 • Transmission M edia, 684 • Cable Television Architecture, 686 • HDTV and Cable, 691 • O ther Video A pplications, 692

Chapter 28 Mobile Fiber-Optic Communications, 699 M obile Systems, 6 99 • Remotely Controlled Robotic Vehicles, 7 00 • Fibers in Aircraft, 703 ’ Shipboard Fiber-Optic Networks, 705 • Automotive Fiber Optics, 706

Chapter 29 Fiber-Optic Sensors, 713 Fiber-Sensing Concepts, 713 • Fiber-Optic Probes, 7 14 • Fiber-Sensing Mechanisms, 716 • Some Fiber Sensor Examples, 719 • Fiber-Optic Gyroscopes, 7 22 • Smart Skins and Structures, 7 24

Chapter 30 Imaging and Illuminating Fiber Optics, 729 Basics o f Fiber Bundles, 7 2 9 • Optics of Bundled Fibers, 7 34 • Imaging Applications, 7 3 7 • Light Piping and Illum ination, 740

Appendices, 745 A ppendix A : Important Constants, Units, Conversion Factors, and Equations, 745 • A ppendix B: Decibels and Equivalents, 749 • A ppendix C: Standard Time-Division M ultiplexing Rates, 751 • A ppendix D: ITU Frequencies and Wavelengths for L- and C-bands, 100-G H z Spacing, 100 Channels, 753 • A ppendix E: Laser and Fiber Safety, 755 • A ppendix F: Fiber-Optic Resources, 7 5 7

Glossary, 761 Index, 775

book is dedicated to the memory of Heather Williamson Messenger, gifted editor, goodfriend, and victim o f domestic violence.

An Introduction to Fiber Optics


About This Chapter This chapter is a starting point to look around and see where you’re going before you dig inro details. The goal is to put fiber optics and communications into context and show how they go together. I start with a personal commentary about the turbulent times o f the past several years, then explain the plan for this book. A brief history of fiber optics follows, which introduces some important concepts. Then a brief history of communications explains the need for bandwidth and how fiber optics filled that need, perhaps too well. Finally, I explain some o f the terminology o f the field to help you in your looking about.

A Personal View: Ups and Downs Fiber optics has come a long way in the nearly three decades I’ve been watching its development. For many years the field grew steadily, with new technology creating new applications, and new applications, in turn, supplying money to develop more new technology. The growth sped out o f control in the late 1990s as the Internet fed a seemingly limitless thirst for bandwidth that only optical fibers could provide. The boom turned into a bubble, and the bubble into a bust as I watched in amazement. We knew the bubble was too good to be true, but none o f us wanted it to end. We told ourselves that the communications industry was in better shape than the dot-coms because it had real hardware, not just web sites. Then the industry ran right off a cliff and landed with an ugly splat. We traded grim jokes, noting that we would have done better to invest in cases o f beer and return the empties in a state with a bottle-deposit law. Employment dropped nearly as badly. The industry seemed a vast, smoking crater.

Fiber revolutionized telecommunications by supplying tremendous bandwidth.


Chapter 1

• Fiber-optic technology remains



That depressing view is as much o f an exaggeration as was the euphoric overenthusiasm 0f tbe bubble. We’ll never see that manic growth again, and that’s just as well. But fiber-optic technology remains healthy, with advances continuing at a more sober rate. Fiber optics has become the backbone o f the global telecommunications network, giving us instant access to Web sites and telephones around the world. That network continues to reach toward homes and businesses. Cable television companies, telephone companies, Internet providers, and power companies have their own fiber-optic networks. When you use a cell phone, your calls usually go wireless only to the tower, where a fiber-optic cable runs to the backbone telephone network. The demand for bandwidth continues to rise, although there’s a lot o f surplus fiber in the ground right now. Fiber revolutionized telecommunications in the twentieth century, just as the railroads revolutionized transportation in the nineteenth century. Overbuilding o f railroads caused spectacular busts in the latter half o f the nineteenth century, but railroads remained the backbone of the national transportation network until the spread o f the interstate highway system in the 1950s and 1960s. Railroads still carry people and freight today— especially in Europe. The fiber-optic gold rush is over, and the field has had a roller-coaster ride o f dramatic ups and downs. We’ve gained some experience and a few gray hairs in the process, but we’ve survived. Fiber has carved itself a vital niche in the communications world and will play a growing role around the world as other countries expand their own communications networks. Fiber is here to stay.

The Roots of Fiber Optics Fiber optics did not begin as a communications technology. Optical fibers evolved from devices developed to guide light for illumination or displays, and were first used to look inside the human body. Bundles o f optical fibers are still used to examine the stomach and the colon because they can reach into otherwise inaccessible areas. It’s worth looking at how this idea began— it will teach you the basic ideas o f light guiding in a fiber.

Piping Light Light normally goes in straight lines, but sometimes we want it to go around corners.

Think o f optical fibers as pipes that carry light. Lenses can bend light and mirrors can deflect it, but otherwise light travels in straight lines. The working o f optical devices, from our eyes to giant telescopes and sensitive microscopes, depends on light going in straight lines. Yet sometimes it is nice to be able to pipe light around corners and look into inaccessible places. The first steps in that direction were taken in the nineteenth century. In 1880, William Wheeler, a young engineer from Concord, Massachusetts, filed for a patent on a way to pipe light through buildings. Thomas Edison had already made the first incandescent light bulbs but hadn’t gotten all the bugs out. Wheeler wanted to distribute light from an electric arc, a light source that was better developed at the time, but was blindingly bright. He planned to put arc lamps in the basements o f buildings and

An Introduction to Fiber Optics

{I* ItfeL)



Ho. 2 4 7 ,2 2 9 .

Patented Sept. 2 0 ,1 8 8 1 .

FIGURE 1.1 W heeler’s plan fo r pipin g light into rooms (U.S. Patent 247,229).

distribute the light to distant rooms through a set of pipes coated with a reflective layer inside, as shown in Figure 1.1. Diffusers at the ends o f the pipes would spread the light out inside each room. Wheeler was a solid engineer who became an expert in designing water works. He later founded a successful company that made reflectors for street lamps. His design was logical at the time since air seemed to be a much clearer medium than any known solid. But his light pipes never caught on, and Edison’s incandescent bulbs eventually worked much better than arc lamps.

Total Internal Reflection Even before Wheeler’s time, scientists knew how to trap light inside a solid. A phenomenon called total internal reflection, described in Chapter 2, can confine light inside glass or other transparent materials. This phenomenon involves sending light through the material in such a way that it strikes the surface exposed to air at a glancing angle. Then the light is reflected back into the solid. You can see the effect in diamond or cut glass, in which one surface acts like a mirror to reflect light to your eye. Glassblowers may have been the first to realize this effect could guide light along a bent glass rod, but it wasn’t widely recognized until 1841 when a Swiss physicist, Daniel Colladon,

Total internal reflection can guide light along a glass rod or water jet.

FIGURE 1.2 Light guided down a water jet.

Light beam becomes more diffuse as it passes down the water jet, because turbulence breaks up surface.

adapted the trick for a jet of flowing water in his popular science lectures. Figure 1.2 shows how he directed a bright light down a horizontal pipe leading out o f a tank o f water. When he opened the spout, water flowed out in a jet and the pull of gravity bent the water jet into a parabolic arc. Total internal reflection trapped the light inside the water jet. The light beam bounced off the top surface, then off the lower surface, until turbulence in the flowing water broke up the beam. Others borrowed the idea for their own demonstrations. The Paris Opera used it on stage in 1853. The great Victorian exhibitions o f the 1880s adapted the idea to make illuminated fountains that fascinated fairgoers who hadn’t seen bright artificial lights. But the water jet remained essentially a parlor trick of little practical use.

Glass Light Guides and Imaging ^ An optical fiber guides light like a very thin

glass rod.

Inventors soon adapted the idea of guiding light to more practical purposes. By the early 1900s, they had patented a scheme for guiding light through a bent glass rod to illuminate the inside of the mouth for dentistry. This technique was much better than sticking a gas lamp into a patient’s mouth, but it was far from perfect. Illuminated tongue depressors followed, A fine glass fiber is actually a very thin, flexible rod, so it can guide light in the same way. Assemble glass fibers into a bundle, and they can carry an image from one end to the other,

An Introduction to Fiber Optics

as you will learn in Chapter 30. Clarence W. Hansell, an American electrical engineer and prolific inventor, was the first to take this logical step and patented the idea in the late 1920s. Hansell thought the bundles could be used for inspecting out-of-the-way places, for medical applications, or even for a facsimile machine. Heinrich Lamm, a German medical student, made the first image-transmitting bundle in 1930 and was able to photograph the bright filament o f a lamp. He combed the fibers to align them, but the bundle didn’t work well because it consisted of bare fibers, in which total internal reflection was at the surface exposed to the air. Light can easily leak through that surface if anything touches or scratches it, and the fibers inevitably touched and scratched each other in Lamm’s bundle. Light even leaked out at places where fingerprint oil was smudged on the surface. Neither Hansell nor Lamm got very far. The same problems bedeviled other men who independently invented fiber bundles for imaging in the early 1950s. These men were a Danish inventor, Holger Moller Hansen, two eminent optics professors, Abraham van Heel and Harold H. Hopkins, and Hopkins’ student, Narinder Kapany. Solving the problem required a fresh look at the requirements for total internal reflection. We normally think o f it as occurring where light is unable to enter the air, but what really matters is a quantity called the refractive index, which you’ll learn about in Chapter 2. Total internal reflection can happen when light travelling in one medium tries to enter another medium with a lower refractive index. Air has a much lower refractive index than glass, but the difference does not have to be large. Oils, beeswax, and many plastics have refractive indexes that are higher than air but lower than glass. Coat the glass fiber with one o f those materials, and total internal reflection can still occur, but the surface is protected from scratches, fingerprints, and leakage o f light into other glass fibers, as shown in Figure 1.3. Moller Hansen tried coating a fiber with margarine, but the results were impractically messy. Brian O ’Brien, a noted American optical physicist, suggested the idea to van Heel, who coated his fibers with plastic and beeswax, which were more practical. In December 1956, Larry Curtiss, an undergraduate student at the University of Michigan, slipped a rod of glass with high refractive index into a tube of glass with lower index and made the first glass-clad fiber. The technology has been refined considerably since then, but glass-clad fiber remains the most common type. Fiber bundles were the key to making flexible endoscopes, gastroscopes, and colonoscopes to examine the throat, stomach, and colon. Other imaging applications soon emerged, as described in Chapter 30. Fiber bundles are also used for illumination; however, this technology has been largely eclipsed by fiber-optic communications.

Clad fibers were the key development in making fiber-optic imaging practical.

The first pr applicatic fiber optic gastrosc

Optical Communications Optical fibers aren’t necessary for optical communications. People have communicated using light since ancient times. The ancient Greeks lit signal fires on hilltops to relay news of the fall o f Troy. The first “telegraph” was an optical one, invented by French engineer Claude Chappe in the 1790s. Operators relayed signals from one hilltop telegraph tower to the next by moving semaphore arms. Samuel Morse’s electric telegraph put the optical telegraph out of business, but it left behind countless “telegraph hills.”

An optical telegraph was invented in the 1790s in France.

Chapter 1

FIGURE 1.3 Light cannot leak out o f clad fibers i f they touch another surface.

Invention of the laser stimulated interest in optical communications.

Charles Kao and George Hockham proposed fiber-optic communications.

After inventing the telephone, Alexander Graham Bell turned to sending voices through the air on beams o f light, demonstrating his “photophone” in 1880. Bell was elated and considered it his greatest invention. The photophone, however, never proved practical, and radio waves eventually provided the means for wireless communications. Other attempts at optical communications followed, but few people took them seriously until Theodore Maiman made the first laser in 1960. The laser generates a tight beam o f coherent light at a single pure wavelength. It’s the optical equivalent o f the pure carrier frequency that is modulated with a radio or television signal. Its coherence made it very attractive for optical communications, and within a few months Bell Laboratories had made their own laser and used it to send light pulses between two towers 25 miles apart. However, other experiments soon showed that fog, rain, snow, and haze could block signals. Bell Labs tried sending laser beams through hollow light pipes, but they didn’t work well either. Optical fibers were available at this time, but they couldn’t send light very far. The clearest fibers used for medical endoscopes lost half o f the light they carried after three meters (10 feet). Go 30 meters (100 feet) and just 0.1% o f the light remains. That loss was acceptable for examining the stomach through several feet o f fiber, but it made optical fiber useless for communications. Two young engineers at Standard Telecommunications Laboratories in England, Charles K. Kao and George Hockham, took a different approach. Instead o f asking how clear was

An Introduction to Fiber Optics

the best fiber, Kao asked what was the fundamental lower limit on the loss o f glass. He and Hockham found that most o f the loss in the glass was caused by impurities, not by the glass itself. In 1966, they predicted that a fiber made o f highly purified glass would be so clear that 10% o f the light entering it would remain after 500 meters (1600 feet). This level o f purity sounded unattainable, but a few laboratories around the world tackled the problem. Kao and Hockham turned out to be too conservative and Robert Maurer, Donald Keck, and Peter Schultz of the Corning Glass Works beat their prediction in 1970. Two years later they had fibers in which 10% o f the light remained after 2.5 kilometers (8000 feet). Better fibers followed and in today’s best optical fibers 10% of the light remains after passing through 50 kilometers (30 miles) o f fiber. That exceptionally low loss lets fibers carry signals much further than copper wires.

Other Fiber Properties Long-distance transmission isn’t the only thing that matters in communications. It’s also important to be able to carry a lot o f information, which in the communications world is called bandwidth. The more bandwidth, the more information a signal can carry. For reasons we’ll explore later, optical signals inherently have a very high bandwidth, which is why the laser first interested communications engineers. Equally important, optical fibers can transmit those signals without seriously limiting their bandwidth. That’s not true for copper wires. Electronic devices can generate signals at high frequencies, carrying lots o f information, but copper cables tend to attenuate those essential high frequencies, so the signals can’t go far. Telecommunications fibers are made o f glass, but they’re not as fragile as they sound. Glass is inherently a strong material as long as it’s not cracked, but it is brittle in bulk. Communication fibers, however, are flexible because they’re relatively thin. Optical fibers are often compared to a human hair. The sizes are close, but fibers are stiffen On a microscopic scale, a well-made optical fiber is also much smoother than a human hair. The cross-section o f a typical communications fiber is shown in Figure 1.4. The glass fiber itself is 125 micrometers (0.005 inch) thick, with the light-guiding core a central

FIGURE 1.4 Cross-section o f a typical communications fiber.

Optical fibers have very high bandwidth.

Glass fibers are inherently strong, allowing their use in outdoor cables.

Chapter 1

region about 9 |Jim in diameter. A plastic coating covers the entire fiber, bringing its overall size to 250 |xm, or 0.25 mm (about 0.01 inch). These dimensions, like those of other fiber-optic components, are usually given in metric units. The plastic layer protects the fiber surface from scratches and microcracks that could cause it to break. The result is a fiber that’s much stronger than you would expect. It’s very hard to snap a fiber with your hands, although you can break one if you trip on it while wearing heavy shoes. Communications fiber works perfectly well in cables used in harsh outdoor environments. Fibers designed for other purposes have different properties, as you will learn in Chapters 4 through 7.

The Very Basics of Communications The basic idea o f communications is very simple: to transmit information from one point to another. Both the technology and the business o f communications are much more complicated, and you should understand a bit about both before you dig deeper into this book. Chapter 3 will give you a more formal introduction to communications, but here you will get some idea o f how communications in general, and fiber communications in particular, really work.

Communications Technology Telecommunications means communications over a distance.

The word “communications” is used in many different ways. We say we are trying to “communicate” with someone when we convey a message in words. Colleges have “communications” departments that focus on writing and broadcasting, not engineering. Here we’re talking about telecommunications, which means communications over a distance by means o f radio waves, electrical signals, or optical signals. Today’s telecommunications are based on electronic technology. Electronic devices generate and process signals. When you talk on the telephone, circuits in the phone convert your voice to electrical signals. If you’re talking on a corded phone, those electrical signals pass along wires into the phone company’s network. If you’re talking on a cell phone, the phone converts its electrical signals into radio signals and transmits them to a tower, which sends them to the telephone network through wires, optical fibers, or radio links. The Internet and cable television systems work similarly. The networks are vast and complex. You’ll learn more about basic concepts in Chapter 3 and learn some details in Chapters 23 through 27. The optics in a fiber-optic network have to talk with the electronics. The basic idea is shown in Figure 1.5. The input signal drives a light source, modulating its intensity. If the input signal is a series o f bits, it turns the light source off and on. In practice, the light source is part o f an optical transmitter. The optical transmitter contains electronic circuits that process the signal so it drives the light source properly, but we won’t worry about those details now. The light then leaves the source and enters an optical fiber, which carries it to a receiver. The receiver converts the light back to electronic form to drive devices on the other end. We’ll cover the details later. This simple example shows transmission between a pair of points. A system that performs that job is often called a lin k between the points. A link that carries digital data is called a data link.

An Introduction to Fiber Optics

FIGURE 1.5 Elements o f a sim ple fiber-optic system.

The telecommunications network is considerably more complex and uses different technologies at different levels. Like a network o f roads or streams, the telecommunications network starts out small and packs more traffic onto main arteries. Fiber optics are the backbone o f these networks, the superhighways that carry the heaviest traffic along the busiest routes. Today, fiber-optic links reach into virtually every community in the United States, although they carry different amounts of traffic. Wireless transmission over radio or microwaves is used for some links, largely for mobile phones and broadcasting from ground antennas or satellites. Plain copper wires and coaxial cables are used for other links. Both copper and wireless usually carry less traffic than fibers. The basic functions of the telecommunications network are to transmit and distribute information. Think of these two functions as being performed by pipes and switches, as shown in Figure 1.6. The pipes carry information from its source to its destination. The switches direct the information through the proper pipes. The principles are the same whatever the network is, and in some cases, such as broadcast television, there isn’t much

FIGURE 1.6 A communications system consists o f pipes that transmit signals an d switches that direct them to their destinations.

The telecommunications network distributes and transmits information.


Chapter 1

switching involved. The same signals go through one big pipe— radio waves transmitted through the air— to everybody. The pipes are called transmission m edia and include optical fibers, copper wires, satellites, and broadcasts through the atmosphere. The switches direct the signals through particular pipes. As you’ll learn later, different networks use different types o f switches (often specialpurpose computers). Most switches are electronic, but some optical switches have been developed. This book is about fiber optics, so it concentrates mainly on optical systems, although it does describe some electronics.

The Business of Telecommunications

M ajor costs are capital expense and operating expense.

Wavelengthdivision multiplexing sends signals at different wavelengths through one fiber.

Telecommunications is a big business with billions o f dollars at stake. It used to be a quiet, orderly business run by government agencies or private monopolies that had to comply with detailed sets o f rules and regulations. The rules divided their turf. The telephone monopoly couldn’t carry video signals, and the cable television monopoly couldn’t carry telephone signals. Then governments began to change the rules, new technology arrived on the scene, new competitors appeared, and the picture became very complicated. Most telecommunications systems today are operated by businesses that make money from charging customers for their services. Their expenses fall into the usual business categories such as providing service, sales, marketing, and overhead. The most important costs in running a network are operating expense or “opex” and cap ital expense or “capex.” Operating expense covers day-to-day operations, the salaries o f equipment operators, technicians and managers, maintenance, and repairs. Some work is done on company sites; other work is done elsewhere and requires sending a technician to a remote site (a “truck roll’’ in industry jargon). Capital expense is buying and installing new equipment. Network operators can trade one expense off against the other, investing capital expense in an expensive new automated system that automatically performs functions to reduce operating expenses. Government agencies that provide communications services to the public handle their expenses in a similar manner, but they aren’t expected to make a profit. Communications companies decide what equipment to use by estimating how much it will cost and what new revenue it will bring in, and balancing these considerations against other budgetary priorities. The decision process can get quite complex, and installation costs can play a big role. Companies have to pay for rights o f way, and construction in downtown urban areas can be extremely expensive. Some of the earliest fiber-optic systems were installed because the small cables fit much better into crowded underground ducts in big cities than did thicker copper cables. The fiber-optic equipment cost more, but the overall project cost less because it didn’t require new construction. The major advantage o f fiber has always been its ability to transmit signals at higher speeds and over longer distances than other transmission media. The demand for bandwidth has increased with the steady growth in telephone and video traffic, and the explosive growth in data traffic over the Internet. Engineers steadily increased the volume of data that fiber systems could transmit. The most explosive growth came in the past decade, as a new technique called wavelength-division m ultiplexing (W DM ) made it possible to

An Introduction to Fiber Optics

transmit separate signals through the same fiber at many wavelengths. W D M multiplies the capacity o f individual fibers. The idea is similar to transmitting signals through the air at many separate radio frequencies, which allows many radio and television stations to transmit simultaneously to homes. Developers also began talking about optical networking, in which signals would be routed around the country optically without being converted into electronic form. Eventually investment in fiber optics got out o f hand, creating an economic “perfect storm,” which survivors call “the bubble.”

Understanding the Bubble The rapid growth o f the Internet and the proliferation of dot-com companies started pumping up the bubble in the late 1990s. A number of factors magnified its impact on the telecommunications industry. For decades, the conventional wisdom o f telecommunications engineers had been that you can’t have enough bandwidth. Networks are designed around the assumption that everyone is not on the phone at the same time. The long-distance network can’t handle all the country’s phone lines at once, which is why you sometimes can’t get a long-distance line on Mother’s Day. As Internet demand started rising, telecommunications carriers installed more fiber to handle the demand. Extra fibers are cheap compared to construction projects, so the carriers added lots o f fibers to make sure they had extra capacity in the future. They also planned on using wavelength-division multiplexing to multiply the capacity o f every fiber. Nobody thought that they could ever have too much bandwidth. But nobody fully grasped how fast the Internet was growing. For a brief period around 1996, Internet traffic seemed to be doubling every three months. Worldcom kept quoting that number for years until it became an Internet myth, widely believed although its origins were dubious. Telecommunications carriers looked at that tremendous growth rate and decided they’d need more fibers to handle the projected traffic. They didn’t know that in reality Internet traffic was doubling only once every year, a fact hard to ascertain because the traffic was divided among many different carriers. A gold rush started in technology stocks, focusing on the Internet, as new companies reported fast growth in sales. In reality, many of the numbers were fudged, and others looked deceptively large because they were starting from zero. Market analysts made wild projections of fantastic growth. A few people made fortunes selling promising new companies. Even when the first dot-coms began to fail, telecommunications looked good compared to selling dog food on the Internet, as I heard leading market analyst John Ryan say at a conference. Venture capitalists kept pumping money into the optical industry. New companies popped up from nowhere. Some had solid ideas, but others seemed to have little more than fancy trade-show booths and a pitch to investors. Yet their valuations kept rising with the growth of the bubble. It was tempting to believe that this growth was real and everyone would get rich. It wasn’t. No one had realized how much market projections or corporate profits had been inflated by wishful thinking and fraud. When the collapse came, it was disastrous. Companies either folded or shrank to shadows o f their former selves. Many fibers installed

Bandwidth was traditionally scarce in telecommunications.

The bubble badly hurt telecommunications and fiber optics.

Chapter 1


The Problem of the Bubble The bubble created a tremendous amount o f paper wealth that largely evaporated in its aftermath and left the telecommunications industry in shambles. Fortunes were made by a few people who sold stock near the peak; many more saw their holdings shrink. The development of new technology requires investment, and the bubble distorted the entire pattern of investment. Cumulative losses are estimated to be in the hundreds o f billions o f dollars. Allegations o f

fraud abound, going far beyond the handful o f criminal charges filed and regulatory actions taken. Both crooks and fools played roles, but we may never know which was more important, nor how we can avoid future bubbles. The historical parallels with the rise o f the railroads during the nineteenth century are striking. You might find it illuminating to read some of the excellent books that chronicle the age o f robber barons and the lives o f railroad tycoons such as Cornelius Vanderbilt, James J. Hill, Jay Gould, and Jim Fisk.

during the bubble remain unused. The industry is still nursing a massive hangover. Troubling questions remain about how telecommunications carriers can make a profit when the price o f bandwidth keeps dropping. Yet telecommunications remains a viable product, and the demand continues to grow. Fiber-optic technology is needed to meet that demand, and to help reduce costs so carriers can make a profit. Unused fibers are likely to be lit in the future as the demand for bandwidth continues to increase. That’s why it’s important to learn about fiber optics.

Fiber Terms: Terminology and Units The appendixes include a glossary, tables listing important units, and other useful data. Many terms are standardized or widely accepted, but others are not. The communications industry is notorious for its many cryptic acronyms and sometimes puzzling buzzwords. I have tried to avoid unclear terms and all but the most widely accepted acronyms. I do use some designations set by international standards organizations, such as types o f optical fiber specified by the International Telecommunications Union (ITU ), because these labels have specific meanings and are widely used in the industry. Terms are explained the first time they appear. The terminology will continue to evolve as the field grows and changes. I try to avoid proprietary terms. Many companies develop their own terminology, and different companies often have different names for the same technology. I do use a few trade names or trademarked terms that are widely used or are descriptive; they are capitalized as proper names to reflect their status. Every writer has their own terminological preferences. I particularly despise meaningless market-speak, such as calling a product or system a “solution,” because it tells nothing about what the thing is. I also prefer to spell out whole words rather than resort to acronyms. The latter conviction comes from reading too many specialized magazines that

An Introduction to Fiber Optics

don’t communicate clearly to readers who are not experts in the field. The acronyms I do use are well accepted. In this introductory chapter, I have used both metric and Imperial units to help you get started. In the rest o f the book I give virtually all measurements only in the metric units that are used throughout the telecommunications industry. You should get used to those units. Standard dimensions for most devices— starting with the fiber itself— are quoted in metric units. The American fiber industry uses Imperial units in only a few cases, usually for the lengths o f cable runs. Appendix A lists the common metric prefixes for units. Because this book is published in the United States, it uses standard American spelling such as “meter” and “fiber” with few exceptions. The only important exception is in the “Fibre channel” set of standards for data transmission.

What Have You Learned? 1. Fiber optics has revolutionized telecommunications by supplying tremendous bandwidth, which previously was in short supply.

. . 4 . Total internal reflection can guide light along a glass rod or water jet. An optical 2 Fiber-optic technology remains healthy, but the business has suffered problems.

3 Light normally goes in straight lines, but optical fibers can guide it around corners. fiber guides light in a manner similar to a very thin glass rod.


5 Clad fibers were crucial in making fiber-optic imaging practical for examining the stomach and colon. The first practical application of fiber optics was gastroscopy.

6. An optical telegraph was invented in the 1790s in France. It was replaced by the electrical telegraph. 7. Optical fibers have very high bandwidth and can transmit signals farther than copper wires.


Glass fibers are inherently strong, allowing their use in outdoor cables.


9 Telecommunications means communications over a distance. 10. The telecommunications network is made o f “pipes” and “switches” that distribute and transmit information. 11. W a v elen g th -d iv isio n m u ltip le x in g tra n sm its m u ltip le signals th ro u g h o n e fib er at d iffe re n t w aveleng ths. 1 2 . T h e te le c o m m u n ic a tio n s b u b b le serio u sly d isru p ted th e fib e r-o p tic s ind ustry.

13. Fiber-optic measurements are made in metric units, although American companies often measure cable lengths in Imperial units (feet or miles).

What's Next? In Chapter 2, you’ll learn basic principles o f physics and optics needed to understand fiber optics. Then you’ll learn basic fiber-optic concepts.

Fiber-optic measurements are metric.

Chapter 1

Further Reading On the evolution o f fiber optics: Jeff Hecht, City o f Light: The Story o f F iber Optics (Oxford University Press, 1999 and 2004) On the Internet bubble: K. G. Coffin and A. M. Odlyzko, “Growth o f the Internet,” in I. P. Kaminow and T. Li, eds., O ptical F iber Telecommunications IV B : Systems an d Im pairm ents (Academic Press, 2002), pp. 17-56; available online at http://w w w .dtc.um n.eduG odlyzko/doc/ o f. internet.grow th.pdf On the development o f communications in general: Arthur C. Clarke, How the World Was One: Beyond the G lobal Village (Bantam, 1992) Anton A. Huurdeman, The Worldwide History o f Telecommunications (Wiley InterScience, 2003) Irwin Lebow, Inform ation Highways & Byways: From the Telegraph to the 21st Century (IEEE Press, 1995) Laszlo Solymar, Getting the Message: A History o f Communications (Oxford University Press, 1999)

Questions to Think About 1. For a bundle o f optical fibers to transmit an image, the fibers must be arranged in the same pattern on both ends o f the bundle. What limits the size o f the smallest details that can be seen?


2 Devise an analogy using common implements found in a kitchen or cafeteria to show how a bundle o f fibers transmits an image.


3 Most o f the light lost in going through a glass window is reflected at the surface. Ignoring this surface reflection loss, suppose that a one-millimeterthick window absorbs 1% o f the light entering it and transmits 99% . Neglecting reflection, how much light would emerge from a one-meter-thick window?


4 If optical fibers transmit signals so much better than wires, why aren’t they used everywhere?


5 During the bubble years, many people in the industry thought Internet traffic was doubling every three months. In reality, it was doubling about every year. How much difference in growth o f Internet traffic would this make over a period o f five years?

6 . Why didn’t anybody wonder how long Internet traffic could continue doubling every three months?

An Introduction to Fiber Optics

Chapter Quiz 1 . Light can be guided around corners best in a. reflective pipes. b. hollow pipes with gas lenses. c. clad optical fibers. d. bare glass fibers.

2 . The first practical use o f optical fibers was a. in communications via optical telegraph. b. in Alexander Graham Bell’s photophone. c. to illuminate flowing jets o f water. d. in bundles to examine the inside o f the stomach.

3 . What is the principal requirement for a cladding on an optical fiber? a. It must have a refractive index lower than the core to produce total internal reflection. b.

It must be opaque so light

doesn’t leak out.


It must be made o f plastic

to keep the fiberflexible.


It must have a refractive index lower than thatof air.

4 . Flexible bundles of optical fibers

can be used to

a. examine the inside o f the stomach without surgery. b. examine the inside o f the colon without surgery. c. illuminate hard-to-reach machinery. d. all o f the above e. none o f the above

5 . A new automated control system costs $1 million. How much will it have to reduce annual operating expenses if company policy says the payback time has to be no more than four years? (Neglect interest rates.) a. $100,000 b. $250,000 c. $400,000 d. $500,000 e. $1 million


The elements of a fiber-optic data link must include a. light source, receiver, and fiber. b. light source and cable. c. fiber and receiver.

Chapter 1

d. fiber only. e. cable only. 7 . You need to install a new cable to handle four years o f growth on a transmission route. The traffic now fills one fiber, and traffic is doubling every three months. How many fibers will you need in four years? a. 4 b. 16 c. 128 d. 65,536 e. over 1 million 8 . As in Problem 7, you need to install a new cable to handle four years o f growth on a transmission route where traffic is doubling every three months. All the traffic now fits in a signal that requires one wavelength in a fiber that can handle 32 wavelengths with wavelength-division multiplexing. How many fibers will you need to carry that traffic if you fill each one with 32 wavelengths? a. 4 b. 16 c. 128 d. 2048 e. 65,536

9 . Reality has set in, and you realize that traffic is doubling every year. How many fibers would you need if each fiber carried only one signal and the first fiber was already full? a. 1 b. 4 c. 16 d. 128 e. 2048 10.

How many fibers would you need to handle the transmission load in Problem 9 if each fiber could transmit signals at 32 wavelengths? a. 1 b. 2 c. 4 d. 16 e.

can’t tell from data given

Fundamentals of FiberOptic Components

About This Chapter Fiber optics started as a branch of optics and evolved into a hybrid field that includes electronics and telecommunications. The basic concept behind a fiber is optical, and some single or bundled optical fibers are used as optical components. However, the most common use of fiber optics is in telecommunications, where many concepts originated in electronics and radio communications. Today, signals are converted back and forth between optical and electronic formats as they pass through the global telecommunications network. Fiber-optic transmitters and receivers are opto-electronic devices, part optical and part electronic. To understand fiber-optic communications, you need to learn about three fields: optics, electronics, and telecommunications. This chapter introduces optical and electronic concepts to lay the groundwork for understanding fiber-optic components. Chapter 3 covers communications systems. Later chapters explain particular devices and systems in more detail.

Basics of Optics Optics is the part o f physics dealing with light and its interaction with matter. The workings of optical fibers depend on optics, so you need to understand basic optical principles and how light interacts with matter. To prepare you, we will review these principles without going into great detail or length. Some parts o f this review may seem unnecessary, but read it anyway because later chapters assume you understand these fundamentals. From a physical standpoint, you can consider light to be either electrom agnetic waves or particles called photons. This is the famous wave-particle duality of modern physics.


Chapter 2

Light can be considered as electromagnetic waves, photons, or rays.

Both viewpoints are valid and valuable. Optical engineers are concerned with the path that light follows, so they often consider light as rays that follow straight lines between or within optical elements, bending only at surfaces. Each of these viewpoints can be useful at different times. The ray model o f light propagation represents how light passes through space and optical devices. Rays are easy to visualize; you can think of them as laser beams drawing straight lines in space. Yet light is not really made up o f rays; it’s made up o f electromagnetic waves or photons. You can learn other things about light by considering it to be waves or photons.

Electromagnetic Waves and Photons

A photon is a quantum of electromagnetic energy.

Viewed as an electromagnetic wave, light is composed o f electric and magnetic fields, which vary in amplitude as they move through space together at the speed of light, denoted c. The two fields are perpendicular to each other and to the direction in which the light travels, as shown in Figure 2.1. The amplitude o f each field varies sinusoidally, like a sine function in trigonometry, rising from zero to a positive peak, going back through zero, hitting a negative peak, then returning to zero. The distance that light travels during that complete cycle is called the wavelength. The usual symbol for wavelength is the Greek letter A. (lambda), and that’s one symbol you should remember. The number o f waves or cycles per second is called tht frequency, and it’s measured in hertz (after Heinrich Hertz, who discovered electromagnetic waves). Frequency usually is denoted by the Greek letter v (nu). Wavelength decreases as frequency increases, and waves can be measured by either. Many light sources such as red laser pointers emit continuous light waves, which oscillate steadily at the same frequency. You can think o f them as sine waves that go on for a very long time. Other sources emit pulses of light, and it’s useful to think of pulses of light as groups of photons. A photon is a quantum o f electromagnetic energy. It’s also a wave packet, a series of a few waves that build quickly to a peak amplitude, then fade back to nothing, as shown in

FIGURE 2.1 A light wave consists o f electric an d magnetic fields.

O n e W a v e le n g th

Fundamentals of Fiber-Optic Components

FIGURE 2.2 A single photon is a short p acket o f

Figure 2.2. Like a continuous wave, a pulse or wave packet has a wavelength and a frequency, but the wavelength and frequency are not as well defined as for a continuous beam. Thanks to the uncertainty principle, the shorter the pulse, the larger the uncertainty in wavelength. The amount o f energy carried by a single photon depends on the oscillation frequency: The faster the wave oscillates, the higher the energy. A continuous wave is a series o f photons, emitted one after the other. Each photon has a unit of energy set by the wavelength or frequency, so the total energy is the number o f photons times that photon energy. In wave terms, this is proportional to the wave amplitude squared.

The Electromagnetic Spectrum What we call “light” is only a small part o f the spectrum o f electromagnetic radiation. The fundamental nature of all electromagnetic radiation is the same. It can be viewed as photons or waves and travels at the speed o f light (c), which is approximately 300,000 kilometers per second (km/s), or 180,000 miles per second (mi/s). The difference between radiation in different parts o f the electromagnetic spectrum is a quantity that can be measured in several ways: as the length of a wave, as the energy of a photon, or as the oscillation frequency of an electromagnetic field. Figure 2.3 compares these three views. Each measurement— wavelength, energy, or frequency— has its own characteristic unit. The preferred unit depends on the part o f the spectrum. The optics world usually talks in wavelength, which is measured in meters, micrometers (p,m or 10 6 m), nanometers (nm or 10 9 m), or sometimes in angstroms (1A = 10 10 m). Don’t even think o f wavelength in inches. (If you absolutely have to know, 1 p,m is 0.00003937 in.) Frequency is measured in cycles per second (cps) or hertz (Hz), with megahertz (MHz) meaning a million hertz and gigahertz (GHz) meaning a billion hertz. (The metric system uses the standard prefixes listed in Appendix A to provide different units o f length, weight, frequency, and other quantities. The prefix makes a unit a multiple o f a standard unit. For example, a millimeter is a thousandth [10~3] o f a meter, and a kilometer is a thousand [ 103] meters.)

The light carried in fiber-optic communications systems can be viewed as either a wave or a particle.

Chapter 2


T h e O p tic a l S p e c tru m

Electro magnetic spectrum.

R a d io

X -R a y s

M ic ro w a v e s

v 's lb le _ * J L ig h t T l In fra re d F re q u e n c y (H z)

i 106

i 107

i 108

I I 1 1 0 9 1 0 '“ 1 0 "

I , p

G am m a R ays

Ultra tv io le t



I I I I i i i i i i i 1 0 12 1 0 13 1 0 14 1 0 16 1 0 16 1 0 17 1 0 18 1 0 19 1 0 20 1 0 21 1 0 22

P h o to n E n e rg y (e V )

W a v e le n g th


10-® 10 7 10-® 1 0 " 1 0 " 10-“ 10 2 10 ' I 1 —





I— 1






10~2 10-3 10 "




1 0 -5 10~8 10 "

I 10 0

I— 10 "

I I 10 3 10 4



I 105

I 10 6

I 10 7




10 -9 10 ~ 1° 10 ~ 1110 - 1210~13

Photon energy can be measured in many ways, but the most convenient here is in electron volts (eV)— the energy that an electron gains in moving through a 1-volt (V) electric field. All the measurement units shown on the spectrum chart are actually different rulers that measure the same thing. There are simple ways to convert between them. Wavelength is inversely proportional to frequency, according to the formula: wavelength


or c V

where c is the speed o f light, X is wavelength, and v is frequency. To get the right answer, all terms must be measured in the same units. Thus cmust be in meters per second (m/s), X must be in meters, and frequency must be in hertz (or cycles per second). Plugging in the approximate value o f c, we have a more useful formula for wavelength: 3 X 1 0 8 m/s

X — -----------------v You can also turn this around to get the frequency if you know the wavelength: 3 X 108 m/s

Not many people talk about photon energy (E ) in fiber optics, but a value can be gotten from Planck’s law, which states: E = hv

Fundamentals of Fiber-Optic Components

where h is Planck’s constant (6.63 X 10 - 34 J-s, or 4.14 X 10~ 15 eV-s) and v is the frequency. Because most interest in photon energy is in the part of the spectrum measured in wavelength, a more useful formula is 1.2399

E{eV) =


( |x m )

which gives energy in electron volts when wavelength is measured in micrometers (|xm). We are mainly interested in a small part o f the spectrum shown in Figure 2.3— the optical region, where optical fibers and other optical devices work. That region includes light visible to the human eye at wavelengths o f 400 to 700 nm and nearby parts o f the infrared and ultraviolet, which have similar properties. Roughly speaking, this means wavelengths of 200 to 20,000 nm (0.2 to 20 |xm). The wavelengths normally used for communications through silica glass optical fibers are 750 to 1700 nm (0.75 to 1.7 p-m) in the near infrared, where silica is the most transparent. Glass and silica fibers can transmit visible light over shorter distances, and special grades o f silica (often called fu sed quartz) can transmit near-ultraviolet light over short distances. Plastic fibers transmit best at visible wavelengths.

Fiber-optic communications systems transmit near-infrared light invisible to the human eye.

Wave Phase and Interference One important consequence o f the wave nature o f light is that light waves have a property called phase, which measures the progress o f the wave in its cyclical variation in amplitude. Figure 2.1 showed one complete cycle, in which the amplitude o f the electric and magnetic fields rises, falls, and returns to the starting point. Light waves from a continuous source repeat this cycle endlessly. Repeating this cycle is like going around a circle, and the phase is measured as an angle between 0° and 360°. The electric and magnetic fields depend on each other, so normally the phase is measured only for the electric field. Electromagnetic waves combine by adding their amplitudes. If you start with a pair of waves with the same wavelength and amplitude, and the peaks and valleys line up perfectly, their amplitudes add, producing an effect called constructive interference (shown in Figure 2.4). However, if the peaks o f one wave line up with the valleys o f the other wave, the sum o f the two intensities at any instant is zero, because one has a positive value and the other has a negative value o f the same amount. This case, also shown in Figure 2.4, is known as destructive interference. Destructive interference occurs when the two waves are 180° out o f phase. W ith intermediate phase shifts, the combined amplitude is between the peak o f constructive interference and the null o f destructive interference. (Because the wave repeats indefinitely, we only measure phase shifts within one cycle, between 0° and 360°.) We normally don’t see this interference effect because most light sources radiate light in all directions at a wide range of wavelengths. Turn on two light bulbs in a dark room, and the total intensity is the sum o f the two intensities. To see interference you need to combine two identical light waves so their amplitudes add or subtract. You can do this by passing light through a pair o f closely spaced slits. Light spreads out from each o f the slits,

Light waves add or subtract in amplitude depending on their relative phase.

Chapter 2

FIGURE 2.4 Constructive and destructive interference.

Constructive Interference

Destructive Interference

1 1 8 0 ° P h a s e S h ift


producing a pattern o f light and dark regions where the waves interfere constructively and destructively, as shown in Figure 2.5. (The pattern arises because the waves travel slightly different distances from the two slits.) If you’re familiar with the law o f conservation o f energy, you may wonder where the energy goes when the light waves interfere destructively. The energy doesn’t disappear; it’s just rearranged in space, appearing where the waves interfere constructively. The total

FIGURE 2.5 Interference o f light waves traveling slightly different paths produces bright an d dark stripes.

D e s tru c tiv e In te rfe re n c e (d a rk) C o n s tru c tiv e In te rfe re n c e (lig h t)

S lits

In te rfe re n c e P a tte rn

Fundamentals of Fiber-Optic Components

amount o f light in the interference pattern is the total passing through the two slits, but it’s not spread evenly.

Refractive Index The speed of light in a vacuum (c) is considered the universal speed limit. Normally nothing travels faster, although sometimes light can go a bit over the speed limit if it carries no information. Light always travels more slowly through transparent materials than through a vacuum. The speed difference for a material is measured by a number called the refractive index, denoted by the letter n in optics, which equals the speed o f light in a vacuum divided by the speed of light in the material:

Refractive index is the speed of light in a vacuum divided by the speed of light in a material.

^m aterial

The refractive index o f a vacuum equals 1.0 by definition. For normal optical materials, the refractive index is greater than 1.0 in the optical part o f the spectrum. (There are some peculiar exceptions you don’t need to worry about.) In practice, the refractive index is measured by comparing the speed of light in a material to the speed o f light in air rather than in a vacuum. This makes little practical difference because the refractive index o f air at normal pressure and temperature is 1.000293. Light changes speed as it goes from one material into another, such as from air into glass. This causes an effect we call refraction. To understand refraction, consider what happens to the peaks of light waves as they enter glass from air, as shown in Figure 2.6. The peaks of the waves line up in air, but when the waves hit the glass at an angle, some o f the light enters the glass while the rest remains in the air. The frequency o f the wave does not change as the waves slow down in the glass, so the wave takes the same time to complete a cycle, but it doesn’t travel as far between peaks in glass as it did in the air. The waves in air continue at the same speed until they reach the surface of the glass, where they also slow down. This process o f slowing down bends the path o f the light, as you can see in Figure 2.6. The same thing would happen if you braked the wheels on only one side o f your car; its path would turn toward the side where the wheels slowed. Figure 2.6 shows the wave view o f light, with a broad w avefront passing through the glass. In practice, it’s more useful to consider refraction from the ray viewpoint. The bold line in the figure represents the light ray, which bends at the surface of the glass. That ray shows how the path o f the light bends as light passes between the two media. The bending of light at a surface depends on the refractive indexes o f the two materials and the angle o f incidence at the surface. Both the angle o f incidence and the angle o f refraction of the transmitted light are measured from a line perpendicular to the surface called the normal. Snell’s law describes this bending: I = n. sin R where and nr are the refractive indexes o f the initial medium and the medium into which the light is refracted, and /and f?are the angles of incidence and refraction, respectively, as shown in Figure 2.6.

Refraction occurs when light changes speed as it goes between two materials.

Chapter 2

FIGURE 2.6 Refraction o f light entering glass. Ray W a v e le n g th in A ir

N o rm a l

G la s s


Total internal reflection occurs when light in a high-index material hits a boundary with a material of lower refractive index at a glancing angle.

Figure 2.6 shows the standard example o f light going from air into glass. The same thing happens in reverse when light emerges into air on the other side o f the glass. If the front and rear surfaces o f the glass are flat, light emerges at the same angle at which it entered, and the net refraction is zero, as when you look through a flat window. (The light is displaced a little bit, but we usually don’t notice that shift.) If one or both surfaces are curved, the light rays emerge at a different angle than when they entered the glass, and you see a net refraction or bending of the light rays, as if you were looking through a lens. Figure 2.7 shows the overall refractive effect. What does this have to do with fiber optics? Stop and consider what happens when light in a medium with a high refractive index (such as glass) comes to an interface with a medium having a lower refractive index (such as air). If the glass has a refractive index of 1.5 and the air an index of 1.0, the equation becomes 1.5 sin / = 1 sin R Instead of being bent closer to the normal, as in Figure 2.6, the light is bent farther from it, as in Figure 2 .8 . T his isn’t a problem if the angle o f incidence is small. For / = 30°, sin I — 0.5, and sin R = 0.75. But a problem does occur when the angle o f incidence becomes too steep. For / = 60°, sin / = 0.866, so Snell’s law says that

Fundamentals of Fiber-Optic Components

In c id e n t L ig h t \

] N o rm a l 11 11 i 11 11 i

A ir

Refraction through a window an d a lens.



G la s s


Light emerges from ! ! \ 11 \ a flat window at an 1 unchanged angle, ' 02 = but is bent by a lens.


Air R e fra c te d L ig h t 01

a. Window

sin R = 1.299. Your pocket calculator will tell you this is an error. That angle can’t exist because the sine can’t be greater than 1.0. Snell’s law indicates that refraction can’t take place when the angle o f incidence is too large, and that’s true. Light cannot get out o f the glass if the angle o f incidence exceeds a value called the critical angle, where the sine o f the angle o f refraction would equal 1.0. (Recall from trigonometry that the maximum value o f the sine is 1.0 at 90°, where the light would be going along the surface.) Instead, total internal reflection bounces the light back into the glass, obeying the law that the angle o f incidence equals the angle o f reflection, as shown in Figure 2.8. It is this total internal reflection that confines light in optical fibers, at least to a first approximation. As you will see in Chapter 4, the mechanism o f light guiding is more complex in modern communication fibers. The critical angle above which total internal reflection takes place, 0crit, can be deduced by turning Snell’s law around, to give 0crit = arcsin (nr/»,•) For the example given, with light trying to emerge from glass with n — 1.5 into air, the critical angle is arcsin (1/1.5), or 41.8°.

Chapter 2

Light here is refracted

FIGURE 2.8 Refraction an d total internal reflection.

Light Guiding Light is guided in the core of an optical fiber by total internal reflection at the boundary.

The two key elements o f an optical fiber— from an optical standpoint— are its core and cladding. The core is the inner part o f the fiber, which guides light. The cladding surrounds it completely. The refractive index o f the core is higher than that o f the cladding, so light in the core that strikes the boundary with the cladding at a glancing angle is confined in the core by total internal reflection, as shown in Figure 2.9. The difference in refractive index between core and cladding need not be large, and is less than 1% in most telecommunications fibers. For a 1% difference, corresponding to = 0.99, the critical angle, 0crit, measured from the normal is about 82°. That means light is confined in the core if it strikes the cladding interface at an angle of 8° or less, as shown in Figure 2.9. The upper limit measured from the interface is called the confinem ent angle, 0COnf> of the fiber.

Total Internal Reflection

FIGURE 2.9 Light guiding in an optical fiber.

L ig h t R a y C la d d in g

Fundamentals of Fiber-Optic Components

Light must fall inside this angle to be guided in the fiber core.

FIGURE 2.10 M easuring the acceptance angle.

Another way to look at light guiding in a fiber is to measure the fiber’s acceptance angle— the angle over which light rays entering the fiber will be guided along its core, shown in Figure 2.10. Because the acceptance angle is measured in air outside the fiber, it differs from the confinement angle in the glass. The acceptance angle normally is measured as num erical aperture (NA), which for light entering a fiber from air is approximately NA =

V(a§ -


The angle over which a fiber accepts light depends on the refractive indexes of the core and cladding.

where «o is the refractive index o f the core and ?i\ is the index o f the cladding. For a fiber with core index o f 1.50 and cladding index o f 1.485 (a 1% difference), NA = 0.21. An alternative but equivalent definition is the sine o f the half-angle over which the fiber can accept light rays, 12° in this example (0 in Figure 2.10). Another alternative definition is NA = Wq sin 0conf, where 0conf is the confinement angle in the fiber core ( 8 ° in this example). These angles are measured from a line drawn through the center of the core, called the fib e r axis. Note that the half-acceptance angle is larger than the largest glancing angle at which light rays must strike the cladding interface to be reflected, which I said earlier was 8°. What does this mean? Look at Snell’s law of refraction again. The difference is the factor Wq, which is the refractive index of the core glass, or 1.5. As you can see in Figure 2.10, refraction bends a light ray entering the fiber so that it is at a smaller angle to the fiber axis than it was in the air. The sine of the angle inside the glass equals that o f the angle outside the glass, divided by the refractive index of the core (« q).

Light Collection Efficiency An optical fiber will pick up some light from any light source. You can see this if you point a single large-core fiber at a lightbulb. Look into the other end o f the fiber, taking care not to get your eye too close. You should see an illuminated spot in the fiber. That light comes from the bulb, but it’s only a tiny fraction o f the total the bulb emits. Appendix E describes eye-safety precautions, which are important if you are using a laser source. Developing ways for small-core fibers to collect light efficiently was an important step in developing practical fiber-optic communications. This includes both collecting light from

Light source size and alignment are critical in collecting light in a fiber core.

Chapter 2

Joining the ends of optical fibers requires careful alignment and fight tolerances.

Transfer losses must be considered in fiber-optic communications systems.

Attenuation, dispersion, and crosstalk can degrade signals transmitted by optical fibers.

external sources and transferring light from one fiber to another. CouplingW^nl efficiently into the fiber requires both focusing it onto the core and aligning it so it falls within the fiber’s acceptance angle. The combination imposes demanding requirements. Simple optics can focus the light from an ordinary bulb so it forms a narrow beam. You can see the results in a flashlight beam or a searchlight. A careful look reveals that the focusing is not perfect, but the beams are strongly directional. However, focusing a large light source into a narrow beam leaves a large spot that spreads far beyond the fiber core. Large light sources can be focused onto small spots with strong magnifying lenses. You’ve probably used that trick to burn a hole in paper with focused sunlight. However, that normally leaves the light spreading at too large an angle for the fiber to collect it efficiently. For communication systems, it’s generally more efficient to find a light source that is close to the fiber core in size. Generally these are semiconductor lasers, which emit light from a small spot, or optical-fiber amplifiers, which emit light from a doped core, as described later. Light-emitting diodes (LEDs) can be used with some larger core fibers because they are less expensive and the larger cores can collect more o f their light. Larger light sources generally are easier to align with fibers, but their lower intensity delivers less light. Chapter 9 describes light sources in more detail. Transferring light between fibers requires careful alignment and tight tolerances. Light transfer is most efficient when the ends of two fibers are permanently joined in a splice (described in Chapter 13). Temporary junctions between two fiber ends, made by connectors (also described in Chapter 13) typically have slightly higher losses but allow much greater flexibility in reconfiguring a fiber-optic network. Special devices called couplers (described in Chapter 14) are needed to join three or more fiber ends. One of the most important functional differences between fiber-optic and wire communications is that fiber couplers are much harder to make than their metal-wire counterparts. Losses in transferring signals between wires are so small that they can normally be neglected. This is not so for fiber optics. As you will see in Chapter 21, system designers should account for coupling losses at each connector, coupler, splice, and light source.

Fiber Transmission Optical fibers inevitably affect light transmitted through them. The same is true for any material transmitting any kind o f signals. You notice these effects most for poor transmitters, like dirty windows or crackling telephone lines. However, they are present even for the tenuous gas dispersed in intergalactic space, which astronomers can spot because it absorbs a tiny fraction o f the light passing through it. Generally these effects degrade signals, and if they become large enough, they can make it impossible to receive the signals. The three principal effects that degrade signals in optical fibers are attenuation, dispersion, and crosstalk. You can see analogous effects when electronic signals go through copper wires or are broadcast as radio or television signals. These effects are critical to the performance o f fiber-optic systems, so I will introduce the concepts here before exploring them in more detail in later chapters.

Fundamentals of Fiber-Optic Components

Fiber Attenuation Attenuation makes signal strength fade with distance. In some cases, such as broadcast radio, distance alone can cause attenuation because signals spread out through space as they travel. As the signal spreads over a larger volume, the intensity drops. This is not the case in optical fibers, which are waveguides that confine light within the core along their entire length. This prevents signals from spreading over a larger volume, but other effects cause different types of attenuation. The three primary effects are absorption, scattering, and leakage o f light from the fiber core. You will learn more about these later, but a basic understanding o f the concepts will help you now. Although optical fibers are made o f extremely pure glass, they absorb a tiny fraction o f the light passing through them. The amount depends on the wavelength and the presence o f impurities. Certain impurities cause strong absorption, but even pure silica has some absorption. Every transparent object absorbs a little light but transmits most o f the light that enters it; opaque materials transmit a little light a little way inside them, but they absorb (or reflect) most o f the incident light. Atoms within the glass also scatter light. The physics are complex, but the atoms act as if they were tiny reflective particles, like droplets in a fog bank. Scattering reflects light in other directions, so it escapes from the fiber core and is lost from the signal. Like absorption, scattering is inherent in all fiber materials, but generally is small. The amount of scattering increases at shorter wavelengths, so it’s higher at visible wavelengths than in the infrared. The physics are the same as for light scattering in the atmosphere, which spreads short-wavelength blue light all over the sky, while allowing longer red wavelengths to reach us as the sun rises and sets. Light leakage occurs when light escapes from the fiber core into the cladding. It’s normally very low unless the fiber is bent sharply, when light can escape by hitting the corecladding boundary at a steep enough angle to avoid total internal reflection. As you will learn later, fiber installation and the environment can bend fibers in ways that allow light to leak out, but normally this loss is the smallest o f the three types. Like leaky plumbing, it’s a rare event that indicates something has gone wrong. Although absorption and scattering are extremely small in optical fibers, total attenuation accumulates when light travels through many kilometers o f fiber. Attenuation normally is measured by comparing the strength o f the input signal to the output. For example, if 99% o f the input light emerges from the other end, a fiber has 1% attenuation. Attenuation is cumulative, and normally uniform through the entire length o f a fiber. Thus every meter o f fiber should have the same attenuation as the previous meter. If 99% of the light emerges from the first meter, 99% o f that light should emerge from the second meter, and so on. For a 10-meter fiber, the light emerging should be Output = Input X 0.99 X 0.99 X 0.99 X 0.99 X 0.99 X 0.99 X 0.99 X 0.99 X 0.99 X 0.99 = 0.904 X Input More generally, the output is Output = Input X (transmission/unit length) r°tal length = Input X (0.99)10

Absorption, scattering, and light leakage are the components of fiber attenuation.

Atoms within the fiber scatter light out of the core.

Attenuation of a fiber is the product of the length times the characteristic loss in decibels per kilometer.


Chapter 2

These sorts o f calculations get messy, so generally attenuation is measured in decibels (dB), which are very useful units, although peculiar ones. The decibel is a logarithmic unit measuring the ratio of output to input power. (It is actually a tenth o f a unit called a bel after Alexander Graham Bell, but that base unit is virtually never used.) Loss in decibels is defined as ( power out dB loss = —10 X log]0 I :— \ power in

Decibel losses are easy to underestimate; every 10 dB decreases signal strength by a factor of 10.

Thus, if output power is 0.001 o f input power, the signal has experienced a 30-dB loss. The minus sign is added to avoid negative numbers in attenuation measurements. It is not used in systems where the signal level might increase, where the sign of the logarithm indicates if the signal has decreased (minus) or increased (plus). Each optical fiber has a characteristic attenuation that is measured in decibels per unit length, normally decibels per kilometer. The total attenuation (in decibels) in the fiber equals the characteristic attenuation times the length. To understand why, consider a simple example with a fiber having the relatively high attenuation o f 10 dB/km. That is, only 10% of the light that enters the fiber emerges from a 1-km length. If that output light was sent through another kilometer o f the same fiber, only 10% o f it would emerge (or 1% of the original signal), for a total loss of 20 dB. As you can see, the decibel scale simplifies calculations o f attenuation. It’s widely used in electronics and acoustics as well as optics. You’ll learn more about decibels later, but you should realize that they are easy to underestimate. Decibels are really exponents, not ordinary numbers. Every additional 10-dB loss reduces the output a factor o f 10. A 20-dB loss is a factor o f 100 (102'°), a 30-dB loss is a factor o f 1000 (103'°), and a 40-dB loss is a factor o f 10,000 (104'°). These numbers can get very big very fast. Appendix B gives some comparisons for decibel units, which you may find surprising.

Bandwidth and Dispersion Optical fibers are unique in transmitting highspeed signals with low attenuation.

Attenuation of copper wires increases with signal frequency.

Low attenuation alone is not enough to make fibers invaluable for telecommunications. The thick wires that transmit electrical power also have very low loss, but they cannot transmit information at high speeds. Optical fibers are attractive because they combine low loss with high bandwidth to allow high-speed signals to travel over long distances. In a communication system, this becomes high bandwidth, the ability to carry billions o f bits per second over many kilometers. Concepts such as ba?idwidth and inform ation capacity are crucial in communications, and the next chapter will tell you more about them. They measure the flow of information through a communication system. For example, television signals have more bandwidth than audio signals. In general, the more bandwidth or information, the better. The more information you want to transmit, the faster the signal has to vary, and it’s the need for rapidly varying signals that can cause problems in transmitting high-bandwidth signals. Different effects limit different types o f communications. The number o f dots and dashes an old-fashioned electrical telegraph could transmit was limited by how fast one operator could hit the transmitting key and how fast another could write down or relay the incoming signals.

Fundamentals of Fiber-Optic Components

FIGURE 2.11 Loss as a function o f frequency.

S ig n a l F re q u e n c y

The speed limit on electrical wires comes from the nature o f electrical currents. Moving electrons induce currents in the copper around them, so the impedance o f a wire increases with the speed at which the signal varies. In practice, that means the higher the frequency, the higher the attenuation. Pairs of copper wires have very low attenuation at the extremely low frequencies used for electrical power transmission, 60 Hz in North America and 50 Hz in Europe, and they can carry audio frequencies over reasonable distances, but not television signals. Coaxial cables can transmit higher frequencies, but their attenuation increases sharply with frequency, as shown in Figure 2.11. In contrast, optical fibers have essentially the same attenuation across a wide range o f operating frequencies, although dispersion does attenuate high signal frequencies, as described below. The main limitation on fiber-optic bandwidth is an effect called dispersion. It is easiest to visualize if you consider a signal as made up of an instantaneous pulse containing many photons. The photons are not perfectly identical, so they spread out a little as they travel, like a group o f race cars on a track. Some spreading occurs because the wavelengths differ slightly, and the refractive index o f the glass varies with wavelength. Other spreading comes because the photons may travel slightly different paths through some types o f fiber. The effects are small, but like attenuation they accumulate with distance. The farther the pulses travel, the more the photons spread out. If the light travels far enough, the first photons in one pulse catch up with the last photons in the previous pulse, and eventually it’s no longer possible to tell the pulses apart, as shown in Figure 2.12. You’ll learn more about dispersion in Chapter 5. Figure 2.12 is an oversimplification in one important aspect: In real fiber-optic systems, digital signals start as boxy square-wave pulses, created by switching a light source on and

Dispersion limits fiber transmission bandwidth.

Chapter 2

FIGURE 2.12 Pulse dispersion.

In itia l n e a r-in s ta n ta n e o u s p u ls e s ...

s p re a d g ra d u a lly a s th e y p a s s a lo n g fib e r...

s ta rtin g to o v e rla p e a c h o th e r...

u n til th e y c a n n o lo n g e r b e re c o g n iz e d .

off very quickly. Gradually the edges round off as some photons get ahead and others fall behind. From a signal-processing standpoint, the sharp edges o f a square wave are really signals at frequencies many times higher than the rate at which the pulses are being switched off and on. As the pulse o f photons blurs out along the fiber, those high-frequency components are lost. Thus the blurring o f sharp square-wave pulses into rounded lumps is really high-frequency attenuation, and optical fibers do have limited transmission bandwidth. But what’s important is that the limit for optical fibers is at much higher frequencies than for copper wires.

Crosstalk and Nonlinear Effects Crosstalk is the leakage of signals between nominally independent channels.

Crosstalk occurs when signals cross the barriers that are supposed to separate them from each other. You have crosstalk on the phone if you hear a radio station or another conversation in the background. The different communication channels— phone lines and radio broadcasts— are supposed to be separate from each other. However, a little bit o f one can leak into another channel. There are many reasons for electrical crosstalk. Phone wires can act as antennas to pick up strong radio signals. Currents in one pair o f wires can induce signals in another pair running beside them. Sometimes other equipment may transmit signals through the air at the same frequency, so you might hear static on your AM radio when a motor operates nearby. Fibers are immune to the usual electronic crosstalk. They don’t carry electrical currents, and the light inside them is unaffected by nearby currents. You can run fibers along power lines and never hear a thing, although the 60-cycle hum would overwhelm telephone wires.

Fundamentals of Fiber-Optic Components

However, fibers carrying multiple signals or optical channels at different wavelengths— an important technique called wavelength-division multiplexing— are vulnerable to crosstalk. Nominally, light signals at different wavelengths passing through the same fiber do not interact because no current flows between them. However, like electrical phone signals passing through parallel wires, there can be secondary interactions called nonlinear effects because they aren’t directly proportional to the strength of a single signal. These nonlinear effects are complex, and are the prime cause of crosstalk. You’ll learn more about them in Chapter 5.

Nonlinear interactions between optical channels in the same fiber can cause crosstalk.

Electro-Optics and Other Components Electronics play important roles in fiber-optic systems. Because this book is about fiber optics, it doesn’t cover electronics in general, but it will cover the electronics used in fiber systems. It assumes only a very general knowledge o f electronics. Many components have both optical and electronic elements, which often are called electro-optics or opto-electronics to emphasize the connection. Sometimes these components are lumped with optical devices and called photonics, a term that originated from the idea o f manipulating photons just as electronics manipulate electrons. These components fall into two very broad categories. One includes devices that convert signals between optical and electronic formats, such as transmitters and receivers. These provide vital connections between fiber-optic systems and other equipment, such as telephones and computers. The second includes devices that manipulate light but are powered or controlled by electronic circuits, such as optical amplifiers that raise the strength of optical signals, and modulators that control the intensity o f light passing through them. We’ll introduce both types o f components briefly here and cover them in more detail later.

Electronics play important roles in fiber-optic equipment.

Transmitters and Light Sources Optical transmitters convert electronic input signals into the optical signals carried by fiber systems. Electronic circuits take the input electronic signal and process it to modulate light generated by a light source. Typically the light source is a semiconductor laser— often called a laser diode— or a light-em itting diode (LED). You’ll learn more about light sources in Chapter 9 and about transmitters in Chapter 10. Different types of light sources are used for different applications. Lasers generate higher power and can be modulated at higher speeds, so they transmit faster signals farther than LEDs. The wavelength is chosen to meet requirements for transmission distance and bandwidth. Most transmission is in a band called the near-infrared, which is invisible to the human eye. High-speed, long-distance systems use a range of wavelengths from 1530 to 1625 nanometers, where optical fibers have low attenuation and optical amplifiers are readily available. Short high-speed systems use 1310 nanometers, a wavelength at which attenuation is somewhat higher but dispersion is lower. Wavelengths o f 750 to 900 nanometers are used for systems spanning no more than a few kilometers. Low-cost systems spanning much shorter distances typically use red LEDs and plastic fibers, which have high attenuation. Some transmitters include stages that combine or m ultiplex different signals to generate a composite signal containing the information in multiple signals.

Transmitter wavelength depends on the application requirements.

Chapter 2


Photonics Newcomers to the world of fiber optics are likely to be confused by the term photonics, which is widely used in some circles but ignored in others. The use of this term reflects a confusing and controversial history. I first heard the term “photonics” about 30 years ago. But it didn’t come into popular use until it was adopted some years later by Bell Labs and one of the industry’s leading trade magazines (formerly called Optical Spectra, now Photonics Spectra). The idea was for “photonics” to describe devices that manipulate light in the same way that electronics describes things

that manipulate electrons. Because I wrote regularly for a competing magazine, I tended to avoid the word. Other optical engineers and scientists also showed little enthusiasm, because “photonics” sounded like another word for “optics,” which they felt was a perfectly adequate description o f their field. Matters came to a head when a group o f leaders attempted to change the name o f the Optical Society of America to the Optics and Photonics Society. The members soundly rejected the proposal, and the community remains divided on “photonics.” Some like its modern sound, but others find it unnecessary or obscure and think “optics” is a better description o f the field.

Receivers and Detectors A receiver converts an optical signal into electronic form.

A receiver converts an optical signal into an electrical signal usable by other equipment. The input light signal is directed into a detector, which produces a current or voltage proportional to the amount of light illuminating it. Electronic circuits in the receiver amplify that signal and convert it into the format required by electronic equipment at the receiver end o f the system. Like transmitters, receivers are designed to operate at specific wavelengths; the usable wavelengths depend on the detector chosen. The receiver also demultiplexes input signals combined at the transmitter, producing separate output signals corresponding to each o f the input signals.

Fiber-Optic Applications The bulk of this book is about fiber-optic applications in communications, but it’s important to remember that there are other uses for fiber optics. Chapter 29 describes the wide variety of fiber-optic sensors, from gyroscopes that sense rotation to acoustic sensors that pick up faint undersea sounds. Chapter 30 shows how bundles o f optical fibers are used for imaging and illumination.

What Have You Learned? 1. Light is one type of electromagnetic radiation. It is a part o f the electromagnetic spectrum with a distinct range o f wavelengths, frequencies, and photon energies. Optical wavelengths include the near-ultraviolet, visible, and near-infrared.


2 Light can be viewed as electromagnetic waves, photons, or rays, depending on the situation. Each view has its advantages.

Fundamentals of Fiber-Optic Components

. . 5. Light waves add or subtract in amplitude depending on their relative phase, an

3 A photon is a quantum of electromagnetic energy. 4 Wavelength equals the speed o f light divided by the frequency o f the wave. effect called interference.

6. Refractive index («) is the speed o f light in a vacuum divided by the speed of light in the material. It is always less than 1.0 for materials at optical wavelengths. 7. Refraction is the bending of light as it changes speed when entering a new material. It depends on the refractive index of the material and the angle of incidence. 8. Total internal reflection can trap light inside a material that has a higher refractive index than its surroundings. The critical angle for total internal reflection depends on the difference between the two indexes.


9 Total internal reflection guides light along the core o f an optical fiber, which has a higher refractive index than the surrounding cladding. 10. Light that falls within the acceptance angle o f a fiber is guided in the core. The numerical aperture is the sine of the acceptance angle. 11. Fiber collection efficiency depends on light source size and alignment to the fiber core. 12. Attenuation reduces the amount o f light transmitted, reducing transmission distance. It depends on wavelength and occurs because the glass scatters and absorbs light. It is measured in decibels.


13 Disp ersion is the spreading out of signal pulses, which limits fiber transmission bandwidth. Optical fibers have much higher bandwidth than copper wires.


14 Both attenuation and dispersion increase with transmission distance. 15. Electronics play important roles in fiber-optic equipment. Opto-electronic or electro-optic devices have both electronic and optical functions. 16. Transmitters convert electronic input signals to optical format by modulating light from an LED or laser. 17. A receiver converts an optical signal into electronic form.

What's Next? In Chapter 3, we will look at how fiber-optic systems are used in communications.

Further Reading Introductory Level: J. Warren Blaker and Peter Schaeffer, O ptics: An In trodu ction fo r T echnicians an d Technologists (Prentice Hall, 2000) David Falk, Dieter Brill, and David Stork, Seeing the Light: Optics in Nature, Photography, Color, Vision an d Holography (Harper & Row, 1986) B.

K. Johnson, Optics an d O ptical Instruments (Dover, I960)

Advanced Treatments: Eugene Hecht, Optics, 4th ed. (Addison-Wesley, 2002) Francis A. Jenkins and Harvey A. White, Fundamentals o f Optics (McGraw-Hill, 1976)

Questions to Think About 1. Interference seems to be a strange effect. The total light intensity from two bulbs is the sum of the two intensities. Yet the light intensity is really the square o f the amplitudes, and if the two waves are in phase, you double the amplitude, which when squared means the intensity should be four times the intensity o f one bulb. Don’t these views contradict each other? 2. One photon is a wave packet that doesn’t last very long. A continuous light source emits a steady or continuous wave. How is the continuous light source emitting photons?


3 The sun emits an energy of about 3.8 X 1033 ergs per second. A photon with wavelength of 1.3 micrometers has an energy of about 1.6 X 10 12 erg. If you assume the sun emits all its energy at 1.3 jam, how much attenuation in decibels do you need to reduce the sun’s entire output to a single 1.3-|xm photon per second?


4 If an entire galaxy contains a billion stars, each one as luminous as the sun, how much attenuation does it take to reduce its entire output to a single 1.3-|am photon per second?


5 Suppose a material has attenuation o f 10 dB/m at 1.3 micrometers. How thick a block o f the material would you need to reduce the sun’s entire output to a single photon as in Problem 3?

6. Medical imaging fiber has attenuation o f 1 dB/meter at optical wavelengths. If the attenuation is the same at 1.3 p.m, and you don’t have to worry about the sun’s energy melting the fiber, how long a fiber would reduce the sun’s output in Problem 3? 7. Atoms and molecules in the atmosphere scatter light in the same way that atoms in glass scatter light in an optical fiber. The shorter the wavelength in the visible spectrum, the stronger the scattering. Where do you think the sky gets its blue color from and why?


Diamond has a refractive index o f 2.4. What is its critical angle in air and what does that have to do with its sparkle?

Chapter Quiz 1 . Which of the following is not electromagnetic radiation? a. radio waves b. light c. infrared radiation

Fundamentals of Fiber-Optic Components

d. X-rays e. acoustic waves

2 . Optical fibers have minimum loss near 1.5 pm. What is the frequency that corresponds to that wavelength? a. 200 MHz b. 20 GHz c. 200 GHz d. 20 THz e. 200 THz

3 . An electron-volt is the energy needed to move an electron across a potential of 1 V. Suppose you could convert all the energy from moving an electron across a potential o f 1.5 V into a photon. What would its wavelength be? a. 0.417 pm b. 0.5 pm c. 0.827 pm d. 1.21 pm e. 1.2399 pm

4 . Light that passes from air into glass is a. reflected. b. refracted. c. absorbed. d. scattered.

5 . Light is confined within the core of a simple clad optical fiber by a. refraction. b. total internal reflection at the outer edge o f the cladding. c. total internal reflection at the core-cladding boundary. d. reflection from the fiber’s plastic coating.


An optical fiber has a core with refractive index o f 1.52 and a cladding with index o f 1.45. Its numerical aperture is a. 0.15. b. 0.20. c. 0.35. d. 0.46. e. 0.70.

7 . Zircon has a refractive index of 2.1. What is its critical angle for total internal reflection in air? a. 8° b. 25°

Chapter 2

c. 32° d. 42° e. 62°


The output o f a 20-km fiber with attenuation o f 0.5 dB/km is 0.005 mW. What is the input power to the fiber? a. 0.5 mW b. 0.1 mW c. 0.05 mW d. 0.03 mW e. 0.01 mW

9 . What fraction o f the input power remains after light travels through 100 km of fiber with 0.3 dB/km attenuation? a. 0.1% b. 0.5% c. 1% d. 5% e. 10%

10. If a 1-cm glass plate transmits 90% o f the light that enters it, how much light will emerge from a 10-cm slab o f the same glass? (Neglect surface reflection.) a. 0% b. 9% c. 12% d. 35% e. 80% 1 1 . What happens to light that is scattered in an optical fiber? a. It escapes from the sides o f the fiber. b. Glass atoms absorb its energy. c. Glass atoms store the light and release it later. d. It is reflected back toward the light source. e. It excites acoustic waves in the glass.

12 . What effect does dispersion cause? a. scattering o f light out the sides o f the fiber b. stretching o f signal pulses that increases with distance c. shrinking o f signal pulses that become shorter with distance d. attenuation o f signal pulses

Fundamentals of Communications


About This Chapter Communications is the most important application o f fiber optics. Optical fibers serve as low-cost, flexible “pipes” that carry signals in environments ranging from climatecontrolled office buildings to the ocean bottom. They span distances ranging from a few meters inside an automobile to nearly 10,000 kilometers across the Pacific ocean. They carry signals at speeds up to trillions o f bits per second and form the backbone o f the global telecommunications network. To understand these uses o f optical fibers, you need to understand the basic concepts behind modern communications. This chapter explains how communications systems function, the types o f signals they transmit, the types o f services they offer, and how the communications industry works. This chapter also shows you how fiber optics fit in with other communications equipment in the global network.

Communications Concepts Communications is the process o f conveying information, and the word is used in two distinct senses. One is communication through the use o f the written or spoken word by writers, public speakers, broadcasters, and public relations specialists. The other is telecommunications, which is sending information over a distance using technology. In this book, I discuss the use o f fiber optics in telecommunications, although I usually simply say “communications.” Many different technologies are used in modern telecommunications. Electrical signals travel through plain copper wires and coaxial cables, also made o f copper. Radio and microwave signals travel through the air from antennas on the ground, in

Telecommunications sends signals over a distance by fiber, wire, or radio waves.


Chapter 3

aircraft, or in satellites. Beams o f light travel through the air or through optical fibers. To learn about these technologies, we’ll quickly examine the history o f communications.

A Short History of Communications

The optical telegraph relayed messages from hilltop to hilltop 200 years ago.

Multiplexing allowed telegraph lines to carry two or more signals at once.

The earliest long-distance communications were made by signal fires that relayed simple information. The ancient Greeks relayed news o f the fall o f Troy by lighting fires on a series of mountain peaks. During the American Revolution, patriots watched the belfry of Boston’s Old North Church for signal lamps. These lamps would reveal the path of British troops leaving Boston to seize weapons stored in Concord. Such signals could be seen for miles, but the people sending them had to arrange the signals’ meaning in advance. So the patriots knew that one lamp meant the British were leaving Boston by crossing a narrow neck o f land leading west; two lamps meant the British were leaving by boat. (Paul Revere was already on his way; the lanterns were used to warn others in case Revere was caught.) The message was vital to the revolution, but it was also simple. The two lanterns conveyed only two bits o f information: that the British were coming and that they were crossing the water. A written letter or a human messenger could carry more information but took time to travel, on foot or by horse. The first system that we might call telecommunications was a series o f hilltop towers built in the 1790s by French engineer Claude Chappe. Each tower was in sight o f the next and had an operator to relay messages by moving a set of wooden arms. The operator in each tower looked through a telescope to read the message coded by the positions o f the arms in the previous tower, recorded the message, then relayed it to the next tower by moving the arms o f his own tower. This process continued on down the line to others, as shown in Figure 3.1. Other countries soon adopted Chappe’s optical telegraph, and it remained in use for decades until it was replaced by Samuel Morse’s electrical telegraph. The electrical telegraph sends signals through wires as a series o f short and long electrical pulses. Known as Morse code, “dots” and “dashes” represent letters. Each dot or dash is a bit of information. Like the optical telegraph, the electrical telegraph requires operators to relay signals; but wires can carry signals further between operators, and transmission isn’t interrupted by darkness or bad weather. Because o f these advantages, the electrical telegraph spread across the continents and, in 1866, across the Atlantic. Telegraph wires formed a network running between major cities. Telegraphers received signals and either delivered them locally or relayed them to more distant stations. People did not have home telegraphs; but the stock ticker, a special-purpose telegraph that printed information on stock trades, was available in the stockbrokers’ offices. The first telegraph lines could carry only one message at a time, but inventors soon discovered ways to make them carry two or more signals at once, a process called multiplexing. Alexander Graham Bell was trying to invent a new type o f multiplexer when he realized that his invention could be used to carry voice. Instead o f sending dots and dashes through the wire, Bell modulated an electrical signal that reproduced a speaker’s voice. He called his invention the telephone. (Bell also invented a device he called the “photophone,” which modulated the brightness o f a beam o f light sent through the air, but it never proved practical.)

Fundamentals of Communications

FIGURE 3.1 Relaying messages by moving the arms on an optical telegraph.

First tower sends signal.

Second tower sends signal.

Third tower sends signal.

Although many telegraph companies saw no future in the telephone, it soon spread to homes and offices. By the 1890s, dense thickets of telephone wires stretched between poles in downtown areas. The early telephone network differed from the telegraph network because voice signals could not travel as far through wires as the dots and dashes o f the telegraph. The telephone worked locally; the telegraph could send signals long distances. Operators and mechanical devices could regenerate telegraph signals when they became weakened by distance; but telephone signals could not be amplified until vacuum tube circuits were developed. Only in 1915 could telephone calls reach across North America,

Chapter 3

Radio moved to higher frequencies as electronics improved.

The demand for transmission capacity has increased steadily.

and not until 1956 did a submarine cable carry telephone conversations under the Atlantic— 90 years after the first transatlantic telegraph cable. Radio communications followed. Guglielmo Marconi showed that radio waves could transmit telegraph signals by a mile in 1896 and steadily increased the distance in the following years. Radio’s advantage was that it required no wires, and the British Admiralty saw its potential for communicating with ships at sea. Ships in trouble could transmit pleas for help (most famously the Titanic). Radio telephones followed and in the 1920s became the first way to send voices across the Atlantic. Radio broadcasting began about the same time. Radio communications began at low frequencies, but improvements in electronic technology opened up higher frequencies. Higher frequencies can carry more information, as we’ll see later. Low frequencies o f tens or hundreds o f kilohertz are fine for voices, but television pictures carry more information and require higher frequencies. Television signals are broadcast at frequencies of tens or hundreds of megahertz, a thousand times higher than the band used for audio-only radio. High-frequency radio signals and microwaves, which are really higher-frequency radio waves, can multiplex voice or video signals, just as electrical telegraph could carry multiple messages. This led telephone networks to use chains o f high-frequency radio and microwave towers to relay signals tens o f kilometers, repeating signals in an electronic version of Chappe’s optical telegraph. By 1970, satellites were relaying voice and video signals around the world and the telecommunications network had become global. That made it possible to make telephone calls around the world, although the rates were dollars per minute or more. Long-distance calls within the United States cost less, but could still add up to budget-busting bills. Then fiber-optic communications arrived. When the American long-distance market was opened to competition in the early 1980s, telephone companies built their new national high-capacity backbone lines with optical fiber. The ever-increasing capacity of fiber now far exceeds that o f any other telecommunications medium. This brief history of communications carries a few important lessons. The demand for transmission capacity has increased steadily. Engineers have turned to new technologies to provide more capacity and to send signals over longer distances. Transmission capacity has always been a key limitation. Distance and connectivity are also important. People have friends and family scattered around the world; businesses need to contact people in farflung places. It’s no longer enough to have phones at home and at work; people want mobile phones in their pockets so they can be reached at anytime and anywhere. Let’s look at the key concepts o f capacity and connectivity.

Transmission Capacity and Bandwidth Bandwidth traditionally has been in short supply.

Transmission capacity or bandw idth tells us how much information a system can carry. Bandwidth has traditionally been in short supply, so communications systems are designed to make the most o f it by various means. Depending on what’s being transmitted, bandwidth can be measured in various ways, and that requires more explanation. The telephone is a familiar example that illustrates the principles of transmission capacity. A corded phone is linked to the telephone network by a pair of copper wires, which normally can carry a voice signal several miles. Higher frequencies suffer greater attenuation

Fundamentals of Communications

5.33 FM C h a n n e ls = 8 0 ,0 0 0 H z 1

1 5,000 H z 1 FM R a dio





1 5 ,0 0 0 Total


C h annel


5,0 0 0

O ne V o ice — , Telepho ne |


20 V oice l-«— C h a n n e ls

C h annel (4000 Hz)

= 8 0 ,0 0 0 Hz

FIGURE 3.2 8 0 ,000 hertz o f bandw idth can carry twenty 4-kH z voice channels, or 5-33 FM radio channels.

8 0 ,0 0 0 H z B a ndw idth

than lower frequencies, so phone lines carry frequencies o f only 300 to 4000 hertz, much less than the human ear’s nominal range o f 20 to 20,000 hertz. The high end o f this band is reserved for control signals, so the actual bandwidth is about 300 to 3400 hertz. This is enough for intelligible conversation, but not for high-fidelity reproduction. The telephone industry settled on this limit long ago, which means that the nominal bandwidth o f a phone line is 4000 hertz. In contrast, FM radio has a bandwidth of 15,000 hertz per channel, which gives it much better fidelity. Frequency range is one way to measure bandwidth for analog signals, such as voice and music, which vary continuously. Bandwidth also can be measured by the number o f standard Analog bandwidth is measured in channels that can fit in the band. For example, a system that can transmit 80,000 hertz, as frequency range shown in Figure 3.2, can carry 20 telephone voice channels or 5.33 FM radio channels. In or number of this case, each channel is a distinct signal. channels. For digital transmission, bandwidth is the number of bits per second passing through the system. (Be aware that computer data storage is measured in bytes rather than in bits; one byte contains eight data bits, and may contain extra bits for error correction.) Digital bandwidth also can be measured as the number o f standard channels transmitted. For example, a standard analog telephone voice channel can be converted to a digital signal carrying 64,000 bits per second. A digital signal carrying 20 digitized standard voice channels has a nominal line rate of 1,280,000 bits per second or 1.28 megabits per second. You will note that the digitized bandwidth seems to be higher than the corresponding analog bandwidth. The numbers indeed are larger, but comparing digital and analog capacity is not as simple as comparing the numbers, as you’ll see when you learn about analog and digital signals later in this chapter.

Multiplexing The builders o f electrical telegraphs realized that it costs less to build one high-capacity system that could carry many signals than build several separate systems with lower capacity. This technique, called multiplexing, can be used for any type o f signal as long as the bandwidth is available. Modern telecommunications systems typically multiplex signals together in three different ways: by frequency, by wavelength, or by time. • Frequency-division multiplexing is the transmission o f signals at different frequencies of which radio and television broadcasts are good examples. Each area broadcaster

Multiplexing combines many signals into one higher-speed signal.

Chapter 3

broadcasts signals at a specific frequency or band. Stations are set at standard frequencies, such as 89.7 MHz for an FM radio station or 204 to 210 MHz assigned to channel 12 on the U.S. television dial. Other broadcasters in the region are assigned other frequencies. You tune your radio or television receiver to select one o f these frequencies, but all o f them are transmitted through the air simultaneously. (For broadcasting, a buffer is usually kept in reserve between adjacent channels, so neither channel 11 nor channel 13 would operate in the same area as channel 12.) Broadcast signals go through the air, but frequency-division multiplexing also works through cable. In fact, frequency-division multiplexing is standard in cable television, where certain frequency bands are reserved for specific video channels, although the frequencies may not match those on the broadcast dial. • W avelength-division m ultiplexing is the optical counterpart o f frequency-division multiplexing. Separate signals modulate light sources emitting at different wavelengths, just as separate signals modulate radio transmitters broadcasting at different frequencies. Each separate wavelength is an optical channel. Light from the separate sources is combined and transmitted through a single optical fiber. Then the wavelengths are separated again, or demultiplexed, at the other end, as shown in Figure 3.3. As with frequency-division multiplexing, the wavelengths can be packed closely together, but the signals must not overlap. • Time-division multiplexing works

' .


FIGURE 3.3 Wavelengthdivision multiplexing combines signals in one fiber.

9 Tim e-division m ultiplexing combines streams o f bits from two or more sources to produce a single stream o f bits at a faster rate, as shown in Figure 3.4. For example, four signals at 10 million bits (megabits) per second can be combined to generate one 40-Mbit/s signal. Figure 3.4 shows data bits being interleaved, with one bit from stream A followed by one from stream B, then one from stream C, one from stream D, and finally the next bit from stream A. As you will learn later, other variations on time-division multiplexing may arrange the incoming data bits differently, such as in blocks called packets. Note that timedivision multiplexing is designed specifically to work with digital signals made o f strings o f incoming bits. Note that although demultiplexing may seem to be merely the opposite o f multiplexing, separating the combined signals often is a more complex and demanding task than

Fiber inputs at four wavelengths... are put together in one fiber.

1 5 4 4 nm

1 5 4 4 nm

\ / \ / \ / \ / 1543 nm




W a v e le n g th D iv is io n M u ltip le x e r

> y v v v v \ / \ / v v \ / \ / 1543 nm W W V \ / V W V W \ A i 5 4 2 nm ' W W W W V W W W W /1 5 4 1 n m

1 5 4 3 nm D e m u ltip le x e r 1542 nm

15 4 2 nm




1541 nm

1541 n m

Fundamentals of Communications

FIGURE 3.4 S ig n a l D -

_____ a

uiMJiniuuuL D

S ig n a l C -

M u ltip le x e r









S ig n a l B ■

S ig n a l a



Time-division multiplexing combines several slow signals into one faster signal.

multiplexing. Imperfect demultiplexing leaves you with too little o f the desired channel, or too many undesired channels scrambling your signal.

Terminology In introducing the broad area of communications, we have covered a lot of ground and introduced some terms that have specific meanings in the field. Telecommunications is full of confusing buzzwords, so let’s pause to review and explain some important terms before going on to cover the field in more detail. Information is what is communicated. It may be a very simple message confirming receipt of some anticipated message, conveyed by lighting a signal fire. Or it can be a huge and complex message, such as digital files that contain an entire book, an album o f music, or a motion picture. Even if the “information” contains something this is not at all informative, such as your least favorite television program; it still counts as information. A signal transmits that information. Signals may take many forms, such as acoustic, electronic, or optical. They may be converted from one form to another, and still contain the same information. For example, when you make a long-distance telephone call, your telephone converts the sound waves from your mouth into an electrical signal. This electrical signal is converted into an optical signal at a telephone switching office, then back to an electronic form at another telephone switching office, and finally back into sound waves at the other person’s telephone. Signals are coded in various formats so they can be understood by both the sender and the receiver, as you’ll learn in the next section. A system is a collection of equipment that performs a task, such as transmitting a signal. It’s also a vague word that can elude precise definition. Systems can contain other systems, sometimes called subsystems, and can range in scale from gigantic to tiny. We often speak of the telephone system as one entity that includes all telephone equipment, but it also includes switching systems that direct phone calls to their proper destinations. Solution is a meaningless marketing buzzword. It often functionally means “system,” or “something that someone gets paid for selling you.” A channel is a distinct signal. An example is the signal from a television station that you select on your dial or remote control. A multiplexed signal may carry many channels. An optical channel is a signal carried on one wavelength; a single fiber may carry many optical channels. Radio from a telecommunications standpoint describes frequencies o f the electromagnetic spectrum from about 10 kilohertz to 100 gigahertz. Different parts o f that spectrum are used for different purposes. Radio waves broadcast both sound (often called simply “radio”)

Radio waves have frequencies from 10 kHz to 100 GHz.

Chapter 3

FIGURE 3.5 Types o f copper cables.

C o m m o n T e le p h o n e W irin g

i S in g le T w is te d P a ir

'v v v v x

Microwaves have frequencies of 1 to 100 GHz.

"Copper" is the generic term for all metal cables.

F o u r In s u la te d W ire s in R ib b o n

and video (television). The radio spectrum is divided into many bands, which carry various services; these bands vary from region to region. M icrowaves are high-frequency radio waves ranging from about 1 to 100 GHz, corresponding to wavelengths o f 30 centimeters to 3 millimeters. Waves at the upper end o f the microwave range and higher sometimes are called m illim eter waves, because their wavelengths are measured in millimeters. “Microwaves” is an old name, given at a time when radio wavelengths less than a meter were considered short. Wireless literally means “without wires.” In practice, it means sending signals through the air, without a physical connection, to a fixed or mobile terminal such as a cellular telephone or pager. Signals sent by radio waves, microwaves, and visible or infrared light through the air are examples o f wireless communications. Coaxial cable is a metal clad cable with a central wire running along its axis. The central wire is surrounded by a nonconductive material called a dielectric (usually plastic) and covered by a metal shield, as shown in Figure 3.5. The central wire carries current, and the shield confines the electromagnetic field generated by the current and blocks external electromagnetic fields that could induce noise. Often called coax, coaxial cable transmits radio and low-frequency microwave signals. Coax is commonly used for cable television and video signals. Twisted p air is— strictly speaking— a pair o f thin insulated copper wires wound around each other in a helical pattern. Twisted pair is the nominal standard for telephone wiring in homes and offices. However, a closer look reveals that telephone wires are often flat strips containing four parallel wires, as shown in Figure 3.5. Only two wires are needed to carry signals for a single phone line. High-performance versions of twisted pair, such as Category 5 cable, can carry higher-frequency signals. Copper is the generic term for metal cables, including twisted pair and coax, because most metal cables are made of copper.

Signals and Formats Telecommunications signals and their formats are crucial elements in any communications system. The sender and receiver o f any message must agree on a format that both can understand. This is true for both one-way communications, like television broadcasts, and

Fundamentals of Communications

two-way communications, like telephone calls. The transmitter must generate the signal in a format that the receiver can convert to a usable form. As mentioned earlier, signals can be generated, transmitted, and received in different ways, including sound waves, electrical currents, electronic voltages, and light. We’ll start with some very general ideas about telecommunications signals and formats, then consider some specific points important for fiber optics.

Carriers and Modulation To understand the basic structure o f a telecommunications signal, let’s start with a very simple example: an AM radio station. AM stands for am plitu de m odulation, which describes its operation. The radio station is licensed to operate at a specific frequency in a certain location. A radio-frequency oscillator at the station generates a single pure wave that oscillates at the exact frequency specified by the license, such as 980 kHz in the AM band. This pure wave is called the carrier because it carries the signal; but it is not a signal because it’s only a pure tone that carries no special information. The input is a signal, which may be an announcer’s voice or a piece o f music. The sound waves are converted to an electronic signal, which varies in intensity with the volume o f the sound wave at any instant. That is, the electrical intensity is proportional to the acoustic intensity. The electronic signal is at the same frequency as the original sound. This is called a baseband signal, meaning that the input signal is at its intended frequency. This baseband signal modulates the amplitude o f the carrier wave. For clarity, Figure 3.6 shows a very simple case, where the modulation simply switches the carrier wave on and off, as if sending Morse code. This is the modulated transmitter output. For a real radio station, this would produce an irregular wave that varies with the sound intensity at any instant.

FIGURE 3.6 B a s e b a n d In p u t S ig n a l

C a rrie r W a v e

T ra n s m itte d S ig n a l

E n v e lo p e o f M o d u la tio n

Input signal modulates a carrier wave.

A telecommunications signal is a modulated carrier wave.

The carrier wave has a higher frequency than the baseband signal.

Chapter 3

Amplitude modulation is standard in fiberoptic systems.

Note that the radio carrier frequency is much higher than the highest frequency in the baseband input signal. AM radio stations broadcast at frequencies of 540 to 1700 kHz, well above the 5-kHz upper limit on the baseband audio-frequency signal. This is a common feature in the modulation of carrier frequencies to generate telecommunications signals. The transmitter modulates the carrier with the input audio signal. The receiver demodulates the received radio signal, effectively removing the carrier frequency to recover the baseband audio signal. Another important point is that the transmitter and receiver have to agree on the same format to transmit a signal properly. For a radio station, one part o f that format is the carrier frequency. Tune your radio to 680 kHz, and you won’t hear a radio station broadcasting at 570 kHz or 890 kHz. The type o f modulation also is critical. AM radio modulates the intensity o f the carrier frequency. FM radio uses frequency m odulation, which modulates the frequency of the carrier signal rather than its intensity. That approach gives a much cleaner signal, as you can easily tell if you compare the sound quality o f AM and FM stations. However, you can’t decode AM radio with an FM tuner. The same principles apply in fiber optics. Light waves are the carriers, and their frequencies are much higher than the frequencies o f the signals they transmit. Amplitude modulation is standard in commercial fiber-optic systems.

Analog and Digital Communications Digital signals transmit a series of bits. Analog signals vary continuously.

Communications signals come in two basic types, analog and digital, as shown in Figure 3.7. The level of an analog signal varies continuously, making it an analog o f the variations of the original input. A digital signal transmits a series o f bits; if the input signal was analog, the bits represent how that input signal varies. Virtually all practical digital signals are binary codes, which are at either a high level (“on” or “one”) or a low level (“off” or “zero”), as shown in Figure 3.7. Analog and digital formats each have advantages. The older analog technology is more readily compatible with our senses and much existing equipment. Our ears, for example, detect continuous variations in sound level, not merely the presence or absence o f sound; likewise our eyes detect levels o f brightness, not merely the presence or absence o f light.

Level varies continuously.

FIGURE 3.7 A nalog an d digital signals.

A n a lo g S ig n a l

Binary— either on or off. D ig ita l S ig n a l

Fundamentals of Communications

Both audio and video communications traditionally have been in analog format. The wires that serve corded telephones deliver continuously varying analog electronic signals to a standard telephone handset, which converts the incoming electronic signals to continuously varying sound waves. Traditional analog television sets likewise receive analog signals, which they decode to display pictures on the screen. Digital signals, on the other hand, are easier to process with electronics and optics: It is much simpler and cheaper to produce circuits that detect whether a signal is at a high or low point (on or off) than it is to produce one that can accurately replicate a continuously varying signal. Digital signals also are much less vulnerable to noise and distortion. For an analog device, the output must increase linearly with respect to the input to accurately reproduce an input signal; and once noise gets into an analog signal, it’s very hard to remove. In contrast, digital signals don’t have to be reproduced accurately; all you need is to be able to tell the ones from the zeroes. It’s like the difference between seeing a person across the street clearly enough to identify them (analog), and merely recognizing that someone is across the street (digital). These advantages are leading engineers to shift increasingly to digital transmission. Recorded music is largely digital, with CDs having replaced phonograph records and audiocassette tapes. Most cellular telephones rely on digital transmission, and the new highdefinition television signals are digital. Cable television transmission is a mixture o f digital and analog signals. All this technology is possible because signals can readily be converted between analog and digital formats. For example, the telephone network includes circuits that convert the analog electrical signals that replicate your voice to digital form, and other circuits that convert those digital signals back to analog form so you can understand them. The idea o f analog-to-digital conversion is simple, as shown in Figure 3.8. A conversion circuit samples an analog signal at regular intervals, measuring its amplitude at that instant. In the telephone system, the samples are taken 8000 times per second, twice the highest frequency (4000 Hz) that must be reproduced. Each measurement is assigned to one o f a number o f possible slots that represent different amplitude levels. The amplitude in today’s phone systems is described by an eight-bit code, so there are a total o f 256

FIGURE 3.8 D igitization o f an analog signal. Average Level in Interval

S a m p lin g In te rv a l (8 0 0 0 /s in te le p h o n e )

Other Possible Levels (256 in telephone)

Digital signals are easy to process with electronics and optics.

Analog telephone signals are digitized at 64,000 bits per second.


Chapter 3

possible signal levels. This produces 64,000 bits per second. (Older telephone systems used a seven-bit code for signal amplitude, which gives 128 signal levels and produces 56,000 bits per second.) Digital-to-analog conversion is straightforward. The circuit sets the signal amplitude at the level measured for each sampling interval, essentially building the analog waveform interval by interval. Comparing the analog bandwidth o f 4000 Hz to the digital bandwidth o f 64,000 bits per second suggests another potential disadvantage o f digital transmission. Accurate digital reproduction o f an analog signal requires sampling at a rate faster than the highest frequency in the analog system. In addition, each sample requires several bits o f data, so digital systems seem to require a higher overhead. It isn’t that bad in reality, because there is no precise equivalence between analog and digital capacity requirements. Digital systems can tolerate limited bandwidth much better than analog systems because they only need to detect the presence o f a pulse, not reproduce its shape accurately. In many cases, the bandwidth needed to carry a digitized version o f an analog signal is comparable to the bandwidth needed by the original analog signal. Fiber optics work well for digital transmission, and initially were mainly used in digital systems. The light sources used with fiber-optic systems are vulnerable to nonlinearities that can distort analog signals at high frequencies, but engineers have succeeded in making highly linear analog fiber systems, which are mainly used to distribute cable television signals.

Electronic and Optical Signals

Optical signals are the number of photons. Electrical signals are the current or voltage.

At first glance, conversion o f signals between optical and electronic formats seems automatic. An electronic signal goes into a transmitter, and an optical signal emerges from the attached optical fiber. Conversely, an optical input to a receiver produces an electronic output. However, the process isn’t quite that simple. In optical devices, the signal is the number of photons, which corresponds to the flow o f current. A laser is switched on by increasing the current passing through it; a detector produces a current proportional to the number o f photons reaching it. Electronic signals can be represented in two ways: as a voltage, or electrical potential, and as a current, or the number o f electrons flowing through a circuit. Voltage and current are related, but in practice one or the other is considered the signal, depending on the circuit elements used. Simple circuits can convert a voltage variation to a current variation. Most electronic circuits in optical systems use voltage signals, which must be converted to current variations to drive the optical system. This conversion is not difficult, but you should realize the signals are not exactly equivalent.

Connectivity Pipes and switches can represent a communications system.

Communications systems provide connectivity by transmitting signals from place to place. A critical job of any communications system is to get the signals to the right place. To understand how this is done, we’ll divide the parts o f a communications system into two basic categories: pipes and switches.

Fundamentals of Communications

All receivers get the same signal.

FIGURE 3.9 A n te n n a


r\ R e c e iv e rs


i M o b ile P h o n e


T ra n s m itte r



Some representative communications systems.

F ix e d P h o n e


6 ° b. Mobile Phones

a. Broadcast

G ro u p S e rv e r


C o m p u te r

P rin te r

C o m p u te r

C o m p u te r

C o m p u te r

c. Local-Area Network d. Switched Network

As you saw in Figure 1.6, pipes transmit signals from one point to another. A point-topoint link, such as the one between your personal computer and its keyboard, is an example o f a pipe. Switches direct signals arriving from one point to other points; the telephone system, for example, contains many switches to direct calls locally, nationally, or globally. With enough pipes and switches, a communications system can make all the connections you need. The pipes and switches view of communications is obviously an oversimplification, but it can be useful for understanding the basic designs o f communications systems. We’ll consider three basic approaches to connectivity: broadcasting, switching, and networking. These approaches are shown in Figure 3.9.

Broadcasting A broadcast system sends the same signal to everyone who receives it. In its usual simple form, broadcast transmission is one-way, from the signal source to many individuals. Local radio and television stations are good examples. Their transmitters radiate signals from a main antenna that can be picked up by receivers throughout the local area. Satellite television works the same way; a satellite broadcasts microwaves, which home receivers detect and decode. Broadcasting may not reach everyone within range, because some people don’t have antennas, televisions, or satellite television service. Broadcasting doesn’t have to be through the air. Cable television systems broadcast signals through a network o f optical fibers and coaxial cables. Most subscribers receive

Broadcasting sends the same signal to all points.

Chapter 3

the same signals; but premium channels are broadcast in a scrambled form that can be decoded only by using special equipment. Cable television systems set aside certain channels for two-way service. These channels enable them to provide telephone and broadband data service, and allow customers to order special services, such as pay-perview programs. Signals transmitted through the air are not broadcast if only one person or receiver can receive them. For example, cellular telephone conversations are transmitted through the air by local towers, but only the person with the proper cell phone can decode signals directed to that phone. Other phones and antennas working in the same frequency range can pick up the signals, but cannot decode them.

Switched Systems A switched system makes temporary connections between points.

A sw itched system makes temporary connections between terminals so they can exchange information. The telephone system is a familiar example. An old-fashioned telephone switchboard made physical connections between phones when the operator plugged wires into the jacks. Today, electronic switches in a telephone company switching office perform the same function for local phone calls, completing a circuit that links your telephone to the phone you are calling. These switches also route long-distance calls through a network of other switches that direct them to their destinations. When these switches make a connection, they dedicate transmission capacity between the two phones for as long as they stay on the line. For a local call, the connection goes through phone wires from your home to the local switching office, then from the switching office through wires to the other caller’s phone. For a long-distance call, the switches reserve a voice channel providing room for one phone call on a multiplexed line that carries many separate conversations on multiple voice channels. Once the call is complete, the line is released and can be used for other calls. The key aspects o f a switched system are the following: They make temporary connections, dedicate transmission capacity between a pair o f nodes or terminals, and make connections between any pair o f terminals attached to the switch. Our example used wired telephones, but cell phones work the same way, except that the signals are transmitted through the air rather than over wires.

Networking Networked computers are interconnected so they can communicate with each other.

The word network has many meanings. One o f these is a specific architecture for connecting computer terminals. A computer network consists o f many terminals connected so they can send signals to one another. These connections are always “on,” so each terminal can send messages to any other terminal at any time. The signal that carries the message may pass through many terminals, but it is addressed specifically to one terminal, so other terminals cannot read it. (For our purposes, we assume that system security can’t be breached.) The permanent connections are like streets on which delivery trucks travel to deliver parcels to your home. An office local-area network (LAN) is a good example o f this type of network, as shown in Figure 3.10. A message from one terminal to another may go through the whole network,

Fundamentals of Communications

P e rs o n a l C o m p u te r

F ile S e rv e r

L a s e r P rin te r

FIGURE 3.10 An office localarea network (LAN).

but the only terminal that receives the message is the one to which it’s addressed. The results look superficially like the switched telephone system, but the details differ considerably, as you will see later. Note, for example, that the networking approach does not set aside any transmission capacity to send a particular signal. Networking lies at the heart o f packet switching, which directs messages by a process called routing. As you will learn later, routing differs greatly from the type o f switching described previously.

Transmission Media and Switching Technologies So far, we’ve largely ignored the media used for pipes and switches. In a general context, the nature o f the pipes doesn’t matter a lot. Telephone calls made over wired phones and cell phones serve the same purpose, even though they are transmitted and directed in different ways. An office local-area network can be implemented over wires, fiber-optic cables, or wireless links. Cable television does differ significantly from over-the-air broadcast television because it allows two-way transmission, but the differences are largely due to the switches. Switching technology matters more because it can constrain the system performance. In the early days o f the telephone system, switching was done mechanically. Now most switching is done electronically. Electronic switches operate in two different ways: by setting the path that signals will follow (as in making a telephone connection), or by reading the address transmitted with a data message and directing that message to the proper address. During the bubble, prospects for optical switching created a concept called optical networking. The idea was overpromoted, but included an important concept that shouldn’t be lost. Because fiber optics have the highest capacity o f any current transmission medium, they form the high-capacity core o f the modern telecommunications network— the communications equivalent o f superhighways in our network o f roads and highways. Optically

Switching technology can constrain performance.

Chapter 3

switched signals could remain in optical form throughout the high-capacity fiber-optic core of the network. Keeping the signals in optical form would allow them to remain organized in the same way they are arranged for optical transmission, with separate signals carried on different wavelengths in the same fiber. Interest in optical networking evaporated when the bubble ended, but may revive as Internet traffic continues to increase.

Communications Services So far we’ve talked about communications from a general standpoint. Now let’s look at specific types o f communications services. We won’t try to cover everything. Rather, we will concentrate on the types of services most likely to reach modern homes and businesses: voice, video, and data. The three were originally quite distinct services, but have come to converge as different networks offer the same services.


Home phones are part of the local loop and connect to a local switching office.

Voice communications, the oldest o f the services, is traditionally defined as telephony and delivered to subscribers through copper wires. The telephone system is a switched network that offers global connectivity, so you can call any continent from your phone. You can loosely divide the telephone system into a hierarchy o f subsystems, shown in simplified form in Figure 3.11. Your home or business phone is part o f the base o f the phone network, variously called the subscriber loop, the local loop, or the access network. These links run from individual

FIGURE 3.11 Elements o f the telephone network.

R e gion al c a rrie rs

Fundamentals of Communications

telephones to local telephone switching offices (called central offices in the industry). A typical central office serves thousands o f homes either directly or through feeder cables that carry signals to neighborhood distribution nodes. Central offices multiplex calls directed outside their local areas on trunk lines, which run between central offices, connecting cities, towns, and suburbs in the area. Together with other links between telephone company facilities and large organizations, trunk lines form the regional or metro network that carries telephone traffic in the area. The access and metro networks— or subscriber loop and regional networks— are the realm o f regional carriers or local exchange carriers that provide services within a region. The best known o f these carriers are the regional B ell operating companies (RBOCs), sometimes called ILECs for Incumbent Local Exchange Carriers: Bell South, SBC, Verizon, and Qwest. Their smaller competitors, called com petitive local exchange carriers (CLECs), offer regional telephone services, usually over the phone lines installed by the incumbents. You’ll learn more about regional phone networks in Chapter 24 and about local phone networks in Chapter 25. Many local phone systems also offer broadband Internet access via digital subscriber line (DSL). Unlike basic phone service, DSL is always on, so you don’t have to dial in as you do with a conventional modem. However, DSL transmission distances are limited, so the service is not available to homes far from the central office. Long-distance service is a separate business, but is now offered by most of the regional carriers as well as companies that specialize in long-distance service, such as A T& T and MCI. Local and regional service areas are distinguished on the basis of area codes that existed when the original seven Bell operating companies split from A T& T in 1984, so the maps of these areas may look strange today. Long-distance carriers pick up signals at regional nodes and transport them around the country via long-distance backbone systems. These backbone systems, in turn, connect with international systems such as submarine cables running to Europe, Asia, and South America. Although you see only one long-distance carrier on your phone bill, in reality calls may pass through multiple carriers to reach their destination. You’ll learn more about the global telephone network in Chapter 23. Cellular phones, pagers, and satellite phones link to this global telecommunications system. In a sense, the mobile services are like regional phone carriers that distribute signals locally through the air on radio waves rather than through cables. However, they generally provide their own long-distance service, which may be billed at the same rate as local service. Fiber-optic cables are used throughout the backbone and regional networks, and run to many of the local distribution nodes and cell phone towers shown in Figure 3.11. Large organizations, such as universities and big office buildings, often have direct fiber-optic connections. With a few exceptions, copper lines run from regional distribution nodes to homes and small businesses.

Optical fibers are the backbone of the global telecommunications network.

Cable Television and Video Cable-television networks now offer telephone service and broadband Internet access in competition with the telephone network. Superficially cable lines resemble telephone lines in that they distribute signals from a central facility (called a head-en d) to neighborhood nodes and individual homes. However, the network designs differ significantly, reflecting the origins of cable TV.

Cable TV systems now offer telephone and broadband Internet access.

Chapter 3

Sometimes called CATV (for Community Antenna TV ), cable T V systems originally distributed signals in areas outside the reach of standard broadcast television. A single tall antenna picked up distant broadcasts, and the system distributed the same signal to all subscribers. (It beat having everyone install their own antenna to pick up distant stations.) Cable later offered additional channels from distant “super stations” and from special services distributed by satellite links. Today virtually all cable systems offer two-way services, but their internal architecture differs from that o f local telephone systems. Video signals require much more bandwidth than sound signals, so cable T V systems use coaxial cable to connect to homes. The coax runs from neighborhood distribution nodes, attached via fiber-optic cables to the cable head-ends, which distribute signals in the community. As in the original cable systems, the basic design distributes the same signals to all subscribers in broadcast style. Premium services are transmitted in a coded form that requires special decoders. Modern cable systems have refined this design by setting aside channels that provide services in addition to broadcast-style video. These channels carry special services to and from individual neighborhood nodes, which distribute them to homes in a networked model. For example, cable modem service is really multiple subscribers sharing the capacity o f one cable channel, which functions as a local-area network serving those homes. As in an office network, each subscriber sees only messages addressed to them, not the other data traveling on the network. Cable systems use similar technology to offer telephone service. Cable TV systems now offer a combination of analog and digital services on separate channels. Digital signals typically are offered at premium rates with the analog services in the basic package. Special cable boxes are needed to receive the digital video signals. You will learn more about cable T V systems in Chapter 27.

Data Communications and the Internet Data communications include local-area networks and the Internet.

Most people think that all data communications goes over the Internet, but that is an illusion arising from the use of personal computers in homes and offices. There are two types o f personal links to the Internet: dial-up connections made by home modems over ordinary phone lines to companies that provide Internet service, or broadband connections using DSL or cable modems. Small businesses also may use these same types o f service, but larger businesses use special higher-speed services. Most organizations have internal localarea networks (LANs), such as the one shown in Figure 3.10, to transfer data among their own computers, and they use these local-area networks to link to the Internet. (Even home offices may have LANs to link computers to each other, a printer, and a cable modem or DSL.) Local-area networks typically run within a building or within a campus o f buildings that houses a large organization. As you will learn in Chapter 26, small LANs may be linked together to form larger wide-area networks that serve the whole organization. Individual personal computers typically connect over short lengths of special high-quality copper cable made for data transmission; Category 5 cables are widely used. Wireless connections also are possible. Fibers run to individual devices only if they require very high-speed connections; are far from the main network; or are in an environment where copper wires don’t work well, such as where heavy machinery generates strong electromagnetic noise.

Fundamentals of Communications

Backbone wide-area networks that serve large buildings, or links between corporate networks and remote sites or the Internet, typically use fibers. The longer the link and the higher the speed, the more likely fibers are to be used. Fibers also provide the backbone o f the Internet, which functions like the long-distance telephone network. The hardware is essentially the same as that used for long-distance telephone traffic, but the data-transmission protocols are different. You’ll learn more about data transmission standards in Chapters 20 and 26.

Convergence Voice, video, and data services have evolved considerably during the past decades. Modern versions of these systems strongly resemble each other and can provide similar services, an effect the industry calls convergence. The similarities are both more and less than they seem, and deserve a brief explanation. Both voice and data are transmitted over digital backbone systems, but the hardware and protocols differ in detail. The traditional telephone network is optimized for voice traffic with a separate voice channel reserved for each telephone call. The Internet is optimized for data traffic without fixed channels for separate data streams. Telephone lines have long carried digital data and now data lines can carry voice signals, a scheme called VoIP for Voice over Internet Protocol. Advocates claim VoIP will replace ordinary phone lines, but it sometimes sounds like a bad cell-phone connection. Connectivity is essential for both telephone and Internet traffic. You don’t want a voice line or an e-mail address that can’t connect to the rest o f the world’s phones and e-mail addresses. Somewhere along the line all voice and data signals must be identified and directed to the right place. Connectivity is not as essential for video, and video feeds to cable T V networks typically are via satellite links, which transmit encoded signals picked up by antennas on the ground. Depending on costs, the video signals may be distributed through fiber in a metropolitan area, but usually they are not transmitted long distances through fiber on the ground. The distinctions among the various long-distance telecommunications networks are more organizational than physical. In reality, one fiber-optic cable may contain fibers carrying different signals for different organizations. For example, a single fiber might carry telephone traffic at one wavelength and Internet traffic on another. Companies may trade capacity on their cables with other organizations that have excess capacity on different routes.

Other Communications Services Voice, video, and Internet data are the three major services of the telecommunications world, but you should be aware of other communications services that fill specific market niches. • B roadband fix ed wireless services use microwave antennas to provide broadband service in regions where companies think it would be too expensive to install cables. The idea has been around for a while, but has never caught on widely. 9 D ata transmission inside vehicles is becoming more important as the number of control systems increases. Sometimes fiber is used for vehicle systems because of its

Voice, video, and data services are converging, and can be offered over one network.

Chapter 3

broad bandwidth and immunity to electromagnetic interference. Fibers are used in ships, some aircraft, the International Space Station, and some high-end automobiles, as described in Chapter 28. • Closed circuit video transmission is needed for surveillance, monitoring, and broadcasting o f sports events. Fibers have high bandwidth, are lightweight, and suffer little interference, so they often are used in these applications.

The Business of Telecommunications Understanding telecommunications technology requires learning a little about the business of telecommunications. Telecommunications is a complex industry that involves different companies offering different goods and services. Business considerations have shaped the present network.

A Very Short History of the Telecommunications Business • Telephone service •I j was consiaerea a na ura monopoy.

Private companies started the telephone industry in the late 1800s, but government agencjes became involved as the telephone became a vital service. Governments considered teler pbone service to be a “natural monopoly” because they felt it only made sense to build a single telephone system to serve all homes and businesses. Through most o f the twentieth century, most telephone service outside the United States was run by government post, telephone, and telegraph agencies. In the United States, private telephone service was heavily regulated by state and federal agencies, and most of the nation’s cities were served by a single giant company, AT&T. This began to change in the late 1970s, as other companies began to offer long-distance service. In 1984, A T& T spun off seven regional operating companies, three o f which have since disappeared in mergers. Overseas, telephone agencies were separated from post offices and privatized. Cellular telephone networks emerged in the same period, and now handle a large share of telephone traffic. Cellular service is competitive, with multiple companies offering service across the United States. The trend in the cable industry is also toward consolidation. Cable T V began as small companies scattered around the country, but is now dominated by a handful o f M ultiple System Operators (MSOs) such as Comcast and Time-Warner. Internet services now are offered by telephone and cable companies, and by other companies including Microsoft, AOL, and Earthlink. The telecommunications bubble pumped a tremendous amount o f money into the industry, which companies used to expand and to buy other companies, often at greatly inflated prices. New companies tried to build “overlay” networks that provided services in competition with existing phone and cable systems, but only a few o f them survived the collapse o f the bubble. Most o f the local “competition” that remains today is based on regulations that require phone companies to lease their transmission lines to other companies that want to provide phone service. Cable companies aren’t required to lease their lines, but sometimes allow other companies to offer broadband service over their cables.

Fundamentals of Communications


Regulations Telecommunications has always been a regulated industry. Although the changes o f recent years have been called “deregulation,” it might be better to call them “changes in regulation.” Governments used to set the prices that carriers could charge. Now they write rules that require carriers to lease their lines to competitors at specified prices and under certain conditions. These regulations assure the public access to essential services at reasonable cost and protect the public from crooks and fools. (The crooks and fools made their presence evident during the bubble years when WorldCom faked its accounts and Enron tried

to create a market for bandwidth trading. Most of us consider access to at least basic telephone service at honest prices to be essential in today’s society.) W hat regulations are proper? Does the Federal Communications Commission (FC C ) act in the public interest, as its charter specifies? Or do corporate lawyers and lobbyists who contribute generously to political campaigns shape its regulations? FCC policies can’t help but shape the future o f the telecommunications industry. Are Congress and the F C C too concerned with prom oting corporate profits and issues like preventing copying o f digital music and video? These are questions that deserve serious debate.

Types of Businesses The telecommunications industry includes distinct types o f companies that earn money in different ways. The most important types include the following: • Carriers transmit information over lines that they own or rent. Local and longdistance telephone companies are both carriers, and some companies provide both services. Cable T V systems also function as carriers. They sell transmission service as well as access to networks. • Carriers’ carriers lease capacity on transmission lines they built to carriers and other companies who need service. They don’t retail to individuals or small businesses. They are wholesalers that provide service to other companies. • Internet Service Providers (ISPs) provide Internet access, and related services such as Web hosting and e-mail. They range from giant corporations to small independent companies, and usually are retailers o f services. • Equipm ent manufacturers make hardware and software that they sell to companies and individuals. Their products range from expensive hardware for long-distance networks to desktop cables and modems. •

Contractors an d installers install hardware for carriers. Many specialize in construction.

What Have You Learned? 1. Telecommunications sends signals over a distance through such media as optical fibers, copper wires, and radio waves.


Chapter 3


2 The optical telegraph was the first form o f telecommunications 200 years ago. The electrical telegraph made it obsolete. Telephones followed, then broadcast radio and television.


3 Improvements in electronic technology allowed operation at increasingly higher radio frequencies, which offered more bandwidth for signal transmission.


4 The demand for transmission capacity has increased steadily, and bandwidth has traditionally been in short supply. 5. Analog bandwidth is measured by the frequency range. Digital bandwidth is measured in bits per second.

6. Multiplexing combines many signals into one higher-speed signal. The important types are frequency-division multiplexing, wavelength-division multiplexing, and time-division multiplexing. Time-division multiplexing works only for digital signals.


7 Copper is a general term that includes twisted pair and coaxial cables.


Telecommunications systems transmit signals by modulating a carrier wave with a baseband signal. The baseband signal is at a lower frequency than the carrier.

. 10. Analog signals vary continuously. Digital signals transmit a series o f bits. Analog 9 Fiber-optic systems use amplitude modulation.

signals can be digitized, then converted back to analog format. Analog telephone signals are digitized at 64,000 bits per second. 11. Electrical signals are current or voltage. Optical signals are the number of photons.

12. A communications system can be viewed as an array of pipes and switches. The pipes carry signals, and the switches direct them. 13. Broadcast communications directs the same signal to many points. 14. A switched system makes temporary connections between terminals. The telephone system is an example. 15. Networking interconnects computers permanently so they can send messages to each other. Each message carries a label so only one computer receives it. 16. The telephone system includes the local loop, which connects subscribers, local switching offices, regional or metro networks, and a long-distance backbone system. Optical fibers provide the backbone. 17. Cable T V systems resemble telephone systems locally. They offer television and broadband Internet access as well as video. 18. Data communications includes local-area networks and the Internet. 19. Convergence is the merging of voice, video, and data services so they can be offered over one network.


20 Teleph one service was considered to be a natural monopoly and was run by government agencies or heavily regulated private monopolies. Most countries’ telephone networks are now private and regulated differently.

Fundamentals of Communications

What's Next? Now you have a general idea how fiber optics and telecommunications work. The rest of the book will present more details about the technology. Chapters 4 through 7 cover optical fibers and their important features.

Further Reading Roger L. Freeman, Fundamentals o f Telecommunications (Wiley InterScience, 1999) Gil Held, Voice an d D ata Internetw orking (McGraw-Hill, 2000) Anton A. Huurdeman, The Worldwide History o f Telecommunications (Wiley InterScience, 2003) Gary M. Miller, M odern Electronic Communication (Prentice Hall, 1999) Tom Standage, The Victorian Internet (Berkeley Books, 1998)

Questions to Think About 1. How does using a higher-frequency carrier affect the amount o f information that can be transmitted?


2 Why does multiplexed transmission o f a combined signal cost less than separate transmission o f each signal?


3 Computer networks, mobile telephones, and broadcast systems all distribute signals to many terminals. How do these systems differ?


4 Why is frequency-division multiplexing equivalent to wavelength-division multiplexing?


5 The bandwidth of digitized signals measured in bits per second is much higher than the bandwidth of the original analog signal measured in hertz. For example, the analog bandwidth of a phone line is 4000 Hz, but the digitized signal is 64,000 bits per second. Why are these two bandwidths usually equivalent in practice?

6. Why was telephone service considered to be a natural monopoly? 7. Data transmission rates to personal computers have increased from 1200 bits per second with a dial-up modem in 1985 to about 400,000 bits per second with a cable modem or DSL in 2000. If bandwidth keeps increasing at the present rate, how fast will transmission be in 2015?

Chapter Quiz 1.

Which came first? a. electrical telegraph b. optical telegraph

Chapter 3

c. telephone d. wireless radio transmission

2 . Which o f the following statements are true for analog signals? a. They vary continuously in intensity. b. They are transmitted in parts of the telephone network. c. They are compatible with human senses. d. They can be processed electronically. e. All o f the above

3 . Which o f the following statements are true for digital signals? a. They can encode analog signals. b. They are transmitted in parts o f the telephone network. c. They can be processed electronically. d. They are used in computer systems. e. All o f the above

4 . You digitize a 10-kHz signal by sampling it at twice the highest frequency (i.e., 20,000 times a second) and encoding the intensity in resulting data rate? a.


8 bits. What

is the


b. 64 kbit/s c. 144 kbit/s d. 160 kbit/s e. 288 kbit/s

5 . What part o f the telephone network is connected directly to your home telephone if you get your telephone service from a local telephone company? a. subscriber loop b. feeder cable c. trunk line d. backbone system

6. What part of the telephone network carries the highest-speed signals? a. subscriber loop b. feeder cable c. trunk line d. backbone system

7 . Time-division multiplexing o f eight signals at 150 Mbit/s each produces a. eight optical channels each carrying 150 Mbit/s. b. one channel carrying 120 Mbit/s. c. one channel carrying



d. eight signals at 150 MHz.

Fundamentals of Communications


. What are the “pipes” used to broadcast television signals from a station on the ground?

a. air b. optical fibers c. coaxial cables d. twisted pair

9 . The carrier signal modulated to produce one optical channel in a fiber-optic system is a. a single wavelength o f light generated in the transmitter. b. a radio-frequency signal supplied electronically to the transmitter. c. an acoustic vibration in the optical fiber. d. a combination o f wavelengths generated by several light sources. 1 0 . Who offers DSL and what service does it normally provide? a. Television broadcasters offer it for Internet access. b. Cable television carriers offer it for Internet access. c. Telephone carriers offer it for Internet access. d. Cable television carriers offer it for telephone service. e. Internet service providers offer it for telephone service. 1 1 . What is the only important telecommunications system that uses fiber to transmit analog signals? a. local telephone service b. long-distance telephone systems c. Internet backbone systems d. cable T V systems e. none 1 2 . What U.S. government agency regulates telecommunications? a. Federal regulations have been abolished. b. Department of Homeland Security c. Department o f Commerce d. Federal Communications Commission e. International Telecommunications Union

Types of Optical Fibers

About This Chapter Not all optical fibers are alike. Several different types, made for different applications, guide light in different ways. This chapter describes the basic concepts behind standard fibers, concentrating on fiber design and light guiding. It is closely linked to the chapters that follow. Chapter 5 describes the important properties o f optical fibers. Chapter 6 covers fiber materials, structures, and manufacturing, which play a vital role in determining fiber properties. Chapter 7 covers specialty fibers used in amplifiers, wavelength selection, and applications other than merely guiding light.

Light Guiding Chapter 2 showed how the total internal reflection o f light rays can guide light along optical fibers. This simple concept is a useful approximation of light guiding in many types o f fiber, but it is not the whole story. The physics of light guiding is considerably more complex, because a fiber is really a waveguide and light is really an electromagnetic wave with frequency in the optical range. Like other waveguides, an optical fiber guides waves in distinct patterns called modes, which describe the distribution of light energy across the waveguide. The precise patterns depend on the wavelength of light transmitted and on the variation in refractive index that shapes the core, which can be much more complex than the simple, single cores described in Chapter 2. In essence, these variations in refractive index create boundary conditions that shape how electromagnetic waves travel through the waveguide, like the walls o f a tunnel affect how sounds echo inside.


Total internal reflection is only a rough approximation of light guiding in optical fibers.

Core-cladding structure and material composition are key factors in determining fiber properties.

It’s possible to calculate the nature o f these transmission modes, but it takes a solid understanding o f advanced calculus and differential equations, which is far beyond the scope of this book. Instead, we’ll look at the characteristics of transmission modes, which are important in fiber-optic systems. By far the most important is the number o f modes the fiber transmits. Fibers with small cores can transmit light in only a single mode. It can be hard to get the light into the fiber, but once it’s inside, the light behaves very uniformly. It’s easier to get light into fibers with larger cores that can support many modes, but light does not behave the same way in all the modes, which can complicate light transmission, as you will learn later in this chapter. This chapter covers the many types of optical fibers that have been developed to meet a variety of functional requirements. Their designs differ in important ways. For example, bundles o f fibers used for imaging need to collect as much light falling on their ends as possible, so their claddings are made thin compared to their cores. Communications fibers have thicker claddings, both to keep light from leaking out over long distances and to simplify handling of single fibers. Various types o f communications have their own requirements. Fibers for short links inside cars or offices typically have large cores to collect as much light as possible. Long-distance fibers have small cores, which can transmit only a single mode, because this well-controlled light can carry signals at the highest speed. The two considerations that affect fiber properties most strongly are the core-cladding structure and the glass composition. The size o f the core and cladding and the nature of the interface between them determine the fiber’s modal properties and how it transmits light at different wavelengths. The simple types o f fiber discussed in Chapter 2 have a step-index structure, where the refractive index changes sharply at the abrupt boundary between a high-index core and a low-index cladding. Replacing that abrupt boundary with a gradual transition between core and cladding, or including a series o f layers, changes fiber properties. Glass composition, covered in Chapter 6 , strongly affects fiber attenuation, as well as influencing pulse spreading. Combined with other minor factors, these parameters determine important fiber characteristics, including ^

Attenuation as a function o f wavelength.

Collection o f light into a fiber (coupling).

Transmission modes.

Pulse spreading and transmission capacity, as a function o f wavelength.

Tolerances for splicing and connecting fibers.

Operating wavelengths.

Tolerance to high temperature and environmental abuse.

Strength and flexibility.

Figure 4.1 shows selected types o f single fibers (as distinct from bundled fibers), along with a plot o f refractive index across the core and cladding, called the index profile. Only the core and cladding are shown for simplicity; actual fibers have an outer plastic coating to protect them from the environment. The coating’s thickness depends on fiber size. For

Types of Optical Fibers

FIGURE 4.1 Common types o f optical fib e r (to scale). IT U designations are standards o f the International Telecommunications Union.



Step-lndex Multimode Fibers





125 p m — »■



\ ______

— In d e x P ro file

125 p m — * -



b. Graded-lndex Fibers (50/125 is ITU G.651)



W -80 p e n



c. Step-lndex Single-Mode Fiber (ITU G.652)

d. Nonzero Dispersion Shifted Single-Mode Fiber (ITU G.655)

In d e x P ro file

e. Reduced Core Step-lndex Single-Mode Fiber

Chapter 4

a typical communications fiber with 125-|xm cladding, the plastic coating is 250 |xm. I will start with the fiber type that is simplest to explain in terms o f total internal reflection, called step-index multimode fiber, because it transmits many modes.

Step-index Multimode Fiber As we saw in Chapter 2, bare, transparent filaments surrounded by air are the simplest type o f optical fiber, but they don’t work well in practice. Cladding the fiber with a transparent material having lower refractive index protects the light-carrying core from surface scratches, fingerprints, and contact with other cores of the same material, so the light will not escape from the surface. This simple fiber consists of two layers of material, the core and cladding, which have different refractive indexes. If you drew a cross section of the fiber and plotted the refractive index, as in Figure 4.1(a), you would see a step at the corecladding boundary, where the index changes abruptly.

Light-Guiding Requirements • To guide light, the fiber coremust ave re rac ive index hiqher than the cladd'ng

As long as the core o f a fiber has a diameter many times larger than the wavelength o f light jt carries, we can calculate fiber properties using the simple model o f light as rays. The fundamental requirement for light guiding is that the core must have a higher refractive index t^ an ^ d a t i n g material. We saw in Chapter 2 that the critical angle for total internal n . „ . , , c j i j j c • ■ j reflection, 0 cr;t, depends on the ratio or core and cladding retractive indexes. © crit

= arcsin


For a typical fiber, the difference is small, about 1%, so the critical angle is arcsin (0.99), or about 82°. This means that light rays must be within 8 ° o f the axis o f the fiber to be confined in the core, as shown in Figure 4.2. This value is called the confinem ent angle, ^confinement’ ar*d equals 90° — 0cr;t. The angle is not very sensitive to the refractive-index

FIGURE 4.2 Light guiding in a large-core stepindex fiber. The confinem ent angle measures the angle between guided light rays an d the fib er axis; the acceptance angle is measured in air.

Only light falling in this angle is guided along the fiber.

Critical Angle for Total Internal Reflection

Types of Optical Fibers

difference. If the difference is doubled to 2%, the confinement angle becomes 11.5°. You can directly calculate the confinement angle measured from the core-cladding boundary using the arc-cosine: ® c o n fin em e n t

arcC O sf


y "core J

The confinement angle gives the maximum angle at which guided light can strike the corecladding boundary once it’s inside the glass. However, refraction occurs when the light enters the glass from air, bending light toward the axis o f the fiber. To calculate the acceptance angle, measured in air, you must account for this refraction using the standard law o f refraction. As long as the light enters from air, you can simplify this to

sin O half-acceptance

^ c o re ^

^ co n fin e m e n t

which gives the sine of the largest possible angle from the axis o f the fiber, called the halfacceptance angle, O half-acceptance- You can calculate the half-acceptance angle directly by juggling the trigonometry a bit more: O h alf-acceptan ce

The confinement angle is the largest angle at which light rays confined to a fiber core strike the core-cladding boundary.

arcsin(/tcore X sin O co n fin em en t)

Doubling the half-acceptance angle gives the full-acceptance angle. The confinement angle is small enough that you can roughly approximate the half-acceptance angle by multiplying the confinement angle by the refractive index o f the core, nCOK.

Imaging Fibers The first clad optical fibers developed for imaging were what we now call step-index multimode fibers. Developers tested a variety of cladding materials with low refractive indexes, including margarine, beeswax, and plastics. However, the key practical development was a way to apply a cladding of glass with lower refractive index than the core. As we will see in Chapter 6 , glass comes in many different formulations with varied refractive indexes. The simplest way to make glass-clad fibers is to slip a rod of high-index glass into a tube with lower refractive index, heat the tube so the softened glass collapses onto the rod, let them fuse together, then heat the whole preform, and pull a fiber from the molten end. The cladding of imaging fibers generally is a thin layer surrounding a thicker core. The reason for this design is that imaging fibers are assembled in bundles, with light focused on one end o f the bundle to emerge at the other. Light falling on the fiber cores is transmitted from one end to the other, but light falling on the cladding is lost. The thinner the cladding, the more light falls on the fiber cores and the higher the transmission efficiency. Reducing the size o f individual fibers increases the resolution o f images transmitted through a bundle, but very fine fibers are hard to handle and vulnerable to breakage. Typically, the smallest loose fibers used in imaging bundles are about 20 |xm (0.02 mm, or 0.0008 in.). Even at this size, they remain large relative to the wavelength o f visible light (0.4 to 0.7 |xm in air), and you can get away with considering light guiding as determined by total internal reflection o f light rays at the core-cladding boundary. (The highest-resolution fiber bundles are made by melting fibers together and stretching the whole solid block.)

Step-index multimode fibers were the first fibers developed for imaging.

Chapter 4

Illuminating and Beam-Delivery Fiber Large-core stepindex fibers are used to deliver laser power.

Single step-index fibers with large cores— typically 400 p,m to 1 mm— can be used to guide a laser beam from the laser to a target or industrial workpiece. The large diameter serves two purposes. First, it can collect power from the laser more efficiently than a smaller core fiber. In addition, it spreads the laser power over a larger area at the ends of the fiber and through a larger volume within the fiber. This is important because some laser power inevitably is lost at the surfaces and within the fiber. If the beam must be focused tightly to concentrate it in the fiber, the power density (power per unit area) may reach levels so high it can damage exposed ends o f the fiber. The design o f these large-core fibers is similar to those in Figure 4.1(a). The core diameters are proportionally larger, whereas cladding thicknesses do not increase as rapidly. As the fibers become thicker, they also become less flexible.

Communications Fibers Light pulses stretch out in length and time as they travel through large-core step-index fiber.

FIGURE 4.3 Light rays that enter multimode step-index fib er at different angles travel different distances through the fiber, causing pulse dispersion.

Step-index multimode fibers with cores not quite as large can be used for some types of communications. One smaller type, shown in Figure 4.1(a), has a 100-p.m core surrounded by a cladding 20 ptm thick, for total diameter of 140 p,m. It is typically called 100/140 fiber, with the core diameter written before the overall diameter o f the cladding. Typically an outer plastic coating covers the whole fiber, protecting it from mechanical damage and making it easier to handle. The large core is attractive for certain types o f communications, because it can collect light efficiently from inexpensive light sources such as LEDs. If you think o f light in terms of rays, you can see an important limitation o f large-core step-index fibers for communication (see Figure 4.3). Light rays enter the fiber at a range o f angles, and rays at different angles travel different paths through the same length of fiber. The larger the angle between the light ray and the axis, the longer the path. For example, a light ray that entered at 8 ° from the axis (the maximum confinement angle in the earlier example) o f a perfectly straight 1-m length o f fiber would travel a distance of 1.0098 m (1 m/cos 8 °) before it emerged from the other end. Thus light just inside the confinement angle would emerge from the fiber shortly after light that traveled down the middle. This pulse-dispersion effect becomes larger with distance and can limit datatransmission speed. In fact, the ray model gives a greatly simplified view o f light transmission down optical fibers. As I mentioned earlier, an optical fiber is a waveguide that transmits lightwaves

Types of Optical Fibers

in one or more transmission modes. Stay tuned for the next section, and I’ll explain more about these modes. The larger the fiber core, the more modes it can transmit, so a stepindex fiber with a core o f 20 |xm or more is a multimode fiber. Light rays enter the fiber at different angles, and the various modes travel down the fiber at different speeds. What you have as a result is modal dispersion, which occurs in all fibers that carry multiple modes. It is largely irrelevant for imaging and guiding illuminating beams, but it is a serious drawback for communications. To understand why, we need to take a closer look at modes.

Modes and Their Effects Modes are stable patterns that waves form as they pass through a waveguide. The number of modes that can travel along a waveguide depends on the wavelength of the wave and the size, shape, and nature o f the waveguide. For an optical fiber, the dominant factor is the core diameter; the larger the core, the more modes the fiber can carry. This leads to a fundamental trade-off between the higher signal quality possible with single-mode transmission and the easier input coupling with larger-core fibers. Waveguide theory, which describes modes, originally was developed for microwaves, but can be applied to any guided electromagnetic waves— including light passing through the core of an optical fiber. You don’t want to worry about the mathematical details o f waveguide theory— and I certainly don’t— but it is important to learn some basic concepts about waveguides and modes. Electromagnetic waves are oscillating electric and magnetic fields, and how they oscillate in a waveguide depends on how they are confined. The best-known microwave waveguides are rectangular metal tubes, but flexible plastic rods called dielectric waveguides also can guide microwaves. (Dielectric means electrically insulating.) A dielectric microwave waveguide is equivalent to a bare optical fiber, with the surface guiding the waves— so anything touching the surface causes losses. In a clad optical fiber, the guiding dielectric surface is the boundary between core and cladding, where the refractive index changes. In the ray model o f light propagation, light guided in the fiber is totally reflected at this boundary. But waveguide theory reveals that a small fraction o f the light actually extends beyond the core into the inner part o f the cladding, which leads to some complications. As long as the fiber core is big enough to accept any light, it can carry light in the lowestorder mode, where the electric field intensity is highest at the center of the core and drops to the sides, as shown at left in Figure 4.4. As the core diameter increases beyond a certain point, called the cu toffwavelength, the fiber can support transmission in additional modes. The two curves at right in Figure 4.4 show the second and third lowest-order modes. Fiber cores support many modes simultaneously, with the number increasing very rapidly with the core diameter. The difference in refractive index between the core and cladding also influences the number o f modes. An optical fiber is a cylindrical waveguide. It’s also possible to make planar optical waveguides as stripes o f high-index material on a substrate with lower refractive index. You will learn more about planar waveguides later.

• Small-core fibers carry arsing e

^ ^ fiber is a ec rlc op ica


Chapter 4

Single-Mode Waveguides Single-mode fibers must have small

Conventional microwave waveguides carry a single mode. Multimode microwave waveguides don’t work well because interactions between the modes generate noise. Singlemode transmission is cleaner and simpler, and it’s also preferred for fiber-optic systems. The main limitation is that the core of the fiber must be small enough to restrict transmission to a single mode, yet large enough to collect most of the input optical signal. The balance is struck by adjusting the difference between core and cladding refractive index. The smaller the core-cladding difference, the larger the core can be. The refractive index difference is large for a bare glass fiber (with n — 1. 5) in air (with n = 1.000293), so the core must be around 1 pm to transmit only a single mode at the usual transmission wavelengths. Standard single-mode telecommunications fibers have a cladding index only about 0.5% lower than the core index; this allows core diameters above 9 pm, which is several times the 1. 5 -pm wavelength used for long-haul transmission. Larger-core fibers carry multiple modes. In practice, transmission is much better when a fiber carries many modes than when it carries a few, so there is a large gap between singlemode fibers with core diameters below 10 pm and multimode fibers with core diameters 50 pm or larger.

Modal Properties Some light penetrates into the cladding.

Although the core-cladding boundary is nominally the surface o f the waveguide in a clad optical fiber, the light energy does not really propagate along that boundary. Some light penetrates the boundary and goes a short distance into the cladding, while most o f the light remains inside the core. This effect occurs in all types o f clad fibers, but is most important in single-mode fibers, where it is characterized by the m ode-field diam eter, which is slightly larger than the core diameter. Technically, the mode-field diameter is the point where light intensity drops to l/e2(0.135) o f the mode’s peak intensity. Figure 4.4 shows the distribution o f light energy in modes, while Figure 4.5 shows the path of light in a single-mode fiber.


M o d e -F ie ld D ia m e te r

Electric field s fo r the lowest-order mode in an optical fib er (left) an d fo r the second-order an d third-order modes (right). Higherorder modes are more complex. M ode

M ode

M ode

Types of Optical Fibers

FIGURE 4.5 C la d d in g

Light penetrates slightly into cladding.

Light penetrates slightly into the cladding o f a single-m ode stepindex fiber.

Mode-field diameter is slightly larger than core. In te n s ity P ro file o f L ig h t in L o w e s t-O rd e r M ode

Light leakage into the cladding makes cladding transmission important, although not as critical as for the core. Guided waves travel mostly in the core in single-mode fibers. In multimode fibers, some modes may spend more time in the cladding than in the core. Modes are sometimes characterized by numbers. Single-mode fibers carry only the lowest-order mode, assigned the number 0. Multimode fibers also carry higher-order modes. The number o f modes that can propagate in a fiber depends on the fiber’s numerical aperture (or acceptance angle) as well as on its core diameter and the wavelength o f the light. For a step-index multimode fiber, the number o f such modes, N m, is approximated by Modes = 0.5

core diameter X NA X ir

N m = 0.5


txD X NA

where X is the wavelength and D is the core diameter. To plug in some representative numbers, a 100-ptm core step-index fiber with NA = 0.29 (a typical value) would transmit thousands o f modes at 850 nm. This formula is only an approximation and does not work for fibers carrying only a few modes.

Leaky Modes Low-order modes are better guided than the higher-order modes in a multimode fiber. Modes that are just beyond the threshold for propagating in a multimode fiber can travel for short distances in the fiber cladding. In this case, the cladding itself acts as an unclad optical fiber to guide those cladding modes.

Some modes can propagate short distances in the cladding of a multimode fiber.

Chapter 4

Because the difference between guided and unguided modes is small, slight changes in conditions may allow light in a normally guided mode to leak out o f the core. Likewise, some light in a cladding mode may be recaptured. Slight bends o f a multimode fiber are enough to allow escape o f these leaky modes.

Modal-Dispersion Effects • Modal dispersion n multimode steplaT ester$ . T argest type o

Each mode has its own characteristic velocity through a step-index optical fiber, as if it were a light ray entering the fiber at a distinct angle. This causes pulses to spread out as they travel along the fiber in what is called m odal dispersion. The more modes the fiber transmitS’ the m ° re PU1S£S Spread 0Ut'

Later we will see that there are other kinds o f dispersion, but modal dispersion is the

largest in multimode step-index fibers. Precise calculations o f how many modes cause how much dispersion are rarely meaningful. However, you can make useful approximations by using the ray model (which works for multimode step-index fibers) to calculate the difference between the travel times of light rays passing straight through a fiber and bouncing along at the confinement angle. For the typical confinement angle o f 8 ° mentioned earlier, the difference in propagation time is about 1%. That means that an instantaneous pulse would stretch out to about 30 ns (30 billionths of a second) after passing through a kilometer of fiber. That doesn’t sound like much, but it becomes a serious restriction on transmission speed, because pulses that overlap can interfere with each other, making it impossible to receive the signal. Thus pulses in a 1-km fiber have to be separated by more than 30 ns. You can estimate the maximum data rate for a given pulse spreading from the equation Data rate =

°J pulse spreading

Plug in a pulse spreading of 30 ns, and you find the maximum data rate is about 23 Mbit/s. In practice, the maximum data rate also depends on other factors. Dispersion also depends on distance. The total modal dispersion is the product of the fibers characteristic modal dispersion per unit length, D q , multiplied by the fiber length, L: D — Dq



Thus a pulse that spreads to 30 ns over 1 km will spread to 60 ns over 2 km and 300 ns over 10 km. (For very accurate calculations, you should replace L with Id , where 7 is a factor normally close to 1, which depends on the fiber type.) Because total dispersion increases with transmission distance, the maximum transmission speed decreases. If the maximum data rate for a 1-km length o f fiber is DR divided by refractive index: _


Cmat ~ ~n'T T ia t Thus the farther the light goes from the axis o f the fiber, the faster its velocity. The difference isn’t great, but it’s enough to compensate for the longer paths followed by the light

FIGURE 4.6 Refractive-index profile o f a with 62.5-fJim core.

R a d iu s (|tm )

Replacing the sharp boundary between core and cladding with a refractive-index gradient nearly eliminates modal dispersion.

Chapter 4

Graded-index fiber bends light back into core as the refractive index decreases (darker shading indicates higher refractive index).

FIGURE 4.7 The refractiveindex gradient in a graded-index fib er bends light rays back toward the center o f the fiber.

C la d d in g

Light goes faster in the low-index outer core, so it catches up with light in the higher-index center.

rays that go farthest from the axis o f the fiber. Careful adjustment o f the refractive-index profile— the variation in refractive index with distance from the fiber axis— can greatly reduce modal dispersion by equalizing the transit times o f different modes.

Practical Graded-index Fiber Standard gradedindex fibers have 50- or 62.5-pm cores.

Graded-index fibers were developed especially for communications. Standard types have core diameters o f 50 or 62.5 pm and cladding diameters o f 125 pm, although some have been made with 85-pm cores and 125-pm claddings. The 50-pm core fiber is covered by the International Telecommunications Union (ITU) G.651 standard. The core diameters are large enough to collect light efficiently from a variety o f sources. The cladding must be at least 20 pm thick to keep light from leaking out. The graded-index fiber is a compromise, able to collect more light than small-core single-mode fiber and able to transmit higher-speed signals than step-index multimode fibers. It was used in telecommunications systems extending farther than a few kilometers until the mid-1980s, but gradually faded from use in telephone systems because single-mode fibers offered much higher bandwidth. Recent improvements have improved the modal dispersion of graded-index fibers so they can carry higher-speed signals, but they remain limited to data communications and networks that carry signals no farther than a few kilometers.

Limitations of Graded-index Fiber Residual dispersion and modal noise limit performance of graded-index fibers.

Graded-index fibers suffer some serious limitations that ultimately made them impractical for high-performance communications. Modal dispersion is not the only effect that spreads out pulses going through optical fibers. Other types o f dispersion arise from the slight variation o f refractive index with the wavelength of light. These are present in graded-index fibers and became increasingly important as transmission moved to higher speeds. Chapter 5 will describe these dispersion effects. Multimode transmission itself proved a serious problem. Different modes can interfere with each other, generating what is called m odal noise. This appears as an uneven distribution of light across the end of the fiber, which continuously changes in response to very minor

Types of Optical Fibers

fluctuations, generating noise. Such modal effects also made it impossible to control precisely how fibers behaved when several were spliced together, because the light in some modes can shift into other modes or leak into the cladding at joints. In addition, ideal refractive-index profiles are very difficult to realize in practice. The refractive-index gradient must be fabricated by depositing many thin layers o f slightly different composition in a precisely controlled sequence. This is expensive, and some fluctuations from the ideal are inevitable. These limitations do not prevent graded-index fibers from being used in short systems, even at high speeds, as long as dispersion does not accumulate to high enough levels to limit data rates. However, single-mode fibers are standard for long-distance, high-performance systems.

Single-Mode Fiber The basic requirement for single-mode fiber is that the core be small enough to restrict transmission to a single mode. This lowest-order mode can propagate in all fibers with smaller cores. Because single-mode transmission avoids modal dispersion, modal noise, and other effects that come with multimode transmission, single-mode fibers can carry signals at much higher speeds than multimode fibers. They are the standard choice for virtually all kinds o f telecommunications that involve high data rates and span distances longer than a couple o f kilometers, and are often used at slower speeds and shorter distances as well. The simplest type o f single-mode fiber, often called standard single mode and designated ITU G .652, has a step-index profile, with an abrupt boundary separating a high-index core and a lower-index cladding. The refractive-index differential is 0.36% for a widely used fiber, and is well under 1% in other standard types. Figure 4.8 shows cross sections o f the two principal types o f step-index single-mode fiber made from fused silica.

* s'm(^es^^Pe ^ S' ^ 6 m° 6 WoffleSw^ith Qn Q[3rUpf boundary between a highindex core and a lower-index cladding,


C o re

D e p re s s e d In d e x In n e r C la d d in g

b. Depressed-Cladding Fiber

Two types o f stepindex single-mode fiber. The difference between core and cladding refractive index is the same, but in the depressed cladding fiber at the bottom, the inner cladding is doped with fluorine to reduce its refractive index.

Chapter 4

The simplest design is the matched-cladding fiber shown at the top o f Figure 4.8. The cladding is pure fused silica; germanium oxide (G e 0 2) is added to the core to increase its refractive index. An alternative design is the depressed cladding fiber shown at the bottom. In this case, the core is fused silica doped with less germanium oxide than is needed for a matched cladding fiber. The inner part o f the cladding surrounding the core is doped with fluorine, which reduces its refractive index below that o f pure fused silica. The outermost part o f the core is pure fused silica, without the fluorine dopant. Both these designs typically are widely used in telecommunications systems operating at 1.31 and 1.5 |JLm; core diameters are around 9 (Jtm.

Conditions for Single-Mode Transmission C*L



riper witn a small enough core iransmits only a . . |' r single mode ot


Earlier in this chapter, you saw that the number o f modes, N m transmitted by a step. index fiber depends on the fiber core diameter, D, the refractive indexes o f core (nq) and cladding («j), and the wavelength of light A.. You can write the formula in terms o f numer. /m a x ical aperture (NA): /t t D X N A V N m = 0.5 y “ ) You also can replace NA with the core and cladding indexes— useful because NA as acceptance angle isn’t very meaningful for single-mode fibers— and reformulate the equation: N m = 0.5 { ^


(n\ - n\)

Reducing the core diameter sufficiently can limit transmission to a single mode. By manipulating the mode-number equation and calculating a constant using Bessel functions, you can find the maximum core diameter, D, which limits transmission to a single mode at a particular wavelength, A: D chromatic


A X )2

For single-mode fiber, this formula is: ^ s in g le mode =


( D chromatic X I X

A X )2 +

( D PM D X

V l)2

A couple o f examples will show how dispersion calculations work.

MULTIMODE DISPERSION EXAMPLE In Example A, considered earlier in this chapter, an 850-nm LED sends 100 Mbit/s through 2 0 0 m o f 50/125-p.m fiber. M odal bandwidth o f a typical com m ercial 50/125-p.m fiber is 400 MHz, which is equivalent to a modal dispersion o f 2.5 ns/km. For a 200-m length, that corresponds to modal dispersion o f 0.5 ns. To that, you add the chromatic dispersion, calculated from the formula: ^ ^chromatic

-^chrom atic X Z. X A X

A typical value o f chromatic dispersion is 100 ps/nm • km at 850 nm, which combined with a linewidth of 50 nm for a typical 850-nm LED gives a chromatic dispersion of 1.0 ns for a 200-m length of fiber. This means that the chromatic dispersion is higher than modal dispersion because of the large LED spectral linewidth. Adding modal and chromatic dispersions together according to the sum-of-squares formula indicates total dispersion is 1.1 ns. That leaves plenty o f room for transmitting 100 Mbit/s, assuming the transmitter and receiver are fast enough.

Single-Channel System Design

If you wanted to transmit 1 Gbit/s, you could try using a VCSEL source with a 1-nm linewidth. In that case, total dispersion is At = \/(0.5 ns)m0dal + (100 ps/nm •km X 0.2 km X 1 nm)2 = 0.50 ns This is essentially equal to the modal dispersion o f 0.5 ns and is adequate for gigabit transmission over 200 m.

SINGLE-MODE TRANSMISSION EXAMPLE In Example B, we considered transmitting a 2.5-Gbit/s signal a total o f 3 00 km through a single-mode fiber at 1550 nm. Assume that chromatic dispersion is specified at 3 ps/nm - km, a relatively low value at 1550 nm. Let’s consider two cases: a FabryPerot laser with linewidth o f 1 nm and an externally modulated distributed-feedback laser with linewidth o f 0.1 nm. For a first approximation, ignore polarization-mode dispersion. For the 1-nm laser, chromatic dispersion is ACiromatic = 3 ps/nm •km X 300 km X 1 nm = 900 ps This value is much too high for a 2.5-Gbit/s system. With the distributed-feedback laser, chromatic dispersion is Achromatic = 3 ps/nm •km X 300 km X 0.1 nm = 90 ps Thus using fiber with low chromatic dispersion near 1550 nm leaves plenty of margin. However, you could not get away with using step-index single-mode fiber with dispersion around 20 ps/nm-km at 1550 nm. Archromatic = 20 ps/nm-km X 300 km X 0.1 m = 600 ps We also need to consider polarization-mode dispersion. A typical value for new nonzero dispersion-shifted fiber is about 0.5 ps/root-km, so for 300 km o f single-mode fiber, the average polarization-mode dispersion is A%md = 0-5 ps/Vkm X \/300 km = 8.7 ps This is only a tenth as large as chromatic dispersion, so it makes a negligible contribution to total dispersion, and does not affect transmission at 2.5 Gbit/s. To fully assess performance o f the system, you need to consider transmitter and receiver rise time as well. Suppose both have rise times o f 100 ps, reasonable for products designed to work at 2.5 Gbit/s. Then the total pulse spreading is A t = V 1 0 0 2 + 1002 + 902 = 168 ps Our earlier formula shows that this is adequate for transmitting 4 Gbit/s.

LONG-DISTANCE SINGLE-MODE EXAMPLE Long-distance systems must span longer distances o f several hundred to several thousand kilometers. Dispersion poses more problems in these systems because the total pulse

Dispersion poses more problems at long distances.

spreading increases with the length o f the system. The seriousness o f dispersion effects also increases with data rate. Suppose you are using nonzero dispersion-shifted fiber with a chromatic dispersion of 3 ps/km-nm and PMD o f 0.5 ps/root-km to span a distance o f 1000 km. As in the last example, you are using an externally modulated laser with spectral width o f 0.1 nm. Plug in the numbers and you find: Achromatic = 3 ps/nm •km X 1000 km X 0.1 nm = 300 ps A^pm d = 0-5 ps/A/km X \/1000 km = 15.8 ps ACotal = A/3002 + 15.82 = 300.4 ps This response time is too long to allow good transmission at 2.5 Gbit/s, according to the criteria described earlier. This means that some form o f dispersion compensation is needed to reduce average chromatic dispersion along the length of the fiber so it can carry 2.5 Gbit/s. Things get even more difficult at 10 Gbit/s. If you flip the formula for maximum NRZ data rate to show the maximum allowable pulse spreading for a given data rate, you get At,'m a x im u m

°-7 Data rate

For a 10 Gbit/s signal, this gives a maximum pulse spreading o f 70 ps, which requires considerably more dispersion compensation. Let’s assume that the transmitter bandwidth remains 0.1 nm, although high-speed modulation can increase this number. Suppose we can reduce the average chromatic dispersion by a factor o f 5, to 0.6 ps/nm-km. Using the same assumptions above gives us: A Chromatic = 0-6 ps/nm ■km X 1000 km X 0.1 nm = 60 ps A%m d = 0-5 ps/A/km X


1000 km = 15.8 ps

= A/602 + 15.82 = 62 ps That figure is adequate if you only consider fiber response, but you also have to consider transmitter and receiver rise times. If they both equal 25 ps— very fast devices— that would bring the total pulse spreading to the 70 ps limit. You might want average chromatic dispersion reduced even further, to about 0.3 ps/nm •km. So far we have assumed polarization-mode dispersion is fairly benign by using specifications for new fibers fresh from the factory. However, older fibers or fibers installed under less than ideal conditions may have higher polarization-mode dispersion. Suppose we want to use older fibers with a realistic value o f 2 ps/root-km for polarization-mode dispersion. Those calculations give us: Achromatic = 0-6 ps/nm •km X 1000 km X 0.1 nm = 60 ps AtPMD = 2 ps/\/km X A/1000 km = 63.2 ps AftotaI = V 6 0 2 + 63.22 = 87 ps This produces polarization-mode dispersion slightly larger than the chromatic dispersion, and yields a total pulse spreading too high for 10 Gbit/s transmission, even ignoring transmitter and receiver rise times.

Single-Channel System Design

Dispersion Compensation As these examples show, dispersion compensation becomes essential as transmission speeds and distances increase. Compensation is easiest for chromatic dispersion. As you learned in Chapter 5, chromatic dispersion has a characteristic sign that indicates whether the shorter or longer wavelengths have gone farther in the fiber. To compensate for chromatic dispersion, you add a length o f fiber (or some other optical component) with dispersion o f the opposite sign. If the shorter wavelengths are falling behind, adding a length o f fiber that makes the longer wavelengths fall behind serves to compress the length o f the output pulse. One approach to chromatic dispersion compensation is by alternating segments o f fiber with a different dispersion sign, as shown in Figure 21.7. In this example, one fiber has a positive dispersion and the other a negative dispersion that is smaller in magnitude. This mixture is possible by using two types o f nonzero dispersion-shifted fibers, one with zero dispersion at wavelengths shorter than the 1550-nm band, the other with zero dispersion at longer wavelengths. Optical signals pass through alternating segments o f the two fibers, so the cumulative chromatic dispersion is first negative, then shifts positive, and so on. This is called distribu ted com pensation, because the fibers that compensate for the pulse spreading are distributed along the fiber segment. In this example, the dispersions o f the two fibers do not differ greatly in magnitude, but do differ in sign. An alternative is to insert dispersion-compensation modules at selected points along the length of the system. Each module has chromatic dispersion opposite in sign to that o f a certain length o f the transmission fiber, so one module might compensate for 20 km of

L a y o u t o f tw o fib e rs

N e g a tiv e

P o s itiv e

D is p e rs io n

D is p e rs io n

Dispersion compensation becomes important at high speeds or over long distances.

FIGURE 21.7 Dispersion compensation distributed along length o f system.


Chapter 21

dispersion. Modules can be built with fiber Bragg gratings that provide dispersion compensation. However, they more typically consist o f easy-to-install coiled fibers designed specifically for dispersion compensation, which have chromatic dispersion higher in magnitude but opposite in sign to that o f the transmission fiber. Chromatic dispersion in compensating fibers typically is several times larger than it is in transmission fibers, so shorter lengths are required. That’s an advantage because compensating fibers generally have poorer transmission. It is relatively simple to compensate chromatic dispersion at one wavelength to give a net chromatic dispersion o f zero along a fiber segment, as shown in Figure 21.7. It is virtually impossible to compensate for chromatic dispersion across a wide range o f wavelengths, as you will learn in Chapter 22. Compensation for polarization-mode dispersion is very difficult. Unlike chromatic dispersion, PMD is a dynamic effect, which means its degree varies randomly over time. Thus PMD compensation also must be dynamic, with automatic adjustment over time. Both optical and electronic techniques have been demonstrated in the lab, but they are not yet widely accepted or in practical use.

Transmitter and Receiver Response Times Transmitters and receivers must be matched to fiber characteristics.

So far I have concentrated on fiber dispersion, but transmitter and receiver response times also play critical roles in system bandwidth budgets. Just because a transmitter and receiver are rated to operate at 10 Gbit/s in some situations does not mean they can transmit at that speed in a ll systems. For example, a transmitter and receiver, both having 40 ps rise times, can transmit 10 Gbit/s signals through fiber with up to 40 ps of cumulative pulse spreading. However, they couldn’t be used in the example I showed earlier, where total pulse spreading in the fiber was 62 ps. That would require a faster transmitter and receiver, with response times around 25 ps. For this reason manufacturers sell different models o f transmitters and receivers for transmission at the same data rates through different types o f fibers. Transmitter-receiver pairs intended for short-distance use do not have to meet the same stringent speed requirements as those used for long-distance systems. In short, you have to match transmitters and receivers to the fiber system used to assure you get the desired transmission speed. To show you how time-response calculations work, I have chosen simple examples. In reality there are a few other complications that come from the nature o f transmitters and receivers. In particular, the range of wavelengths from transmitters is broadened by a couple of distinct effects.

Transmitter Spectral Broadening As you learned earlier, directly modulating a semiconductor laser changes its refractive index as the density of current carriers changes. This effect, called chirp, causes the resonant wavelength to shift during a pulse, effectively spreading the range o f output wavelengths. A laser that emits a bandwidth o f 0.001 nm in a continuous beam spans a much larger range when directly modulated. External modulation can avoid chirp.

Single-Channel System Design

193.1 T H z C a rrie r

FIGURE 21.8 M odulation broadening is caused by side bands generated by the m odulating signal.

F re q u e n c y ra n g e o f m o d u la tio n

External modulation cannot prevent a second type o f spectral broadening that arises from any amplitude modulation o f a pure carrier signal. The same effect occurs with radio. Modulation generates new frequency components by adding (and subtracting) the frequency o f the modulating signal to the carrier frequency, as shown in Figure 21.8. This process generates side bands at frequencies both above and below the carrier. The two side bands contain identical information, so radio transmitters generally suppress one side band to conserve scarce frequency space. Side-band suppression is difficult at optical wavelengths, so normally both side bands are present. The result is an effective broadening o f the transmitter spectrum that increases directly with the bandwidth. If modulation produces a 10-GHz range o f frequencies, it generates 10-GHz side bands on each side for a total spectral range o f 20 GHz. This affects both chromatic dispersion in the fiber and the channel spacing possible in D W DM systems.

Modulation broadens the wavelengths in an optical carrier.

Cost/Performance Trade-offs So far I have only mentioned in passing a critical concern in real-world system design— cost. Minimizing cost is an implicit goal in all system design. I list some simple guidelines below, but it is impossible to give hard-and-fast rules for the tough job o f making tradeoffs between cost and performance. Ultimately it is your judgement as a system buyer or designer whether pushing error rate from 10 9 to 10-12 is worth an extra $1000. The best I can give you are some ideas to apply in working situations.

Users must make cost/performance trade-offs.

Chapter 21

Choice of Fiber Type The choice of fiber type is crucial.

The choice o f fiber type has a tremendous impact on the cost and performance o f any system. A fundamental choice is between single- and multimode fiber, but several variations on both types are available. Installation is expensive, so you want to allow room for your system to expand. Bandwidth requirements inevitably increase, just as each new generation o f software demands more computer memory and hard-disk space. You need single-mode fiber if your system spans more than a couple o f kilometers. Premium types, preferred for amplified W D M systems, are nonzero dispersion-shifted fibers, with low dispersion through the 1550-nm erbium amplifier window. Some singlemode fibers are optimized for “metro” applications over distances to a few hundred kilometers; others are optimized for terrestrial long-haul or submarine systems. Step-index single-mode fiber, with zero dispersion near 1310 nm, is still common, but normally is used at shorter distances because of its high dispersion at 1550 nm. Low-water fibers have a broader transmission window where amplification is not critical. Single-mode fiber sometimes is used at distances shorter than a couple of kilometers, especially for high-speed transmission, such as 10-Gigabit Ethernet. Graded-index multimode fibers are normally preferred for networks where transmission distances are up to a couple o f kilometers, speeds are moderate, and many connections are required. Their big advantage is easier coupling to each other and to low-cost light sources. Bandwidth is significantly higher for 50/125-fJtm fiber than for 62.5/125-(xm fiber. Remember that transmission speeds are sure to increase, so plan for future capacity expansion. Plastic fibers and large-core step-index multimode fibers have very limited distance ranges. Plastic fibers have high attenuation, and large-core multimode fibers have low bandwidth. Their applications remain limited, but they can be valuable in certain situations, such as short data links inside equipment.

Other Guidelines It’s impossible to give a comprehensive set o f guidelines for fiber system design, but I can give a set of rough-and-ready suggestions, starting with a few common-sense rules: Don't forget to apply common sense in system design. Labor is never free.

9 Your time is valuable. If you spend an entire day trying to save $5 on hardware, the result will be a net loss unless you are mass-producing the design. 9 Installation, assembly, operation, and support are not free. For a surprising number o f fiber-optic systems, installation and maintenance cost more than the hardware. You may save money in the long term by paying extra for hardware that is easier to install and service. 9 It can cost less to pay an expert to do it than to learn how yourself. Unless you need to practice installing connectors, it’s much easier to buy connectorized cables or hire a fiber-optic contractor for your first fiber-optic system. ^ You can save money by using standard mass-produced components rather than developing special-purpose components optimized for a particular application.

Single-Channel System Design

Some basic cost trade-offs are common in designing fiber-optic systems.

9 The performance o f low-loss fiber, high-sensitivity detectors, and powerful transmitters must be balanced against price advantages o f lower-performance devices. 9 Low-loss, high-bandwidth fibers generally accept less light than higher-loss, lowerbandwidth fibers. Over short distances, you can save money and overall attenuation by using a higher-loss, more costly cable that collects light more efficiently from lowercost LEDs. (Because o f the economics o f production and material requirements, largecore multimode fibers are considerably more expensive than step-index single-mode fibers.) 9 The marginal costs o f adding extra fibers to a cable are modest and much cheaper than installing a second parallel cable. However, if reliability is important, the extra cost of a second cable on a different route is a worthwhile insurance premium. 9 LEDs are much cheaper and require less environmental protection than lasers, but they produce much less power and are harder to couple to small-core fibers. Their broad range o f wavelengths and their limited modulation speed limit system bandwidth. • Fiber attenuation contributes less to losses o f short systems than do losses in transferring light into and between fibers. 9 Topology o f multiterminal networks can have a large impact on system requirements and cost because o f their differences in component requirements. Coupler losses may severely restrict options in some designs. • Light sources and detectors for 1300 and 1550 nm cost more than those for the 650- or 800- to 900-nm windows, although fiber and cable for the longer wavelength may be less expensive. •

1550-nm light sources cost more than 1300-nm sources.

• Fiber and cable become a larger fraction of total cost— and have more impact on performance— the longer the system. 9 Balance the advantages of eliminating extra components with the higher costs of the components needed to eliminate them. For example, it’s hard to justify the expense of two-way transmission through a single fiber over short distances. • Optical amplifiers or high-power laser transmitters make sense in systems distributing signals to many terminals. 9 Compare costs o f high-speed TD M on single channels or lower-speed T D M at multiple wavelengths. • D ark fibers— extra fibers installed in the original cable that were never hooked up to light sources— are often available in existing cables. 9 Remember that human actions— not defective equipment— cause most fiber-optic failures. Consider ring or mesh topologies that can survive a single break. Take the extra time and spend the extra money to make important systems less vulnerable to damage. This means labeling and documenting the system carefully, as well as not leaving cables where people can trip over them.

Always leave room for future upgrades. Bandwidth requirements are sure to increase, and it costs less overall to install higher-capacity fiber now than to install a cheap one that you have to replace with a more expensive one later. ^

Install extra fibers in your cables; they’re a lot cheaper than going back later to install more cables.


Leave margin for repair, such as slack in cables. It costs much less than complete replacement later.


Leave room for expansion by adding W D M to your system.

Watch for potential bottlenecks that might prevent future expansion.

As you grow more familiar with fiber optics, you will develop your own guidelines based on your own experience.

What Have You Learned? 1. Design of fiber-optic systems requires balancing sometimes-conflicting performance goals as well as costs.

2. The system loss budget is calculated by subtracting all system losses from the transmitter output power plus the gain o f any optical amplifiers. The resulting output power should equal the input power required by the receiver plus system margin.

3. Significant losses can occur in coupling light from sources into fibers. You need multimode fibers to collect light from LED sources and single-mode fibers to collect light from edge-emitting laser sources. Large-core fibers are more efficient for large-area LEDs.

4. Total fiber loss equals attenuation (dB/km) multiplied by transmission distance. Multimode fibers may suffer transient losses in the first 100 to 200 m.

5. Total loss from connectors, couplers, and splices is their characteristic loss multiplied by the number o f each in the system. You calculate the most likely loss using average loss and the worst case using maximum specified loss.

6. System margin is a safety factor that allows for repairs and aging o f components. Typical values are 5 to 10 dB.

7. Optical amplifiers boost signal strength, but because o f their high cost they are best used in long, high-speed systems or systems that distribute signals from one source to many receivers.


. Transmission capacity budgets calculate bandwidth or bit rate; they depend only on source, fiber, and receiver characteristics. You can estimate capacity by calculating response time.

9. Response time o f a system is the square root o f the sum o f the squares of component response times. Calculations must include transmitter and receiver response times as well as fiber dispersion.

Single-Channel System Design

10. Modal dispersion and chromatic dispersion combine to limit capacity of multimode fibers. Because o f these capacity limits, multimode fibers are rarely used over more than a couple o f kilometers. 11. Chromatic dispersion and polarization-mode dispersion limit capacity o f singlemode fibers, which usually transmit over a kilometer or more. Chromatic dispersion depends on source spectral width as well as fiber dispersion. 12. Dispersion compensation becomes important at high speeds or over long distances. The compensating elements can be fibers distributed through the length o f the system, or lumped at certain points. 13. Transmitters and receivers must be matched to fiber characteristics.

14. Modulation broadens the range of wavelengths in the optical carrier. The largest effect is chirp for directly modulated lasers, but external modulation also broadens transmitter spectral range. 15. Installation can cost much more than hardware. With demand for transmission capacity rising steadily, you should keep your upgrade paths open.

What's Next? Chapter 22 covers the fundamentals o f optical networking design, including wavelengthdivision multiplexing.

Further Reading Vivek Alwayn, O ptical Network Design an d Im plem entation (Cisco Press, 2004) Gerd Keiser, O ptical F iber Communications, 3rd ed. (McGraw-Hill, 2000) Rajiv Ramaswami and Kumar N. Sivarajan, O ptical N etw orks: A P ractical Perspective (Morgan Kaufmann, 2002)

Questions to Think About 1. Why is the relative light-collection efficiency o f fibers in decibels proportional to 20 times, rather than 10 times, the log o f the ratio o f their diameters?

2. Suppose the amplifiers in a transatlantic cable are limited to 12 dB o f gain to limit noise. How far apart can they be spaced if the fiber attenuation averages 0.24 dB? You can neglect splices, and the system contains no connectors.

3. You are installing a fiber-optic data link between a laboratory and a remote datacollection center 5 km away. You want to use a VCSEL transmitter with —5 dBm output and a fiber with loss o f 2.5 dB/km at 850 nm. The system includes patch panels on each end, with two connector pairs at each patch panel, and additional connectors on the terminal equipment. If connector loss is 0.5 dB, and you want a 10 dB margin, how sensitive must your receiver be?

4. You have to design a system with 1 Gbit/s data rate using return-to-zero (RZ) digital coding. What is the 3-dB analog bandwidth o f this system? What NRZ data rate could this system transmit?

5. You want to transmit 1 Gbit/s through 100 km o f single-mode fiber, using a transmitter and receiver that each have response times of 0.4 ns. The transmitter has line width o f 0.1 nm. Neglecting polarization mode dispersion, what is the maximum chromatic dispersion allowable in the fiber?

6. Can you get away with using a 1550 nm V CSEL in the system o f Problem 5 if the VCSEL has linewidth o f 0.5 nm, output o f —5 dBm, and the fiber loss is 0.25 dBm? (This assumes you can find such a VCSEL.) Check both the pulse spreading and the power level, assuming receiver sensitivity o f —30 dBm.

Chapter Quiz 1 . A large-area LED transfers 10 p W (10 dBp) into an optical fiber with core diameter o f 100 pm and numerical aperture o f 0.30. What power should it couple into a fiber with 50 pm core and NA o f 0.2? a. lO dBp b. 9.5 dBp c. 3 dBp. d. 1.0 dBp, e. 0.4 dBp

2. A connector is specified as having loss o f 0.6 dB ± 0.2 dB. What is the maximum connector loss in a system containing five such connector pairs? a. 0.6 dB b. 3.0 dB c. 4.0 dB d. 5.0 dB e. none o f the above

3. A 100-Mbit/s signal must be sent through a 100-m length o f fiber with eight connector pairs to a receiver with sensitivity o f —30 dBm. The fiber loss is 4 dB/km, and the average connector loss is 1.0 dB. If the system margin is 5 dB, what is the minimum power that the light source must couple into the fiber? a. —13.0 dBm b. —13.4 dBm c. —16.0 dBm d. —16.6 dBm e. —20.0 dBm

4. A system is designed to transmit 622 Mbit/s through 50 km o f cable with attenuation o f 0.4 dB/km. The system contains two connector pairs with 1.5 dB

Single-Channel System Design

loss, a laser source that couples 0 dBm into the fiber, and a receiver with sensitivity o f —34 dBm. How many splices with average loss o f 0.15 dB can the system contain if the system margin must be at least 8 dB? a. none b. 10 c. 20 d. 30 e. 40 f. none o f the above

5. A 2.5-Gbit/system must span a distance of 2000 km, with optical amplifiers every 80 km. If the fiber loss is 0.3 dB/km at 1550 nm and there is one 0.1 dB splice every 16 km, what must the amplifier gain be if the system is not to gain or lose signal strength across its entire length? a. 20 dB b. 24.4 dB c. 26.4 dB d. 30 dB e. 34.4 dB


You need to transmit identical 1-Gbit/s signals to 200 homes using a 1310-nm laser source. The homes are 1 to 4 km from your transmitter and use receivers sensitive to 30 dBm. What transmitter power do you need to achieve a 5-dB system margin if your fiber has a 0.4-dB/km loss at 1310 nm, each signal path from transmitter to home includes 6 connectors with a 0.5-dB average loss, and you split the signal in a 1 X 200 tree coupler with no excess loss? a. 4.6 dBm b. 9.6 dBm c. 2.6 dBm d. 0.0 dBm e. —0.4 dBm

7. What is the duration o f a single-bit interval in a 1.7-Gbit/s signal? a. 1.7 ns b. 1 ns c. 0.588 ns d. 0.294 ns f. 0.170 ns


. What is the response time o f a system with transmitter response of 2 ns, receiver response o f 1 ns, and 100 m o f multimode fiber with dispersion o f 20 ns/km (including both modal and chromatic dispersions)? a.

2 ns


2.236 ns


Chapter 21

c. 2.646 ns d. 2.828 ns e. 3 ns

9. What is the total dispersion of 10 km of graded-index fiber with modal dispersion o f 2.5 ns/km and chromatic dispersion o f 100 ps/nm •km when it is used with an 850-nm LED having a 50-nm spectral width? a. 5 ns b. 25 ns c. 50 ns d. 55.9 ns e. 75 ns

10 . What is the total dispersion of 10 km of single-mode fiber with chromatic dispersion of 17 ps/nm •km and average polarization-mode dispersion o f 0.5 ps/Vkrn at 1550 nm when used with a laser source with spectral width of 1 nm? a. 1.58 ps b. 10 ps c. 17 ps d. 170 ps e. 172 ps 1 1 . You are designing a 1-Gbit/s system using NRZ-coded signals with a transmitter with 0.3 ns rise time and a receiver that also has 0.3 ns rise time. What is the maximum total dispersion allowable through the entire length o f the fiber? a. 0.1 ns b. 0.3 ns c. 0.44 ns d. 0.56 ns e. 0.7 ns 1 2 . You generate a 2.5-Gbit/s NRZ signal with rise time o f 0.15 ns and spectral width o f 0.1 nm. You have to send it through 80 km o f nonzero dispersionshifted fiber with chromatic dispersion o f 6 ps/km-nm at the transmitter wavelength. Average polarization-mode dispersion is 0.5 ps/root-km. What is the total pulse dispersion in the fiber? a. 4.5 ps b. 45 ps c. 48 ps d. 52.5 ps e. 80 ps

Optical Networking System Design

About This Chapter Optical networking adds another dimension to the concepts o f system design you learned in Chapter 21 for the simple case o f a single optical channel per fiber. This chapter covers systems that transmit two or more optical channels per fiber, where signals are managed by wavelength or optical channel. The extra optical channels increase the complexity o f system design, and the number o f factors that must be considered. This chapter opens with a review o f optical networking concepts. It then explains how optical channels are packed together, contrasting wavelength-division and timedivision multiplexing, and dense and coarse channel spacing. Then it covers the properties o f optical fibers and optical amplifiers that affect optical networking design. Finally it covers optical switching and channel management, including the importance o f wavelength conversion. Optical network design is still a young field, so we will not cover it in as much detail as we did single-channel design in Chapter 21.

Optical Networking Concepts Optical networking organizes signals by wavelength as well as by the time sequence of digital data. Wavelength-division multiplexing packs a number of optical channels into a transmitting fiber, with each optical channel transmitted at a different wavelength. In an ideal optical network, signals can readily be converted to different wavelengths to rearrange or redistribute them, as cars shift between lanes on a highway. W D M began as a way to squeeze more data through an optical fiber, but is evolving into a new way of managing data by wavelength as well as by digital coding.


Chapter 22

There are several key concepts in optical networking.

Granularity is the subdivision of signals in a network.

• Signal granularity measures how signals are subdivided within an optical network. The greater the granularity, the more potential ways there are to organize the signals. In general granularity is a good thing because data transmitted over most networks is assembled from many small data streams, not massive flows between two points. Think o f the traffic as being like many automobiles, not a few 200-car freight trains. • Total transmission capacity of a system measures how much information a fiber can transmit. It equals the sum of the data rates on all the individual optical channels carried by the system. To maximize transmission capacity, you pack channels as closely together as possible and transmit the highest possible data rate on each channel. Raising the data rate increases the bandwidth required for each optical channel, so trade-offs are inevitable. Cost-performance trade-offs also are inevitable because not all fiber-optic routes require the greatest possible bandwidth, and packing high-speed channels tightly together can be very expensive. 9 Fiber transmission capacity depends on attenuation and dispersion, which are functions of wavelength. Variations in attenuation and dispersion limit the transmission capacity of certain parts o f the spectrum more than others. The degree of limitation depends on overall transmission distance. • A m plification capacity also varies across the spectrum. Good amplifiers are not available at all wavelengths, and amplifiers do not have uniform gain across their operating ranges. Amplification is a must for long-distance optical networks, but may not be required in other types. 9 Switching capacity for optical networking depends on the ability to manipulate signal wavelength. Wavelength conversion technology is still in development. Two broadly different families of optical networking technologies have been developed for different types of applications, and further variations may emerge in coming years.

DWDM is used for long-haul systems; CWDM is used for shorter links.

^ Dense- WDM packs as much transmission capacity as possible into as few fibers as possible. Its main applications are in spanning distances of hundreds of kilometers or more, where infrastructure costs are relatively high. Sharing the capacity over more channels reduces overall costs significantly even if the cost per channel is high. • Coarse- WDM reduces equipment costs for W D M systems that span moderate distances— tens of kilometers or less. Infrastructure costs over these distances are relatively low, so it’s important to limit the cost per channel. These concepts are the cornerstones o f optical network design. Let’s look at how they are used.

Optical Channel Density W D M divides a block of spectrum among optical channels.

A W D M optical network divides a block o f spectrum among multiple optical channels. The space required for an optical channel depends on its data rate; the higher the data rate, the broader the bandwidth the channel requires. For example, the signal produced

Optical Networking System Design

by modulating a single-line light source at 10 Gbit/s occupies four times more bandwidth than does an identical light source modulated at 2.5 Gbit/s. Allocating the spectrum among optical channels is a fundamental step in optical network design.

WDM Compared to High-Speed TDM Before we slice up the optical spectrum, we should compare W D M to high-speed timedivision multiplexing. In principle, a single 160-Gbit/s data stream could carry 16 10-Gbit/s signals or 64 2.5-Gbit/s signals. Why not time-division multiplex the slower signals to higher speeds and avoid the need for so many separate transmitters at different wavelengths? Part of the answer is that high data rates impose limitations on fiber transmission. Higherspeed TD M signals run into transmission distance limits; increase the data rate by a factor of four, and the maximum transmission distance drops by a factor of 16 because of chromatic dispersion. The pulse durations are four times shorter, and modulation effects broaden the transmitter spectrum by a factor o f four, multiplying pulse spreading caused by chromatic dispersion by a second factor of four. A signal that can travel through 1600 km of fiber at 2.5 Gbit/s can travel only 100 km at 10 Gbit/s, and a mere 6.25 km at 40 Gbit/s. Combine 16 2.5-Gbit/s signals into one signal at 40 Gbit/s and dispersion limits it to 6.25 km; but transmit 16 separate 2.5-Gbit/s signals at different wavelengths and they can all travel 1600 km. Current technology performs well at 10 Gbit/s, but transmission at 40 Gbit/s is difficult, and transmission at 160 Gbit/s is extremely difficult, even in the laboratory. Another part o f the answer is granularity. There’s more need for separate 2.5- or 10-Gbit/s channels than for 40-Gbit/s channels. Internet traffic maps mostly show 2.5-Gbit/s links, with 10-Gbit/s transmission only on the highest-traffic routes between the biggest cities. The companies that provide transmission capacity usually subdivide the capacity o f their fibers into sizes that their customers want. W D M enables them to split fiber capacity into 2.5- and 10-Gbit/s slices, providing the granularity needed for today’s traffic. Customers can transmit independent traffic on each wavelength, and often can choose the transmission format. Overall, W D M offers better transmission distance and granularity than high-speed TD M for today’s technology and transmission demands.

Spectral Range and Optical Channels Only a limited spectral range is usable in any fiber system, depending on factors such as attenuation, dispersion, and amplification that will be described later. The available spectrum is divided among multiple optical channel slots, which normally are o f an identical width set by international standards. For example, a range o f 3200 GHz, corresponding to 1562.42 to 1535.82 nm (192.0 to 195.2 THz), can be divided into 32 100-GHz channels, 64 50-GHz channels, or 16 200-GH z channels. The space that each channel requires depends on the modulation rate. As shown in Figure 22.1, a 10-Gbit/s signal spreads the modulation spectrum o f a carrier signal across a broader range than does a 2.5-Gbit/s signal, so it can’t fit in as narrow an optical channel. The degree of separation possible depends on the optics as well as the modulation bandwidth.

• ^ er Iransr9'ssi°r' d is ta n c e U


im ite d

Chapter 22


10-G bit/s signals usually go in 100-GHz channels in practical systems.

Channel Spacing

Advanced technology can reduce the modulation bandwidth and sharpen the selectivity of the optics to improve channel spacing. However, limits on total transmission capacity are unavoidable. So far the best results obtained in the laboratory without using elaborate polarization schemes have squeezed about 0.8 bit per second into 1 Hz of bandwidth. This figure of merit, called spectral efficiency, corresponds to fitting a 40-Gbit/s signal into a 50-GHz optical channel. Spectral efficiencies of practical commercial systems are much lower; they typically assign 10-Gbit/s signals to 100-GHz channels and 2.5-Gbit/s signals to 50-GHz slots.

Populating Channels Not all wavelength slots are populated.

One confusing dichotomy in optical networking is the often vast difference between the actual operating load o f a fiber-optic system and its stated capacity. The difference arises because W D M systems are designed with slots to accommodate a certain number o f optical channels, but carriers do not immediately populate all these channels with operating transmitters and receivers. Also optical systems often are modular, so only part o f the hardware is installed initially, with the remaining optics added as transmitters and receivers are installed on new channels. Figure 22.2 shows an example o f this dichotomy, a W D M system that has potential slots for 40 optical channels. Its p oten tial capacity is the number o f available slots times the

Optical Networking System Design

FIGURE 22.2 \1-X40

Reflects wavelengths longer than \ 8.

H ig h - o r L o w -p a s s F ilte rs

Reflectfwavqlengths shorter than \ 17.

P artial provisioning o f a 40-channel DW DM system.

| X 1 7 - X 40/

Reflects wavelengths shorter than \ 33 r ------------- 1 D e m u ltip le x e s j X33- X 40

' R e s e rv e d fo r F u tu re



' E x p a n s io n (n o t y e t in s ta lle d )

Only three receivers and transmitters actually installed.

maximum data rate per channel. The carrier operating the system, however, has installed transmitters and receivers on only three o f the optical channels; all the others, shown in dashed lines, are reserved for future expansion. In this example, the carrier has bought optics that separate input signals into five groups o f optical channels, plus optics that divide one of those five groups of optical channels into eight separate channels. Only three of those eight channels are populated, and the transmitters and receivers may not be operating at the maximum rate. Thus this system could be carrying only 7.5 Gbit/s (2.5 Gbit/s on each of the three populated channels), but have a potential capacity o f 400 Gbit/s (10 Gbit/s on each of 40 channels). This example is typical of the incremental approach that carriers take to installing transmission capacity. Operating companies do not need all the potential capacity immediately, but they do want to have room for future expansion. Transmitters and receivers currently are expensive, but their prices are coming down, so the carrier populates only the channel slots that are needed immediately. If more capacity is needed later, the carrier can populate more slots with cheaper transmitters and receivers, so the carrier buys only the capacity it needs today. In fact, transmission loads vary, and during the telecommunications bubble many carriers vastly overestimated the maximum transmission capacity they would need. The result was that the actual peak load o f a system like the one in Figure 22.2 would fall far short of its 7.5-Gbit/s capacity.

Dense- and Coarse-WDM So far we have focused on the practice of packing optical channels closely together to squeeze as much bandwidth as possible through a single optical fiber. This is called dense-W D M or DWDM. In practice, D W DM systems have channel spacing o f 200 GHz or less. They

Coarse-WDM can reduce transmission costs.

Chapter 22

10-Gigabit Ethernet uses four-channel CWDM.

provide huge transmission capacity over long distances, and during the telecommunications bubble they filled the tremendous demand for bandwidth. Today most of these systems are underutilized. No one seriously disputes the advantages o f transmitting signals through one fiber on many separate optical channels. However, they do question the advantages o f D W D M , which was developed specifically to provide high-speed, long-distance transmission. Long-haul applications could justify the high costs o f precision optics to split the spectrum into narrow slices and o f cooled high-speed laser transmitters to provide precise wavelengths that fit into the narrow channel slots. However, those high costs prevented D W D M technology from being used in other applications that could benefit from wavelength-division multiplexing. An alternative approach, called coarse-W DM or CWDM, was developed to avoid the high costs associated with precise wavelength control. It divides the spectrum into fewer slices, wide enough to accommodate the wide wavelength variations o f much cheaper uncooled diode lasers. Each CW DM channel can carry many D W DM channels, as shown in Figure 22.3. Two variations o f CW DM have been developed: one for dividing a single high-speed signal into four slower data streams, and the other for sending up to 18 separate signals through a single fiber without amplification. The 10-Gigabit Ethernet standard has an option for splitting the data stream into four separate signals transmitted at 2.5 Gbit/s, for a total o f 10 Gbit/s. This allows the use of graded-index multimode fibers over longer distances than those possible at 10 Gbit/s and the use o f lower-cost 2.5-Gbit/s transmitters with single-mode fiber. The coarse-WDM transmitters use uncooled distributed-feedback lasers emitting at center wavelengths of 1275.7, 1300.2, 1324.7, and 1349.2 nm. The broad range o f wavelengths is possible because no optical amplifiers are needed for the distances used in 10-Gigabit Ethernet. Loose wavelength tolerances and using uncooled lasers lower costs. The lasers could operate at temperatures o f 0 to 70°C with wavelength drifting no more than 5 nm, which would keep the signals in the proper coarse-WDM slots.


13 n m P e a k -

CW DM an d DW DM channel widths.

■ N


iS i

E x


c 0 O ^o ^ "S o Q

2 0 n m C o a rs e -W D M C h a n n e l

Optical Networking System Design


FIGURE 22.4 CW DM channels specified in IT U G 694.2 standard.

I 0.9 c0 1