kaminow, i. p. (1997). optical fiber telecommunications iii-a

OPTICAL FIBER TELECOMMUNICATIONS IIIA Edited by Lucent Technologies, Bell Laboratories Holmdel, New Jersey I>ucent Technologies.. Some Practical Codes Applications of Coding to Futur

OPTICAL FIBER

TELECOMMUNICATIONS IIIA

OPTICAL FIBER

TELECOMMUNICATIONS IIIA

Edited by

Lucent Technologies, Bell Laboratories

Holmdel, New Jersey

I>ucent Technologies Bell Laboratories

Holmdel New Jersey

San Diego London Boston

New York Sydney Tokyo Toronto

Copyright 0 I997 by Lucent Technologies

All Rights Reserved

No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher

Requests for permission to make copies of any part of the work should be mailed to:

Permissions Department Harcourt, Inc., 6277 Sea Harbor Drive,

Orlando, Florida 32887-6777

Academic Press

A Harcourt Science and Technology Company

525 B Street Suite 1900, San Diego, California 92101-4495, USA

http:l/www.academicpress.com

Academic Press

Harcourt Place, 32 Jamestown Road, London NWI 7BY, UK

http://www.academicpress.com

Library of Congress Cataloging-in-Publication Data

Optical fiber telecommunications 111 / [edited by] Ivan P Kaminow, Thomas L Koch Includes bibliogmphical references and index

I Optical communications 2 Fiber optics I Kaminow, Ivan P II Koch Thomas L

p cm

ISBN 0-12-395170-4(~ A)- lSBN0-12-395171-2 (v B)

96-43812 TK5103.59.H3516 1997

PRlNTED IN THE UNITED STATES OF AMERICA

01 02 03 04 05 06 MM 8 7 6 5 4 3

For o u r dear grandchildren:

Sarah, Joseph Rafael, Nicolas, Gabriel, Sophia, and M a i m - IPK

For Peggy, Brian, and Marianne - TLK

The New Volume

Survey of Volumes IIIA and IIIB

Common Modulation Formats for Fiber Svstems-Direct

Detection and Coherent Systems

Fundamental Detection Performances and Deviations of

Currently Available Systems

Potential Role of Forward Error-Correcting Codes in Fiber

Systems and Its Beneficial Ripple Effects o n System and

Some Practical Codes

Applications of Coding to Future Lightwave Systems Such as

Summary

References

WDM Systems and Optical Networks

Chapter 4 Advances in Fiber Design and Processing

David J DiGiovanni, Donald P Jablonowski, and Man F Y a n

Introduction

Erbium Doped Fiber AmplifieI

High-Power Fiber Amplifier

Chapter 5 Advances in Cable Design

Kenneth W Jackson, T Don Mathis, P D Patel,

Manuel R Santana, and Phillip M Thomas

Introduction

Performance, Reliability, and Standards

Outside Plant Cables

References

Chapter 6 Polarization Effects in Lightwave Systems

Craig D Poole and Jonathan Nagel

A H Gnauck and R M Jopson

Chapter 9 Terrestrial Amplified Lightwave System Design

Chiirigpeng Fan and J P Kiinz

Introduction

Architecture and Building Blocks of Lightwave Networks

Optical ,4mplifier System Design

WDM Issues

Optical Fiber Constraints

Advanced Experiments and Field Trials

Measures of System Margin

Chapter 11 Advances in High Bit-Rate Transmission Systems

Kinichiro Ogawa, Liang D Tzeng, Yong Kwan Park,

and Eiichi Sano

Pulse Propagation and Solitons in Optical Fibers: A Tutorial

Spontaneous Emission and Other Noise Effects

Frequency Guiding Filters

Models of Current Access Architectures

Considerations for Network Providers

Examples of Current Access Architecture Models

Use of WDM in Passive Optical Networks for Access

Contents xi

Chapter 14 Lightwave Analog Video Transmission

Mary R Phillips and Thomas E Darcie

Introduction

Analog Lightwave Systems

Analog Lightwave Technology

WDM Local Access Networks

Wavelength-Routing Network Test Beds

Contributors

Neal S Bergano (Ch 10) AT&T Laboratories 101 Crawfords Corner Road Holmdel, New Jersey 07733

Joseph E Berthold (Ch 2) Bellcore 331 Newman Springs Road, Room

32331, Red Bank, New Jersey 07701

Vincent W S Chan (Ch 3) Massachusetts Institute of Technology Lincoln

Laboratory, 244 Wood Street Room D-309 Lexington Massachu- setts 02173

A R Chraplyvy (Ch S), Lucent Technologies, Bell Laboratories Crawford

Hill Laboratory 791 Holmdel-Keyport Road, Holmdel, New Jersey

07733

Thomas E Darcie (Ch 14) AT&T Laboratories-Research Communica-

tion Infrastructure Research Laboratory, 791 Holmdel-Keyport Road Room R219, Holmdel, New Jersey 07733

David J DiGiovanni (Ch 4), Lucent Technologies, Bell Laboratories 600

Mountain Avenue, Room MH 6C-312, Murray Hill New Jersey 07974

Chungpeng Fan (Ch 9) Lucent Technologies, Bell Laboratories, 101 Craw-

fords Corner Road, Room H O 2G-601A Holmdel, New Jersey 07733

F Forghieri (Ch 8), AT&T Laboratories-Research, Crawford Hill Labora-

tory 791 Holmdel-Keyport Road Holmdel New Jersey 07733

Nicholas J Frigo (Ch 13), AT&T Laboratories-Research, Crawford Hill

Laboratory, 791 Holmdel-Keyport Road, Room H O H L-175, Holmdel New Jersey 07733

A H Gnauck (Ch 7), AT&T Laboratories-Research Crawford Hill Labo-

Holmdel New Jersey 07733

J P Gordon (Ch 12), Lucent Technologies, Bell Laboratories 101 Craw-

fords Corner Road, Holmdel, New Jersey 07733

xiii

Donald P Jablonowski (Ch 4), Lucent Technologies, Bell Laboratories,

Ivan P Kaminow (Ch 1,15), Lucent Technologies, Bell Laboratories, 791

Holmdel-Keyport Road, Holmdel, New Jersey 07733

J P Kunz (Ch 9), Lucent Technologies, Bell Laboratories, 101 Crawfords

Corner Road, Room 4B-611, Holmdel, New Jersey 07733

P V Mamyshev (Ch 12), Lucent Technologies, Bell Laboratories, 101

Crawfords Corner Road, Holmdel, New Jersey 07733

T Don Mathis (Ch 5), Lucent Technologies, Bell Laboratories, 2000 North- east Expressway, Room 1D-51, Norcross, Georgia 30071

L F Mollenauer (Ch 12), Lucent Technologies, Bell Laboratories, 101

Crawfords Corner Road, Room 4C-306, Holmdel, New Jersey 07733

Jonathan Nagel (Ch 6), AT&T Laboratories-Research, Crawford Hill Laboratory, Holmdel-Keyport Road, Room L-137, Holmdel, New Jer- sey 07733

Kinichiro Ogawa (Ch ll), Lucent Technologies, Bell Laboratories, 9999 Hamilton Boulevard, Breinigsville, Pennsylvania 18031

Hamilton Boulevard, Breinigsville, Pennsylvania 18031

P D Patel (Ch 5 ) , Lucent Technologies, Bell Laboratories, 2000 Northeast

Expressway, Room 1D-40, Norcross, Georgia 30071

Mary R Phillips (Ch 14), ATx Telecom Systems, 1251 Frontenac Road,

Naperville, Illinois 60563

Craig D Poole (Ch 6), EigenLight Corporation, 8 Wemrock Drive, Ocean,

New Jersey 07712

Eiichi Sano (Ch ll), NIT LSI Laboratories, Atsugi, Japan

Manuel R Santana (Ch 5 ) , Lucent Technologies, Bell Laboratories, 2000 Northeast Expressway, Room 1D-32, Norcross, Georgia 30071

Contributors xv

Phillip M Thomas (Ch 5 ) , Lucent Technologies, Bell Laboratories, 2000 Northeast Expressway Room 1C-55 Norcross, Georgia 30071

R W Tkach (Ch 8), AT&T Laboratories-Research, Crawford Hill Labo-

ratory, 791 Holmdel-Keyport Road Holmdel, New Jersey 07733

Liang D Tzeng (Ch 1 l), Lucent Technologies, Bell Laboratories 9999 Hamilton Boulevard Breinigsville, Pennsylvania 18031

Man F Yan (Ch 4) Lucent Technologies, Bell Laboratories 600 Mountain Avenue Room MH 6C-308, Murray Hill New Jersey 07974

Ivan P Kaminow

A T& T Bell Laboratories (retired) Holnidel Nebt, Jerse?

History

Optical Fiber Telecommunications, edited by Stewart E Miller and Alan

G Chynoweth, was published in 1979, at the dawn of the revolution in lightwave telecommunications This book was a stand-alone volume that collected all available information for designing a lightwave system Miller was Director of the Lightwave Systems Research Laboratory and, together with Rudi Kornpfner, the Associate Executive Director, provided much

of the leadership at the Crawford Hill Laboratory of Bell Laboratories: Chynoweth was an Executive Director in the Murray Hill Laboratory leading the optical component development Many research and develop- ment (R&D) groups were active at other laboratories in the United States Europe, and Japan The book, however, was written exclusively by Bell Laboratories authors although it incorporated the global results

Looking back at that volume, I find it interesting that the topics are quite basic but in some ways dated The largest group of chapters covers the theory materials, measurement techniques, and properties of fibers and cables for the most part, multimode fibers A single chapter covers

900-nm band The remaining chapters cover direct and external modulation techniques, photodetectors and receiver design, and system design and applications Still the basic elements for the present-day systems are there: low-loss vapor-phase silica fiber and double-heterostructure lasers Although a few system trials took place beginning in 1979 it required several years before a commercially attractive lightwave telecommunica- tions system was installed in the United States This was the AT&T North-

1

OPTICAL FIBER 1 ELECOMLIUNICATIONS

L'OLCIME IIIA

Copyright C' 1497 hy Lucent Technologlei

411 rights of rrproduction In an) form reserved

n-

that began service in January 1983, operating at a wavelength of 820 nm and a bit rate of 45 Mb/s in multimode fiber Lightwave systems were upgraded in 1984 to 1310 nm and about 500 Mb/s in single-mode fiber in the United States, as well as in Europe and Japan

Tremendous progress was made during the next few years, and the choice

of lightwave over copper for all long-haul systems was ensured The drive was to improve performance, such as bit rate and repeater spacing, and to

nications ZZ (OFT ZZ), edited by Stewart E Miller and me, was published

in 1988 to summarize the lightwave design information known at the time

To broaden the coverage, we included some non-Bell Laboratories authors, including several authors from Bellcore, which had been divested from Bell Laboratories in 1984 as a result of the court-imposed “Modified Final Judgment.” Corning, Nippon Electric Corporation, and several universities were represented among the contributors Although research results are

tions than in the previous volume

The early chapters of OFTZZcover fibers, cables, and connectors, dealing with both single- and multimode fiber Topics include vapor-phase meth- ods for fabricating low-loss fiber operating at 1310 and 1550 nm, under- standing chromatic dispersion and various nonlinear effects, and designing polarization-maintaining fiber Another large group of chapters deals with

a wide geographic scope of systems for loop, intercity, interoffice, and undersea applications A research-oriented chapter deals with coherent systems and another with possible local area network applications, including

a comparison of time-division multiplexing (TDM) and wavelength-division multiplexing (WDM) to effectively utilize the fiber bandwidth Several chapters cover practical subsystem components, such as receivers and trans- mitters, and their reliability Other chapters cover the photonic devices, such

as lasers, photodiodes, modulators, and integrated electronic and integrated optic circuits, that compose the subsystems In particular, epitaxial growth methods for InGaAsP materials suitable for 1310- and 1550-nm applica- tions, and the design of high-speed single-mode lasers are discussed

The New Volume

By 1995, it was clear that the time for a new volume to address the recent research advances and the maturing of lightwave systems had arrived The contrast with the research and business climates of 1979 was dramatic System experiments of extreme sophistication were being performed

1 Overview 3

by building on the commercial and research components funded for a proven multibillion-dollar global industry For example, 10.000 km of high- performance fiber was assembled in several laboratories around the world

for NRZ (non-return-to-zero), soliton, and WDM system experiments The

competition in both the service and hardware ends of the telecommunica- tions business was stimulated by worldwide regulatory relief The success

in the long-haul market and the availability of relatively inexpensive compo- ncnts led to a wider quest for other lightwave applications in cable television and local access network markets The development of the diode-pumped erbium-doped fiber amplifier (EDFA) played a crucial role in enhancing the feasibility and performance of long-distance and WDM applications

In planning the new volume, Tom Koch and I looked for authors to update the topics of the previous volumes, such as fibers, cables, and laser sources But a much larger list of topics contained fields not previously included, such as SONET (synchronous optical network) standards

EDFAs, fiber nonlinearities, solitons, and passive optical networks (PONS)

Throughout the volume, erbium amplifiers, WDM, and associated compo- nents are common themes

Again, most of the authors come from Bell Laboratories and Bellcore, where much of the research and development was concentrated and where

we knew many potential authors Still, we attempted to find a few authors from elsewhere for balance Soon after laying out the table of contents and lining up the authors, however, a bombshell and a few hand grenades struck AT&T decided to split into three independent companies, Bellcore was put up for sale, and several authors changed jobs, including Tom Koch and I The resulting turmoil and uncertainty made the job of getting the chapters completed tougher than for the earlier volumes which enjoyed a climate of relative tranquillity

In the end, we assembled a complete set of chapters for Optical Fiber Telecommunications III, and can offer another timely and definitive survey

of the field Because of the large number of pages, the publisher recom- mended separating the volume into two sections, A and B This format should prove more manageable and convenient for the reader The chapters are numbered from Chapter 1 in each section, with this Overview repeated

to buy just one book

Survey of Volumes IIIA and IIIB

The chapters of Volumes IIIA and IIIB are briefly surveyed as follows in

an attempt to put the elements of the book in context

VOLUME IIIA

The market forces of deregulation and globalization have driven the need for telecommunications standards Domestically, the breakup of AT&T meant that service providers and equipment suppliers no longer accepted

de facto standards set by “Ma Bell.” They wanted to buy and sell equipment competitively and to be sure that components from many providers would interoperate successfully The globalization of markets extended these needs worldwide And the remarkable capability of silicon integrated circuits to perform extremely complex operations at low cost with high volume has made it possible to provide standard interfaces economi- cally

The digital transmission standard developed by Bellcore and employed

in all new domestic circuit-switched networks is SONET, and a similar international standard is SDH (synchronous digital hierarchy) In the same period, a telecommunications standard was devised to satisfy the needs

of the data market for statistical multiplexing and switching of bursty com-

puter traffic It is called A T M (asynchronous transfer mode) and is being em-

braced by the computer industry as well as by digital local access provid-

Information Coding and Error Correction in Optical Fiber

Communications Systems (Chapter 3)

The ultimate capacity of a communication channel is governed by the rules

of information theory The choice of modulation format and coding scheme determines how closely the actual performance approaches the theoretical limit The added cost and complexity of coding is often the deciding factor

in balancing the enhanced performance provided by this technology So

far, coding has not been required in high-performance lightwave systems

mance of high-speed electronics improves, we can expect to see more uses

of sophisticated coding schemes In particular, forward error-correcting codes (FECs) may soon find applications in long-distance, repeaterless

undersea systems A review of coding techniques, as they apply to lightwave systems, is given by Vincent W S Chan in Chapter 3

1 Overview 5

Advances in Fiber Design and Processing (Chapter 4)

The design and processing of fibers for special applications are presented

in Chapter 4 by David J DiGiovanni Donald P Jablonowski, and Man

F Yan Erbium-doped silica fibers for amplifiers at 1550 nm, which are described in detail in Chapter 2, Volume IIIB, are covered first Rare- earth-doped fluoride fibers for 1300-nm amplifiers are described later as

are fibers for cladding-pumped high-power fiber amplifiers

Dispersion management is essential for the long-haul high-speed systems described in later chapters The design and fabrication of these fibers for new WDM installations at 1550-nm and for 1550-nm upgrades of 1310-nm systems are also reviewed

Advances in Cable Design (Chapter 5)

Chapter 5 , by Kenneth W Jackson T Don Mathis P D Patel Manuel

R Santana and Phillip M Thomas, expands on related chapters in the

two previous volumes, OFT and OFT IZ The emphasis is on practical applications of production cables in a range of situations involving long- distance and local telephony, cable television broadband computer net- works premises cables and jumpers Field splicing of ribbon cable and the division of applications that lead to a bimodal distribution of low and high fiber count cables are detailed

Polarization Effects in Lightwave Systems (Chapter 6)

Modern optical fibers possess an extremely circular symmetry, yet they retain a tiny optical birefringence leading to polarization mode dispersion (PMD) that can have severe effects on the performance of very long digital systems as well as high-performance analog video systems Systems that contain polarization-sensitive components also suffer from polarization-

Nagel review the origins measurement and system implications of remnant birefringence in fibers

Dispersion Compensation for Optical Fiber Systems (Chapter 7)

Lightwave systems are not monochromatic: chirp in lasers leads to a finite range of wavelengths for the transmitter in single-wavelength systems, whereas WDM systems intrinsically cover a wide spectrum At the same time the propagation velocity in fiber is a function of wavelength that

can be controlled to some extent by fiber design, as noted in Chapter 4 To avoid pulse broadening, it is necessary to compensate for this fiber chro- matic dispersion Various approaches for dealing with this problem are pre- sented in Chapter 7, by A H Gnauck and R M Jopson Additional system approaches to dispersion management by fiber planning are given

in Chapter 8

Fiber Nonlinearities and Their Impact on Transmission Systems

(Chapter 8)

Just a few years ago, the study of nonlinear effects in fiber was regarded

as “blue sky” research because the effects are quite small The advance of technology has changed the picture dramatically as unrepeatered undersea spans reach 10,000 km, bit rates approach 10 Gb/s, and the number of WDM channels exceeds 10 In these cases, an appreciation of subtle nonlin- ear effects is crucial to system design The various nonlinearities represent perturbations in the real and imaginary parts of the refractive index of silica as a function of optical field In Chapter 8, Fabrizio Forghieri, Rob- ert W Tkach, and Andrew R Chraplyvy review the relevant nonlineari- ties, then develop design rules for accommodating the limitations of non- linearities on practical systems at the extremes of performance

Terrestrial Amplified Lightwave System Design (Chapter 9)

planning lightwave networks and designing transmission equipment, respec-

tively In Chapter 9, they review the practical problems encountered in

designing commercial terrestrial systems taking advantage of the technolo- gies described elsewhere in the book In particular, they consider such engineering requirements as reliability and restoration in systems with EDFAs, with dense WDM and wavelength routing, and in SONET-

SDH rings

Undersea Amplified Lightwave Systems Design (Chapter 10)

Because of their extreme requirements, transoceanic systems have been the most adventurous in applying new technology EDFAs have had an especially beneficial economic effect in replacing the more expensive and less reliable submarine electronic regenerators Wideband cable systems have reduced the cost and improved the quality of overseas connections

1 Overview 7

to be on a par with domestic communications In Chapter 10, Neal S

Bergano reviews the design criteria for installed and planned systems around the world

Advances in High Bit-Rate Transmission Systems (Chapter 11)

limit is defined in part by the availability of electronic devices and circuits

In Chapter 11, Kinichiro Ogawa, Liang D Tzeng, Yong K Park, and Eiichi Sano explore three high-speed topics: the design of high-speed receivers performance of lO-Gb/s field experiments, and research on devices and integrated circuits at 10 Gb/s and beyond

Solitons in High Bit-Rate, Long-Distance Transmission (Chapter 12)

Chromatic dispersion broadens pulses and therefore limits bit rate; the Kerr nonlinear effect can compress pulses and compensate for the disper- sion When these two effects are balanced, the normal mode of propagation

is a soliton pulse that is invariant with distance Thus, solitons have seemed

to be the natural transmission format, rather than the conventional NRZ format, for the long spans encountered in undersea systems Still, a number

of hurdles have manifested as researchers explored this approach more deeply Perhaps the most relentless and resourceful workers in meeting and overcoming these challenges have been Linn Mollenauer and his associ- ates L F Mollenauer, J P Gordon, and P V Mamyshev provide a defini- tive review of the current R&D status for soliton transmission systems in Chapter 12 Typical of a hurdle recognized, confronted, and leaped is the Gordon-Haus pulse jitter; the sliding filter solution is described at length

A Survey of Fiber Optics in Local Access Architectures (Chapter 13)

The Telecommunications Act of 1996 has opened the local access market

to competition and turmoil New applications based on switched broadband digital networks, as well as conventional telephone and broadcast analog video networks are adding to the mix of options Furthermore, husiness

factors, such as the projected customer rake rare, far outweigh technol-

ogy issues

In Chapter 13, Nicholas J Frigo discusses the economics, new architec- tures and novel components that enter the access debate The architectural

proposals include fiber to the home (FTTH), TDM PON, WDM PON, hybrid fiber coax (HFC), and switched digital video (SDV) networks The critical optical components, described in Volume IIIB, include WDM lasers and receivers, waveguide grating routers, and low-cost modulators

Cable television brings the analog broadcast video spectrum to conventional television receivers in the home During the last few years, it was found that the noise and linearity of lightwave components are sufficiently good

to transport this rf signal over wide areas by intensity modulation of a laser carrier at 1310, 1060, or 1550 nm The fiber optic approach has had a dramatic effect on the penetration and performance of cable systems, lower- ing cost, improving reliability, and extending the number of channels New multilevel coding schemes make rf cable modems an attractive method for distributing interactive digital signals by means of HFC and related architectures Thus, cable distribution looks like an economic technology for bringing high-speed data and compressed video applications, such as the Internet, to homes and offices Now, in the bright new world of deregu- lation and wide-open competition, cable may also carry telephone ser- vice more readily than telephone pairs can carry video In Chapter 14,

Mary R Phillips and Thomas E Darcie examine the hardware require- ments and network architectures for practical approaches to modern lightwave cable systems

The final chapter in Volume IIIA looks at novel architectures for routing

in high bit-rate, multiple-access networks For the most part, the emphasis

is on wavelength routing, which relies on the novel wavelength-sensitive elements described in Volume IIIB Such networks offer the prospect of

“optical transparency,” a concept that enhances flexibility in network de- sign Commercial undersea and terrestrial networks are already incorporat- ing preliminary aspects of wavelength routing by the provision of WDM add-drop multiplexing Further, the proposed WDM PON networks in

Chapter 13 also employ wavelength routing

Chapter 15, however, considers a wider range of architectures and appli- cations of this technology After reviewing optical transparency, it treats WDM rings for local networks, metropolitan distribution, and continental undersea telecommunications (AfricaONE) Then it reviews several multi-

Important considerations in the basics, design, and performance of

and Howard D Kidorf Designs are optimized for digital terrestrial and undersea systems, as well as for applications to analog cable television and wavelength-routed WDM networks, which are covered in Chapters 13 14 and 15 in Volume IIIA Performance monitoring and the higher order effects that come into play for the extreme distances encountered in under- sea systems are also discussed

Transmitter and Receiver Design for Amplified Lightwave Systems (Chapter 3)

Chapter 3 by Daniel A Fishman and B Scott Jackson, defines the engi- neering requirements for transmitters and receivers in amplified systems mainly operating at 2.5 Gbls and satisfying the SONET-SDH standards Topics that are essential for commercial networks, such as performance monitoring, are included

(Chapter 4)

the laser sources have become more stringent than those described in Chapter 13 of OFT ZI The greater fiber spans and the introduction of EDFA and WDM technologies require both improved performance and

totally new functionality In Chapter 4, Thomas L Koch reviews lasers and subsystems designed for low-chirp applications, employing direct modula- tion, external modulation, and integrated laser-modulators He also covers

a variety of laser structures designed to satisfy the special needs of WDM systems for precise fixed wavelengths, tunable wavelengths, and multiple wavelengths These structures include fixed DFB (distributed feedback) lasers, tunable DBR (distributed Bragg reflector) lasers, multifrequency waveguide grating router lasers (MFL), and array lasers

Advances in Semiconductor Laser Growth and Fabrication Technology (Chapter 5)

Some of the greatest advances in laser performance in recent years can be traced to advances in materials growth In Chapter 5, Charles H Joyner covers such advances as strained quantum wells, selective area growth, selective etching, and beam expanded lasers

Vertical-Cavity Surface-Emitting Lasers (Chapter 6)

The edge-emitting lasers employed in today’s lightwave systems are de- scribed in Chapter 4 In Chapter 6 , L A Coldren and B J Thibeault

update progress on a different structure Vertical-cavity surface-emitting lasers (VCSELs) are largely research devices today but may find a role in telecommunications systems by the time of the next volume of this series Because of their unique structure, VCSELs lend themselves to array and WDM applications

Optical Fiber Components and Devices (Chapter 7)

Although fiber serves mainly as a transmission line, it is also an extremely

convenient form for passive and active components that couple into fiber

transmission lines A key example is the EDFA, which is described in

Chapter 4, Volume IIIA, and Chapter 2, Volume IIIB In Chapter 7, Alice

of UV-induced fiber gratings, which have important uses as WDM multiple- xers and add-drop filters, narrow band filters, dispersion compensators, EDFA gain equalizers, and selective laser mirrors

Special fibers also serve as the vehicles for high-power lasers and ampli- fiers in the 1550- and 1310-nm bands High-power sources are needed for

1 Overview 11

long repeaterless systems and passively split cable television distribution networks Among the lasers and amplifiers discussed are 1550-nm Er/Yb cladding-pumped, 1300-nm Raman, and Pr and Tm up-conversion devices

Silicon Optical Bench Waveguide Technology (Chapter 8)

A useful technology for making passive planar waveguide devices has been developed in several laboratories around the world: at AT&T Bell Labora- tories, the technology is called silicon optical bench ( S U B ) Waveguide patterns are formed photolithographically in a silica layer deposited on a

silicon substrate In Chapter 8, Yuan P Li and Charles H Henry describe

the SiOB fabrication process and design rules suitable for realizing a variety

of components The planar components include bends, splitters, directional couplers, star couplers, Bragg filters, multiplexers and add-drop filters Different design options are available for the more complex devices i.e

a chain of Fourier filters or an arrayed waveguide approach The latter technique has been pioneered to Corrado Dragone of Bell Laboratories (Dragone, Edwards and Kistler 1991) to design commercial WDM compo-

and add-drop filters

Lithium Niobate Integrated Optics: Selected Contemporary Devices and System Applications (Chapter 9)

More than 20 years have passed since the invention of titanium-diffused waveguides in lithium niobate (Schmidt and Kaminow 1974) and the associ- ated integrated optic waveguide electrooptic modulators (Kaminow, Stulz, and Turner 1975) During that period, external modulators have competed with direct laser modulation, and electrooptic modulators have competed with electroabsorption modulators Each has found its niche: the external modulator is needed in high-speed, long-distance digital, and high-linearity analog systems where chirp is a limitation; internal modulation is used for economy, when performance permits (See Chapter 4 in Volume IIIB.)

In Chapter 9, Fred Heismann, Steven K Korotky and John J Veselka review advances in lithium niobate integrated optic devices The design and performance, including reliability and stability, of phase and amplitude modulators and switches polarization controllers and modulators and elec- trooptic and acoustooptic tunable wavelength filters are covered

Photonic Switching (Chapter 10)

Whereas Chapter 9 deals with the modulation or switching of a single input, Chapter 10 deals with switching arrays These arrays have not yet found commercial application, but they are being engineered for forward-looking system demonstrations such as the D A R P A MONET project (Multiwave- length Optical Network), as mentioned in Chapter 15, Volume IIIA In Chapter 10, Edmond J Murphy reviews advances in lithium niobate, semi- conductor, and acoustooptic switch elements and arrays Murphy also cov- ers designs for various device demonstrations

References

Dragone, C., C A Edwards, and R C Kistler, 1991 Integrated optics N X N

Kaminow I P L W Stulz, and E H Turner 1975 Efficient strip-waveguide Schmidt, R V., and I P Kaminow 1974 Metal-diffused optical waveguides in

multiplexer on silicon I E E E Photon Techn Lett 3:896-899

modulator Appl Phys Lett 27:555-557

LiNb03 Appl Phys Lett 25458-460

Chapter 2

Joseph E Berthold

Bellcore Red Bank, Nebr Jer-se!

SONET and ATM

The mid-1980s to the mid-1990s has seen unprecedented change in commu- nications technology This chapter deals with two great technological ad-

vances The first is the synchronous digital hierarchy (SDH) an Interna-

tional Telecommunications Union (ITU) standard, and the closely related synchronous optical network (SONET) an American National Standards Institute (ANSI) standard Although there are differences between the two standards for the purposes of this chapter the distinctions are unimportant

We use the terms SDH and S O N E T interchangeably The second is the

asynchronous transfer mode (ATM) which also is standardized in the ITU These are the major standards that specify how to make use of optical communications technology in that they specify how information is carried and manipulated through optical networks

Optical communications technological advances were necessary, but not sufficient, for the creation of the SDH and ATM They arose as a result

ning in the early 1980s They were also stimulated by the needs of businesses that were beginning to understand and apply information technology, and

This chapter gives an overview of SONET and ATM mainly to show what else is needed beyond high-capacity optical links to create broadband networks We first examine the forces in the marketplace that helped bring these concepts forth and the technological advances that made them possi-

new standards have redefined what is needed to make major advances in networking technology This should be instructive as researchers to seek

13

OPTICAL F I B E R TE LECOhlMUNIC-\TIONS

A E MERGING COMPETITION IN TELECO MMUNICA TIONS

communications industries throughout the world In the United States, the

1982 agreement to divest AT&T of its local-exchange business fractured the Bell System into eight independent companies The seven local-exchange companies formed had nonoverlapping service areas They were not al- lowed to transport signals across their entire service areas, but were re- stricted to stay within local access and transport areas (LATAs), of which several hundred were formed The local-exchange companies were required

to offer equal access to their LATA networks to interexchange telephone companies AT&T was the largest interexchange company in the United States, but it had a growing number of competitors What had been a nationwide network largely designed, built, and operated by a vertically integrated company, AT&T, was to be transformed into a nationwide net- work composed of many competing but interoperable networks In this new environment, neither the network operators nor the network equipment suppliers were willing to have the new AT&T unilaterally make the technol-

ogy decisions for the U.S communications industry The original SONET

proposal was a direct outcome of the AT&T divestiture There was a need

to have a standardized, high-capacity interface, because so many interfaces were now required

Although AT&T divestiture is often used as the prime example of the changing telecommunications business environment, many similar upheav- als are in progress throughout the world Nippon Telephone and Telegraph (NTT) is the process of privatization, and competition has begun in the Japanese market In Europe, the nationally owned telecommunications organizations, the PTTs (posts, telephone, and telegraphs), are on a time- table for privatization and competition England was among the first to introduce competition, and British Telecom now has competition for local telephone service by companies that also offer cable television services

B TECHNOLOGICAL FORCES THAT INFLUENCED SONET AND ATM

Besides market demand, major advances were made in three key technolo- gies: optical transmission technology, high-speed integrated circuit technol- ogy, and microprocessor technology All these technologies had a major

The history of optical communications has been one of ever-increasing performance and increasing performance-cost ratio The measure of trans-

2 SONETand ATM 15

mission performance, the distance-bit-rate product, has seen exponential growth during several decades, and this growth is expected to continue well into the future The challenge to networking systems architects, re- searchers, and developers is to provide the networking technology needed

to exploit the capability of optical communications technology - that is

to create the correct signal structures and networking principles that will make the most of optical transmission performance now and in the future Later in this chapter we see that the SONET and ATM standards are distinguished from most previous data networking and telecommunications standards in that they are scalable in bit rate They were designed with that relentless exponential in mind!

Another impact of the expected continuing growth of optical transmis- sion performance is that bandwidth was no longer viewed as a scarce commodity, one to be carefully guarded Efficiency was in some cases sacrificed to keep a simple, scalable multiplexing structure Bandwidth was also dedicated to support operations functions

Besides optical transmission technology, high-speed very large scale inte-

on the basic structure of SONET and ATM VLSI makes highly complex signal structures possible and cost-effective, because whatever can be ac- complished "on a chip" can be mass-produced at low cost If high levels

of integration are achievable it is possible to make trade-offs in the execu- tion of signal processing functions by choosing between high-speed serial and parallel logic, or by choosing the appropriate combination of the two Functions that were associated previously with serial implementations, such

as high-speed signal scrambling (used to make clock recovery in a digital receiver easier and more reliable), can also be done by using parallel algorithms With the capability of doing all the complex logic required

in parallel, the only required high-speed electronics function in an SDH

multiplexer is the final bit-interleaving process that creates the high-speed transmission signal The VLSI technology used for signal processing may even have adequate speed for all multiplexing functions, including serializa-

tion, for some signal rates This comes about as a by-product of semiconduc-

tor technology evolution to increase circuit density As circuit dimensions decrease, gate delays also decrease For the highest line rates the relatively few logic gates required for the sterilization function may be included on

a separate chip, perhaps the same chip responsible for optical signal modu- lation

Many assumptions went into the development of SONET and ATM that were not manifest directly in the digital signal formats but are nonetheless

groundbreaking in the context of previous systems One of the most impor- tant other technological impacts was that of the microprocessor Micropro- cessor and semiconductor memory technologies are on the same relentless exponential performance-improvement curves as optical transmission tech- nology is Early transmission systems were relatively static They were monitored and controlled, whenever control was possible, by centralized management systems Systems in the future could be foreseen whereby a local processor could be as powerful as one of the processors in the older central monitoring systems Although this technology will have a profound impact on the way in which management processes are accomplished in the future, and in the architecture of network management systems, it did also have an impact on the signal format for SONET, through the addition

of an embedded data communications channel in the signal to allow remote management and the download of software into network elements Dis- tributing intelligence and control also raised new security and network reliability issues

11 Network Solutions

A APPLICATION NEEDS

Before we discuss the details of the SDH and ATM, it is instructive to

consider some examples of applications that should be supported by wide- area networks The ones chosen here are in use in private networks, predom- inantly in local area networks (LANs) The choices of the following four applications, listed in Table 2.1, were made to illustrate their differences and to show that a network optimized for specific services is unlikely to

be flexible enough to meet the needs of as-yet undiscovered applications

The first example is distributed computing, a broad term that is meant

to include the execution of tasks by computers where there is a need to

share processing power or data across a network of computers The traffic generated by distributed computing applications is characterized by the transfer of variable-sized packets of information In some cases they will

be extremely short, in other cases very long The traffic patterns are unpre- dictable, except for a very specific task The system designer would most like to have low latency - that is, low delay from when data are inserted into the network and when they are delivered Systems can deal with any level of delay, but the designer prefers it to be as low as possible If the

2 SONET and ATM 17

Table 2.1 Characteristics of Applications and the

Requirements They Place on Networks

What Is Desired Data Loss Symmetrical Application Transported Delay Impact Traffic?

ms )

Retransmit

Retransmit

Accept impairment

conceal

Accept impairment, conceal

High-resolution image delivery is characterized by the transfer of large data blocks A 2000- by 2000-pixel image with 24-bit pixel resolution re- quires 100 Mb of data If it were transmitted over a 100-Mbk network, it would consume the full bandwidth for 1 s Assuming that the image retrieval

is done by a person, delays of several seconds would be tolerable Any transmission errors would require retransmission of the image, or part if

it was segmented for delivery If the source of the images was a data archive, the outbound bandwidth would be much larger than the inbound bandwidth, which would consist mainly of short request packets

Audio and video database access brings the first real-time requirement

to the network The traffic is a large block of data, but it is data that must

rate required may vary over many orders of magnitude, depending on the audio and video fidelity The audio and video must be synchronized The

delay to start playback can be on the order of seconds, which allows a portion of the data stream to be stored locally to compensate for later delays in network transmission If errors are introduced there is no time for retransmission and the signal would appear impaired unless the decoder employs mechanisms to disguise the loss, such as replaying the previous video frame The traffic pattern would be asymmetrical

The final example involves human interaction, and psychology comes into play In an audiovisual conference, absolute delay is important Re- searchers found that a delay in the communications path is noticeable when

it becomes significantly longer than 50 ms Satellite telephone links were found to be objectionable to many users The delays between questions or statements and reactions to them can be interpreted as less than frank and honest responses, which can lead to a lack of trust Again, loss of data must

be tolerated, because it would be impossible to retransmit errored data in time for it to be useful Techniques for error mitigation in codecs may lessen the impact of errors

There was no single networking approach that was able to support these example services over the wide area cost-effectively We had a mixture of bursty data, some with low delay requirements and others without, and continuous media requirements where guaranteed bandwidth and low delay were essential Not all applications were symmetrical, so a symmetrical network might waste half the network resources There was a need for a flexible approach that could also handle the mix of application demands ef- ficiently

B MAJOR NETWORKING PARADIGMS: CIRCUITS, PACKETS, AND CELLS

1 Circuit Networks

Circuit technology is used in telecommunications networks worldwide Cir- cuits are connections between communication endpoints and provide users with access to a network resource for the duration of the connection Some

in contrast to shorter lived connections, or switched connecrions Permanent

connections are provided by dedicating a transmission circuit, and they are

managed through transport facilities rather than switching facilities A very

common use of permanent connections is for circuits used by businesses to interconnect their local computing facilities across wide areas The Internet

2 SONET and ATM 19

makes use of permanent circuits leased from telecommunications carriers

to interconnect their switches Permanent circuits are leased for relatively long periods, months to years Switched connections are established on demand, on the basis of end-user signaling They are provided through service switches, and the networkwide connection is established on the basis of available resources, according to a routing policy

In early days, a telephony circuit was just a pair of wires, dedicated or connected through a switch With the advent of digital technology and time-division multiplexing (TDM), many circuits could share a transmission medium but even so they were a dedicated resource, a time slot If a circuit was left connected even though no data flowed through it the transmission resource could not be used by another connection that had data to send

SONET and the SDH are circuit networks and they are administered

as permanenf circuit networks It would be possible to create SDH circuit

switches where connections could be set up in response to user signaling

as our telephone circuits are today, and proposals to do so were debated at some length in the late 1980s However, the advent of ATM as a networking solution that could support the same type of application needs but was also more flexible in bit rate and could additionally support packet traffic relegated SONET and the S DH to the role of a transport rather than a service switching infrastructure

Packet networks grew out of the computing world and used computing technology for their switching Packet networks rely on circuit networks for transmission, and packets provide a means of sharing the circuit bandwidth Most simply put, a packet is a block of information with some overhead

most important purpose is to distinguish one packet from another so that they can be delivered to the correct destination Another important function included in the header is a mechanism for error checking The header needs

to have error protection so that errors do not lead to misrouted packets The information content needs error checking so that the computers can

be sure that the information is transmitted error free In the event that errors are introduced in transmission the packets can be retransmitted There are two important types of packet networks: connectionless packet networks and connection-oriented packet networks Connectionless packet

networks operate in a way analogous to the post office, and their packets

best effort packet-delivery service Connectionless packets are launched into the network with a large and powerful header that contains all the information necessary to deliver the packet, whatever its destination As

a by-product of their connectionless nature, it is not possible to guarantee performance or reserve network resources for any particular communica- tion, so connectionless networks do not support real-time services reliably The switches in the network are stateless When the network is subjected

to overload, packets are buffered for later delivery If the buffer capacity

is exceeded, the packets are discarded In connectionless networks there

is no guarantee that all packets will traverse the same network path and

be subject to the same delays If a communication requires the transmission

of a series of packets, it is up to the end stations to ensure that the sequence

is preserved

The Internet is an example of a connectionless packet network, because

it is based on the Internet protocol (IP), which is a connectionless protocol

IP packets have a complete source and destination address in their header capable of delivering information to anyone on the Internet around the world

Connection-oriented packet networks have similarities with both circuit networks and packet networks They are like circuit networks in that a connection, called a virtual circuit, must be established before data can be transported Network nodes are prepared to support virtual circuits, and they devote resources to them The resources include buffer memory and link bandwidth Unlike circuits, when packets are not submitted to the network for some of the virtual circuits, the resources can be devoted to carry other virtual circuits This is called statistical multiplexing Assump-

tions are made about the traffic likely to be contributed by all the sources, and the network may overload if all sources begin sending packets at their peak rate Occasional pileups of traffic are to be expected and are handled

by maintaining local memory to buffer packets until it is possible to transmit

them The network is not stateless, as a connectionless packet network is, but stores the state of all connections that are present When all the re- sources are committed, the network may deny requests to establish new virtual circuits The headers of connection-type packet networks can be shorter than those of connectionless networks because the addressing por- tion of the header needs only to distinguish each virtual circuit from each other established on common links and switch ports

2 SONET and ATM 21

Cell networks, sometimes called fasr packer networks, were designed to

support both continuous high-bandwidth data streams and bursty packet traffic Cell networks are a special type of connection-oriented packet net- work with a protocol so simple that it can be easily placed in hardware to allow very fast operation The ATM is a cell-based networking protocol Its cells are short and of a fixed length, an important simplification There

is a connection setup process before traffic can be sent, and as a part of that process there can be a negotiation of resources which in the ATM case is link and switch port bandwidth Depending on how the connection setup process is managed by a network operator, resources can be allocated

to guarantee that all established circuits can continually transmit at their peak allowable bit rates In this case it is a circuit network that is freed from the rigid bit rates of the TDM hierarchies It can also be administered

as a network that allows statistical multiplexing where different connections vie for a share of the bandwidth available There are a number of service classes supported that would allow both high-priority circuits and loss- or delay-tolerant packets to coexist ATM is not tied to any physical transmis- sion medium although it was designed with the assumption of low error rate fiber optics transport We describe ATM in more detail later

C NEEDS L E A D I N G T O SONET A N D T H E S D H

There are a number of advantages that would result from a new system of

as follows:

Internationally standard broadband signals The digital transmission

standards that preceded the SDH were lacking in several ways They arose from a need to time-division multiplex voice signals and were therefore based on a unit of bandwidth of 64 kb/s No other rate had any special significance from a service perspective TDM systems were deployed first in the United States and created a 24- channel multiplex European systems came somewhat later and they were able to achieve a larger multiplex because of advances in electronics technology in the intervening period Different multiplex hierarchies grew, but none were universal In the United States, stan- dards were in use for signal levels up to 45 Mb/s, but for levels above that different vendors diverged in their approaches The re-