Optical Fiber Telecommunications V [Volume B, 1 ed.] 9780080569628

Citation preview

Optical Fiber Telecommunications V B

About the Editors

Ivan P. Kaminow retired from Bell Labs in 1996 after a 42-year career. He conducted seminal studies on electrooptic modulators and materials, Raman scattering in ferroelectrics, integrated optics, semiconductor lasers (DBR, ridge-waveguide InGaAsP, and multi-frequency), birefringent optical fibers, and WDM networks. Later, he led research on WDM components (EDFAs, AWGs, and fiber Fabry-Perot Filters), and on WDM local and wide area networks. He is a member of the National Academy of Engineering and a recipient of the IEEE/OSA John Tyndall, OSA Charles Townes, and IEEE/LEOS Quantum Electronics Awards. Since 2004, he has been Adjunct Professor of Electrical Engineering at the University of California, Berkeley. Tingye Li retired from AT&T in 1998 after a 41-year career at Bell Labs and AT&T Labs. His seminal work on laser resonator modes is considered a classic. Since the late 1960s, he and his groups have conducted pioneering studies on lightwave technologies and systems. He led the work on amplified WDM transmission systems and championed their deployment for upgrading network capacity. He is a member of the National Academy of Engineering and a foreign member of the Chinese Academy of Engineering. He is also a recipient of the IEEE David Sarnoff Award, IEEE/OSA John Tyndall Award, OSA Ives Medal/ Quinn Endowment, AT&T Science and Technology Medal, and IEEE Photonics Award. Alan E. Willner has worked at AT&T Bell Labs and Bellcore, and he is Professor of Electrical Engineering at the University of Southern California. He received the NSF Presidential Faculty Fellows Award from the White House, Packard Foundation Fellowship, NSF National Young Investigator Award, Fulbright Foundation Senior Scholar, IEEE LEOS Distinguished Lecturer, and USC University-Wide Award for Excellence in Teaching. He is a Fellow of IEEE and OSA, and he has been President of the IEEE LEOS, Editor-in-Chief of the IEEE/OSA J. of Lightwave Technology, Editor-in-Chief of Optics Letters, Co-Chair of the OSA Science & Engineering Council, and General Co-Chair of the Conference on Lasers and Electro-Optics.

Optical Fiber Telecommunications V B Systems and Networks Edited by

Ivan P. Kaminow Tingye Li Alan E. Willner

AMSTERDAM • BOSTON • HEIDELBERG • LONDON NEW YORK • OXFORD • PARIS • SAN DIEGO SAN FRANCISCO • SINGAPORE • SYDNEY • TOKYO Academic Press is an imprint of Elsevier

Academic Press is an imprint of Elsevier 30 Corporate Drive, Suite 400, Burlington, MA 01803, USA 525 B Street, Suite 1900, San Diego, California 92101-4495, USA 84 Theobald’s Road, London WC1X 8RR, UK ¥ This book is printed on acid-free paper. 

Copyright  2008, Elsevier Inc. 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. Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone: (þ 44) 1865 843830, fax: (þ 44) 1865 853333, E-mail: [email protected]. You may also complete your request on-line via the Elsevier homepage (http://elsevier.com), by selecting ‘‘Support & Contact’’ then ‘‘Copyright and Permission’’ and then ‘‘Obtaining Permissions.’’ Library of Congress Cataloging-in-Publication Data Application submitted British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library. ISBN: 978-0-12-374172-1 For information on all Academic Press publications visit our Web site at www.books.elsevier.com Printed in the United States of America 08 09 10 11 12 8 7 6 5 4 3 2 1

Working together to grow libraries in developing countries www.elsevier.com | www.bookaid.org | www.sabre.org

For Florence, Paula, Leonard, and Ellen with Love—IPK For Edith, Debbie, and Kathy with Love—TL For Michelle, our Children (Moshe, Asher, Ari, Jacob), and my Parents with Love—AEW

This page intentionally left blank

Contents

Contributors

ix

Chapter 1

Overview of OFT V volumes A & B Ivan P. Kaminow, Tingye Li, and Alan E. Willner

1

Chapter 2

Advanced optical modulation formats Peter J. Winzer and Rene´-Jean Essiambre

23

Chapter 3

Coherent optical communication systems Kazuro Kikuchi

95

Chapter 4

Self-coherent optical transport systems Xiang Liu, Sethumadhavan Chandrasekhar, and Andreas Leven

131

Chapter 5

High-bit-rate ETDM transmission systems Karsten Schuh and Eugen Lach

179

Chapter 6

Ultra-high-speed OTDM transmission technology Hans-Georg Weber and Reinhold Ludwig

201

Chapter 7

Optical performance monitoring Alan E. Willner, Zhongqi Pan, and Changyuan Yu

233

Chapter 8

ROADMs and their system applications Mark D. Feuer, Daniel C. Kilper, and Sheryl L. Woodward

293

Chapter 9

Optical Ethernet: Protocols, management, and 1–100 G technologies Cedric F. Lam and Winston I. Way

Chapter 10 Fiber-based broadband access technology and deployment Richard E. Wagner

345

401

vii

viii

Chapter 11 Global landscape in broadband: Politics, economics, and applications Richard Mack Chapter 12 Metro networks: Services and technologies Loukas Paraschis, Ori Gerstel, and Michael Y. Frankel

Contents

437

477

Chapter 13 Commercial optical networks, overlay networks, and services Robert Doverspike and Peter Magill

511

Chapter 14 Technologies for global telecommunications using undersea cables Se´bastien Bigo

561

Chapter 15 Future optical networks Michael O’Mahony

611

Chapter 16 Optical burst and packet switching S. J. Ben Yoo

641

Chapter 17 Optical and electronic technologies for packet switching Rodney S. Tucker Chapter 18 Microwave-over-fiber systems Alwyn J. Seeds Chapter 19 Optical interconnection networks in advanced computing systems Keren Bergman

695

739

765

Chapter 20 Simulation tools for devices, systems, and networks Robert Scarmozzino

803

Index to Volumes VA and VB

865

Contributors

Keren Bergman, Department of Electrical Engineering, Columbia University, 1312 S.W. Mudd, 500 West 120th Street, New York, NY 10027, USA, [email protected] Se´bastien Bigo, Alcatel-Lucent Bell Labs, Centre de Villarceaux, Route de Villejust, 91620, Nozay, France, [email protected] Sethumadhavan Chandrasekhar, Bell Laboratories, Alcatel-Lucent, Holmdel, NJ, USA, [email protected] Robert Doverspike, Transport Network Evolution Research AT&T Labs Research – Room A5-1G12, 200 S. Laurel Ave, Middletown, NJ, USA, [email protected] Rene´-Jean Essiambre, Bell Laboratories, Alcatel-Lucent, 791 Holmdel-Keyport Road, Holmdel, NJ 07733, USA, [email protected] Mark D. Feuer, Optical Systems Research, AT&T Labs, 200 S. Laurel Ave, Middletown, NJ, USA, [email protected] Michael Y. Frankel, CTO Office, Ciena, 920 Elkridge Landing Rd, Linthicum, MD, USA, [email protected] Ori Gerstel, Cisco Systems, 170 West Tasman Drive, San Jose, CA 95134, USA, [email protected] Ivan P. Kaminow, University of California, Berkeley, CA, USA, [email protected] Kazuro Kikuchi, Department of Frontier Informatics, University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8561, Japan, [email protected] Daniel C. Kilper, Bell Laboratories, Alcatel-Lucent, 791 Holmdel-Keyport Road, L-137, Holmdel, NJ, USA, [email protected] Eugen Lach, Alcatel-Lucent Deutschland AG, Bell Labs Germany, Lorenzstrasse 10, D-70435 Stuttgart, Germany, [email protected] ix

x

Contributors

Cedric F. Lam, OpVista, 870 N. McCarthy Blvd, Milpitas, CA, USA, [email protected] Andreas Leven, Bell Laboratories, Alcatel-Lucent, Holmdel, NJ, USA, [email protected] Tingye Li, Boulder, CO, USA, [email protected] Xiang Liu, Bell Laboratories, Alcatel-Lucent, Department of Optical Networks, Holmdel, NJ, USA, [email protected] Reinhold Ludwig, Fraunhofer Institute for Telecommunications, Heinrich-Hertz-Institut, Einsteinufer 37, D-10587 Berlin, Germany, [email protected] Richard Mack, KMI Research, CRU Group, Mount Pleasant, London, UK, [email protected] Peter Magill, Optical Systems Research, AT&T Labs, 200 S. Laurel Ave; Middletown, NJ, USA, [email protected] Michael O’Mahony, Department of Electronic System Engineering, University of Essex, Colchester, CO43SQ, UK, [email protected] Zhongqi Pan, Department of Electrical and Computer Engineering, University of Louisiana at Lafayette, USA, [email protected] Loukas Paraschis, Cisco Systems, 170 West Tasman Drive, San Jose, CA 95134, USA, [email protected] Robert Scarmozzino, RSoft Design Group, 400 Executive Blvd., Ossining, NY 10562, USA, [email protected] Karsten Schuh, Alcatel-Lucent Deutschland AG, Bell Labs Germany, Lorenzstrasse 10, D-70435 Stuttgart, Germany, [email protected] Alwyn J. Seeds, Department of Electronic and Electrical Engineering, University College London, Torrington Place, London WC1E 7JE, England, [email protected] Rodney S. Tucker, Department of Electrical and Electronic Engineering, University of Melbourne, Vic., 3010, Australia, [email protected] Richard E. Wagner, Corning International, Corning, New York, USA, [email protected] Winston I. Way, OpVista, 870 N. McCarthy Blvd, Milpitas, CA, USA, [email protected]

Contributors

xi

Hans-Georg Weber, Fraunhofer Institute for Telecommunications, Heinrich-Hertz-Institut, Einsteinufer 37, D-10587 Berlin, Germany, [email protected] Alan E. Willner, University of Southern California, Los Angeles, CA, USA, [email protected] Peter J. Winzer, Bell Laboratories, Alcatel-Lucent, 791 Holmdel-Keyport Road, Holmdel, NJ 07733, USA, [email protected] Sheryl L. Woodward, Optical Systems Research, AT&T Labs, 200 S. Laurel Ave, Middletown, NJ, USA, [email protected] S. J. Ben Yoo, Department of Electrical and Computer Engineering, University of California, Davis, CA, USA, [email protected] Changyuan Yu, Department of Electrical and Computer Engineering, National University of Singapore, Singapore, [email protected]

This page intentionally left blank

1 Overview of OFT V volumes A & B Ivan P. Kaminow*, Tingye Li†, and Alan E. Willner‡ * University †

of California, Berkeley, CA, USA Boulder, CO, USA ‡ University of Southern California, Los Angeles, CA, USA

Optical Fiber Telecommunications V (OFT V) is the fifth installment of the OFT series. Now 29 years old, the series is a compilation by the research and development community of progress in optical fiber communications. Each edition reflects the current state-of-the-art at the time. As Editors, we started with a clean slate of chapters and authors. Our goal was to update topics from OFT IV that are still relevant as well as to elucidate topics that have emerged since the last edition.

1.1 FIVE EDITIONS Installments of the series have been published roughly every 6–8 years and chronicle the natural evolution of the field: • In the late 1970s, the original OFT (Chenoweth and Miller, 1979) was concerned with enabling a simple optical link, in which reliable fibers, connectors, lasers, and detectors played the major roles. • In the late 1980s, OFT II (Miller and Kaminow, 1988) was published after the first field trials and deployments of simple optical links. By this time, the advantages of multiuser optical networking had captured the imagination of the community and were highlighted in the book. • OFT III (Kaminow and Koch, 1997) explored the explosion in transmission capacity in the early-to-mid 1990s, made possible by the erbium-doped fiber amplifier (EDFA), wavelength division multiplexing (WDM), and dispersion management. Optical Fiber Telecommunications V B: Systems and Networks Copyright  2008, Elsevier Inc. All rights reserved. ISBN: 978-0-12-374172-1

1

Ivan P. Kaminow et al.

2

• By 2002, OFT IV (Kaminow and Li, 2002) dealt with extending the distance and capacity envelope of transmission systems. Subtle nonlinear and dispersive effects, requiring mitigation or compensation in the optical and electrical domains, were explored. • The present edition of OFT, V, (Kaminow, Li, and Willner, 2008) moves the series into the realm of network management and services, as well as employing optical communications for ever-shorter distances. Using the high bandwidth capacity in a cost-effective manner for customer applications takes center stage. In addition, many of the topics from earlier volumes are brought up to date; and new areas of research which show promise of impact are featured. Although each edition has added new topics, it is also true that new challenges emerge as they relate to older topics. Typically, certain devices may have adequately solved transmission problems for the systems of that era. However, as systems become more complex, critical device technologies that might have been considered a “solved problem” previously have new requirements placed upon them and need a fresh technical treatment. For this reason, each edition has grown in sheer size, i.e., adding the new and, if necessary, reexamining the old. An example of this circular feedback mechanism relates to the fiber itself. At first, systems simply required low-loss fiber. However, long-distance transmission enabled by EDFAs drove research on low-dispersion fiber. Further, advances in WDM and the problems of nonlinear effects necessitated development of nonzero dispersion fiber. Cost considerations and ultra-high-performance systems, respectively, are driving research in plastic fibers and ultra-low-polarization-dependent fibers. We believe that these cycles will continue. Each volume includes a CD-ROM with all the figures from that volume. Select figures are in color. The volume B CD-ROM also has some supplementary Powerpoint slides to accompany Chapter 19 of that volume.

1.2 PERSPECTIVE OF THE PAST 6 YEARS Our field has experienced an unprecedented upheaval since 2002. The irrational exuberance and despair of the technology “bubble-and-bust” had poured untold sums of money into development and supply of optical technologies, which was followed by a depression-like period of over supply. We are happy to say that, by nearly all accounts, the field is gaining strength again and appears to be entering a stage of rational growth. What caused this upheaval? A basis seems to be related to a fundamental discontinuity in economic drivers. Around 2001, worldwide telecom traffic ceased being dominated by the slow-growing voice traffic and was overtaken by the rapidly growing Internet traffic. The business community over-estimated the

1. Overview of OFT V Volumes A & B

3

growth rate, which generated enthusiasm and demand, leading to unsustainable expectations. Could such a discontinuity happen again? Perhaps, but chastened investors now seem to be following a more gradual and sensible path. Throughout the “bubble-and-bust” until the present, the actual demand for bandwidth has grown at a very healthy 80% per year globally; thus, real traffic demand experienced no bubble at all. The growth and capacity needs are real, and should continue in the future. As a final comment, we note that optical fiber communications is firmly entrenched as part of the global information infrastructure. The only question is how deeply will it penetrate and complement other forms of communications, e.g., wireless, access, and on-premises networks, interconnects, satellites, etc. This prospect is in stark contrast to the voice-based future seen by OFT, published in 1979, before the first commercial intercontinental or transatlantic cable systems were deployed in the 1980s. We now have Tb/s systems for metro and long-haul networks. It is interesting to contemplate what topics and concerns might appear in OFT VI.

1.3 OFT V VOLUME A: COMPONENTS AND SUBSYSTEMS 1.3.1 Chapter 1. Overview of OFT V volumes A & B (Ivan P. Kaminow, Tingye Li, and Alan E. Willner) This chapter briefly reviews herewith all the chapters contained in both volumes of OFT V.

1.3.2 Chapter 2. Semiconductor quantum dots: Genesis—The Excitonic Zoo—novel devices for future applications (Dieter Bimberg) The ultimate class of semiconductor nanostructures, i.e., “quantum dots” (QDs), is based on “dots” smaller than the de Broglie wavelength in all three dimensions. They constitute nanometer-sized clusters that are embedded in the dielectric matrix of another semiconductor. They are often self-similar and can be formed by self-organized growth on surfaces. Single or few quantum dots enable novel devices for quantum information processing, and billions of them enable active centers in optoelectronic devices like QD lasers or QD optical amplifiers. This chapter covers the area of quantum dots from growth via various band structures to optoelectronic device applications. In addition, high-speed laser and amplifier operations are described.

4

Ivan P. Kaminow et al.

1.3.3 Chapter 3. High-speed low-chirp semiconductor lasers (Shun Lien Chuang, Guobin Liu, and Piotr Konrad Kondratko) One advantage of using quantum wells and quantum dots for the active region of lasers is the lower induced chirp when such lasers are directly modulated, permitting direct laser modulation that can save on the cost of separate external modulators. This chapter provides a comparison of InAlGaAs with InGaAsP long-wavelength quantum-well lasers in terms of high-speed performance, and extracts the important parameters such as gain, differential gain, photon lifetime, temperature dependence, and chirp. Both DC characteristics and high-speed direct modulation of quantum-well lasers are presented, and a comparison with theoretical models is made. The chapter also provides insights into novel quantum-dot lasers for high-speed operation, including the ideas of p-type doping vs tunneling injection for broadband operation.

1.3.4 Chapter 4. Recent advances in surface-emitting lasers (Fumio Koyama) Vertical cavity surface-emitting lasers (VCSELs) have a number of special properties (compared with the more familiar edge-emitting lasers) that permit some novel applications. This chapter begins with an introduction which briefly surveys recent advances in VCSELs, several of that are then treated in detail. These include techniques for realizing long-wavelength operation (as earlier VCSELs were limited to operation near 850 nm), the performance of dense VCSEL arrays that emit a range of discrete wavelengths (as large as 110 in number), and MEMSbased athermal VCSELs. Also, plasmonic VCSELs that produce subwavelength spots for high-density data storage and detection are examined. Finally, work on all-optical signal processing and slow light is presented.

1.3.5 Chapter 5. Pump diode lasers (Christoph Harder) Erbium-doped fiber amplifiers (EDFAs) pumped by bulky argon lasers were known for several years before telecom system designers took them seriously; the key development was a compact, high-power semiconductor pump laser. Considerable effort and investment have gone into today’s practical pump lasers, driven by the importance of EDFAs in realizing dense wavelength division multiplexed (DWDM) systems. The emphasis has been on high power, efficiency, and reliability. The two main wavelength ranges are in the neighborhood of 980 nm for low noise and 1400 nm for remote pumping of EDFAs. The 1400-nm band is also suitable for Raman amplifiers, for which very high power is needed.

1. Overview of OFT V Volumes A & B

5

This chapter details the many lessons learned in the design for manufacture of commercial pump lasers in the two bands. Based on the performance developed for telecom, numerous other commercial applications for high-power lasers have emerged in manufacturing and printing; these applications are also discussed.

1.3.6 Chapter 6. Ultrahigh-speed laser modulation by injection locking (Connie J. Chang-Hasnain and Xiaoxue Zhao) It has been known for decades that one oscillator (the slave) can be locked in frequency and phase to an external oscillator (the master) coupled to it. Current studies of injection-locked lasers show that the dynamic characteristics of the directly modulated slave are much improved over the same laser when freely running. Substantial improvements are found in modulation bandwidth for analog and digital modulation, in linearity, in chirp reduction and in noise performance. In this chapter, theoretical and experimental aspects of injection locking in all lasers are reviewed with emphasis on the authors’ research on VCSELs (vertical cavity surface-emitting lasers). A recent promising application in passive optical networks for fiber to the home (FTTH) is also discussed.

1.3.7 Chapter 7. Recent developments in high-speed optical modulators (Lars Thyle´n, Urban Westergren, Petter Holmstro¨m, Richard Schatz, and Peter Ja¨nes) Current high-speed lightwave systems make use of electro-optic modulators based on lithium niobate or electroabsorption modulators based on semiconductor materials. In commercial systems, the very high-speed lithium niobate devices often require a traveling wave structure, while the semiconductor devices are usually lumped. This chapter reviews the theory of high-speed modulators (at rates of 100 Gb/s) and then considers practical design approaches, including comparison of lumped and traveling-wave designs. The main emphasis is on electroabsorption devices based on Franz–Keldysh effect, quantum-confined Stark effect and intersubband absorption. A number of novel designs are described and experimental results given.

1.3.8 Chapter 8. Advances in photodetectors (Joe Charles Campbell) As a key element in optical fiber communications systems, photodetectors belong to a well developed sector of photonics technology. Silicon p–i–n and avalanche photodiodes deployed in first-generation lightwave transmission systems operating at 0.82-mm wavelength performed very close to theory. In the 1980s, InP photodiodes

6

Ivan P. Kaminow et al.

were developed and commercialized for systems that operated at 1.3- and 1.5-mm wavelengths, albeit the avalanche photodiodes (APDs) were expensive and nonideal. Introduction of erbium-doped fiber amplifiers and WDM technology in the 1990s relegated APDs to the background, as p–i–n photoreceivers performed well in amplified systems, whereas APDs were plagued by the amplified spontaneous emission noise. Future advanced systems and special applications will require sophisticated devices involving deep understanding of device physics and technology. This chapter focuses on three primary topics: high-speed waveguide photodiodes for systems that operate at 100 Gb/s and beyond, photodiodes with high saturation current for high-power applications, and recent advances of APDs for applications in telecommunications.

1.3.9 Chapter 9. Planar lightwave circuits in fiber-optic communications (Christopher R. Doerr and Katsunari Okamoto) The realization of one or more optical waveguide components on a planar substrate has been under study for over 35 years. Today, individual components such as splitters and arrayed waveguide grating routers (AWGRs) are in widespread commercial use. Sophisticated functions, such as reconfigurable add–drop multiplexers (ROADMs) and high-performance filters, have been demonstrated by integrating elaborate combinations of such components on a single chip. For the most part, these photonic integrated circuits (PICs), or planar lightwave circuits (PLCs), are based on passive waveguides in lower index materials, such as silica. This chapter deals with the theory and design of such PICs. The following two chapters (Chapters 11 and 12) also deal with PICs; however, they are designed to be integrated with silicon electronic ICs, either in hybrid fashion by short wire bonds to an InP PIC or directly to a silicon PIC.

1.3.10 Chapter 10. III–V photonic integrated circuits and their impact on optical network architectures (Dave Welch, Chuck Joyner, Damien Lambert, Peter W. Evans, and Maura Raburn) InP-based semiconductors are unique in their capability to support all the photonic components required for wavelength division multiplexed (WDM) transmitters and receivers in the telecom band at 1550 nm. Present subsystems have connected these individual components by fibers or lenses to form hybrid transmitters and receivers for each channel. Recently, integrated InP WDM transmitter and receiver chips that provide 10 channels, each operating at 10 Gb/s, have been shown to be technically and

1. Overview of OFT V Volumes A & B

7

economically viable for deployment in commercial WDM systems. The photonic integrated circuits are wire-bonded to adjacent silicon ICs. Thus a single board provides optoelectronic regeneration for 10 channels, dramatically reducing interconnection complexity and equipment space. In addition, as in legacy singlechannel systems, the “digital” approach for transmission (as compared to “all-optical”) offers ease of network monitoring and management. This chapter covers the technology of InP photonic integrated circuits (PICs) and their commercial application. The impact on optical network architecture and operation is discussed and technology advances for future systems are presented.

1.3.11 Chapter 11. Silicon photonics (Cary Gunn and Thomas L. Koch) Huge amounts of money have been invested in silicon processing technology, thanks to a steady stream of applications that justified the next stage of processing development. In addition to investment, innovative design, process discipline and large-volume runs made for economic success. The InP PICs described in the previous chapter owe their success to lessons learned in silicon IC processing. Many people have been attracted by the prospects of fabricating PICs using silicon alone to capitalize on the investment and success of silicon ICs. To succeed one requires a large-volume application and a design that can be made in an operating silicon IC foundry facility. A further potential advantage is the opportunity to incorporate on the same photonic chip electronic signal processing. The application to interconnects for high-performance computers is a foremost motivation for this work. While silicon has proven to be the ideal material for electronic ICs, it is far from ideal for PICs. The main shortcoming is the inability so far to make a good light source or photodetector in silicon. This chapter discusses the successes and challenges encountered in realizing silicon PICs to date.

1.3.12 Chapter 12. Photonic crystal theory: Temporal coupled-mode formalism (Shanhui Fan) Photonic crystal structures have an artificially created optical bandgap that is introduced by a periodic array of perturbations, and different types of waveguides and cavities can be fabricated that uniquely use the band gap-based confinement. These artificially created materials have been of great interest for potential optical information processing applications, in part because they provide a common platform to miniaturize a large number of optical components on-chip down to single wavelength scale. For this purpose, many devices can be

8

Ivan P. Kaminow et al.

designed using such a material with a photonic bandgap and, subsequently, introducing line and point defect states into the gap. Various functional devices, such as filters, switches, modulators and delay lines, can be created by controlling the coupling between these defect states. This chapter reviews the temporal coupled-mode theory formalism that provides the theoretical foundation of many of these devices.

1.3.13 Chapter 13. Photonic crystal technologies: Experiment (Susumu Noda) Photonic crystals belong to a class of optical nanostructures characterized by the formation of band structures with respect to photon energy. In 3D photonic crystals, a complete photonic band gap is formed; the presence of light with frequencies lying in the band gap is not allowed. This chapter describes the application of various types of materials engineering to photonic crystals, with particular focus on the band gap/defect, the band edge, and the transmission band within each band structure. The manipulation of photons in a variety of ways becomes possible. Moreover, this chapter discusses the recent introduction of “photonic heterostructures” as well as recent developments concerning two- and three-dimensional photonic crystals.

1.3.14 Chapter 14. Photonic crystal fibers: Basics and applications (Philip St John Russell) Photonic crystal fibers (PCFs)—fibers with a periodic transverse microstructure— first emerged as practical low-loss waveguides in early 1996. The initial demonstration took 4 years of technological development, and since then the fabrication techniques have become more and more sophisticated. It is now possible to manufacture the microstructure in air–glass PCFs to accuracies of 10 nm on the scale of 1 mm, which allows remarkable control of key optical properties such as dispersion, birefringence, nonlinearity and the position and width of the photonic band gaps (PBGs) in the periodic “photonic crystal” cladding. PCF has in this way extended the range of possibilities in optical fibers, both by improving wellestablished properties and introducing new features such as low-loss guidance in a hollow core. In this chapter, the properties of the various types of PCFs are introduced, followed by a detailed discussion of their established or emerging applications. The chapter describes in detail the fabrication, theory, numerical modeling, optical properties, and guiding mechanisms of PCFs. Applications of photonic crystal fibers include lasers, amplifiers, dispersion compensators, and nonlinear processing.

1. Overview of OFT V Volumes A & B

9

1.3.15 Chapter 15. Specialty fibers for optical communication systems (Ming-Jun Li, Xin Chen, Daniel A. Nolan, Ji Wang, James A. West, and Karl W. Koch) Specialty fibers are designed by changing fiber glass composition, refractive index profile, or coating to achieve certain unique properties and functionalities. Some of the common specialty fibers include active fibers, polarization control fibers, dispersion compensation fibers, highly nonlinear fibers, coupling or bridge fibers, high-numerical-aperture fibers, fiber Bragg gratings, and special single mode fibers. In this chapter, the design and performance of various specialty fibers are discussed. Special attention is paid to dispersion compensation fibers, polarizationmaintaining and single-polarization fibers, highly nonlinear fibers, double clad fiber for high-power lasers and amplifiers, and photonic crystal fibers. Moreover, there is a brief discussion of the applications of these specialty fibers.

1.3.16 Chapter 16. Plastic optical fibers: Technologies and communication links (Yasuhiro Koike and Satoshi Takahashi) Plastic optical fiber (POF) consists of a plastic core that is surrounded by a plastic cladding of a refractive index lower than that of the core. POFs have very large core diameters compared to glass optical fibers, and yet they are quite flexible. These features enable easy installation and safe handling. Moreover, the large-core fibers can be connected without high-precision accuracy and with low cost. POFs have been used extensively in short-distance datacom applications, such as in digital audio interfaces. POFs are also used for data transmission within equipment and for control signal transmission in machine tools. During the late 1990s, POFs were used as the transmission medium in the data bus within automobiles. As we move into the future, high-speed communication will be required in the home, and POFs are a promising candidate for home network wiring. This chapter describes the POF design and fabrication, the specific fiber properties of attenuation, bandwidth and thermal stability, and various communications applications, concluding with a discussion of recent developments in graded-index POFs.

1.3.17 Chapter 17. Polarization mode dispersion (Misha Brodsky, Nicholas J. Frigo, and Moshe Tur) Polarization-mode dispersion (PMD) has been well recognized for sometime as an impairment factor that limits the transmission speed and distance in high-speed lightwave systems. The complex properties of PMD have enjoyed scrutiny by

10

Ivan P. Kaminow et al.

theorists, experimentalists, network designers, field engineers and, during the “bubble” years, entrepreneurial technologists. A comprehensive treatment of the subject up to year 2002 is given in a chapter bearing the same title in Optical Fiber Telecommunications IVB, System and Impairments. The present chapter is an overview of PMD with special emphasis on the knowledge accumulated in the past 5 years. It begins with a review of PMD concepts, and proceeds to consider the “hinge” model used to describe field test results, which are presented and analyzed. The important subject of system penalties and outages due to first-order PMD is then examined, followed by deliberations of higher-order PMD, and interaction between fiber nonlinear effects and PMD.

1.3.18 Chapter 18. Electronic signal processing for dispersion compensation and error mitigation in optical transmission networks (Abhijit Shanbhag, Qian Yu, and John Choma) Dispersion equalization has its origin in the early days of analog transmission of voice over copper wires where loading coils (filters) were distributed in the network to equalize the frequency response of the transmission line. Digital transmission over twisted pairs was enabled by the invention of the transversal equalizer which extended greatly the bandwidth and reach. Sophisticated signal processing and modulation techniques have now made mobile telephones ubiquitous. However, it was not until the mid 1990s that wide deployment of Gigabit Ethernet rendered silicon CMOS ICs economical for application in high-speed lightwave transmission. Most, if not all lightwave transmission systems deployed today, use electronic forward error correction and dispersion compensation to alleviate signal degradation due to noise and fiber dispersive effects. This chapter presents an overview of various electronic equalization and adaptation techniques, and discusses their high-speed implementation, specifically addressing 10-Gb/s applications for local-area, metro, and long-haul networks. It comprises a comprehensive survey of the role, scope, limitations, trends, and challenges of this very important and compelling technology.

1.3.19 Chapter 19. Microelectromechanical systems for lightwave communication (Ming C. Wu, Olav Solgaard, and Joseph E. Ford) The earliest commercial applications of microelectromechanical systems (MEMS) were in digital displays employing arrays of tiny mirrors and in accelerometers for airbag sensors. This technology has now found a host of applications in lightwave communications. These applications usually require movable components, such as mirrors, with response times in the neighborhood of 10–6 s, although fixed

1. Overview of OFT V Volumes A & B

11

elements may be called for in some applications. Either a free-space or integrated layout may be used. This chapter describes the recent lightwave system applications of MEMS. In telecommunications, MEMS switches can provide cross-connects with large numbers of ports. A variety of wavelength selective devices, such as reconfigurable optical add–drop multiplexers (ROADM) employ MEMS. More recent devices include tunable lasers and microdisk resonators.

1.3.20 Chapter 20. Nonlinear optics in communications: from crippling impairment to ultrafast tools (Stojan Radic, David J. Moss, and Benjamin J. Eggleton) It is perhaps somewhat paradoxical that optical nonlinearities, whilst having posed significant limitations for long-haul WDM systems, also offer the promise of addressing the bandwidth bottleneck for signal processing for future optical networks as they evolve beyond 40 Gb/s. In particular, all-optical devices based on the 3rd order (3) optical nonlinearity offer a significant promise in this regard, not only because the intrinsic nonresonant (3) is nearly instantaneous, but also because (3) is responsible for a wide range of phenomena, including 3rd harmonic generation, stimulated Raman gain, four-wave mixing, optical phase conjugation, two-photon absorption, and the nonlinear refractive index. This plethora of physical processes has been the basis for a wide range of activity on all-optical signal processing devices. This chapter focuses on breakthroughs in the past few years on approaches based on highly nonlinear silica fiber as well as chalcogenide-glass-based fiber and waveguide devices. The chapter contrasts two qualitatively different approaches to all-optical signal processing based on nonphase-matched and phase-matched processes. All-optical applications of 2R and 3R regeneration, wavelength conversion, parametric amplification, phase conjugation, delay, performance monitoring, and switching are reviewed.

1.3.21 Chapter 21. Fiber-optic quantum information technologies (Prem Kumar, Jun Chen, Paul L. Voss, Xiaoying Li, Kim Fook Lee, and Jay E. Sharping) Quantum-mechanical (QM) rules are surprisingly simple: linear algebra and firstorder partial differential equations. Yet, QM predictions are unimaginably precise and accurate when compared with experimental data. A “mysterious” feature of QM is the superposition principle and the ensuing quantum entanglement. The fundamental difference between quantum entanglement and classical correlation lies in the fact that particles are quantum-mechanical objects which can exist not only in states | 0 > and | 1> but also in states described by  | 0 > þ b | 1 >, while classical objects

12

Ivan P. Kaminow et al.

can only exist in one of two deterministic states (i.e., “heads” or “tails”), and not something in between. In other words, the individual particle in quantum entanglement does not have a well-defined pure state before measurement. Since the beginning of the 1990s, the field of quantum information and communication has expanded rapidly, with quantum entanglement being a critical aspect. Entanglement is still an unresolved “mystery,” but a new world of “quantum ideas” has been ignited and is actively being pursued. The focus of this chapter is the generation of correlated and entangled photons in the telecom band using the Kerr nonlinearity in dispersion-shifted fiber. Of particular interest are microstructure fibers, in which tailorable dispersion properties have allowed phase-matching and entanglement to be obtained over a wide range of wavelengths.

1.4 OFT V VOLUME B: SYSTEMS AND NETWORKS 1.4.1 Chapter 1. Overview of OFT V volumes A & B (Ivan P. Kaminow, Tingye Li, and Alan E. Willner) This chapter briefly reviews herewith all the chapters contained in both volumes of OFT V.

1.4.2 Chapter 2. Advanced optical modulation formats (Peter J. Winzer and Rene´-Jean Essiambre) Today, digital radio-frequency (rf) communication equipment employs sophisticated signal processing and communication theory technology to realize amazing performance; wireless telephones are a prime example. These implementations are made possible by the capabilities and low cost of silicon integrated circuits in highvolume consumer applications. Some of these techniques, such as forward error correction (FEC) and electronic dispersion compensation (EDC) are currently in use in lightwave communications to enhance signal-to-noise ratio and mitigate signal degradation. (See the chapter on “Electronic Signal Processing for Dispersion Compensation and Error Mitigation in Optical Transmission Networks” by Abhijit Shanbhag, Qian Yu, and John Choma.) Advanced modulation formats that are robust to transmission impairments or able to improve spectral efficiency are being considered for next-generation lightwave systems. This chapter provides a taxonomy of optical modulation formats, along with experimental techniques for realizing them. The discussion makes clear the substantial distinctions between design conditions for optical and rf applications. Demodulation concepts for coherent and delay demodulation are also covered analytically.

1. Overview of OFT V Volumes A & B

13

1.4.3 Chapter 3. Coherent optical communication systems (Kazuro Kikuchi) The first generation of single-channel fiber optic networks used on-off keying and direct detection. Later, coherent systems, employing homodyne and heterodyne detection, were intensely researched with the aim of taking advantage of their improved sensitivity and WDM frequency selectivity. However, the quick success of EDFAs in the 1990s cut short the prospects for coherent systems. Now, interest in coherent is being renewed as the need for greater spectral efficiency in achieving greater bandwidth per fiber has become apparent. This chapter reviews the theory of multilevel modulation formats that permit multiple bits/s of data per Hz of bandwidth. (See the chapter on “Advanced Modulation Formats” by Winzer and Essiambre.) The growing capabilities of silicon data signal processing (DSP) can be combined with digital coherent detection to provide dramatic improvements in spectral efficiency. Experimental results for such receivers are presented.

1.4.4 Chapter 4. Self-coherent optical transport systems (Xiang Liu, Sethumadhavan Chandrasekhar, and Andreas Leven) As stated above, coherent detection transmission systems were investigated in the 1980s for their improved receiver sensitivity and selectivity, and for the promise of possible postdetection dispersion compensation. However, the emergence of EDFAs and amplified WDM systems relegated the technically difficult coherent technology to the background. Now, as high-speed signal processing technology becomes technically and economically feasible, there is renewed interest in studying coherent and selfcoherent systems, especially for their capability to increase spectral efficiency through the use of advanced multilevel modulation techniques and, more important, for the possibility of implementing postdetection equalization functionalities. Self-coherent systems utilize differential direct detection that does not require a local oscillator. With high-speed analog-to-digital conversion and digital signal processing, both phase and amplitude of the received optical field can be reconstructed, thus offering unprecedented capability for implementing adaptive equalization of transmission impairments. This chapter is a comprehensive and in-depth treatment of self-coherent transmission systems, including theoretical considerations, receiver technologies, modulation formats, adaptive equalization techniques, and applications for capacity upgrades and cost reduction in future optical networks.

1.4.5 Chapter 5. High-bit-rate ETDM transmission systems (Karsten Schuh and Eugen Lach) Historically, it has been observed that the first cost of a (single-channel) transmission system tended to increase as the square root of its bandwidth or bit rate. This

14

Ivan P. Kaminow et al.

observation has prompted the telecom industry to develop higher-speed systems for upgrading transport capacity. Indeed, there is a relentless drive to explore higher speed for multichannel amplified WDM transmission where, for a given speed of operation, the total system cost is roughly proportional to the number of channels plus a fixed cost. It is important to note that the cost of equalizing for signal impairment at higher speeds must be taken into account. This chapter is an up-to-date review of high-speed transmission using electronic time division multiplexing (ETDM), a time-honored approach for upgrading system capacity. The emphasis is on 100-Gb/s bit rate and beyond, as 40-Gb/s systems are already being deployed and 100-Gb/s Ethernet (100 GE) is expected to be the next dominant transport technology. The chapter includes a basic treatment of ETDM technology, followed by a description of the concepts of high-speed ETDM systems. Requirements of optical and electronic components and the stateof-the-art technologies are then examined in detail, and an up-to-date overview of ultra-high-speed systems experiments is presented. Finally, prospects of the various approaches for rendering cost-effective 100 GE are contemplated.

1.4.6 Chapter 6. Ultra-high-speed OTDM transmission technology (Hans-Georg Weber and Reinhold Ludwig) The expected increase of transmission capacity in optical fiber networks will involve an optimized combination of WDM and TDM. TDM may be realized by electrical multiplexing (ETDM) or by optical multiplexing (OTDM). Dispersion impairment notwithstanding, OTDM offers a means to increase the single-channel bit rate beyond the capability of ETDM. Thus OTDM transmission technology is often considered to be a research means with which to investigate the feasibility of ultra-high-speed transmission. Historically, the highest speed commercial systems have been ETDM systems. Latest examples are 40 G systems being deployed at present and (serial) 100 G systems expected to be commercially available in a few years. In the past 10 years, OTDM transmission technology has made considerable progress towards much higher bit rates and much longer transmission links. This chapter discusses ultra-high-speed data transmission in optical fibers based on OTDM technology. The chapter gives a general description of an OTDM system, the OTDM transmitter, the OTDM receiver, and the fiber transmission line. WDM/OTDM transmission experiments are also described.

1.4.7 Chapter 7. Optical performance monitoring (Alan E. Willner, Zhongqi Pan, and Changyuan Yu) Today’s optical networks function in a fairly static fashion and are built to operate within well-defined specifications. This scenario is quite challenging for

1. Overview of OFT V Volumes A & B

15

higher-capacity systems, since network paths are not static and channel-degrading effects can change with temperature, component drift, aging and fiber plant maintenance. In order to enable robust and cost-effective automated operation, the network should be able to: (i) intelligently monitor the physical state of the network as well as the quality of propagating data signals, (ii) automatically diagnose and repair the network, and (iii) redirect traffic. To achieve this, optical performance monitoring should isolate the specific cause of the problem. Furthermore, it can be quite advantageous to determine when a data signal is beginning to degrade, so that the network can take action to correct the problem or to route the traffic around the degraded area. This chapter explores optical performance monitoring and its potential for enabling higher stability, reconfigurability, and flexibility in an optical network. Moreover, this chapter describes the specific parameters that a network might want to monitor, such as chromatic dispersion, polarization-mode dispersion, and optical SNR. Promising monitoring techniques are reviewed.

1.4.8 Chapter 8. ROADMs and their system applications (Mark D. Feuer, Daniel C. Kilper, and Sheryl L. Woodward) As service providers begin to offer IPTV services in addition to data and voice, the need for fast and flexible provisioning of mixed services and for meeting unpredictable traffic demand becomes compelling. Reconfigurable optical add/drop multiplexers (ROADMs) have emerged as the network element that can satisfy this need. Indeed, subsystem and system vendors are rapidly developing and producing ROADMs, and carriers are installing and deploying them in their networks. This chapter is a comprehensive treatment of ROADMs and their application in WDM transmission systems and networks, comprising a review of various ROADM technologies and architectures; analyses of their routing functionalities and economic advantages; considerations of design features and other requirements; and discussions of the design of ROADM transmission systems and the interplay between the ROADM and transmission performance. The chapter ends with some thoughts on the remaining challenges to enable ROADMs to achieve their potential.

1.4.9 Chapter 9. Optical Ethernet: Protocols, management, and 1–100 G technologies (Cedric F. Lam and Winston I. Way) As the Internet becomes the de facto platform for the delivery of voice, data, and video services, Ethernet has become the technology of choice for access and metro

16

Ivan P. Kaminow et al.

networks, and for next-generation long-haul networks. As stated concisely by the authors, “The success of Ethernet is attributed to its simplicity, low cost, standard implementation, and interoperability guarantee,” attributes that helped the “networking community it serves to prosper, hence producing the economy of scale.” This chapter is an in-depth review of the evolution and development of Ethernet technology for application in optical fiber telecommunications networks. Topics covered include: point-to-point Ethernet development, Layer-2 functions, Carrier Ethernet, Ethernet in access PONs, Ethernet OAM (Operation, Administration, Maintenance), development of 10 GE for PON and 100 GE for core applications, and examples of high-speed Ethernet.

1.4.10 Chapter 10. Fiber-based broadband access technology and deployment (Richard E. Wagner) One of the earliest long-haul commercial optical fiber telecom systems was the AT&T Northeast Corridor link from Boston to New York to Washington in 1983. In this application, the large capital investment could be amortized among many users. The prospect of economically bringing fiber all the way to a large number of end users, where cost sharing is not available, has continuously appealed to and challenged the telecom industry. Presently, technology advances and volume manufacture are reducing costs/user, fabulous broadband applications are luring subscribers, and government legislation and subsidies are encouraging growth worldwide. This chapter tracks the history of broadband access, compares the competing access technologies, and projects the roadmap to future deployment in the US, Asia, Europe and the rest of the world. The economic driver for widespread deployment is the explosive growth of Internet traffic, which doubles annually in developed countries and grows even faster in developing countries, such as China. In developed countries, growth is due to new broadband applications; in developing countries, both new users and new applications drive traffic growth. This chapter focuses on the fiber-based approaches to broadband access worldwide, including some of the drivers for deployment, the architectural options, the capital and operational costs, the technological advances, and the future potential of these systems. Three variants of fiber-based broadband access, collectively called FTTx in this chapter, have emerged as particularly important. They are: hybrid-fiber-coax (HFC) systems, fiber-to-the-cabinet (FTTC) systems, and fiber-to-the-home (FTTH) systems.

1.4.11 Chapter 11. Global landscape in broadband: Politics, economics, and applications (Richard Mack) The technology of choice that predominates in a specific telecom arena depends ultimately upon the competitive economics: the normalized capital and

1. Overview of OFT V Volumes A & B

17

operational costs in dollars per unit bandwidth (per unit distance). For metro, regional, and long-haul arenas, lightwave technology is indisputably the king. However, in the access arena, the competitive unit cost of lightwave technology has not favored rational deployment. Indeed, the history of FTTH has followed a tortuous path; the early trials in the 1980s and 1990s did not lead to massive deployment. Globally, Japanese and Korean telecom companies have been leading the installation of FTTH (with as yet unknown economic consequences). In the meantime, the cost of FTTH equipment has been decreasing steadily. Recently, relief from “unbundling” (exemption from requirement for incumbent carriers to share facilities with competitive carriers, as ruled by FCC) and competition from cable TV companies have prompted incumbent carriers in the US to install FTTH with competitive (triple-play) service offerings. As the demand for broadband services grows and revenue improves, the return from the vast investment in FTTH may be realized in the not-to-distant future. This chapter is a fascinating, data-laden account of the history of deployment of optical fiber telecommunications, with emphasis on economics, growth landscape, and broadband services in the access arena. The discussion includes historical highlights, demographics, costs and revenues, fiber installations, services scenarios, competition and growth, regulatory policies, applications and bandwidth requirements, technology and network architecture choices, market scenarios, etc. The interplay of these issues is discussed and summarized in the concluding section.

1.4.12 Chapter 12. Metro networks: Services and technologies (Loukas Paraschis, Ori Gerstel, and Michael Y. Frankel) Metropolitan networks operate in the environs of a major city, transporting and managing traffic to meet the diverse service needs of enterprise and residential applications. Typically, metro networks have a reach below a few hundred kilometers with node traffic capacities and traffic granularity that require amplified dense WDM technology with optical add/drop, although the more economical coarse WDM technology has also been deployed. At present, convergence of IP services and traditional time-division multiplexed (TDM) traffic with low operational cost is an important issue. This chapter reviews the architecture and optical transport of metro networks, which have evolved to meet the demand of various applications and services, including discussions of the evolution of network architecture, physical building blocks of the WDM network layer, requirements of network automation, and convergence of packetized IP with traditional TDM traffic, and ending with a brief perspective on the future outlook.

18

Ivan P. Kaminow et al.

1.4.13 Chapter 13. Commercial optical networks, overlay networks, and services (Robert Doverspike and Peter Magill) As service providers are the ultimate users of novel technologies and systems in their networks, it is important that the innovators have a sound understanding of the structure and workings of the carriers’ networks: the architecture and layers, traffic and capacity demands, management of reliability and services, etc. Commercial networks are continuously upgraded to provide more capacity, new services and reduced capital and operational cost; seamless network evolution is essential for obvious economic reasons. Even when “disruptive” technologies and platforms are introduced, smooth integration within the existing infrastructure is imperative. This chapter reviews the important aspects of current commercial optical networks in all three segments or layers: access, metro, and core. Topics include 1. relationship of services to layers, covering service requirements, layer technology, quality of service, Service Level Agreements, network availability, and network restoration; and 2. network and services evolution, covering demand and capacity, and applications and technologies in all three segments of the network. In the summary section, the authors point out that while much of the “industry focuses on advanced optical technologies for the long-distance network, most of the investment and opportunity for growth resides in the metro/ access [sector].”

1.4.14 Chapter 14. Technologies for global telecommunications using undersea cables (Se´bastien Bigo) The introduction of WDM has enabled a tremendous capacity growth in undersea systems, both by the increase in the number of carrier wavelengths and by the increase in the channel bit rate. Starting from 2.5 Gbit/s in the mid-1990s, the bit rate was upgraded in commercial products to 10 Gbit/s at the end of the last century. The next generation of undersea systems will likely be based on 40-Gbit/s bit-rate channels. However, transmission at 40 Gbit/s is significantly more challenging than at 10 Gbit/s. This chapter gives an overview of the specificities of submarine links with respect to terrestrial links and provides a few examples of recently deployed undersea systems. Moreover, it describes the key technologies involved in undersea systems, explaining particular implementations of early optical systems, today’s 10-Gbit/s systems, and future 40-Gbit/s links. The key technologies are related to fiber selection and arrangement, to amplifier design, to modulation formats, to detection techniques, and to advanced impairment-mitigation solutions.

1. Overview of OFT V Volumes A & B

19

1.4.15 Chapter 15. Future optical networks (Michael O’Mahony) In the past few years Internet traffic, doubling annually, has dominated the network capacity demand, which has been met by the advances in lightwave communications. The transformation from a circuit-switched, voice-centric to a packet-switched, IP-centric network is well underway; amplified WDM transmission systems with terabits-per-second capacity are being deployed; rapid reconfigurable networking and automatic service provisioning are being implemented. The drive to reduce cost and increase revenue has being inexorable. In the meantime, carriers are installing FTTH and offering IPTV services, which will undoubtedly change network traffic characteristics and boost the traffic growth rate. What will the future networks look like? This chapter reviews the growth of the data traffic and evolution of the optical network, including user communities, global regional activities, and service requirements. Discussions cover the diversity of architectures, the evolution of switching, cross-connecting and routing technologies, and the transformation to carrier-grade (100 G) Ethernet. The author notes that device integration “will enable the realization of many of the key functionalities for optical networking.”

1.4.16 Chapter 16. Optical burst and packet switching (S. J. Ben Yoo) Optical switching has the potential of providing more-efficient and higher-throughput networking than its electronic counterpart. This chapter discusses optical burst and packet switching technologies and examines their roles in future optical networks. It covers the roles of optical circuit, burst, and packet switching systems in optical networks, as well as their respective benefits and trade-offs. A description is given of the networking architecture/protocols, systems, and technologies pertaining to optical burst and packet switching. Furthermore, this chapter introduces opticallabel switching technology, which provides a unified platform for interoperating optical circuit, burst, and packet switching techniques. By exploiting contention resolution in wavelength, time, and space domains, the optical-label switching routers can achieve high-throughput without resorting to a store-and-forward method associated with large buffer requirements. Testbed demonstrations in support of multimedia and data communications applications are reviewed.

1.4.17 Chapter 17. Optical and electronic technologies for packet switching (Rodney S. Tucker) The Internet provides most of the traffic and growth in lightwave systems today. Unlike the circuit-switched telephone network, the Internet is based on packet

20

Ivan P. Kaminow et al.

switching, which provides statistical multiplexing of the many data streams that pass through a given switch node. A key element of any packet switch is the ability to buffer (or store) packets temporarily to avoid collisions. Since it is easy to store bits as electronic charge in silicon, commercial packet switches are electronic. This chapter introduces the basics of practical packet routers, indicating the requirements of a switch based on either electronics or photonics. The author compares the physical limits on routers based on storing and switching electrons vs photons. While many in the optics field would like to see photonic switches dominate, it appears that the physical inability to provide a large optical random access buffer means that packet switches will continue to be optoelectronic rather than all-optical for some time.

1.4.18 Chapter 18. Microwave-over-fiber systems (Alwyn J. Seeds) The low-loss, wide-bandwidth capability of optical transmission systems makes them attractive for the transmission and processing of microwave signals, while the development of high-capacity optical communication systems has required the import of microwave techniques in optical transmitters and receivers. These two strands have led to the development of the research area of microwave photonics. Following a summary of the historic development of the field and the development of microwave photonic devices, systems applications in telecommunications, and likely areas for future development are discussed. Among the applications reviewed are wireless-over-fiber access systems, broadband signal distribution and access systems, and communications antenna remoting systems.

1.4.19 Chapter 19. Optical interconnection networks in advanced computing systems (Keren Bergman) High-performance computers today use multicore architectures involving multiple parallel processors interconnected to enhance the ultimate computational power. One of the current supercomputers contains over one hundred thousand individual PowerPC nodes. Thus the challenges of computer design have shifted from “computation-bound” to “communication-bound.” Photonic technology offers the promise of drastically extending the communications bound by using the concepts and technologies of lightwave communications for interconnecting and networking the individual multi-processor chips and memory elements. Silicon photonics (see the chapter on “Silicon Photonics” by C. Gunn and T. L. Koch) looms as a possible subsystem technology having the cost-effectiveness of silicon CMOS manufacturing process.

1. Overview of OFT V Volumes A & B

21

This chapter is a review of the subject of interconnection networks for highperformance computers. Performance issues including latency, bandwidth and power consumption are first presented, followed by discussions of design considerations including technology, topology, packet switching nodes, message structure and formation, performance analysis, and evaluation. Network design implementation, architectures, and system demonstrations are then covered and thoughts are offered on future directions, including optical interconnection networks on a chip.

1.4.20 Chapter 20. Simulation tools for devices, systems, and networks (Robert Scarmozzino) The ability of optical fiber telecommunications to satisfy the enormous demand for network capacity comes from thorough understanding of the physics and other disciplines underlying the technology, as well as the abilities to recognize sources for limitation, develop ideas for solution, and predict, test, and demonstrate those ideas. The field of numerical modeling has already been an important facilitator in this process, and its influence is expected to increase further as photonics matures. This chapter discusses the broad scope of numerical modeling and specifically describes three overarching topics: (i) active and passive device/componentlevel modeling with emphasis on physical behavior, (ii) transmission-systemlevel modeling to evaluate data integrity, and (iii) network-level modeling for evaluating capacity planning and network protocols. The chapter offers an overview of selected numerical algorithms available to simulate photonic devices, communication systems, and networks. For each method, the mathematical formulation is presented along with application examples.

ACKNOWLEDGMENTS We wish sincerely to thank Tim Pitts and Melanie Benson of Elsevier for their gracious and invaluable support throughout the publishing process. We are also deeply grateful to all the authors for their laudable efforts in submitting their scholarly works of distinction. Finally, we owe a debt of appreciation to the many people whose insightful suggestions were of great assistance. We hope our readers learn and enjoy from all the exciting chapters.

This page intentionally left blank

2 Advanced optical modulation formats Peter J. Winzer and Rene´-Jean Essiambre Bell Laboratories, Alcatel-Lucent, Holmdel, NJ, USA

2.1 INTRODUCTION Since the publication of the fourth volume of the book series Optical Fiber Telecommunications in 2002, high-speed optical transmission systems have increasingly been building on techniques that are well established in radiofrequency (RF) communication systems, and in particular in wireless communication. Examples are advanced modulation formats, line coding, enhanced forward error correction (FEC), and digital signal processing at transmitter and receiver. The adoption and extension of these techniques from the Mb/s regime into the multi-Gb/s realm is driven by the desire to steadily lower the cost per end-to-end networked information bit in an environment of continuously increasing data traffic. This is being done by extending the regeneration-free reach of optical line systems at the highest possible per-fiber transmission capacities, while at the same time allowing for optical wavelength routing in optically transparent mesh networks. In this context, communication engineering techniques are now supplementing established methods from optical physics to increase transmission reach, system capacity, and network flexibility. Today, research and commercial implementation of digital optical communication techniques fall into two main areas, distinguished by the per-wavelength symbol rates they operate at: • At per-channel symbol rates of 10 Gbaud, electronic signal processing ranging from simple feed-forward and decision-feedback equalizers (FFE, DFE) at the receiver all the way to maximum-likelihood sequence estimation (MLSE) is commercially available today [1–3], and FEC is found in most commercial 10-Gb/s transport products [4, 5]. Digital signal predistortion at the transmitter is being actively pursued [6], and coherent detection is Optical Fiber Telecommunications V B: Systems and Networks Copyright  2008, Elsevier Inc. All rights reserved. ISBN: 978-0-12-374172-1

23

Peter J. Winzer and Rene´-Jean Essiambre

24

experiencing renewed1 interest [14–16]. In contrast to widely deployed direct detection, coherent detection allows electronic signal processing to make use of the full optical field (magnitude and phase), which can be exploited for more efficient mitigation of fiber transmission impairments such as chromatic dispersion (CD) or polarization mode dispersion (PMD). Coherent detection research is also aiming at increasing the bit rate through multilevel modulation and polarization multiplexing while keeping the signal bandwidth similar to that of a 10-Gb/s binary signal to reuse 10-Gb/s optical line infrastructure and to increase the system’s spectral efficiency (SE). In general, narrow signal bandwidths are desired in high-SE optically routed networks to minimize excessive filtering penalties due to the concatenation of reconfigurable optical add/drop multiplexers (ROADMs) and to avoid excessive cross talk between adjacent wavelength-division multiplexed (WDM) channels. • At per-channel symbol rates of 40 Gbaud, and up to 100 Gbaud, the capabilities of electronic equalization are still limited to low complexity electronic [17–21] and optical [22–24] FFE structures. In this area, modulation formats and line coding are at the center of interest. They are used to mitigate linear and nonlinear impairments of fiber-optic transmission, as well as to achieve high SEs in optically routed networks. At 40 Gb/s, FEC is a standard feature of commercially deployed optical transport systems [25]. Despite the many benefits brought by electronic signal processing and digital communication techniques, it should be noted that an optical fiber network differs substantially from an RF communication link, both fundamentally and technologically. Some key differences relevant to the discussions in this chapter are summarized in Table 2.1.

Fiber Nonlinearities The most important fundamental difference between optical and RF communications is the presence of nonlinear distortions in the optical fiber communication channel. The high transversal confinement of the optical signal field in a fiber core with effective areas between 20 and 110 mm2 causes light intensities to reach or exceed a megawatt/cm2. At such high optical intensities, the fiber’s index of refraction is affected by the presence of optical signals through the optical Kerr effect [26], and signal-induced refractive index changes translate into changes of the signals’ optical phases. Over optically amplified distances of many hundred or even several thousand

1

Coherent detection using advanced optical modulation formats was widely discussed in the context of unamplified lightwave systems in the 1980s [7–12], where attenuation-limited single-span transmission required utmost receiver sensitivities. With the advent of efficient optical amplifiers, allowing for comparable direct-detection receiver sensitivities [9, 13], coherent systems research decayed in the early 1990s.

2. Advanced Optical Modulation Formats

25

kilometers, these phase rotations, in conjunction with fiber dispersion, result in a host of different waveform distortions that increase with signal power. As a consequence, and in stark contrast to classical RF systems, the performance of an optical communication link exhibits a maximum at a certain signal power level, which represents the optimum trade-off between optical amplifier noise (amplified spontaneous emission, ASE) and fiber Kerr nonlinearity. As an example, Figure 2.1(a) [27] shows the Table 2.1 Optical vs radio-frequency communication systems. Optical communications

RF communications

Quantum noise Multipath, WDM crosstalk Fiber nonlinearity þ dispersion kHz (const. for107 bits)

Thermal noise Multipath, multiuser Linear channel, fading kHz (const. for 101–103 bits)

Technological Electrical bandwidth

60 GHz – limited by technology (bandwidth/carrier104 to 106)

Detection

Predominantly square-law

Limited by spectral regulations (bandwidth/carrier102 to 104) Predominantly coherent (I/Q) demodulation

Digital processing

Fairly simple equalization and FEC

Extensively used, sophisticated

Processing power per information bit

Fairly low

Very high

–4

–2

0

2

Signal launch power (dBm) (a)

4

n no

ise

No

ity

–6 –7 –8 –9 –4

ar ine onl

–5

an

–3

Maximum spectral efficiency er n

Capacity C (bits/s)

33% RZ-OOK 67% CSRZ 33% RZ-AMI 33% π/2-AP-RZ 33% RZ-DPSK, single-ended det. 33% RZ-DPSK, balanced det.

Fib

Bit error ratio (BER)

–2

Sh

Fundamental Noise Interference Channel Channel dynamics

SNR (dB) Signal launch power (dBm) (b)

Figure 2.1 (a) Measured BER vs signal launch power for various optical modulation formats transmitted over 1980 km of SSMF [27]; (b) Schematic representation of fiber channel capacity vs received SNR. Both cases exhibit maximum performance at a specific signal power level, representing the optimum trade-off between optical amplifier noise and fiber Kerr nonlinearity (this figure may be seen in color on the included CD-ROM).

26

Peter J. Winzer and Rene´-Jean Essiambre

measured bit error ratio (BER) as a function of launched signal power for various optical modulation formats transmitted over 1980 km of standard single-mode fiber (SSMF). The figure reveals the trade-off between ASE-dominated performance at low signal launch powers and dominating fiber nonlinearity at high powers, indicating different performance optima at format-specific optimum launch powers. Figure 2.1(b) shows the schematic behavior of fiber channel capacity as a function of the received signal-to-noise ratio (SNR) [28]. Due to quantum-mechanical lower bounds on optical amplifier noise, the SNR can only be increased by launching higher signal power, which eventually leads to nonlinear signal distortions. Note that when operated at optimum performance, every fiber-optic communication system shows signs of nonlinear signal distortions.

Electronic Bandwidth Limitations The second major difference between optical and RF communication systems, with both fundamental and technological consequences, is the high absolute bandwidth of optical communication signals. In general, and as evidenced by the historic evolution of lightwave systems [29], the highest possible per-channel bit rates have always yielded the lowest cost, footprint, and power consumption per end-to-end networked information bit, once the underlying technologies were sufficiently mature. Therefore, optical communication systems have always been pushing the limits of high-speed electronic and optoelectronic components, with 100-Gb/s binary transmission systems representing the current limit of electronic multiplexing and demultiplexing capabilities (see Refs [24, 30] and Chapter 5 of this book). The sheer per-channel bandwidth has left little room for the implementation of sophisticated signal processing algorithms, including coherent detection with digital phase locking (“intradyne” detection [14, 15, 31], cf. Sec. 2.4.1 and Chapter 3 of this book). The last 5 years, however, have seen a period of stagnation due to the burst of the telecom bubble, which has allowed electronic processing to catch up with 10-Gb/s line rates, still the most dominantly deployed per-channel bit rate. It remains to be seen whether future systems will be based on 10-Gbaud technology and its evolution to higher rates, with substantial digital processing complexity, or whether systems will continue to use the highest possible symbol rates in conjunction with direct detection and advanced optical technologies such as adaptive dispersion management or optical phase conjugation but with modest electronic processing. In any case, optical communication technologies will always be facing the limits of high-speed signal processing and modulation, which is an important factor to take into account when discussing advanced optical modulation formats. In this chapter, we review, on an advanced level, important communication engineering techniques that have been discussed in the context of optical

2. Advanced Optical Modulation Formats

27

networking over the past 5 years and that are likely to be the subject of further studies over the next 5 years. We also want to point the reader to additional reviews on the topic [27, 32–36]. On a more introductory level, optical modulation formats are discussed in, e.g., [37–39]. The communication engineering background is covered in textbooks such as [40–42], and the background in optical physics can be found in, e.g., [26, 43, 44]. A list of acronyms used in this chapter can be found on pages 80-81.

2.2 COMMUNICATION ENGINEERING AND OPTICAL COMMUNICATIONS The broad field of communication engineering has developed a variety of techniques that can be used to beneficially impact digital data transmission and to approach the capacity [45, 28] of a given communication channel. Figure 2.2 visualizes those methods and ingredients that are relevant to the understanding of advanced optical communication systems, and in particular to the understanding of Modulation Symbol constellation Binary, Multi-level Orthogonal or linearly dependent

Signal space

Line coding Without data correlation Carrier-suppressed return-to-zero (CSRZ)

With data correlation

Intensity/Phase (ASK, OOK, PSK) Frequency (FSK) Polarization (PolSK) Time (PPM)

Modulation waveform [Non] return-to-zero ([N]RZ) Chirped modulation (C-NRZ, CRZ) Vestigial sideband (VSB)

Multiplexing

Duobinary (DB, PSBT) Alternate mark inversion (AMI)

Polarization Polarization multiplexing Polarization interleaving

Wavelength (Frequency)

Error correcting coding

Wavelength-division multiplexing (WDM) Coherent WDM (CoWDM) Orthogonal frequency division multiplexing (OFDM)

Forward error correction Hard-decision decoding Soft-decision decoding

Code Optical code division multiple access (OCDMA)

Coded modulation Trellis-coded modulation (TCM)

Transmission line Nonlinearity management Dispersion map Fiber type

Detection and equalization Receive equalization Feed forward equalization (FFE) Decision feedback equalization (DFE) Optical equalization (OEQ)

Transmit pre-distortion Symbol-by-symbol detection Threshold detection

Multi-symbol detection Maximum-likelihood sequence estimation (MLSE)

Multi-symbol phase estimation

Network aspects ROADM support Optical routing

Demodulation Direct detection With/without phase-to-intensity conversion

Coherent detection Heterodyne detection Homodyne detection Intradyne detection (Multi-symbol phase estimation)

Figure 2.2 Overview of communication engineering techniques discussed in optical networks research today.

Peter J. Winzer and Rene´-Jean Essiambre

28

advanced modulation formats in the broader context of optical transport networks. The aspect of modulation comprises the choice of symbol constellations and signal spaces used to encode information onto an optical carrier. Of particular importance to the optical communication channel is also the appropriate choice of modulation waveforms. In addition to choosing a modulation scheme, redundancy may be introduced into the symbol sequence. This may either be done by line coding, which shapes the spectrum or avoids certain adverse bit patterns to beneficially impact a format’s transmission performance, or by error correcting coding, which uses redundancy to correct symbol errors at the receiver. In either case, redundancy can be added in the time domain by increasing the signal’s symbol rate (typically by 7% in FEC schemes employed in standardized terrestrial optical communications [46]) or in signal space by adding more points or even more dimensions to the symbol constellation. If used for error correcting purposes, the latter strategy is generally referred to as coded modulation. To exploit the vast bandwidth provided by a single optical fiber, multiplexing of many channels is typically performed, with wavelength-division multiplexing (WDM) being the most prominent technique. At the receiver, the optical signal is demultiplexed, and the channel of interest is demodulated from optical carrier frequencies to baseband. The majority of optical communication systems uses demodulation by direct (i.e., square-law) photodetection. Recently, coherent demodulation has also received renewed interest, fueled by advances in high-speed digital signal processing hardware. Along the same lines, signal processing at both transmitter and receiver as well as advanced detection techniques have become important fields of research and commercialization. All these communication engineering techniques, however, have to be seen in the bigger picture of established optical line system design, such as dispersion management, which effectively solves problems particular to the optical communication channel, most notably mitigating impairments brought by fiber nonlinearity. In this chapter, we will discuss several of the techniques shown in Figure 2.2. We will start by looking at modulation waveforms and their properties, which form the basis for classifying modulation formats as well as multiplexing techniques.

2.2.1 Signal Spaces for Modulation and Multiplexing Any digital modulation format is composed of a discrete set {x1(t), x2(t), . . . , xM(t)} of analog modulation waveforms that represent an alphabet of M ‡ 2 communication symbols.2 If M = 2 we speak of binary modulation, which is the prevailing

2

This general concept is very similar to written text, which is based on an alphabet of letters (e.g., 26 discrete letters in the standard Roman alphabet). To convey information, each letter is represented by an analog “waveform”. For example, the letter “A” can be represented by the “waveforms” A, A, A, a, a, or even by ‘ –’ in Morse code. As long as transmitter and receiver use the same alphabet and are able to correctly read the letters’ analog waveform representations, communication may take place.

2. Advanced Optical Modulation Formats

29

alphabet size used in commercially deployed optical communication systems. To convey digital information, the transmitter picks one symbol per symbol interval of duration TS and transmits a string of symbols at the symbol rate RS = 1/TS. The symbol rate is measured in Baud, where 1 Baud = 1 symbol/s. If all symbols in the alphabet occur with equal probability and are all used to carry information, each symbol conveys log2 M bits of information, and bit rate RB and symbol rate RS are related by RB = RS log2 M:

(2.1)

For the analysis and classification of modulation formats and multiplexing schemes, it is convenient to express the set of modulation waveforms as “vectors” in a linear vector space. In analogy to the geometric vector space, where a vector is represented as a two- or three-dimensional “arrow”, a signal may be thought of as an object in an N-dimensional space of continuous functions, called the signal space. The “inner product” between two (generally complex) signals is defined as [40] hxi ðtÞjxk ðtÞi =

Z1

1

xi ðtÞxk ðtÞdt:

(2.2)

In the context of optical communication signals, the modulation waveforms xi;k (t) rÞ, where ~ r denotes the are physically represented by optical field vectors ~ ei;k ðt;~ transverse coordinate, and the definition of the inner product can be extended to

rÞj~ ek ðt;~ rÞi = ei ðt;~ h~

Z1

1

dt

Z

rÞ; d~ r~ ei ðt;~ rÞ ~ ekðt;~

(2.3)

A

where “” is the inner product of two vectors in the geometric sense, and A denotes the transverse area covered by the two optical fields. This definition extends Eqn (2.2) into the modal ð~ rÞ and polarization ð~ eÞ domain. The “length” (or “norm”) of a signal vector equals the signal’s energy ek ðt;~ rÞj~ ek ðt;~ rÞi = Ek : ek ðtÞk = h~ k~

(2.4)

Importantly, two signals are orthogonal if their inner product is zero,3 ~ ek ðtÞ , h~ ei ðt;~ rÞj~ ek ðt;~ rÞi = 0: ei ðtÞ ?~

(2.5)

3 Note that for two real-valued bandpass signals (such as the optical field jjejj2 ) to be orthogonal, it suffices that the real part of the inner product of the two fields’ complex envelopes is zero [40].

Peter J. Winzer and Rene´-Jean Essiambre

30

A digital communication receiver generally projects the received waveform into the signal space used for modulation by evaluating the inner product of the received waveform and the set of known modulation waveforms. By acknowledging the fact that the inner product, Eqn (2.2), can also be written as a convolution (*) sampled at t = 0 [40], hxi ðtÞjxk ðtÞi =

Z1

1

  xi ðÞxk ð½t  Þd t = 0 = xi ðtÞ  xk ðtÞt = 0 ;

(2.6)

we see that this projection can be performed by a matched filter (or a bank of matched filters working in parallel for the case of multiple orthogonal modulation waveforms), followed by a sampling device [40–42]. The impulse response h(t) of the respective matched filter is given by the complex conjugate of the temporally reversed modulation waveform, h(t) = xk (t). Figure 2.3 shows a matched filter front-end for a three-dimensional signal space, together with its graphical representation. The input signal waveform is y(t) and the three orthogonal modulation waveforms are x1,2,3(t). A matched filter bank can be shown to be the theoretically optimum receiver frontend for a wide class of communication scenarios [40], including the detection of optical signals impaired by additive white Gaussian noise, properly modeling amplified spontaneous emission (ASE) of optical amplifiers. Note that in square-law (direct detection) receivers the optical filter (OFs) need to be matched to the modulation waveforms [10, 47]. In general, as many matched filters are needed as there are dimensions of the signal space underlying the modulation format. For example, binary optical intensity modulation (IM), also referred to as on/off keying (OOK), requires a single real-valued filter; quadrature modulation, i.e., separate modulation of the complex optical field’s real and imaginary parts, requires two real-valued (or, equivalently, one complex-valued) filter; polarization shift keying (PolSK) asks for a polarization beam splitter that separates the two orthogonal polarization dimensions. By performing matched filtering, any noise components outside the signal space defined by the matched filter bank are filtered out and have no impact on detection. For example, by introducing a polarization filter, which is the filter that is matched to the signal’s

y (t )

x1* (–t )

y (t )|x 1(t )

x2* (–t )

y (t )|x 2(t )

x3* (–t )

Sample at t = 0

y (t )|x 3(t )

3

y (t )|x 3(t ) y (t )|x 2(t ) y (t )|x 1(t )

2

y (t )

1

Figure 2.3 Matched filter front-end and its graphical representation for a three-dimensional signal space. The input signal waveform is y(t) and the three orthogonal modulation symbols have waveforms x1,2,3(t).

2. Advanced Optical Modulation Formats

31

polarization, the optical noise in the orthogonal polarization is filtered out, thereby improving detection performance. However, due to many practical reasons faced in high-speed optical communication systems, including the effects of frequency offsets, residual CD nonlinear signal distortions, the impact of non-negligible postdetection electronic filtering, the presence of intersymbol interference (ISI) together with the lack of sufficiently powerful equalization or sequence detection devices, or the sheer difficulties associated with the design of commercially attractive matched optical filters, matched filtering is generally not used in optical communications. One rather optimizes manufacturable optical filters for best overall receiver performance in the presence of all practical limitations and shortcomings [48–50]. Nevertheless, the general philosophy of matched filtering is still applicable in optics, even if it is used in a relaxed sense in practice, which we may refer to as “quasi-matched” filtering. Since matched (or quasi-matched) filtering eliminates signal dimensions outside the signal space defined by the matched filter bank, these dimensions can be used for data streams belonging to different users of the same communication channel. The process of sharing a common channel by exploiting signal orthogonality is referred to as multiplexing. Orthogonality can also be used for other purposes, such as optical label encoding, where the payload of a data packet is encoded in one dimension and a packet overhead used for routing purposes is encoded in an orthogonal dimension [51]. Figure 2.4 summarizes the most

Time

Frequency

0

x 1(t )

t TS 3 x 3(t )

x 3(t )

2 x 2(t )

0

TS x 2(t )

t

x -pol.

1

1 x 1(t )

t TS 3 x 3(t )

x 3(t )

x 2(t )

2

0 TS (a) Pulse position modulation (PPM) TDM frame

y-

t

TDM tributary

0 TS (b) Frequency shift keying (FSK) WDM channel

l. po B

f

0 x 2(t )

x 2(t ) TS

0

(f) Wavelength division multiplexing (WDM)

2 jx 1(t )

2 jx 1(t )

t

0

x 1(t )

t TS

(d) Quadrature modulation

A

C D

G

E F

H

I f

J

(h) Polarization multiplexing

OFDM tributary f

l.

po

y-

(g) Orthogonal frequency division multiplexing (OFDM) Coherent WDM (CoWDM)

t TS

WDM channel

y-

x -pol.

(e) Time division multiplexing (TDM)

t

1

x 1(t ) x 1(t )

(c) Polarization shift keying (PolSK)

AB … Z AB … Z t

TS

l.

po

t

Multiplexing

Quadrature

1

x 1(t )

l. po y- 0 x -pol.

0

t TS x 2(t )

x -pol.

Modulation

0

Polarization

x 1(t )

x 1(t )

WDM channel C

A B

E D

f

(i) Polarization interleaving

Figure 2.4 Signal orthogonality can be exploited to construct modulation formats (a–d), or to multiplex signals onto a shared communication channel (e–i).

32

Peter J. Winzer and Rene´-Jean Essiambre

important ways orthogonality is used in optical communications for modulation and multiplexing purposes: Orthogonality in time is achieved by having disjoint time slots within the symbol period TS represent different modulation symbols. The resulting modulation format is called pulse position modulation (PPM) [52, 53] and is depicted in Figure 2.4(a) for the case of a ternary modulation alphabet carrying log23  1.6 bits/symbol. By assigning non-overlapping time slots to different users of a common communication channel, one arrives at time division multiplexing (TDM) shown in Figure 2.4(e), where users A through Z share slots in a TDM frame. The synchronous optical transport network (SONET) as well as the synchronous digital hierarchy (SDH) are examples of TDM systems used in optical networking. If TDM tributaries are added and dropped to a TDM frame fully electronically, one speaks of electronic TDM (ETDM), and of optical TDM (OTDM) if these operations are performed through optical means. Historically, only ETDM systems have lent themselves to successful commercialization, but OTDM systems have proven to be valuable research tools to explore bit rates far beyond the capabilities of current electronic multiplexing technologies. Today, ETDM systems operate at bit rates up to 100 Gb/s (cf. Refs [24, 30, 54] and Chapter 5 of this book); OTDM systems with bit rates of 160 and 320 Gb/s have been reported [55, 56], with some demonstrations going up to 2.56 Tb/s on a single optical carrier wavelength [57, 58], as reviewed in Chapter 6 of this book. Orthogonality in frequency can be achieved by choosing strictly disjoint frequency bands, which is used in WDM as shown in Figure 2.4(f). Dense WDM (DWDM) systems used in transport networks typically have 50- or 100-GHz wide frequency bands, while coarse WDM (CWDM) access systems have 20-nm (2.5 THz) wide wavelength bands [59]. Any residual spectral overlap between WDM channels degrades their orthogonality and manifests itself in coherent WDM crosstalk if adjacent channels are at least partially copolarized. The use of orthogonal polarizations of adjacent channels (polarization interleaving) can re-establish orthogonality using the polarization domain, as discussed below. Alternatively to choosing completely disjoint frequency bands, one can also exploit the fact that two pulses of duration TS are orthogonal if their carrier frequencies are separated by integer multiples of 1/TS, even if they are temporally fully overlapping4 [40, 42]. Hence, different data signals may be modulated onto such orthogonal frequencies, which is at the heart of orthogonal frequency division multiplexing (OFDM) [60–62] or coherent WDM (CoWDM) [63] shown in Figure 2.4(g). In optical communications, the term “OFDM” is typically used whenever the orthogonal subcarriers are generated and modulated electronically, followed by a single electro-optic modulator to convert the frequency multiplex to the optical

4 This is easily seen by evaluating the inner product, Eqn (2.2), for two rectangular pulses of duration TS that have a carrier frequency difference of 1/TS.

2. Advanced Optical Modulation Formats

33

domain. In contrast, the term “CoWDM” is used whenever a set of orthogonal optical subcarriers is generated from a single continuous-wave laser, followed by individual electro-optic modulation on each subcarrier. Both methods are presently being studied in optical communications research. In the context of modulation, orthogonality in frequency may be exploited using frequency shift keying (FSK), as shown in Figure 2.4(b) for a ternary alphabet. Note in this context that the ternary alphabets for PPM and FSK are identical in Figure 2.4, but the analog modulation waveforms are very different. This underlines the fact that digital communication requires knowledge of both the symbol alphabet and the set of analog modulation waveforms. The generation and detection of binary optical FSK transmission at multi-Gb/s rates has been reported [8, 64, 65] but neither FSK nor its continuous-phase line-coded version, minimum shift keying (MSK) [40–42, 66, 67], has yet been studied in the context of high SE long-haul transport systems. Orthogonality in polarization is achieved by choosing one of two orthogonal polarization states. When used for data modulation, one speaks of PolSK shown in Figure 2.4(c). Although PolSK has been studied for optical communications applications (see, e.g., [9, 68–70]) it has not been intensively pursued, primarily due to the need for active polarization management at the receiver, necessitated by random polarization changes in optical fiber [71]. The additional receiver complexity associated with polarization management could be acceptable if PolSK offered significant transmission advantages. As for receiver sensitivity, PolSK is equivalent to OOK [9, 68]. The polarization degree of freedom can also be used to improve the propagation properties of a non-PolSK format by line coding [72, 73] (see Section 2.3.3). With the increasing popularity of coherent detection, which is inherently polarization-sensitive and is therefore typically implemented with polarization diversity at the receiver (cf. Section 2.4.1), the polarization dimension may be further exploited for the construction of modulation formats in the future. When applied to multiplexing, one may either transmit different data streams in the same WDM frequency band but in orthogonal polarizations [polarization multiplexing, Figure 2.4(h)], or one may alternate the polarization of adjacent WDM channels to avoid coherent WDM crosstalk, since coherent interference (“beating”) only occurs if two optical fields have the same polarization [polarization interleaving, Figure 2.4(i)]. Both techniques are used in research experiments to increase SE, with a current experimental record of 3.2 b/s/Hz per wavelength using polarization multiplexed differential quadrature phase shift keying (DQPSK) [74, 75]. However, none of these methods is currently used in commercial direct detection systems due to the complexity associated with polarization maintaining system components and the wavelengthdependent random polarization rotations in transmission fiber [76]. However, due to its inherent need for polarization diversity, coherent detection allows for a straightforward implementation of polarization multiplexing (cf. Section 2.5.9). A significant amount of research is expected to take place in this area over the next 5 years.

Peter J. Winzer and Rene´-Jean Essiambre

34

Orthogonality of the two quadratures of a narrowband bandpass signal, i.e., the real and the imaginary parts of the signal’s complex envelope, is the basis for the signal space most widely used in digital communications. The resulting formats are generally referred to as quadrature modulation formats and the resulting signal space is shown in Figure 2.4(d). The format uses an arbitrary complex-valued optical modulation waveform x1(t) which can, e.g., be pulsed or chirped (i.e., contain some analog, non data-bearing phase modulation). Within each symbol interval (i), the modulation waveform is multiplied by a symbol from an alphabet {ak} of complex numbers, leading to the complex communication signal xðtÞ =

1 X

i = 1

akðiÞ x1 ðt  iTS Þ:

(2.7)

Figure 2.5 shows three important examples for symbol constellations using quadrature modulation. If the alphabet {ak} modulates only along a single quadrature of the signal (e.g., only along the real part), we speak of amplitude shift keying (ASK), as visualized in Figure 2.5(a) for a 4-ary alphabet. In optical communications, the term “ASK” is sometimes used as a synonym for IM. However, M-ary IM only distinguishes between M optical intensity levels and not between M levels of the optical field amplitude. If only the phase is modulated (|ak| = const.), we speak of phase shift keying (PSK), shown for an 8-ary alphabet in Figure 2.5(b). If ak can take on other sets of complex values, which are often chosen on a Cartesian grid, the modulation format is referred to as quadrature amplitude modulation (QAM), as shown for a 16-ary alphabet in Figure 2.5(c). The average energy per symbol is related to the average energy per bit EB by EB =

ES log2 M

(2.8)

provided that all symbols are sent with equal probability of 1/M. All constellations in Figure 2.5 are shown with the same energy per bit. Im{ak }

Im{ak } 000

Im{ak } 00

01

11

001 10

Re{ak }

0000

0100

1100

1000

0001

0101

1101

1001

0011

0111

1111

1011

0010

0110

1110

1010

100 101

011

Re{ak }

010

111

Re{ak }

110

(a) 4-ASK

(b) 8-PSK

(c) 16-QAM

Figure 2.5 Three examples of frequently encountered symbol alphabets in a two-dimensional complex signal space; (a) 4-ary amplitude shift keying (ASK); (b) 8-ary phase shift keying (PSK); (c) 16-ary quadrature amplitude modulation (QAM). Bit combinations are mapped to modulation symbols using Gray coding.

2. Advanced Optical Modulation Formats

35

The assignment of bit combinations to modulation symbols in Figure 2.5 is not arbitrary but is chosen using Gray coding [40]. This coding method describes a mapping of bit combinations to modulation symbols in such a way that only a single bit is changed when transitioning from any symbol to an immediately adjacent symbol separated by the minimum distance in the symbol constellation. For example, the symbol “0101” in 16-QAM is surrounded by the nearest neighbors “0100,” “0001,” “0111,” and “1101,” which differ from the original symbol only in a single digit. Detecting one of these nearest-neighbor symbols is by far the most likely error event in the presence of circularly symmetric noise. Hence, Gray coding ensures that the prevailing class of symbol errors causes no more than a single bit error, which implies that the bit error probability is very similar to the symbol error probability. If Gray coding is not done, a single symbol error can produce as many as log2 M bit errors. In an alternative view of quadrature modulated signals, one can interpret the modulation symbols ak as real and imaginary parts of the analog modulation waveform x(t) sampled at TS-spaced sampling instants. These sample values can then be plotted in the complex plane representing the complex optical field. Figure 2.6 visualizes this process for the simple example of OOK, where the sample values for a 0 and a 1 bit are transferred to the complex plane. Note that this representation of complex envelope signals is not a signal space representation in the strict sense, because the two orthogonal axes are only associated with sampled symbol values rather than with entire symbol waveforms that would form a signal space. Nevertheless, for ISI-free signals that can be written in the form of Eqn (2.7), the symbol constellations obtained this way are equivalent to the ones of the corresponding signal space. Displaying constellations in the complex plane of the sampled optical field offers several distinct advantages over the related signal space representation discussed above. First, it allows the representation of signals that cannot be easily written in the form of Eqn (2.7), which is the case for frequently encountered waveforms with pattern dependencies. For example, non-return-to-zero (NRZ)– OOK with non rectangular pulses can only be written in the form of Eqn (2.7) if the optical power of several adjacent pulses adds up to a constant 1-bit level, which places severe restrictions on the allowed NRZ pulse shapes. Second, the complex plane representation captures a digital waveform at discrete points in time, separated by one symbol interval. In a natural extension, one can then visualize the a1

x (t )

Im{x (t )} a0

Re{x (t )} t

TS Figure 2.6 As an alternative to a strict signal space representation, one can visualize the complex amplitudes of TS-spaced signal samples (open circles) in the complex plane of the optical field (black dots). This representation can be extended by showing the continuous-time transitions between symbols (dashed lines).

Peter J. Winzer and Rene´-Jean Essiambre

36 x 1(t )

TS

t 0

TChip

x 3(t )

x 2(t )

Chip

TS

0

x 4(t )

t

TS

t 0

TS

t

0

Symbol Figure 2.7 Orthogonality in code is established on a subsymbol (chip) granularity. Chips are shown as dotted lines. All four modulation waveforms x1(t) through x4(t), shown as solid lines, are orthogonal to each other.

complex amplitude of the signal transitioning between symbols, as indicated by the dashed curves joining the modulation symbols in Figure 2.6. The particular example in the figure applies to chirped OOK. Note, however, that this way of visualizing symbol transitions does not specify dynamic aspects of the signal, such as pulse rise and fall times or the duration of signal pulses. Orthogonality in code captures the notion of temporally and spectrally fully overlapping signals that are made orthogonal by proper choices of their individual waveforms. Waveform orthogonality is established on a subsymbol granularity where a subsymbol is referred to as a chip. A simple example is visualized in Figure 2.7. The four modulation waveforms shown here are all mutually orthogonal, which is easily verified using Eqn (2.2). Orthogonal waveforms of this kind are also called orthogonal codes and can use other dimensions than the temporal as well. Oftentimes, orthogonal codes are constructed in the time/frequency plane [40, 42]. By assigning orthogonal codes to different users and letting each user perform modulation according to Eqn (2.7) with “his” waveform, we arrive at code division multiple access (CDMA), also called OCDMA in its optical implementation [77]. Note that OCDMA is not an optical modulation format but an optical multiplexing technique. OCDMA is studied in the optical research community for access networks, but has not yet found its way into commercial optical networks. Orthogonality in space can be achieved by allowing no spatial overlap between copropagating signals, e.g., by placing them on separate fibers. This trivial multiplexing technique is sometimes referred to as space division multiplexing and is frequently encountered on short optical links, e.g., for optical interconnects between racks of equipment, where fiber ribbons can be used. Less trivially, different transverse modes propagating in a multimode optical fiber are all orthogonal [78, 79] and may be assigned to different users, which leads to the notion of mode division multiplexing. However, multimode fibers are prone to random mode coupling, and the recovery of any particular transverse fiber mode from a mixture of modes is not easily achieved [80, 81]. Modal multiplexing is an interesting research topic, including the study of multiple-input multiple-output (MIMO) communication techniques [82, 83].

2. Advanced Optical Modulation Formats

37

2.2.2 Memory-Less Multilevel Modulation Modulation formats whose symbol alphabets consist of more than the absolute minimum of two symbols are generally referred to as multilevel formats. If the assignment of symbols to bits or to groups of bits is independent of the symbols sent before or after, one speaks of memory-less modulation, discussed in this section. In contrast, if the assignment is made according to certain rules for the succession of different symbols, one speaks of modulation with memory [40], discussed in Section 2.2.3. Introducing more symbols to a modulation alphabet can either be done by expanding the dimensionality of the signal space or by placing more symbols in an already existing signal space of given dimensionality [40–42, 84, 85]. The former approach typically increases the resilience to detection noise, which is the reason for the exploitation of PPM in deep-space optical links, where receiver sensitivity is of utmost concern [53]. However, an increased signal dimensionality comes at the expense of an increased signal bandwidth if dimensions are added in time or frequency. An important orthogonal dimension that does not increase the signal bandwidth is the polarization dimension. Its exploitation comes at the expense of an increased transmitter and receiver complexity as well as with a limited increase of orthogonal dimensions to be added. The latter approach of placing more symbols in an already existing signal space5 increases the bit rate at constant symbol rate, which has several important implications: • The spectral extent of the multilevel modulated optical signal is less than that of a binary modulated signal at the same bit rate, which allows for stronger filtering in optically routed networks at higher SE. • The reduced symbol rates compared to binary modulation increase the formats’ robustness to some propagation impairments such as CD and PMD. • The reduced symbol rates help to overcome potential limits of high-speed optoelectronic components. For example, the required components to build a 50-Gbaud DQPSK transponder [86–88] are more mature today than those needed to build a 100-Gb/s OOK system [24, 30, 54]. • The reduced symbol rates enable the use of digital electronic signal processing, albeit at the cost of multi-GSamples/s analog-to-digital and/or digital-to-analog conversion. • As a downside, multilevel modulation is generally associated with a reduced tolerance to noise, since adding points to the symbol constellation reduces the minimum distance for a fixed average signal power.

5

All symbol alphabets shown in Figure 2.5 are of this kind.

Peter J. Winzer and Rene´-Jean Essiambre

38

• As another downside, multilevel modulation generally leads to a reduced tolerance to fiber nonlinearity due to the reduced distance between the constellation points. Furthermore, for sufficiently high SE, the impact of fiber nonlinearity is generally more severe for the associated lower symbol rates. In the context of multi-Gb/s optical communications, multilevel IM [32, 89], multilevel phase modulation [90–92], as well as hybrid multilevel intensity/phase and quadrature amplitude modulation [93–97] have been discussed and demonstrated. Multilevel IM has not proven beneficial for fiber-optic transport applications, mainly due to a substantial back-to-back receiver sensitivity penalty compared to binary OOK. For example, the four-level 4-ary IM format incurs a penalty of about 8 dB due to the unequal amplitude level spacings necessitated by signal-dependent noise arising from squarelaw detection [32, 89]. The multilevel modulation format that undoubtedly has received the most attention, including recent transponder commercializations, is (differential) quadrature phase shift keying (QPSK, DQPSK, see Sections 2.5.8 and 2.5.9).

2.2.3 Modulation with Memory and Coding If redundancy is incorporated into a digital modulation signal, we generally speak of coding. As visualized in Figure 2.8, we distinguish between two forms of coding: line coding and error correcting coding. Line coding denotes the introduction of redundancy with the aim to prevent detection errors and to mitigate transmission impairments, e.g., by shaping the modulation spectrum or by avoiding certain adverse symbol patterns.6 This class of coding is also referred to as modulation with memory (i.e., the succession of symbols follows Coding

Line coding

Error correcting coding

Redundancy

Redundancy

Symbol rate

Symbol alphabet Symbol rate

Symbol alphabet

Figure 2.8 Introducing redundancy into a digital signal is referred to as coding. If coding is used to prevent errors, one speaks of line coding. If coding is used to correct for errors, one speaks of error correcting coding. In each case, redundancy can be introduced either in the time domain by increasing the symbol rate or in signal space by expanding the symbol alpabet.

6

An example for line coding is the avoidance of certain words in English spoken by non-native speakers. For example, people with a strong German accent may substitute words containing “th” by a suitable synonym; people with an Italian or French accent may avoid words starting with the letter “h”; and people with a Japanese accent may find themselves not understood when using words containing the letter “r”.

2. Advanced Optical Modulation Formats

39

specific rules dictated by the code) or constrained coding (i.e., the nature of the channel constrains the number of allowed symbol sequences). The required overhead to accommodate redundancy can either be added in the time domain by increasing the symbol rate or in signal space by increasing the symbol alphabet size [40, 98–102]. Alternatively, or in addition to line coding, error correcting coding can add redundancy at the transmitter in order for the receiver to be able to correct or at least to detect transmission errors.7 Like in line coding, redundancy can be added in the time domain by increasing the symbol rate (by 7% when using FEC schemes employed in standardized terrestrial optical communication systems [46]) or in signal space by adding more points to the symbol constellation. The latter strategy is generally referred to as coded modulation [85, 103–105]. Modulation schemes with memory are conveniently characterized using the concept of a finite state machine, represented by state diagrams as shown in Figure 2.9. A state diagram captures the set of possible states of an abstract machine, called an “encoder” in our context, together with a set of inputs and a set of outputs for each state transition. A state machine is fully and uniquely defined if all possible inputs are represented on exactly one transition leaving each state. As two simple yet highly relevant examples for optical modulation formats, consider the two finite state machines shown in Figure 2.9. Both encoders possess two states (A and B) and accept single bits {0, 1} as their input, i.e., they process the input information stream bit by bit. The output of both encoders is the ternary symbol alphabet {1, 0, þ1}, where one symbol is put out for Current input bit

Corresponding output symbol

1/–1 State A

0/0 0/0

0/0 State B

State A

1/+1

State B

1/–1

0/0

1/+1 (a) Pseudo-multilevel modulation (CSRZ)

(b) Correlative coding (DB, PSBT)

Data bits

0

0

1

0

1

1

1

0

0

1

0

1

Pseudo-multilevel symbols Correlative coding symbols

0 0

0 0

+1 +1

0 0

+1 –1

–1 –1

+1 –1

0 0

0 0

–1 –1

0 0

–1 +1

Figure 2.9 Finite state machine representations of two examples of modulation with memory. In pseudo-multilevel modulation (a), the state transitions are independent of the input bit stream, while in correlative coding (b) the value of the current input bit determines the encoder’s new state.

7 Most languages contain a high degree of redundancy, with the human brain acting as a powerful FEC decoder.

40

Peter J. Winzer and Rene´-Jean Essiambre

each input information bit, i.e., redundancy is introduced in the symbol domain without adding any temporal overhead. Both encoders represent a 0 bit by the symbol 0, whereas a 1 bit may be represented either by symbol þ1 or by 1. The symbol set {1, 0, þ1} is highly convenient for the dominant class of direct-detection (square-law) optical receivers, which only respond to the optie. As a cal power P = j~ ej2 , the squared magnitude of the complex optical field ~ consequence, a direct-detection receiver is unable to distinguish between the two received symbols j~ ej, since they both have the same optical power, ej2 . Therefore, the ternary symbol set ½fþj~ ej; 0; j~ ejg is automatiP1 = P2 = j ~ cally mapped to the binary set f0; j~ ej2 g at the receiver, amenable to singlethreshold detection. The difference in the two encoders shown in Figure 2.9 lies in the set of rules that determine the next output symbol, and more specifically whether a 1 bit is represented by a þ1 or a 1. If the new state after a state transition depends on the current input bit, we speak of correlative or constrained coding, while we speak of pseudo-multilevel modulation if the new state after a state transition does not depend on the current input bit.

Pseudo-Multilevel Modulation The encoder in Figure 2.9(a) represents a pseudo-multilevel modulation format, since the new state does not depend on the input bit; starting at state A, we arrive at state B for both a 0 bit and a 1 bit at the input, and vice versa when starting at state B. As a result, the alternation between the symbols þ1 and 1 is completely independent of the input data stream but only depends on whether the input bit falls in an “even” or an “odd” time slot. An example symbol sequence at the encoder output is shown in the table in Figure 2.9 for a given data sequence at the encoder input, assuming that the encoder is in state A initially. With a pulse-like modulation waveform carrying the amplitude-modulated symbols, the resulting format is known as carrier-suppressed return-to-zero (CSRZ, Section 2.5.5) in optical communications. Although CSRZ is by far the best studied pseudo-multilevel modulation scheme in optical communications, other pseudo-multilevel formats have also been discussed, predominantly in the context of mitigating fiber nonlinearities. For example, Ref. 106 shows that an optimum choice of relative phases between adjacent pulses in a set of four consecutive bits can be either {0,/2, 0,/2} or {0,0,,} to suppress intrachannel four-wave mixing (IFWM). The mechanism for the reduction of IFWM using these coding schemes is by destructive interference of IFWM effects [106, 107]. The effectiveness of pseudo-level modulation has also been studied when implemented using a single phase modulator operating at a fraction of the symbol rate [108–111]. An advanced transmitter based on nonlinear polarization rotations coupled with polarizers is described in [112]. Another example of pseudo-multilevel coding is auxiliary polarization modulation [73] where the

2. Advanced Optical Modulation Formats

41

polarization state is rotated back and forth by 90 every other symbol. This type of pseudo-multilevel modulation is particularly efficient to reduce the impact of IFWM [73]. We refer to these modulation schemes as alternate phase (AP) or alternate polarization (APol) modulation. Correlative and Constrained Coding The encoder in Figure 2.9(b) represents a correlative coding scheme, in which the new state depends on the input bit. For a 1 bit at the input, the encoder does not change state. Hence, a continuous string of 1 bits is represented by a continuous succession of þ1 or 1 symbols, depending on the encoder’s current state. Every 0 bit changes state, which implies that an odd number of 0 bits between any two 1 bits alternates the sign of the output symbol representing a 1 bit, while an even number of 0 bits between any two 1 bits does not have an effect. An example symbol sequence at the encoder output is shown in the table in Figure 2.9 for a given data sequence at the encoder input, assuming that the encoder is in state A initially. The resulting modulation format is known as duobinary modulation (DB) or phase-shaped binary transmission (PSBT), discussed further in Section 2.5.3. Another easily generated (and hence fairly well studied) line coded modulation format in optical communications is alternate mark inversion (AMI), with the same symbol set as DB but with a different correlation rule: Each 1 bit is associated with a sign change, irrespective of the number of zeros in between. The example sequence listed in Figure 2.9 would read {0, 0,þ1, 0,1,þ1,1, 0, 0,þ1, 0,1} for AMI. Interestingly, AMI can be generated by applying the pseudo-multilevel CSRZ correlation rule to a DB signal (cf. Figure 2.19). This fact has lead to the name duobinary carrier suppressed (DCS) as a sometimes encountered synonym for AMI [113–115]. Transmission of AMI has been investigated at 40 Gb/s [116–118], where its resistance to fiber nonlinearity has been found to be similar to OOK formats for nonzero dispersion fiber (NZDF) [117]. A moderate advantage has been found for nonlinear transmission over SSMF [116, 119, 118]. In recent literature, the term “modified duobinary” has also been used to denote AMI. Note, however, that modified duobinary is not equivalent to AMI but represents a distinctly different line coded modulation format [40, 84, 120, 121]. Line coding has also been studied in the context of pattern-dependent intrachannel fiber nonlinearities. Here, one can either introduce bit-correlated phase modulation to mitigate transmission impairments [116, 122, 123], or one can remove certain “bad” bit patterns that are responsible for a large part of the nonlinear signal distortions due to intrachannel nonlinearities, which is generally referred to as constrained coding. For example, the impact of IFWM depends on the transmitted data patterns [124] and the relative phases between symbols [125]. For binary OOK, the bit patterns impacting the BER most severely are generally the ones that include very isolated “0’s.” Removing such isolated “0’s” can significantly improve the BER [124]. A variety of line codes have been devised

42

Peter J. Winzer and Rene´-Jean Essiambre

to minimize the impact of IFWM [100, 126–128] and significant improvements in transmission have been predicted. One should point out, however, that constrained coding generally requires an increase in symbol rate, which may offset the advantages brought by the code because of a higher required optical signalto-noise ratio (OSNR) or, at high SEs, because of a broader signal spectrum.

2.3 ELECTRO-OPTIC MODULATOR TECHNOLOGIES As data rates in optical communication systems have traditionally been limited by the speed of available opto-electronic components, it is of utmost importance to always consider practical aspects of modulation and detection hardware when designing optical modulation formats. Finding the most cost-effective modulation technique for a particular system application involves aspects of modulation format and modulator technology. Three basic modulator technologies are widely in use today: directly modulated lasers (DMLs), electroabsorption modulators (EAMs), and Mach–Zehnder modulators (MZMs), together with their extension to nested MZMs.

2.3.1 Directly Modulated Lasers Direct modulation of lasers is the easiest way to imprint data onto an optical carrier. Here, the transmit data is modulated onto the laser drive current, which then switches on and off the light emerging from the laser [129, 130]. The resulting modulation format is binary IM (OOK). Due to their compactness and low cost, DMLs ideally lend themselves to dense integration in small form factor pluggable transponders (SFPs at 2.5 Gb/s; XFPs at 10 Gb/s). Today, DMLs are widely available up to modulation speeds of 2.5 Gb/s, with some availability up to 10 Gb/s and research demonstrations up to 40 Gb/s [131, 132]. The main drawback of DMLs for high bit rate transmission beyond short-reach access applications is their inherent, highly component-specific chirp, i.e., a residual phase modulation accompanying the desired IM [26, 129, 130]; laser chirp broadens the optical spectrum, which impedes dense WDM channel packing and can lead to increased signal distortions caused by the interaction with fiber CD [129, 130] and fiber nonlinearity [133]. On the other hand, the residual frequency modulation of DMLs can be exploited for formats related to binary FSK, such as dispersion supported transmission (DST) [134], where a DML is used to predominantly modulate the frequency of the optical field, with the intent to have the correct amount of CD convert this frequency modulation into IM upon transmission. Transmission distances of 250 km at 10 Gb/s have been achieved using DST [135]. More recently, chirp-controlled DML designs, promoted under the name chirp-managed laser (CML), have achieved transmission distances in excess of 200 km at 10 Gb/s over SSMF [136].

2. Advanced Optical Modulation Formats

43

2.3.2 Electroabsorption Modulators (EAMs, EMLs) EAMs are pin semiconductor structures whose bandgap can be modulated by applying an external voltage, thus changing the device’s absorption properties [130]. EAMs feature relatively low drive voltages (typ. 2V), and are cost-effective in volume production. They are available for modulation up to 40 Gb/s today, with research demonstrations up to 80 Gb/s [137]. However, similar to DMLs, they produce some residual chirp, which can be engineered to increase dispersion limited transmission distances. By carefully controlling the chirp, an EAM may also be used as an optical in-phase/quadrature (I/Q) modulator to generate an electronically predistorted optical field [138], or it can be embedded in an interferometric waveguide structure similar to a nested MZM (Section 2.3.4) [139]. As a downside, EAMs have wavelength-dependent absorption characteristics, dynamic extinction ratios (maximum-to-minimum-modulated light power) typically not exceeding 10 dB, and limited optical power-handling capabilities. Their fiber-tofiber insertion loss is about 10 dB. “On-chip” integration with laser diodes avoids the high loss at the input fiber-to-chip interface, and leads to compact transmitter packages. The resulting electroabsorption-modulated lasers (EMLs), with output powers on the order of 0 dBm, are widely available today for modulation up to 10 Gb/s and integration into XFP transponders. Another way of eliminating the high insertion losses of EAMs is the integration with semiconductor optical amplifiers (SOAs), which can even yield some net fiber-to-fiber gain [140]. Figure 2.10(a) shows the typical exponential transmission characteristics of an EAM as a function of drive voltage.

2.3.3 Mach–Zehnder Modulators Unlike EAMs, which work by the principle of absorption, MZMs work by the principle of interference, controlled by modulating the optical phase. The Im{E } Power transmission (%)

Power transmission (dB) 0

100

–10

V1(t )

Vπ in

V1(t ) out

–2 V

in

V2(t )

–20 Drive voltage

(a) Electroabsorption modulator

0

Voltage difference (ΔV )

(b) Mach–Zehnder modulator

Re{E } Im{E }

V3(t )

–V1(t ) V2(t ) –V2(t ) –V3(t )

out

Re{E }

Im{E } Re{E }

(c) Quadtrature modulator

Figure 2.10 Overview of modulator technologies used for advanced optical modulation formats: (a) typical transmission function of an EAM; (b) sinusoidal power transmission function of a MZM, whose structure is shown in the inset; (c) structure of a quadrature modulator (or nested MZM), together with its main application as a modulator for QPSK signals. (Solid symbols: full drive voltage; open symbols: reduced drive voltage.)

Peter J. Winzer and Rene´-Jean Essiambre

44

modulator structure is shown in the inset to Figure 2.10(b): The incoming light is split into two paths at an input coupler. One (or both) paths are equipped with phase modulators that let the two optical fields acquire some phase difference relative to each other, controlled by the applied phase modulation voltages V1,2(t). Finally, the two fields interfere at an output coupler. Depending on the applied electrical voltage, the interference varies from destructive to constructive, thereby producing IM. The optical field transfer function TE(V1, V2) of the MZM reads (see, e.g., [146, 32]) TE ðV1 ; V2 Þ =

1 jðV1 Þ fe þ ejðV2 Þþj g 2

= ejððV1 ÞþðV2 Þþ

Þ=2

cos½ððV1 Þ  ðV2 ÞÞ=2  =2;

(2.9)

where (V1,2) are the voltage-modulated optical phases of the two MZM arms, and c is an additional, temporally constant phase shift in one of the arms, referred to as the modulator bias. If the phase modulation depends linearly on the drive voltage ( = V), which is true for most materials used for MZMs, the MZM power transfer function depends only on the drive voltage difference DV: TP ðV1 ,V2 Þ = jTE ðV1 ; V2 Þj2 = TP ðDVÞ = cos2 ðDV=2 þ Vbias =2Þ. The characteristic sinusoidal MZM power transfer function is shown in Figure 2.10(b). Its periodicity is fundamentally related to the 2 periodicity of the optical phase. The modulation voltage that is required to change the phase in one modulator arm by , and thereby lets the MZM switch from full transmission to full extinction, is called switching voltage V. For a given drive voltage difference DV according to the desired modulated intensity, the additional degree of freedom in choosing V(t) = V1(t)þ V2(t) can be exploited to imprint phase modulation on the signal. By separately controlling V(t) and DV(t), one can operate a MZM as an I/Q modulator to independently modulate both quadratures of the optical field, e.g., to produce electronically predistorted optical waveforms [141]. One may also use V(t) to imprint an analog phase modulation (chirp) onto a digital signal [142–145]. If chirp is not desired (which is often the case in digital optical modulation), the two modulator arms are driven by the same amount but in opposite directions [V1(t) = V2(t)], which eliminates the phase term in Eqn (2.9). This driving condition is known as balanced driving or push–pull operation. Some modulator structures, e.g., x-cut lithium niobate (LiNbO3) use only a single drive signal which modulates both arms of the MZM simultaneously, having the condition V1(t) = V2(t) for chirpfree operation built in by appropriate drive electrode design. MZMs are most conveniently implemented in LiNbO3 [64, 146, 147], gallium arsenide (GaAs) [148], or indium phosphide (InP) [149–152]. LiNbO3-based devices are bulkier than their semiconductor equivalents, but offer easier fiberto-chip coupling, less wavelength dependence, and no residual IM accompanying the desired phase modulation. In contrast, semiconductor devices, and in

2. Advanced Optical Modulation Formats

45

particular InP, lend themselves to much denser integration [151, 153]. MZMs, especially LiNbO3-based devices, feature wavelength-independent modulation characteristics, excellent extinction performance (typ. 20 dB), and lower insertion losses (typ. 5 dB) than EAMs. The required (high-speed) switching voltages of up to 6 V, however, require broadband driver amplifiers, which can be challenging to build at data rates in excess of 10 Gb/s. Today, LiNbO3-based MZMs are widely available for modulation up to 40 Gb/s. Using optical equalization or duobinary modulation, bit rates of 100 Gb/s have been achieved in research labs [24]. InP-based modulators, available for 10-Gb/s modulation today, feature lower switching voltages (typ. 2–3 V) but exhibit residual absorption in addition to phase modulation in the modulator arms which can lead to chirp unless counteracted by proper electrode design [152]. Due to their well controllable modulation performance and the possibility of independently modulating intensity and phase of the optical field, MZMs form the basis of many advanced optical modulation formats.

2.3.4 Nested MZMs—Quadrature Modulators A single MZM is in principle capable of covering the full complex plane of the optical field, but this requires a drive voltage swing of 2V on each arm [141], which can be difficult to achieve in practice. Furthermore, according to Eqn (2.9), modulation of the two quadratures is not accomplished independently of each other which can complicate the generation of the required high-speed drive waveforms. A solution to both problems is provided by a true quadrature modulator (QM) [64]. The structure of a QM, also referred to as a “nested MZM,” a “double-nested MZM,” “I/Q modulator,” or “Cartesian modulator” and initially proposed for FSK and singlesideband (SSB) modulation [154, 155], is shown in Figure 2.10(c). The symbol diagrams (solid dots) illustrate its most prominent application as a QPSK modulator [90, 91]: first, the incoming light is split into two arms. The sub-MZMs in each arm imprint two independent bit streams on the lightwave, typically using chirp-free (push–pull) modulation with voltages –V1(t) and –V2(t). In the case of QPSK modulation, these two signals are binary PSK signals. A third modulator in the outer MZM structure (V3) introduces a temporally constant 90 phase shift between the two signals, which puts them in quadrature to each other. Upon combination in the output coupler, the four-phase states characteristic of QPSK modulation are obtained. Being a true I/Q modulator, the nested MZM can be used for a variety of other applications [64]. By driving one sub-MZM with the data sequence and the quadrature sub-MZM with the sequence’s Hilbert transform generates SSB modulation [155]. Allowing the third electrode [V3(t)] to be RF-modulated as well, one can implement FSK and MSK, a line-coded version of FSK with continuous optical phase [40, 42, 66, 67]. Furthermore, a QM can be used to generate electronically predistorted optical waveforms [141] with independent control of real and imaginary parts of the optical field as well as with reduced drive amplitude requirements: the whole

Peter J. Winzer and Rene´-Jean Essiambre

46

complex plane can be covered by arbitrarily small modulation voltages, albeit at the expense of an increased insertion loss, as visualized by the open symbols in the constellation diagrams in Figure 2.10(c), showing the set of QPSK symbols obtained if only half the drive voltage is available. QMs are commercially available in LiNbO3 technology for modulation speeds of 20 Gb/s on each sub-MZM, allowing for the generation of 40-Gb/s QPSK signals. Research prototypes have been reported for QPSK modulation at data rates as high as 111 Gb/s [86–88].

2.4 OPTOELECTRONIC DEMODULATION CONCEPTS The conversion process of an (optical) bandpass signal to an (electrical) baseband signal is called demodulation and relies on photodetectors as its key component. Photodetectors used in optical fiber communications are implemented as reversebiased semiconductor pin diodes [43] with demonstrated bandwidths of 100 GHz and beyond [54, 156, 157]. Often, transimpedance amplifiers are copackaged with the photodiodes (PDs) to form a photoreceiver [158]. Diode structures with a builtin high-field carrier multiplication region to produce a larger output current while adding a minimum of extra noise are referred to as avalanche photodiodes (APDs) [43, 159, 160]. APDs are widely available for 10-Gb/s applications today. Across a wide range of optical input powers, photodetectors generate an electrical current that is linearly proportional to the optical intensity impinging on the diode, but is inherently independent of the optical phase and also independent of polarization. Hence, any modulation format that carries information by means of a physical quantity other than the optical intensity requires additional demodulation elements prior to photodetection [9]: • PolSK requires demodulation by means of a polarization beam splitter that is aligned to the two orthogonal polarizations used as modulation waveforms. • FSK requires an optical filter that either passes or rejects the symbol waveform depending on its frequency. • PSK or, more generally, any modulation format in quadrature space that transports information using the optical phase, requires appropriate phaseto-intensity conversion prior to photodetection, which can either be based on a local oscillator (LO) laser (coherent demodulation) or on a delay interferometer (DI) (delay demodulation or differentially coherent demodulation).

2.4.1 Coherent Demodulation Figure 2.11(a) shows the conceptual structure of a coherent receiver [9–12], and discussed in more detail in Chapter 3 of this book. A sufficiently phase-stable (coherent) LO laser acts as a phase reference for the received signal. If the received

2. Advanced Optical Modulation Formats

47 Delayed signal

Signal Signal Local oscillator (a) Coherent demodulation

(b) Delay demodulation

Figure 2.11 Demodulation of a phase-modulated format can either be accomplished using a local oscillator laser as a phase reference (a) or differentially by comparing the phase of two different symbols of a symbol sequence (b).

signal is in-phase with the LO, the photodetector records constructive interference, while it records destructive interference if the two optical fields have opposite phase, thus converting phase modulation into IM amenable to photodetection. The resulting optical power P(t) at the photodetector reads  2 eSig ðtÞ þ2Re f~ eLO ~ eSig ðtÞg PðtÞ = j~ eLO j2 þ~ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi = PLO þ PSig ðtÞ þ 2 PLO PSig ðtÞ cosð2fIF t þ ’Sig ðtÞÞ ; |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

(2.10)

Beat term

where 0 £  £ 1, the heterodyne efficiency, denotes the spatial, modal, and polarization alignment of signal and LO fields. In a single-mode fiber system,  captures the eLO . polarization mismatch between the signal optical field ~ eSig and the LO field ~ Phase modulation is included in the optical signal’s phase ’Sig(t). Typically, the first two terms on the right-hand side of Eqn (2.10) can be ignored in favor of the beat term, either because the temporally constant LOpis chosen much ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi  stronger than the signal, or because the signal PLO þ PSig ðtÞ  2 PLO PSig ðtÞ cos 2fIF t þ ’Sig ðtÞ from the second output port of the beam splitter is subtracted using balanced detection. Importantly, note that in contrast to delay demodulation, balanced detection does not impact the sensitivity of a coherent receiver in a fundamental way. Nevertheless, balanced detection is preferred in practice because it allows to suppress the first two terms on the right-hand side of Eqn (2.10), i.e., any LO relative intensity noise and the direct-detection signal component. Balanced detection also makes use of the entire available signal and LO power. To make a coherent receiver polarization-independent, which is critical in a fiber-optic communication system with inherently randomly varying polarization, one either needs to track the signal polarization by the LO polarization at the receiver (polarization control), or one has to use two parallel coherent receiver setups, one for each of the two orthogonal polarizations (polarization diversity). The latter approach is the more practical one, even though it requires more hardware components. Furthermore, once polarization diversity is implemented, polarization multiplexing can be added without additional complexity in the optoelectronic receiver front-end [15, 161, 162].

Peter J. Winzer and Rene´-Jean Essiambre

48

f

f

fIF = 0 (a) Homodyning

fIF > BSig (b) Heterodyning

f fIF < BSig (c) Intradyning

Figure 2.12 (a) Homodyning implies an intermediate frequency fIF of exactly zero; (b) heterodyning uses an IF typically in excess of three times the signal bandwidth; (c) intradyning uses an IF that falls within the signal band, followed by digital phase locking. The positive-frequency portions of the IF spectra are shown solid and the negative-frequency portions are shown dashed.

Note from Eqn (2.10) and with reference to Figure 2.12(a) that stable interference across the symbol stream is only obtained if the optical frequencies of the LO and the signal carrier are precisely the same (homodyne detection), resulting in fIF = |fSig  fLO| = 0. Robust analog phase-locking of two optical signals is a nontrivial task, which is one of the main reasons why homodyne detection was never commercialized by the optical fiber telecommunications industry. If the two frequencies are offset, the detected photocurrent shows oscillations at the intermediate frequency fIF. If the IF is chosen significantly higher than the signal bandwidth (typically at least by a factor of three), the entire optical signal spectrum (in magnitude and phase) is directly translated from its optical carrier of around 195 THz to an electrical bandpass signal centered at the IF for further electronic processing, as visualized in Figure 2.12(b). Alternatively, one may choose the IF to fall within the signal band by roughly aligning the LO frequency with the signal frequency (intradyning, Figure 2.12(c) [31]). In this case, one saves on optoelectronic front-end bandwidth and processing bandwidth compared to heterodyning, but due to the overlap of positive (solid) and negative (dashed) signal frequency components one has to use digital phase locking algorithms to recover the modulation signal from its sampled I and Q components, which requires high-speed analogto-digital conversion and digital signal processing hardware [14–16, 161–163]. Note also that a second coherent receiver setup with a 90 phase-shifted LO is needed to access both I and Q components simultaneously. The resulting dual-output combination optics for signal and LO is generally referred to as a 90 hybrid.

2.4.2 Delay Demodulation Figure 2.11(b) shows the second method for phase-to-intensity conversion at the receiver. Here, a portion of the received signal is tapped off, delayed, and reinserted into the signal path. That way, the signal interferes with a delayed version of itself upon detection to produce constructive interference if the phase difference between the two interfering portions of the signal is zero, and destructive interference if it is . Hence, the signal acts as a phase reference to itself and takes the role of the LO in a coherent receiver; this similarity is indicated by the hatched and white arrows in Figure 2.11. The demodulation delay is typically chosen to be an

2. Advanced Optical Modulation Formats

49

integer multiple of the symbol duration to provide clean interference properties. Due to stability reasons, one almost exclusively chooses a demodulation delay of a single symbol period.8 If both signal quadratures (I and Q) are to be detected, a second interferometer with 90 shifted interference properties is typically used, in analogy to a 90 hybrid in coherent detection. If a phase modulation format is detected differentially, i.e., by delay demodulation, one generally speaks of differential phase shift keying (DPSK), although the word “differential” characterizes the detection technique and not the format itself. From a strict modulation format point of view, there is no difference between a PSK and a DPSK signal. Finally, we note that one can also use delay demodulation in combination with signal processing to estimate the phase of the received signal by averaging the differential phases over many symbols (multisymbol phase estimation). This averaged phase estimate then takes the role of the LO in coherent detection [166, 167] and Chapter 4 of this book.

2.5 FREQUENTLY ENCOUNTERED OPTICAL MODULATION FORMATS Figure 2.13 summarizes the wealth of possibilities in modulation and multiplexing in a single-mode optical fiber system, which supports a single transverse mode in two polarizations. While not yet extensively pursued, any combination of the orthogonal

PolSK

Pol multiplexing Pol interleaving

Modulation

Multplexing

Memory-less FSK

With memory MSK

Modulation Frequency

Polarization

Multplexing PPM

Signal orthogonality for modulation and multiplexing

Modulation Time

ETDM OTDM

Multplexing

Code (multiplexing only) Quadrature (modulation only)

OCDMA

Intensity

Binary Chirp-free NRZ

RZ

WDM CoWDM OFDM

OOK VSB, SSB

Memory-less Multilevel Chirped

M-ASK

NRZ

RZ

C-NRZ, DST

CRZ, ACRZ

Phase With memory Pseudo-multilevel

Memory-less

Correlative coding

'PSBT' CSRZ Partial response 'PASS' VSB-CSRZ 'CAPS' AP formats DB AMI ('DCS')

Binary NRZ

Multilevel RZ

(D)PSK

NRZ

RZ

(D)QPSK

Figure 2.13 Classification of the most important modulation formats and multiplexing schemes discussed in optical communications today.

8 For signals that undergo strong optical filtering, best receiver performance is obtained if the delay is chosen somewhat smaller than one symbol duration [164, 165].

Peter J. Winzer and Rene´-Jean Essiambre

50

(f) VSB-CSRZ-OOK

Optical spectrum

Optical spectrum

(e) VSB-NRZ-OOK

Frequency

Frequency

Optical spectrum

(d) Duobinary (DB, PSBT)

Frequency (g) DPSK, RZ-DPSK

Frequency

Frequency (h) DQPSK, RZ-DQPSK

Optical spectrum

Frequency

Optical spectrum

RB

(c) 67% CSRZ-OOK

Optical spectrum

10 dB

(b) 33% (50%) RZ-OOK

Optical spectrum

Optical spectrum

(a) NRZ-OOK

Frequency

Figure 2.14 Optical spectra and optical intensity eye diagrams of important modulation formats.

signal spaces discussed above may be used for future advanced optical modulation schemes. For example, combining the quadrature space with the space of orthogonal polarizations yields a four-dimensional signal space with interesting properties. In this section, we will limit ourselves to a discussion of the most frequently encountered modulation formats in the context of optical communications today, which are all based on the quadrature signal space. These and more formats are reviewed in more detail in, e.g., [27, 32–38, 64]. Optical spectra and intensity eye diagrams for the formats discussed here are shown in Figure 2.14. Some key performance characteristics are listed in Table 2.2.

2.5.1 Non-Return-to-Zero Modulation (NRZ and NRZ-OOK) The simplest optical modulation format to generate is NRZ-OOK, often just referred to as “NRZ”. Because of its simplicity, this is also the most widely deployed format in fiber-optic communication systems today. The name “NRZ” takes account of the fact that the optical power does not return to zero between two successive “1” bits. More generally, the term “NRZ” is employed as a qualifier for other than binary intensity modulated (i.e., OOK) formats if they do not use optical pulses as symbol waveforms. For example, “NRZ-DPSK” denotes differential phase shift keying (DPSK) where the optical intensity may stay constant over multiple bits. In contrast, return-to-zero DPSK (RZ-DPSK) denotes binary phase modulation of distinct optical pulses.

2. Advanced Optical Modulation Formats

51

NRZ-OOK

NRZ-DPSK

Drive signal ΔV

0 1 1 0 1 1 1 0 1

+1 0 0 +1+1 0 –1 –1 0

0 π π π 0 π π 0

DB eye

NRZ eye

Data 1

DB (PSBT)

Precoded data 1 0 0 1 1 0 1 1 0 1 0 0 1 1 0 1 1 0

ISI

DB eye NRZ eye +1 0

0 NRZ-OOK

Time

Square-law

OOK

NRZ-DPSK

Time

NRZ-OOK

DB (PSBT) DPSK

DB (PSBT)

Power

Optical power

Transmission

Field

–1 Strongly filtered NRZ-OOK (= DB drive waveform)

Delay & Add vs. Bessel

Photo-detection 1

10 dB

RB

Frequency

0

DB (PSBT) intensity

Figure 2.15 Overview of data modulation using a MZM for NRZ-OOK, DB (PSBT), and NRZ-DPSK. The electrical drive signals in the lower left are applied to the sinusoidal MZM transfer characteristic (upper left; solid: optical power; dashed: optical field) to yield the optical intensity waveforms in the upper right. The MZM bias points are indicated by open circles. Insets: Eye diagrams for NRZ (left) and DB (right) measured at 10.7 Gb/s; filter characteristics of delay-and-add filtering (dashed) and Bessel low-pass filtering (solid), two common options for DB prefiltering.

At bit rates of 10 Gb/s and above, NRZ-OOK is most conveniently generated using DMLs or EAMs (for short-reach and intermediate reach transponders) or using chirp-free MZMs (for long-haul applications). When using a MZM, the modulator is biased at 50% transmission (often referred to as the quadrature point) and is driven from minimum to maximum transmission with a voltage swing of V, as visualized in Figure 2.15. The inset shows a typical measured NRZ eye diagram at 10.7 Gb/s. Note that when fully driven between its transmission minimum and maximum, the nonlinear (sinusoidal) compression of the MZM transfer function steepens the drive waveform and suppresses overshoots and ripple upon conversion to optical power. A controlled amount of chirp may be introduced to increase the dispersion-limited propagation distance, as discussed in Section 2.4. Figure 2.14(a) shows the optical spectrum and the optical intensity eye diagram of an idealized NRZ signal. The NRZ optical spectrum is composed of a continuous portion, which reflects the shape of the individual NRZ data pulses, and a strong discrete tone at the carrier wavelength. The residual tones at multiples of the bit rate RB are generally much weaker than the tone at the carrier frequency; their strength is determined by the spectrum of an individual NRZ data pulse.

Peter J. Winzer and Rene´-Jean Essiambre

52

As evident from Table 2.2, NRZ has a modest receiver sensitivity (fairly high required OSNR) and despite its low signal bandwidth exhibits a comparatively low tolerance to narrow-band optical filtering, which is mainly due to the pronounced impact of ISI on this format, arising whenever a symbol waveform spreads across symbol boundaries and affects detection of the neighboring symbols. The inset to Figure 2.15 shows the onset of ISI in a measured 10.7-Gb/s NRZ waveform, becoming more pronounced under additional optical or electrical filtering. Regarding its robustness to fiber nonlinearity, NRZ generally performs well at 10 Gb/s as long as optimized dispersion maps are used [168]. The optimum dispersion map is largely determined by single-channel transmission, with crossphase modulation (XPM) and four-wave mixing (FWM) forcing a reduction of the signal power by a few dB over single-channel transmission for most installed fibers. At 40 Gb/s and 100 Gb/s, advanced modulation formats have generally better nonlinear transmission properties [118]. It is important to note that the resistance to fiber nonlinearity depends greatly on the exact system parameters such as signal power, distance, dispersion map, fiber type, channel spacing, or on the level of copropagating noise (especially in the case of phase shift keyed formats). In this context, all comments made in this chapter on the nonlinear transmission performance of various formats should be interpreted as guidelines (or “rules of thumb”) rather than having universal scope.

2.5.2 Vestigial Sideband and Single Sideband Modulation When narrow-band optical filtering is desired, as is the case in optically routed networks at high SEs one can take advantage of the fact that real-valued baseband waveforms have symmetric9 spectra around the carrier frequency [84]. Hence, one half of the spectrum can (at least in principle) be perfectly reconstructed from knowledge of the other half. As an important caveat, note that the vast majority of optical receivers still use square-law rather than coherent detection. To be a successful candidate for vestigial sideband (VSB) or single sideband (SSB) filtering, a real-valued modulation format should therefore maintain square-law detectability after conversion to VSB or SSB. In SSB modulation [84], one sideband is suppressed by modulating one quadrature of the optical field with the data signal itself and the other with the signal’s Hilbert transform. While the practical generation of high-quality SSB signals is difficult due to problems in implementing broadband Hilbert transformers [32, 64, 169, 170], VSB signals have been successfully realized in optical communications. A VSB signal is generated [84] by suppressing major parts of one sideband by the gradual roll-off of an optical VSB filter, offset by half its bandwidth from the

9

Strictly speaking, the spectral magnitude is symmetric while the spectral phase is antisymmetric.

2. Advanced Optical Modulation Formats

53

carrier frequency, while performing some filter action on the remaining sideband [cf. dashed filter characteristics in Figure 2.14(e, f)]. VSB-NRZ has been successfully transmitted at 40 Gb/s [33, 171–173] and 100 Gb/s [174, 175], and VSB-RZ [176] and VSB-CSRZ [177, 178] have been demonstrated at 40 Gb/s. Figure 2.14(e) and (f) shows optical spectra and intensity eye diagrams of VSB-NRZ and VSB-CSRZ. In a WDM system, VSB filtering can either be done at the transmitter (i.e., before or in combination with multiplexing the WDM channels) [174, 177, 178] or at the receiver (i.e., after or in combination with demultiplexing) [33, 171–173]. Filtering at the transmitter allows for ultimate spectral compression and spectrally efficient transmission. The advantage from VSB when filtered at the receiver comes from reduced WDM crosstalk for the desired sideband if unequal WDM channel spacings are employed [33, 171, 172, 173], however, at a somewhat lower overall SE. Nonlinear transmission of VSB and SSB formats is similar to that of their double-sideband counterparts [33, 176] for 20 Gb/s and above. At 10 Gb/s and below, the missing sideband has been shown to partly reconstruct itself due to fiber nonlinearity [33], limiting the SE advantage of VSB.

2.5.3 Duobinary Modulation (DB, PSBT) As an alternative to VSB filtering, the modulation spectrum of an optical signal can be compressed by correlative coding. Within the class of line-coded modulation, optical duobinary (DB) is the most frequently used format. It belongs to the subclass of partial response signaling [32, 40, 84, 121, 179, 180]. In optical communications, duobinary modulation [181, 182] is also referred to as phaseshaped binary transmission (PSBT) [101] and phased amplitude-shift signaling (PASS) [102]; although PSBT and PASS, like the combined amplitude phase shift signaling “CAPS” [99], are meant to comprise more general correlative coding rules between amplitude and phase, PSBT and PASS are mostly used as synonyms for DB. The state diagram of DB line coding is shown in Figure 2.9 together with example bit and symbol sequences. The generation of an optical DB signal is visualized in Figure 2.15. The transmitter first uses a differentially precoded version of the data signal [84, 183] at the input. The precoded data stream exhibits a level change for every 0 bit contained in the original data sequence and is used to let the squarelaw detected DB signal represent the original data sequence. Precoding should be implemented at the transmitter to prevent error propagation at the receiver [84, 121]. The precoded sequence is converted into a three-level electrical DB signal by means of severe electrical low-pass filtering, which can be implemented, e.g., as a delay-and-add filter (dashed filter transfer function in the inset to Figure 2.15) or a Bessel filter (solid in the inset). The 3-dB bandwidth of the DB low-pass filter should be about 25% of the bit rate. Delay-and-add filters typically result in a better back-to-back sensitivity of the optical DB signal,

54

Peter J. Winzer and Rene´-Jean Essiambre

while carefully tailored low-pass characteristics yield higher tolerances to CD at the expense of a typically 1- to 2-dB back-to-back sensitivity penalty relative to NRZ-OOK [184] (cf. Table 2.2). A typical low-pass filtered electrical signal, measured at 10.7 Gb/s, is shown as an inset to Figure 2.15. This three-level signal is used to drive a chirp-free MZM between its transmission maxima, where the outer levels generate the –1 symbols and the center level generates the 0 symbol. When detected by a square-law photodetector, the negative portion of the optical field is mapped on top of the positive portion, resulting in the binary optical intensity eye diagram shown as an inset to Figure 2.15 (measured at 10.7 Gb/s) and in Figure 2.14 (d, simulated). DB and NRZ are intimately related to each other, as visualized along with the measured eyes in Figure 2.15. Any NRZ eye diagram simultaneously exhibits an “NRZ eye opening” as well as a “DB eye opening.” Depending on whether detection takes place at the centers of the NRZ eye or at the TS/2-shifted centers of the DB eye, the receiver detects NRZ or DB. By narrow-band (optical or electrical) filtering of the NRZ signal, the NRZ eye gradually closes in while the DB eye opens up. Alternatively to using a separate LP filter preceding the MZM, a MZM designed for 1/4-th of the desired data rate may be used to combine the functionality of low-pass filtering and modulation [185–187], which has allowed MZMbased DB modulation up to 107 Gb/s [188]. Equivalently, low-pass filtering can be performed in the optical domain, with generally less ISI penalty introduced by amplitude and phase ripple of real-world, imperfect electrical prefiltering: passing a binary signal with levels {1, þ1}, i.e., a (D)PSK signal, through a narrow-band optical bandpass filter produces optical duobinary [189]. The main benefit of DB signals is their higher tolerance to CD and narrow-band optical filtering compared to binary signaling formats. This can be understood both in the time domain [102, 190] and in the frequency domain [32, 183, 184]. The time-domain explanation considers a “1 0 1” bit pattern, which for duobinary is encoded as “þ1 0 1”. If, due to CD or optical filtering, the two 1-bit pulses spread into the 0 bit between them, duobinary encoding lets them interfere destructively, thus maintaining a low 0-bit level. Conventional OOK formats, in contrast, let the pulses interfere constructively, which raises the 0-bit level and closes the eye. The frequency-domain explanation is based on the narrower spectral extent of properly filtered DB signals [cf. Figure 2.14(e)], which reduces CD-induced signal distortions.10 The spectral compression of DB results from the smoother “þ1 0 1” transitions of properly filtered duobinary as compared to the sharper “þ1 0 þ1” transitions of conventional OOK. Regarding the impact of nonlinear transmission at 10 Gb/s, DB exhibits roughly the same properties as OOK as long as the dispersion maps are optimized for each

10 Note that it is the entire spectral shape rather than just the 3-dB bandwidth that influences the robustness of a format to CD.

2. Advanced Optical Modulation Formats

55

format. However, the large dispersion tolerance advantage of DB over other formats such as NRZ can be dramatically reduced after nonlinear transmission [191]. At 40 Gb/s, DB has a modestly better transmission performance compared to OOK [116, 118] but is inferior to some other line-coded formats such as AMI [116, 118] or other pseudo-multilevel formats [106].

2.5.4 Return-to-Zero Modulation (RZ and RZ-OOK)

Time

33% RZ

50% RZ

67% CSRZ

33% RZ Power Field

50% RZ

50% RZ

33% RZ

Optical power

Transmission

Like NRZ modulation, return-to-zero (RZ) modulation can be used in connection with almost any digital modulation format by having the modulation waveform take on a pulse-like shape within the symbol duration. This is visualized by measured eye diagrams in Figure 2.16, showing NRZ-OOK and RZ-OOK at 10 Gb/s, NRZ-DPSK and RZ-DPSK at 40 Gb/s, as well as NRZ-DQPSK and RZ-DQPSK at 100 Gb/s. As evident from Table 2.2, RZ formats generally require 1–3 dB less OSNR for a given BER than their NRZ equivalents, even for identical

67% CSRZ

+1 –1 +1

Drive signal ΔV

Time Data Laser

Data mod.

Pulse carver

10 Gb/s OOK

67% CSRZ

40 Gb/s DPSK

100 Gb/s DQPSK

Figure 2.16 Sinusoidally driven MZM as pulse carver for 33% duty-cycle RZ, 50% duty cycle RZ, and 67% duty cycle CSRZ. The MZM bias points are indicated by an open circle. Lower right: measured optical intensity eye diagrams for NRZ and RZ versions of OOK, DPSK, and DQPSK at various bit rates (this figure may be seen in color on the included CD-ROM).

56

Peter J. Winzer and Rene´-Jean Essiambre

receiver bandwidths [48, 192–194]. For beat noise limited direct-detection receivers, this is mostly due to the reduced impact of ISI on RZ formats. As evident from Figure 2.14, the OSNR advantage of RZ waveforms comes at the expense of a broader optical spectrum. However, experiments at high SEs have shown [74, 75] that optical prefiltering can be advantageously used to convert an RZ signal back to NRZ while at the same time introducing much less ISI than would be found by passing an NRZ signal through the same narrow-band OF. This phenomenon is due to the fact that OFs generally possess smoother and better defined transfer characteristics than electrical filters across the same absolute bandwidth. Using RZ symbol waveforms as the basis for a modulation format has important consequences on nonlinear transmission. The signal spectrum is broader than its NRZ equivalent, which generally favors nonlinear transmission, especially at high speeds of 10 Gb/s and above [125]. This fact is discussed further in the context of nonlinear transmission impairments (cf. Figure 2.21). RZ transmitters can be implemented either by electronically generating RZ drive waveforms, which are then modulated onto an optical carrier, or by carving out pulses from an NRZ signal using an additional modulator, called pulse carver. While the first option [195] is feasible up to data rates of 10 Gb/s with today’s technology [196, 197], a pulse carver has to be employed at 40 Gb/s and beyond. Typically, pulse carvers are implemented as sinusoidally driven EAMs or MZMs, since multiGHz sinusoidal signals of appreciable drive amplitudes are easily generated. Using sinusoidally driven EAMs, low duty cycle optical pulses can be realized. This makes EAM-based pulse carvers well suited for OTDM experiments [125, 198]. Due to the variable absorption characteristics and residual chirp of EAMs, advanced RZ modulation formats are typically implemented using MZM-based pulse carvers, where one of the following three carving methods is commonly employed (cf. Figure 2.16): • Sinusoidally driving a MZM at the data rate between the minimum and the maximum transmission results in optical pulses with a full-width at halfmaximum (FWHM) of 50% of the bit duration (a duty cycle of 50%). Decreasing the modulation swing while lowering the modulator bias to still reach good extinction between pulses, the duty cycle can in principle be reduced to 36%, however, with significant excess insertion loss, since the modulator is then no longer driven to its transmission maximum. • Sinusoidally driving a MZM at half the data rate between its transmission minima produces a pulse whenever the drive voltage passes a transmission maximum. This way, duty cycles of 33% can be realized. The pulses may be broadened at the expense of a reduced extinction ratio between pulses by lowering the drive voltage. • Sinusoidally driving a MZM at half the data rate between its transmission maxima results in pulses with 67% duty cycle and with alternating phase. The resulting format [199, 200] is called carrier-suppressed return-to-zero (CSRZ), cf. Section 2.5.5.

2. Advanced Optical Modulation Formats

57

Spectra and intensity eye diagrams of 50% duty cycle RZ (gray) and 33% duty cycle RZ (black), as produced by a MZM in push–pull operation, are shown in Figure 2.14(b).

2.5.5 Carrier-Suppressed Return-to-Zero As introduced in Section 2.2.3, CSRZ [199, 200] is a modulation format with memory, more specifically a pseudo-multilevel format. It is characterized by reversing the sign of the optical field at each bit transition. In contrast to formats using correlative coding, the sign reversals occur at every bit transition, and are completely independent of the information-carrying part of the signal. CSRZ is most conveniently realized by sinusoidally driving a MZM pulse carver at half the data rate between its transmission maxima, as visualized in Figure 2.16. Since the optical field transfer function TE (DV) (dashed) of the MZM changes its sign at the transmission minimum [cf. Eqn (2.9)], phase inversions between adjacent bits are produced. Thus, on average, the optical field of half the 1 bits has positive sign, while the other half has negative sign, resulting in a zero-mean optical field envelope. As a consequence, the carrier at the optical center frequency vanishes, giving the format its name. Since the optical phase in a CSRZ signal is periodic at half the data rate, the CSRZ spectrum exhibits characteristic tones at –RB/2. Spectrum and eye diagram of 67% duty cycle CSRZ, as generated by a MZM in push–pull configuration, are shown in Figure 2.14(c). Using a MZM to generate CSRZ results in a duty cycle of 67%, which can be brought down to 50% at the expense of excess insertion loss by reducing the drive voltage swing. It is important to note that, due to its most widely used practical implementation with MZMs, the duty cycle of CSRZ signals usually differs from the ones of standard RZ. Thus, care has to be taken when comparing the two formats, since some performance differences result from the carrier-suppressed nature of CSRZ, while others simply arise from the different duty cycles. When used in conjunction with a priori carrier-free formats (such as (D)PSK) the carrier-suppressing effect of CSRZ does not come into play. A CSRZ pulse carver then merely generates 67% duty cycle versions of the underlying modulation formats, where the CSRZ phase flips only impact the data encoding onto the format. For example, applying a CSRZ pulse carver to a DPSK signal generates 67% RZ-DPSK with inverted data logic, sometimes still referred to as “CSRZDPSK.” Regarding its transmission performance, CSRZ generally provides increased resistance to fiber nonlinearity relative to plain OOK formats at 40 Gb/s [27, 201]. More advanced pseudo-multilevel formats, e.g., using the alternate /2 modulation [106] formats discussed in Section 2.2.3, can show superior nonlinear transmission performance than CSRZ [107]. An advantage of CSRZ is that it can be strongly filtered [202] or even VSB-filtered [178] while largely preserving its good transmission properties [203, 204].

58

Peter J. Winzer and Rene´-Jean Essiambre

2.5.6 Chirped Return-to-Zero If a symbol waveform contains a controlled amount of analog phase modulation rather than being purely real-valued, the qualifier chirped is added to describe the resulting modulation format. In the case of chirped return-to-zero (CRZ) [205, 206] or alternatechirp return-to-zero (ACRZ) [145, 207], bit-synchronous periodic chirp spectrally broadens the signal bandwidth. Although this reduces the format’s suitability for high SE WDM systems, it generally increases its robustness to fiber nonlinearity [208, 209]. CRZ is predominantly used in ultralong-haul point-to-point fiber communications, as found in transoceanic (submarine) systems [209–212], with a typical phase modulation amplitude around 1 rad [205]. Submarine systems typically use lowdispersion fibers with nearly full periodic dispersion compensation [212]. A significant value of uncompensated dispersion slope remains in these transmission lines [209]. As a result, the dispersion map varies widely across the amplification band. The ability to generate a tunable chirp at the transmitter by using CRZ accommodates the variations of dispersion maps across the amplification band and, at the same time, broadens the spectrum which improves nonlinear transmission (similarly to pseudo-linear transmission at higher bit rates or with shorter pulses). The format CRZ can generally outperform NRZ and RZ in the context of submarine systems [209, 212]. CRZ signals are typically generated by sinusoidally modulating the phase of a RZ signal at the symbol rate using a separate phase modulator. This leads to a complex, three-modulator transmitter architecture, which requires careful synchronization of the three drive signals. Integrated GaAs/AlGaAs modulators for CRZ, combining NRZ data modulator, RZ pulse carver, and CRZ phase modulator in one package, have been reported [213]. Alternative implementations that take advantage of the phase-modulation properties of dual-drive MZMs [cf. phase term in Eqn (2.9)] have also been demonstrated [145].

2.5.7 Binary Differential Phase Shift Keying (DBPSK or DPSK) As introduced in connection with Figure 2.5, modulation formats that multiply a generally complex symbol waveform by an alphabet of complex numbers that all lie on the unit circle in the complex plane are referred to as phase shift keying (PSK) formats. Due to the dominance of square-law detection in optical communications today, the vast majority of phase modulated systems employs delay demodulation or differential detection (cf. Section 2.4), and the formats are thus referred to as differential phase shift keying (DPSK) [214]. Differential binary phase shift keying (DBPSK, or often just “DPSK”) encodes information on the binary phase change between two adjacent bits: A 1 bit is encoded onto a  phase change, whereas a 0 bit is represented by the absence of a phase change. To avoid error propagation at the receiver, differential precoding is used at the transmitter, in analogy to a DB transmitter [84, 121]. Note that differential

2. Advanced Optical Modulation Formats Im{E}

59 Im{E}

Re{E}

Re{E}

√2 1 (a) OOK

√2

(b) DPSK

Figure 2.17 diagrams for (a) OOK and (b) DPSK. The 3-dB sensitivity advantage of DPSK is pffiffiSymbol ffi due to the 2-increased symbol spacing for equal average optical power. The dashed double arrows in (b) illustrate two possible phase modulator implementations.

precoding can be omitted in laboratory experiments using pseudo-random bit sequences (PRBS), since differential precoding does not change a PRBS [215]. The main advantage from using DPSK instead of OOK comes from a 3-dB receiver sensitivity improvement [9, 216], which can be intuitively understood from pffiffiffi Figure 2.17, showing that the symbol spacing for DPSK is increased by 2 compared to OOK for fixed average pffiffiffi optical power [214]. This increased symbol distance makes DPSK accept a 2 larger standard deviation of the optical field noise than OOK for equal BER, which translates into a 3-dB reduction in the required OSNR. Depending on modulation waveforms, extinction ratios, and optical as well as electrical filters, the gain of DPSK can also exceed 3 dB in practice (cf. NRZ-OOK and NRZ-DPSK in Table 2.2). Like for OOK a pulse carver can be inserted to convert the NRZ-DPSK signal to RZ-DPSK (cf. Figure 2.16). However, special care has to be taken to avoid chirp from imperfect pulse carving, which impacts DPSK much more severely than OOK [217]. To perform optical phase modulation, one can either use a straight-line phase modulator or a MZM [214, 218]. The difference between the two phase modulation schemes is indicated by the dashed double arrows in Figure 2.17(b): a straight-line phase modulator modulates the phase along the unit circle in the complex plane, leaving constant the intensity of the phase-modulated light. However, since the optical phase directly follows the electrical drive signal in this modulation scheme, the speed of phase transitions is limited by the combined bandwidth of driver amplifier and phase modulator, and any overshoot or ringing in the drive waveform manifests itself in phase distortions. A MZM, symmetrically driven around zero transmission, modulates along the real axis through the origin of the complex optical field plane [cf. Figure 2.17(b)], which always produces exact  phase jumps at the expense of residual, pattern-dependent optical intensity dips at the locations of the phase transitions (cf. the NRZ-DPSK waveform in the upper right corner of Figure 2.15). These intensity dips get narrower with increasing modulation bandwidth.11 Since exact phase modulation is more important for DPSK than 11 Exploiting DPSK (or DQPSK) intensity dips for data modulation on top of a separately imposed phase modulation has also been studied, predominantly under the name inverse RZ [219–221].

Peter J. Winzer and Rene´-Jean Essiambre

60

T

A

DI

B

–

(a)

DI transm.[dB]

a constant optical intensity, practical DPSK transmitters are most conveniently implemented using a MZM as a phase modulator [214]. Note, though, that the pattern-dependent intensity dips can have a negative impact on nonlinear transmission performance compared to modulation with a straight-line phase modulator, especially at low symbol rates such as at 10 Gbaud [222]. Figure 2.14(g) shows optical spectra and intensity eye diagrams for NRZ-DPSK generated by a MZM as a phase modulator (black) and 33% duty cycle RZ-DPSK (gray), respectively. Note the absence of a zero-intensity rail in the eye diagrams, which is characteristic of phase-modulated formats. The nonlinear transmission properties of DPSK vary greatly with system parameters. At 10 Gb/s, nonlinear transmission is similar to OOK in terms of the optimum signal launch power levels [223]. Due to the substantially reduced pattern-dependent IM of DPSK, one expects a reduced impact of XPM, which should favor DPSK at high SEs (‡ 0.4 bits/s/Hz) where the impact of XPM typically increases. Note however that XPM-induced nonlinear phase noise (NPN) [224] may partially offset this advantage of DPSK. At 40 Gb/s, DPSK has nonlinear transmission thresholds similar or slightly better than CSRZ, as shown in Fig. 2.1(a) [27]. Note, though, that these results are obtained for widely differing BER values, from near the FEC threshold to a few decades below. The performance assessment may be somewhat different if two systems of different lengths but operating at the same BER would be compared. Further optimization of dispersion maps may also impact a performance comparison. Finally, one should note that at equal resistance to nonlinear distortion, the linear sensitivity advantage of DPSK allows improved overall system performance. Figure 2.18(a) shows a balanced DPSK receiver. As discussed in Section 2.4.2, a DI is used to convert phase modulation into intensity modulation prior to photodetection. The delay is typically chosen equal to (or slightly shorter than) the bit duration, T <  TB (25 ps at 40 Gb/s). In practice, the differential delay is implemented by different physical path lengths within the DI. In addition, the DI path length difference has to be fine-tuned with subwavelength accuracy (i.e., on Orthogonal polarizations

A 1/T B

Optical frequency

PDF (PDλ)

(b)

(c)

Figure 2.18 (a) Balanced DPSK receiver using an optical delay interferometer (DI) to convert phase modulation to intensity modulation. (b) Periodic DI power transmission characteristics for the destructive port (A) and the constructive port (B). (c) Polarization-dependent interference conditions within the DI lead to a polarization-dependent shift of the DI transfer characteristic (solid vs dashed), quantified by the parameters PDF or PD .

2. Advanced Optical Modulation Formats

61

the order of 10 nm, corresponding to less than 0.1 fs, in the 1550-nm wavelength range) to control the interference conditions at the DI output [214, 225]. Maintaining good interference is the most critical aspect in the design of high-performance DPSK receivers, as discussed in more detail in the context of DQPSK receivers in Section 2.5.8. At the DI output port A (the “destructive port”), the two optical fields interfere destructively whenever there is no phase change, and constructively whenever there is a phase change between subsequent bits, in agreement with the differential precoding rule (cf. Figure 2.15). Due to energy conservation within the DI, the second DI output port B (the “constructive port”) yields the logically inverted data pattern. In principle, one of the two DI output ports is sufficient to fully detect the DPSK signal (“single-ended detection”). However, the 3-dB sensitivity advantage of DPSK is only seen for balanced detection [214, 226, 227]; as shown in Figure 2.18(a), a balanced receiver forms the difference of ports A and B to obtain the electrical decision variable. The reason for the superior performance of balanced detection compared to singleended detection is the non-Gaussian beat noise statistics in combination with the differential detection mechanism (as opposed to fixed threshold detection for OOK) [214, 226, 227]. Figure 2.18(b) shows, on a logarithmic scale, the sinusoidal transmission characteristics of ports A and B of a DI as a function of frequency. The laser carrier frequency is indicated by a dotted line. Both transmission spectra are periodic with period 1/T. The destructive DI output port A acts as a delay-andsubtract filter (high-pass characteristics to first order), while the constructive port B is a delay-and-add filter (low-pass characteristics to first order) [189]. Therefore, the formats seen at ports A and B are AMI and DB, respectively. Figure 2.19 summarizes the interesting relationship between the modulation formats discussed so far, all starting from the same binary transmit data stream. The apparent lack of symmetry in Figure 2.19 (no arrow from NRZ to DB) is due to the fact that

VSB

Lo

w-

Offset opt. filter

NRZ

t. f ilte r .h igh -pa

uiv

MZM bias@quadrature

ss)

Opt. filter (equiv. high-pass)

ss

C-NRZ

-pa

(eq

h Hig

Imbalanced modulator

DPSK

Binary data

Op

Opt. filter (equiv. low-pass)

Z pulse carving CSR

pa

ss

DB (PSBT)

AMI, DCS MZM bias@extinction

Figure 2.19 Relationship between NRZ (including C-NRZ and VSB-NRZ), DPSK, DB, and AMI. Arrows indicate how to transition from one format to another, starting from the same binary electrical data stream.

62

Peter J. Winzer and Rene´-Jean Essiambre

narrow-band optical filtering of NRZ-OOK results in an optical duobinary signal with three intensity levels [228], in analogy to the three-level electrical drive waveform shown in Figure 2.15. Such a signal cannot take advantage of the square-law detecting property of PD and requires special detection schemes [229–231].

2.5.8 Differential Quadrature Phase Shift Keying Differential quadrature phase shift keying (DQPSK) is the only true multilevel modulation format (more than one bit per symbol) that has received appreciable attention in optical communications so far [86–88, 90–92, 222, 232–236]. It transmits the four phase shifts {0, þ/2, /2, } at a symbol rate of half the aggregate bit rate. As in the case of DPSK, a DQPSK transmitter is most conveniently implemented by two nested MZMs operated as phase modulators, as shown in Figure 2.10(c) [91, 213]. The symbol constellations of the upper and lower modulator paths as well as at the modulator output are also shown, together with the symbol transitions. Using this transmitter structure, one first takes advantage of the exact -phase shifts produced by MZMs, independent of drive signal overshoot and ringing. Second, this transmitter structure requires only binary electronic drive signals, which are much easier to generate at high speeds than multilevel drive waveforms. As in the case of NRZDPSK, the NRZ-DQPSK signal generated with a nested MZM exhibits patterndependent intensity dips at the bit transitions [cf. Figure 2.14(h)], where the deeper dip corresponds to a phase transition of both I and Q components and the shallower dip represents a phase transition in either I or Q. Optionally, a pulse carver can be added to the structure to produce RZ-DQPSK as shown in Figure 2.16. Like at the transmitter, one also strives to work with binary electrical signals at the DQPSK receiver due to implementation benefits in high-speed electronics. At the receiver, the DQPSK signal is thus first split into two equal parts, and two balanced receivers of the form depicted in Figure 2.18(a), but with differently biased DIs, are used in parallel to simultaneously demodulate the two binary data streams contained in the DQPSK signal [90]. Note that the DI delay is in the order of the symbol duration for DQPSK demodulation, which is twice the bit duration. The price for using DQPSK is a six times lower tolerance to frequency drifts between transmit laser and DI compared to DPSK [214, 237]. For example, at 40 Gb/s and for a 1-dB penalty, DPSK tolerates –1.2 GHz of laser-to-DI frequency mismatch, whereas DQPSK only allows for –200 MHz. At 10 Gb/s, the tolerances are tighter by a factor of four, and at 100 Gb/s the tolerances are relaxed by a factor of 2.5. Note in this context that the end-of-life stability of wavelength-locked distributed feedback (DFB) lasers used in commercial optical transmitters amounts to –2.5 GHz, which requires feedback-controlled DI tuning within the receiver. Furthermore, any polarization-dependent frequency shift of the DI transfer characteristics [“PDF” or “PD ,” Figure 2.18(c)] has to be kept below the respective format’s tolerance to frequency offsets [214, 225, 237]. An innovative DI design that simultaneously

2. Advanced Optical Modulation Formats

63

demodulates both binary bit streams of a DQPSK signal mitigates the problem [238]. Also, digital signal processing can be used to circumvent the frequency stability issues, albeit at the expense of electronic processing hardware [239], as discussed in Chapter 4 of this book. Optical spectrum and intensity eye diagram for NRZ-DQPSK and 50% duty cycle RZ-DQPSK are shown in Figure 2.17(h). Note that the shape of the DQPSK optical spectrum is identical to that of DPSK, but the DQPSK spectrum is compressed in frequency by a factor of two due to the halved symbol rate for transmission at fixed bit rate. The compressed spectrum is beneficial for achieving high SEs in WDM systems [74, 75, 85, 88, 90, 234], with a single-polarization record of 1.6 b/s/Hz reported today [74, 75], and optically routed networking with concatenated OADMs reported at 1.0 b/s/Hz [240]. As discussed along with Table 2.2, the spectral compactness of DQPSK also increases the format’s tolerance to CD [90, 241], and the longer symbol duration compared to binary modulation formats makes DQPSK more robust to PMD [90, 241]. Furthermore, it is worth noting from Table 2.2 that DQPSK requires only 1–1.5 dB more OSNR than DPSK at poor BER (e.g., 103). At good BER (e.g., 1012), the OSNR gap between DPSK and DQPSK increases, and DQPSK approaches the performance of OOK [40]. Despite the more complex transmitter and receiver, the good OSNR performance at FEC error ratios makes DQPSK an attractive candidate for optically routed networks that require narrow optical signal spectra. In the nonlinear transmission regime, the performance of DQPSK depends on the specific system conditions of operation such as the presence of tight optical filtering [242], the dispersion map parameters [243] and the presence of neighboring WDM channels [222, 242, 244, 245]. In general, DQPSK suffers more than DPSK from NPN at the same bit rate due to the halved symbol rate and the doubled number of constellation points. More studies remain to be done to better understand all trade-offs associated with nonlinear DQPSK transmission.

2.5.9 Polarization-Multiplexed QPSK with Coherent Detection As detailed in Section 2.4.1, intradyne detection naturally lends itself to polarization multiplexing, which makes polarization-multiplexed QPSK at 10 Gbaud an attractive modulation solution for 40-Gb/s traffic (or 25 Gbaud modulation for 100 Gb/s traffic). The advantage of this approach lies in the reduced signal bandwidth, which for 40-Gb/s polarization-multiplexed QPSK is comparable to that of a 10-Gb/s OOK signal; narrow signal spectra are important in the context of optically routed networks with dense WDM channel packing. Furthermore, the possibility of phase-sensitive digital signal processing enabled by intradyne detection opens up elegant ways for the mitigation of transmission impairments such as CD, PMD, and partially even fiber nonlinearity. Note, though, that the OSNR requirements for polarization-multiplexed modulation are identical to those for a

Peter J. Winzer and Rene´-Jean Essiambre

64

system at twice the symbol rate but without polarization multiplexing, which follows directly from the definition of the OSNR, Eqn (2.11). Note also that at equal OSNR the signal launch power in each polarization is lower by 3 dB compared to single-polarization transmission at twice the symbol rate. Compared to DQPSK with delay demodulation, coherently detected QPSK shows a theoretical sensitivity advantage of 1.5–2.4 dB [40]. From a nonlinear transmission point of view, polarization multiplexing may add nonlinear depolarization, which results from XPM for polarization multiplexed OOK [246, 247]. Using phase-modulated formats such as QPSK can significantly reduce nonlinear depolarization [247]. Using coherent detection in principle allows for efficient compensation of some nonlinear effects [162, 163, 248], but effects such as nonlinear depolarization in the presence of PMD may prove more difficult to compensate. Detailed studies are expected to determine the nonlinear performance of polarization multiplexed signals over the next few years. From an implementation point of view, intradyne detection at multi-Gbaud rates relies on the availability of sufficiently fast analog-to-digital converters and digital signal processing engines. Recent research demonstrations at 10 Gbaud and up to 28 Gbaud have predominantly used real-time oscilloscopes instead of integrated A/D converters to sample a portion of the optical waveform for processing, performed off-line on personal computers [162, 163, 248]. Real-time demonstrations of intradyne detection have been performed at 4.4 Gbaud [15], with ASICs being currently developed for 10-Gbaud operation [163]. As a further trade-off aspect, we note that polarization-multiplexed QPSK with intradyne detection requires significantly more transponder hardware than direct detection, including an additional QPSK modulator at the transmitter to cover the second polarization as well as the coherent receiver optics (LO, 90 hybrids for signal-LO combination). Optoelectronic integration is expected to consolidate components and to allow for better integrated and hence more cost-effective systems in the future.

2.6 SYSTEM DESIGN AND ADVANCED MODULATION FORMATS From a system perspective, the performance trade-offs entering into the choice of a modulation format include • • • • • • •

Receiver sensitivity Resistance to fiber nonlinearity Tolerance to dispersion accumulation and dispersion map Resistance to PMD Achievable SE Resistance to concatenated filtering in optically routed networks Complexity and cost of terminal equipment

2. Advanced Optical Modulation Formats

65

The importance of each of these items varies widely between network types. Metropolitan area systems (reach of 300 km or less) generally emphasize the dispersion tolerance since the impact of fiber nonlinearity is often small for such systems. One should point out however that that some metropolitan as well as regional networks (300 to 1000 km) are composed of very lossy spans and include many ROADMs. For such networks, increasing the signal powers and operating in the nonlinear regime may allow the network to remain optically transparent. In long-haul (1000 to 3000 km) and ultralong-haul (>  3000 km) communication systems the emphasis is almost always put on receiver sensitivity and resistance to fiber nonlinearity. At high speeds (40 Gb/s and above), SE becomes particularly important, enabled by stateof-the-art OF technologies and laser frequency stability. PMD can also be an important issue with high-speed systems, especially on older fiber plant. To better understand the performance of optically amplified lightwave systems and how modulation formats impact system design, we now consider relevant performance criteria and transmission impairments. Many important impairments and modulation formats are summarized in Table 2.2 and discussed in the following sections. The implementation complexities of transmitter and receiver are summarized in the second and third column of the table.

2.6.1 Delivered and Required OSNR The ultimate measure of system performance is the BER after FEC decoding at the receiver. Typical requirements for the post-FEC BER for carrier-grade telecommunication equipment are on the order of or better than 1016. Enhanced FEC for terrestrial applications has a correction threshold of around 103, below which a post-FEC BER of better than 1016 can be guaranteed as long as errors occur statistically independently of each other. Note that this prerequisite for achieving a theoretically expected FEC performance is often only assumed in experimental practice, either because FEC chips cannot be used and pre-FEC error ratios have to be measured, or because post-FEC error ratios cannot be measured over sufficiently long time intervals to reliably predict BERs on the order of 1016. To link BER measurements to system design, in which the dominant random signal impairment is noise from in-line optical amplifiers (ASE), one typically uses the optical signal-to-noise ratio (OSNR). The OSNR is defined as OSNR =

P ; 2 BRef NASE

(2.11)

where P is the average signal power (in both polarizations for polarization-multiplexed systems), BRef is an optical reference bandwidth (typically chosen as 0.1 nm, or 12.5 GHz at 1550 nm), and NASE is the power spectral density of the ASE in each polarization. Another quantity commonly used in digital communication is the SNR per bit [40]. It is defined as the ratio Eb / N0, where Eb is the energy

66

Table 2.2 Overview of modulation formats and some performance values at 42.7 Gb/s, using direct detection (required OSNR at BER = 103) [38]. OSNRReq (dB)

TX complexity

RX complexity (direct detection)

NRZ-OOK 50% RZ-OOK 67% CSRZ-OOK DB VSB-NRZ-OOK VSB-CSRZ NRZ-DPSK 50% RZ-DPSK NRZ-DQPSK 50% RZ-DQPSK

1 MZM 1–2 MZMs 2 MZMs 1 MZM 1 MZM þ 1 OF 2 MZMs þ 1 OF 1 MZM 1–2 MZMs 2 QMs 2 QMs þ 1 MZM

1 PD 1 PD 1 PD 1 PD 1 PD 1 PD 1 DI þ 2 PDs 1 DI þ 2 PDs 2 DIs þ 4 PDs 2 DIs þ 4 PDs

Back-toback 15.9 14.4 14.9 16.6 16.4 14.8 11.7 11.1 13.2 12.2

10 OADMs (0.4 b/s/Hz) 18.2 15.8 14.2 14.2 15.6 14.7 12.1 11.5 12.6 12.0

5 OADMs (0.8 b/s/Hz) n/a n/a n/a 18.4 17.3 16.7 17.6 17.0 12.9 12.0

CD (ps/nm) (2-dB penalty)

DGD (ps) (1-dB penalty)

54 48 42 211 (152) 63 (155) 51 (154) 74 (161) 50 (161) 168 (176) 161 (186)

8 10 11 6 6 11 10 10 20 21

Peter J. Winzer and Rene´-Jean Essiambre

Modulation format

2. Advanced Optical Modulation Formats

67

per bit and N0 is the noise power spectral density. The relation between the SNR and OSNR is given by OSNR =

RS SNR; 2 BRef

(2.12)

where is the RS is the symbol rate. To specify the tolerance of a modulation format to ASE, the required OSNR (OSNRReq) is introduced. It specifies the minimum necessary OSNR that is required to achieve a certain target BER. The fourth column of Table 2.2 lists the required OSNR, based on system simulations described in Ref. 38. The assumed rate of 42.7 Gb/s is representative of a 40-Gb/s per-channel bit rate, including a 7% FEC overhead. Since enhanced FEC schemes for multi-Gigabit/s optical communications are able to correct BER values of 103 to values below 1016 [4], Table 2.2 is based on 103 as the target BER for stating OSNRReq. Actually measured values for OSNRReq may differ somewhat from the numbers given in Table 2.2 due to various optical and electronic hardware implementation aspects at different bit rates, including drive waveforms, filter characteristics, and modulator extinction ratios. Nevertheless, the general trends and most of the quantitative differences predicted by simulation agree well with measurement. As evident from the table, differences in excess of 5 dB between various modulation formats can be observed. Note that in the context of system design, an improvement in the backto-back sensitivity of a modulation format only leads to improved system performance if the improved back-to-back performance can be retained after transmission. Finally, note that the required OSNR, in contrast to the SNR (or Eb / N0), scales linearly with bit rate, increasing by 3 dB for every doubling in bit rate. This is caused by the fixed reference noise bandwidth used in the definition of the OSNR. A bit rate independent performance parameter that is sometimes used in the context of optical communication systems is “photons/bit,” which is numerically equivalent to Eb / N0 in an optically preamplified receiver. Knowing the required OSNR that guarantees satisfactory system performance, system design needs to ensure that the OSNR delivered to the receiver is sufficiently high. The delivered OSNR (OSNRDel) is the OSNR at the end of a transmission line and depends only on line parameters and on the launched signal power but not on the modulation format or on transmission impairments. Evaluating signal and noise at the end of a transmission line composed of N identical amplification sections (“spans”), the delivered OSNR can be expressed on a logarithmic scale in the useful form OSNRDel



 1000 ðdBÞ = 10 log10 þ Pin  NF  LSP  10 log10 N; h BRef |fflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} 58 at = 1550 nm

(2.13)

68

Peter J. Winzer and Rene´-Jean Essiambre

where the factor 1000 captures the conversion of Watts to mW to have the signal power appear in “dBm.” The parameters Pin, NF, and LSP are the launch power (dBm), effective noise figure of a span (dB), and the span loss (dB).

2.6.2 System Penalties and System Margins The transmission of signals over fibers can suffer from a variety of impairments, such as uncompensated CD, PMD, optical filtering from ROADMs, and fiber nonlinearity. The effect of these impairments on transmission performance can be captured by the difference between the required OSNR after transmission and the required OSNR in the absence of the respective impairment. This difference is referred to as the OSNR penalty. In most cases, once the OSNR penalty has reached a few dB for a given impairment, one finds a rapid increase in penalty with any further increase in the impairment. In some cases, a signal corrupted by one transmission impairment may also become more vulnerable to other transmission impairments. One therefore strives to operate a system at low individual impairments (typically less than 2 dB per impairment). The OSNR margin is the ratio of delivered OSNR to required OSNR and accommodates various system penalties. In the early days of system design, the management of system penalties was performed with margin allocation tables [212, 249], similar to link budgets in RF communications, where a fixed penalty is associated with each transmission impairment as well as with a mismatch between system components and with component aging. In a simple treatment, the total allocated system margin is calculated by adding the penalties due to individual impairments, in analogy to a classical RF link budget. However, this margin allocation captures the improbable case where all impairments simultaneously have their worst-case impact. Also, it treats penalties as strictly additive, which is not always the case. Therefore, it is often preferable to apply a statistical treatment to margin allocations. In a statistical approach, one calculates the probability that the OSNR margin falls below a predetermined value and guarantees a maximum system outage probability, i.e., one specifies a very low yet finite probability that the system will not meet the allocated OSNR margin.

Filter Concatenation The fifth and sixth columns in Table 2.2 give, respectively, the required OSNR for the concatenation of 10 OADMs in a system suitable for 0.4 b/s/Hz SE, and 5 OADMs for 0.8 b/s/Hz SE [38]. Note that these simulation results represent the effect of filter concatenation only, as they were carried out for a single WDM channel. Strictly speaking, the values therefore apply to a WDM system using polarization interleaving, where coherent WDM crosstalk is negligible. Depending on the modulation format, coherent WDM crosstalk can lead to further penalties,

2. Advanced Optical Modulation Formats

69

especially at 0.8 b/s/Hz SE; properly capturing coherent WDM crosstalk in numerical simulations for various modulation formats is still a topic of active research [250, 251]. Another aspect of filter narrowing that is not taken into account in Table 2.2 is closely related to nonlinear fiber propagation: since OADMs can be spaced by several hundred kilometers in optically routed networks, a modulation format suitable for high-SE networks has to simultaneously lend itself to multiple passes through OADMs and to nonlinear fiber propagation over appreciable distances. The impact of joint impairments incurred in this scenario has been experimentally studied for 40- and 100-Gb/s systems with SEs between 0.4 and 1.0 b/s/Hz using a variety of modulation formats [27, 162, 203, 204, 240, 252, 253]. Chromatic Dispersion The seventh column in Table 2.2 quantifies the accumulated CD (ps/nm) that results in a 2-dB OSNR penalty at 42.7 Gb/s, assuming no OADMs and mux/demux bandwidths of 85 GHz. Most 40-Gb/s modulation formats exhibit dispersion tolerances on the order of 50 ps/nm, the exception being some spectrally narrow formats, which in general yield significantly better dispersion tolerance [32, 89]. Note that CD tolerance values, like back-to-back OSNR requirements, can depend to an appreciable extent on the waveforms and filters used in the system. Where applicable, the numbers in parentheses in the seventh column of Table 2.2 refer to the dispersion tolerance in a system with 5 OADMs at 0.8 b/s/Hz SE. Since filter narrowing curtails the signal spectrum, an increase in dispersion tolerance is typically observed. Only DB, tailored (by optimized electrical lowpass filtering at the receiver) for high dispersion tolerance in the absence of severe optical filtering, sees a reduction in dispersion tolerance (albeit an improvement in sensitivity) when tightly filtered. This discussion shows that the performance of a modulation format to various impairments cannot be taken in isolation, but has to be evaluated in the context of the system in which it is operating in. Another important example is the format-dependent shrinkage of dispersion tolerance in the presence of fiber nonlinearity, which, for example, is more pronounced for DB than for NRZ-OOK [38]. Advanced digital signal processing at the receiver can significantly increase the dispersion tolerance. For example, by using a MLSE at 10.7 Gb/s, an increase in dispersion tolerance by up to a factor of 2.6 has been demonstrated experimentally for NRZ-OOK [2, 254], with virtually unlimited tolerance if the MLSE complexity can be substantially increased [255]. Note, however, that the amount of increase in dispersion tolerance due to equalization depends on the equalization technique as well as on the modulation format [1, 2, 254, 256, 257]. For example, MLSE reception of DB improves the dispersion tolerance by a factor of 1.3 [254]. Furthermore, the use of coherent demodulation with subsequent digital signal processing can greatly improve dispersion tolerance, albeit at the expense of more complex digital signal processing hardware [163].

70

Peter J. Winzer and Rene´-Jean Essiambre

Finally, the CD values given in Table 2.2 apply for 42.7 Gb/s and scale quadratically with bit rate: a four-fold increase in bit rate is accompanied by a 16-fold reduction in dispersion tolerance. As a useful coincidence, and as an aid to remembering CD tolerances, the numerical value of the dispersion tolerance in ps/nm at 40 Gb/s is almost identical to the dispersion tolerance in km SSMF (D = 17 ps/nm) at 10 Gb/s. For example, if a signal tolerates 50 ps/nm at 42.7 Gb/s, it can propagate over roughly 50 km of SSMF at 10.7 Gb/s for the same OSNR penalty. Polarization Mode Dispersion The eighth column in Table 2.2 quantifies the tolerance of different modulation formats to first-order PMD [258]. It shows the differential group delay (DGD) (ps) that leads to a 1-dB OSNR penalty. For most modulation formats, a 1-dB penalty occurs at a DGD between 30% and 40% of the symbol duration, with RZ formats being in general more resilient to PMD than NRZ formats [258, 259]. Note that the resilience to PMD, in addition, depends to an appreciable extent on the waveforms [260, 261] and filters [260, 262], as well as on other residual distortions. The tolerance to first-order PMD scales linearly with symbol duration. Therefore, DQPSK has about twice the PMD tolerance of binary modulation formats at the same bit rate. The tolerance to first-order PMD shrinks linearly with bit rate. The main problem with PMD in optical fiber systems is its stochastic nature, letting the principal state of polarization (PSP) and the DGD vary on timescales between milliseconds (acoustic vibrations) and months (temperature variations of buried fiber) [263]. The rare occurrence of exceedingly high DGD values prohibits worst-case system design by allocating a fixed OSNR margin to accommodate all possible PMD-induced signal distortions. Instead, systems are allocated some reasonable margin (e.g., 1 dB), and the rare occurrence of DGDs exceeding the margin results in system outage. Properly specifying the outage probability is an important interface between fiber manufacturers, systems integrators, and service providers [258, 263]. If outage requirements cannot be met, PMD has to be compensated on a per-channel basis at the receiver, or the PMD tolerance of the respective modulation format has to be increased using optical [22] or electronic [2, 256, 264, 265] equalization or mitigation [266, 267] techniques. Fiber Nonlinearity The instantaneous Kerr nonlinearity in optical fibers [26] can lead to a host of elementary nonlinear interactions. Knowing which fundamental nonlinear interaction dominates transmission is helpful to conceive techniques that improve transmission, including advanced modulation formats, digital signal processing, and distributed optical nonlinearity management. Determining the fundamental nonlinear interactions distorting the signal in a specific system is an arduous task as it depends on a large number of fiber, signal, and system parameters. The most important fiber parameters impacting the nature

2. Advanced Optical Modulation Formats

71

Fiber nonlinearities Intrachannel Signal-noise NPN

Parametric amplification

SPM-induced NL phase noise

MI

Interchannel Signal-signal

Signal-noise

Signal-signal

SPM

NPN

WDM nonlinearities

Isolated pulse IXPM SPM

IFWM

XPM-induced NL phase noise

XPM

FWM

Figure 2.20 Classification of nonlinearities in optical fibers. Intrachannel and interchannel stand for nonlinearities occurring within or between WDM channels, respectively. SPM: self-phase modulation, (I)XPM: (intrachannel) cross-phase modulation, (I)FWM: (intrachannel) four-wave mixing, MI: modulation instability, NPN: nonlinear phase noise.

of nonlinear transmission are the fiber dispersion, nonlinear coefficient, dispersion slope, and PMD. For the signal it is the symbol rate, the modulation format, and the WDM channel spacing. The predominant system parameters are signal power, transmission distance and amplification scheme (determining the noise level throughout the system, especially important for phase shift keying). As a result of the enormous parameter space impacting nonlinear fiber transmission, accurate predictions of nonlinear transmission penalties require a combination of extensive system modeling with careful experimental measurements. Figure 2.20 summarizes the main fundamental nonlinear interactions in fiber-optic communication systems. Fiber nonlinearities that occur between different symbols of the same WDM channel or between a WDM channel and a spectrally overlapping second optical field (such as ASE or multipath interference, MPI) are referred to as intrachannel nonlinearities. When the nonlinear interactions occur between different WDM channels or between a WDM channel and a spectrally nonoverlapping second optical field, we speak of interchannel nonlinearities. The importance of each class of nonlinearities depends significantly on the per-channel symbol rate. As a “rule of thumb,” for systems using conventional periodic optical dispersion compensation (ODC) as a powerful tool for nonlinearity management, interchannel effects affect WDM systems most strongly at per-channel rates of 10 Gbaud and below, i.e. dispersion-managed solitons [268] and quasi-linear transmission regime (see discussion in Ref. 38), while intrachannel nonlinearities affect systems most strongly at rates above 10 Gbaud (i.e., pseudo-linear transmission [125]). The impact of fiber nonlinearity also depends on the local fiber dispersion: in general, lower-dispersion fibers have stronger interchannel effects than fibers with high local dispersion. These statements are quantified in Figure 2.21, based on extensive sets of numerical simulations for NRZ-OOK and RZ-OOK. The figure indicates the dominant nonlinear impairment as a function of fiber dispersion (y-axis) and bit rate (lower x-axis). Because Figure 2.21 applies to OOK formats, signal–signal nonlinear interactions are dominating transmission. For a given SE (upper x-axis) the figure reveals the bit rate per WDM channel (lower x-axis) that allows for the maximum transmitted energy per

Peter J. Winzer and Rene´-Jean Essiambre

72

Modulation format (OOK )

Fiber dispersion [ps/(km-nm)] SSMF NZDF

NRZ

20

RZ Spectral efficiency (bit/s/Hz) 0.025 0.1 0.4

{

15 10

XPM

SPM

IXPM

IFWM

{

5

0

FWM 2.5

10 40 Bit rate per channel (Gb/s)

160

Figure 2.21 Significance of inter- and intrachannel nonlinear impairments in WDM systems of different per-channel bit rates and OOK modulation. For high-speed TDM systems exceeding 10 Gb/s per channel, the dominant nonlinear interactions are IXPM and IFWM (this figure may be seen in color on the included CD-ROM).

bit over a distance of 1000–2000 km. A transition from NRZ-OOK to RZ-OOK as the optimum OOK format occurs around 20 Gb/s. In addition to the signal–signal interactions, we also find nonlinear signal–noise interactions in transmission. These interactions depend significantly on the noise level during propagation and become stronger if the OSNR is poor during propagation, which has become a common situation with the use of strong FEC [269]. The most important signal–noise interaction found in currently discussed transmission systems is SPM-induced nonlinear phase noise (SPM-NPN) [270–272] and XPM-induced NPN (XPM-NPN) [224]. These NPN phenomena are most important for modulation formats that use the optical phase to carry information (e.g., DPSK, DQPSK) and of less importance for formats where the optical phase does not carry information but is used for line coding purposes (e.g., CSRZ, DB). In general, the effect of NPN is largest when the signal waveform evolution is slow during propagation (e.g., when the symbol rate is low) or when fibers with low values of CD are used. Formats with a large number of phase levels are also generally more sensitive to NPN. One should also point out that NPN depends on both the signal and noise powers. Therefore, systems with identical signal power evolution but different noise power evolution (and most likely different delivered OSNRs) will generally have different OSNR penalties if they are affected by NPN. An example of the impact of SPM-NPN on transmission of DPSK at 42.7 Gb/s is shown in Figure 2.22. Each of the three plots shows the required OSNR as a function of input power to each of the 32 100-km spans of a nonzero dispersion shifted fiber that constitutes the link [273]. In Figure 2.22(a), a balanced detector is used while in Figure 2.22(b) and (c), single-ended detection is used.

2. Advanced Optical Modulation Formats

Intrachannel 16 nonlinearties only 14 12 –3

0 2 3 (a) Launch power pin (dBm)

20 18 16 14 12

Intrachannel nonlinearties only –3 0 2 3 (b) Launch power pin (dBm)

(dB)

22

–3

(dB)

Noise at Tx Noise distr. Noise at Rx

Constructive port

OSNRreq for BER = 10

OSNRreq for BER = 10

18

22

–3

Noise at Tx Noise distr. Noise at Rx

20

Destructive port

OSNRreq for BER = 10

22

–3

(dB)

Balanced detection

73

20

Noise at Tx Noise distr. Noise at Rx

18 16 14 12

Intrachannel nonlinearties only –3 0 2 3 (c) Launch power pin (dBm)

Figure 2.22 Required OSNR (103 BER) as a function of fiber launch power Pin for RZ-DPSK using (a) balanced detection and (b, c) single-ended detection: (b) destructive and (c) constructive port. The lower, middle, and upper curves of each graph, respectively, apply to the cases where noise is added just prior to the receiver, distributed along the line, and at the transmitter [273] (this figure may be seen in color on the included CD-ROM).

The middle curves on each graph represent the case of distributed noise, i.e., the physically realistic case where ASE is generated by an optical amplifier placed after each span. The lower and upper curves of each graph represent the hypothetical cases where the same amount of ASE is added at the receiver only (lowest curves) or at the transmitter only (uppermost curves). The difference between the lowest curves and the middle curves represents the SPM-NPN penalty. Figure 2.22(b) and (c) shows that the benefit of balanced detection over single-ended detection is preserved even for strong intrachannel nonlinearities and nonlinear phase noise. Figure 2.23 demonstrates how a system generally behaves in terms of nonlinearity, showing required and delivered OSNR at a target BER of 103 for signal power levels between 6 and 0 dBm. One can see that there is an optimum power (4 dBm in this example) that maximizes system reach (28 spans). At lower signal launch powers, the system is limited by ASE and hence by the delivered OSNR, while at higher powers it is limited by fiber nonlinearity. An example of an achievable distance (16 spans) for a certain OSNR margin (4 dB) at the power level of 2 dBm is also shown (bold curves). The system parameters for Figure 2.23 are single-channel transmission of 33% RZ-OOK at 42.8 Gb/s. The fiber is NZDF and the amplifier spacing is 100 km. Note that Figure 2.23 applies for a given system configuration and a given signal. Changing the system configuration can change both the delivered and required OSNR. Most system designs use a fixed launch power per channel which may not be the optimum power for a given optical network infrastructure (e.g., average span loss or maximum span length). An important difference between modulation formats is the required OSNR backto-back. In the absence of nonlinear effects (sufficiently low powers or sufficiently short distance), the signal remains undistorted after transmission and the improvement of the required OSNR back-to-back by a certain factor translates directly into a proportional increase in reach by the same factor. This can be seen by considering that the delivered OSNR in Eqn (2.13) increases “dB for dB” with Pin. For instance, if the required OSNR back-to-back improves by 3 dB and does not degrade with

Peter J. Winzer and Rene´-Jean Essiambre

74

26 0 dBm

OSNRDel OSNRReq Max. reach

OSNR (dB)

24 22

Limited by fiber nonlinearity

20 –6 dBm

18

Optimum signal power for maximum reach

OSNRMar

16 14

–6 dBm 0

5

10 15 20 Number of spans, N

25

Limited by noise

Figure 2.23 Required (lower curves) and delivered OSNR (upper curves) as a function of the number of spans for a range of per-channel signal launch powers. At a given power, the maximum number of spans can be bridged at the point where the required OSNR crosses the delivered OSNR (full circles). In this example, a maximum distance of 28 spans is achieved using 4 dBm per channel. The OSNR margin at 16 spans for a power of 2 dBm per channel is also shown (double-headed arrow) (this figure may be seen in color on the included CD-ROM).

25

D

De

OSNR (dB)

red

Syst

em A

d

re

ive

el

20

3 dBm –1 dBm

live

OEO

OEO Syst

em B

Margin

15

Required 10

0

5

10

15 20 Number of spans

25

30

Figure 2.24 Delivered and required OSNR evolution for two systems, A and B, that have the same backto-back required OSNR but different resistances to fiber nonlinearity. An increased distance is possible for system A. OEO: optical–electronic–optical (this figure may be seen in color on the included CD-ROM).

distance (linear system), the same BER can be achieved with 3 dB worse OSNR. Therefore, twice the number of optical amplifiers can be tolerated, and hence twice the number of spans can be bridged. As a result, in a system where transmission suffers less from fiber nonlinearity, the reach can be extended considerably. Many techniques can be used to reduce the impact of fiber nonlinearity, including advanced modulation formats, new fiber designs and advanced dispersion mapping. A comparison of two systems having identical back-to-back required OSNRs but different tolerances to fiber nonlinearity is represented in Figure 2.24. System B has a

2. Advanced Optical Modulation Formats

75

lower tolerance to nonlinearity and is limited in power to 1 dBm for a certain distance that guarantees a certain OSNR margin. System A suffers less from nonlinear distortions and can operate at þ4 dBm over a larger distance than system B at the same OSNR margin. In Figure 2.24, an improvement of 11–26 spans is brought by greater resistance to nonlinearity of system A, even though the required (backto-back) OSNR is the same in both cases.

2.7 SYSTEM TECHNOLOGIES AND MODULATION FORMATS In this section, we discuss the interplay between advanced modulation formats and optical and electronic system technologies. This topic is at the frontier of the understanding of systems and therefore is likely to evolve further over the next years.

2.7.1 Optical Dispersion Compensation (ODC) ODC is used in transmission lines to reduce the accumulation of dispersion so as to minimize signal distortions at the receiver. Probably the most important role of ODC, as an all-optical impairment mitigation technique, is its ability to be inserted periodically in the optical path for optimum mitigation of fiber nonlinearity. The process of finding the best location of the ODC is referred to as optimization of the dispersion map [125, 168]. The dispersion map of a transmission line specifies the precise accumulation of dispersion with distance, as visualized in Figure 2.25(a). A general principle of dispersion mapping relates to how various elementary nonlinear interactions (see Figure 2.20) are managed in the transmission line. For instance, one effect of dispersion mapping is to provide a partial cancelation of intrachannel nonlinearities (IXPM and IFWM) generated at different locations in a

Distance

Distance

(b)

Coherent detection full dispersion post-compensation Cum. dispersion

(a)

Electronic pre-distortion full dispersion pre-compensation Cum. dispersion

Cum. dispersion

Optical dispersion compensation dispersion map

(c)

Distance

Figure 2.25 Three dispersion maps associated with different transponder technologies: (a) optical dispersion compensation with dispersion map optimized for the mitigation of fiber nonlinearity; (b) electronic predistortion with full precompensation of dispersion at the transmitter; and (c) coherent detection with full compensation of dispersion at the receiver.

76

Peter J. Winzer and Rene´-Jean Essiambre

fiber segment or in the transmission line [125]. As discussed along with Figure 2.21, the relative importance of each elementary nonlinear interaction depends on the modulation format as well as on other system parameters (symbol rate, channel spacing, power levels, distance, etc.), making dispersion mapping generally depend on the system specifications.

2.7.2 Electronic Predistortions (EPD) Electronic predistortion (EPD) is a technique to generate arbitrary waveforms (amplitude and phase) at the transmitter [6, 274]. In combination with digital signal preprocessing, it allows the generation of a variety of modulation formats [275] and opens up the possibility of compensation of some propagation effects such as dispersion accumulation [276, 277] and fiber nonlinearity [276, 278]. Electronic predistortion enables the compensation of large amounts of fiber dispersion at the transmitter, which at a first glance could allow the design of transmission systems without ODC. The resulting “all-precompensation” dispersion map is shown in Figure 2.25(b). Similarly, an “all-postcompensation” dispersion map for coherent detection with digital signal processing is depicted in Figure 2.25(c). The impact of fiber nonlinearity on single-channel transmission of advanced modulation formats using EPD with full precompensation of dispersion (but no compensation of fiber nonlinearity) is shown in Figure 2.26 [279]. The system is composed of 14 spans 80 km of SSMF and operates at a bit rate of 10.7 Gb/s. The modulation formats considered are DB, 33% RZ-OOK, NRZ-DPSK, and 33% RZ-DPSK. Nonlinear phase noise is not taken into account here. Duobinary, shown in Figure 2.26(a), exhibits the lowest threshold to fiber nonlinearity. Higher signal launch powers are permissible for RZ-OOK and NRZDPSK, and the highest power levels can be used for RZ-DPSK. The low resistance of DB to fiber nonlinearity is consistent with this format’s slower waveform evolution during transmission compared to formats with broader spectra. In many cases, a faster waveform evolution helps to minimize signal distortions by distributing nonlinear distortions more uniformly across different bits [125]. Evaluating the impact of WDM transmission on nonlinear distortions for 10-Gb/s systems with large accumulated dispersion is more challenging than for systems with frequent ODC that limits the maximum excursion of dispersion. When using frequent ODC, and at symbol rates of 20 Gbaud and lower, variations in signal distortions between short and long pattern lengths and between different data patterns remain typically small. In contrast, when all the dispersion of the entire line is compensated at the transmitter, large variations in nonlinear distortions can be observed when changing the data patterns [276, 280]. This is represented in Figure 2.27 in the form of OSNR penalty histograms for the RZ-DPSK modulation format [279]. At low power per channel (2 dBm/ch), the average penalty is low (1 dB) with a moderate spread in penalty (1.5 dB). At

2. Advanced Optical Modulation Formats

Required OSNR (dB)

20

18

DB 0

18

77

+4

14

16 –1

14

12

+3 +2

–2 12 10

10

–3 Linear –24

–22

–20

–18

–16

8 –22

NRZ-DPSK 12

+4

12

–21

–20

(b)

–19

–18

RZDPSK +6

10 +3

10

6 –22

0

Linear 14

14

8

+1

(a)

16 Required OSNR (dB)

RZ

16

+1 0 Linear

+5

8

+2

+4 +3

6 (c)

Linear

0

+2 +1

(d)

4 –21 –20 –19 –18 –22 –21 –20 –19 –18 Dispersion pre-comp. at Tx (ns/nm) Dispersion pre-comp. at Tx (ns/nm)

Figure 2.26 Required OSNR (103 BER) after transmission as a function of dispersion precompensation for single-channel transmission: (a) duobinary (DB), (b) 33% RZ-OOK, (c) NRZDPSK and (d) RZ-DPSK. The labels refer to the signal launch power (dBm). “Linear” means “in the absence of fiber nonlinearity.” Nonlinear phase noise is not included (this figure may be seen in color on the included CD-ROM).

higher powers (3 dBm/ch and higher), the average penalty rises and the penalty spread increases considerably with different data (or time delays) between neighboring channels. The spread in penalty slowly shrinks with increasing pattern length [280]. Finally, we note that the compensation of transmission impairments at the transmitter generally requires a feedback loop between transmitter and receiver. As CD is varying slowly (if at all) in a lightwave system, the respective feedback loop can be slow. Since transmission distances can be considerable in optical networks, the feedback time (i.e., the round trip time between transmitter and receiver) can be as large as 10 ms for a 1000-km transmission line. Rapidly varying impairments that change over the timescale of the feedback loop (including the required number of iterations for the predistortion algorithms to converge) can therefore not be handled by EPD. Examples for such impairments are PMD or nonlinear distortions caused by sudden power transients in a network.

Peter J. Winzer and Rene´-Jean Essiambre

78

40

Number of occurrences

80 2 dBm/ch

3 dBm/ch

60

30

40

20

20

10

0

0

1

2

3

4

5

Number of occurrences

25

0

4 dBm/ch

20

1

2

3

4

5

6

10

4

5

2 0

2 4 6 8 Required OSNR penalty (dB)

5 dBm/ch

8

15

0

0

10

10

0

0

4 8 12 16 Required OSNR penalty (dB)

20

Figure 2.27 Histograms of required OSNR penalties (103 BER) of a channel in a 50-GHz spaced WDM system using RZ-DPSK at 10.7 Gb/s. Histogram bins are 0.4 dB wide on all graphs. A large spread in penalties from XPM is observed. At 5 dBm per channel, 10 timing realizations had penalties exceeding 20 dB and are shown on the right side of 6-dBm histogram. Nonlinear phase noise is not included (this figure may be seen in color on the included CD-ROM).

2.7.3 Coherent Detection and Digital Signal Processing at the Receiver Coherent detection is a technique that is particularly efficient to detect advanced multilevel modulation formats in quadrature signal space, since the optical field is directly translated into the electrical domain. Furthermore, as outlined in Sections 2.4.1 and 2.5.9, polarization multiplexing is a natural consequence of using coherent detection, and the use of elegant equalization strategies based on the optical field information is made possible. In contrast to conventional direct detection systems, both electronic predistortion and coherent detection use digital signal processing based on the optical field. This lets the compensation of transmission impairments exhibit some similarities between both techniques [141]. One notable difference however is that coherent detection does not require a transmitter-receiver feedback loop. Hence, rapidly varying impairments can also be handled by coherent detection.

2. Advanced Optical Modulation Formats

79

In the case of signal distortions from nonlinear propagation, the behavior of a system based on coherent detection in the absence of ODC is very similar to the case of EPD discussed in previous section. As shown in Figure 2.25(c), a dispersion compensation free system using coherent detection implements an “all-postcompensation” dispersion map, whereas EPD implements an “allprecompensation” map, both of which are, in general, not optimum for nonlinear transmission. Nonlinearity compensation using digital signal processing at the receiver can be based on backward propagation of single-channel nonlinear transmission [278, 280]. Knowing the transmission line parameters and the signal power evolution along the link, reverse propagation can be computed using the signal field detected at the receiver. Note, though, that the detected channel at the receiver is generally corrupted by noise and can be distorted not only by fiber nonlinearity but also by PMD. The simultaneous presence of multiple sources of impairments may limit the efficient compensation of fiber nonlinearity. Moreover, as discussed in previous sections, it is not possible in the context of optically routed networks to perform full compensation of interchannel nonlinearities such as XPM [280] or of signal-noise nonlinearities such as NPN.

2.8 CONCLUSION High-speed optical transmission systems are increasingly building on digital signal processing and digital communication techniques that are well established in RF communication systems, including efficient FEC, advanced modulation, line coding, polarization multiplexing, digital signal processing at both transmitter and receiver, and coherent detection. Nevertheless, an optically routed network exhibits important fundamental as well as technological differences from a radiofrequency system, and not all techniques that have proven successful in radiofrequency communication should be expected to work well for optical networks. Detailed studies over the next few years will screen digital communication techniques with respect to their applicability in high bit rate, high capacity, and inherently nonlinear optical networks, with the goal to lower the overall cost per end-to-end transmitted information bit in an environment of exponentially increasing needs for transport capacity.

ACKNOWLEDGMENTS The authors thank S. Bigo, S. Chandrasekhar, G. Charlet, A.R. Chraplyvy, C.R. Doerr, F. Fidler, G.J. Foschini, A.H. Gnauck, R. Griffin, T. Kawanishi, H. Kogelnik, G. Kramer, J. Proakis, G. Raybon, and R.W. Tkach for valuable discussions.

Peter J. Winzer and Rene´-Jean Essiambre

80

LIST OF ACRONYMS ACRZ AMI AP APD APol ASE ASK BER CAPS CD C-NRZ CoWDM CRZ CSRZ DB DCS DFE DGD DI DML DPSK DQPSK DST DWDM EAM EML EPD ETDM FEC FFE FSK FWHM FWM IFWM IM I/Q ISI IXPM LO MLSE MSK MZM

Alternate-chirp return-to-zero Alternate mark inversion Alternate phase Avalanche photodiode Alternate polarization Amplified spontaneous emission Amplitude shift keying Bit error ratio Combined amplitude phase shift signaling Chromatic dispersion Chirped non-return-to-zero Coherent wavelength division multiplexing Chirped return-to-zero Carrier-suppressed return-to-zero Duobinary Duobinary carrier suppressed Decision-feedback equalizer Differential group delay Delay interferometer Directly modulated laser Differential phase shift keying Differential quadrature phase shift keying Dispersion supported transmission Dense wavelength-division multiplexing Electroabsorption modulator Electroabsorption-modulated laser Electronic predistortion Electronic time division multiplexing Forward error correction Feed-forward equalizer Frequency shift keying Full-width at half-maximum Four-wave mixing Intrachannel four-wave mixing Intensity modulation In-phase/quadrature Intersymbol interference Intrachannel cross-phase modulation Local oscillator Maximum-likelihood sequence estimation Minimum shift keying Mach–Zehnder modulator

2. Advanced Optical Modulation Formats

NPN NRZ NZDF OCDMA ODC OEQ OF OFDM OOK OSNR OTDM PASS PD PMD PolSK PPM PSBT PSK QAM QM QPSK RF (R)OADM RZ SE SNR SPM SSB SSMF TCM TDM VSB WDM XFP XPM

81

Nonlinear phase noise Non-return-to-zero Nonzero dispersion fiber Optical code division multiple access Optical dispersion compensation Optical equalizer Optical filter Orthogonal frequency division multiplexing On/off keying Optical signal-to-noise ratio Optical time division multiplexing Phased amplitude-shift signaling Photodiode Polarization mode dispersion Polarization shift keying Pulse position modulation Phase-shaped binary transmission Phase shift keying Quadrature amplitude modulation Quadrature modulator Quadrature phase shift keying Radio frequency (Reconfigurable) optical add/drop multiplexer Return-to-zero Spectral efficiency Signal-to-noise ratio Self-phase modulation Single sideband Standard single-mode fiber Trellis-coded modulation Time division multiplexing Vestigial sideband Wavelength-division multiplexing Small form factor pluggable transponder (10 Gb/s) Cross-phase modulation

REFERENCES [1] T. Nielsen and S. Chandrasekhar, “OFC 2004 workshop on optical and electronic mitigation of impairments,” J. Lightw. Technol., 23(1), 131–142, 2005. [2] A. Fa¨rbert, S. Langenbach, N. Stojanovic et al., “Performance of a 10.7-Gb/s Receiver with Digital Equalizer using Maximum Likelihood Sequence Estimation,” in Proc. ECOC, page Th4.1.5, 2004.

82

Peter J. Winzer and Rene´-Jean Essiambre

[3] Q. Yu and A. Shanbhag, “Electronic data processing for error and dispersion compensation,” J. Lightw. Technol., 24, 4514–4525, 2006. [4] T. Mizuochi, K. Kubo, H. Yoshida et al., “Next Generation FEC for Optical Transmission Systems,” in Proc. OFC, page ThN1, 2003. [5] T. Mizuochi, “Recent progress in forward error correction and its interplay with transmission impairments,” IEEE J. Sel. Top. Quantum Electron., 12, 544–554, 2006. [6] D. McGhan, C. Laperle, A. Savchenko et al., “5120 km RZ-DPSK Transmission over G.652 Fiber at 10 Gb/s with no Optical Dispersion Compensation,” in Proc. OFC, page PDP27, 2005. [7] Y. Yamamoto, “Receiver performance evaluation of various digital optical modulationdemodulation systems in the 0.5–10 mm wavelength region,” IEEE J. Quantum Electron., 16(11), 1251–1259, 1980. [8] R. A. Linke and A. H. Gnauck, “High-capacity coherent lightwave systems,” J. Lightw. Technol., 6(11), 1750–1769, 1988. [9] G. Jacobsen, “Noise in Digital Optical Transmission Systems,” Artech House, 1994. [10] L. G. Kazovsky, S. Benedetto, and A. E. Willner, “Optical Fiber Communication Systems, Artech House, 1996. [11] S. Betti, G. De Marchis, and E. Iannone, “Coherent Optical Communications Systems,” 1995, John Wiley and Sons Inc. [12] S. Ryu, “Coherent Lightwave Communication Systems,” Artech House, 1995. [13] O. K. Tonguz and R. E. Wagner, “Equivalence between preamplified direct detection and heterodyne receivers,” IEEE Photon. Technol. Lett., 3(9), 835–837, 1991. [14] K. Kikuchi, “Coherent Detection of Phase-Shift Keying Signals using Digital Carrier-Phase Estimation,” in Proc. OFC, page OTuI4, 2006. [15] A. Leven, N. Kaneda, U.-V. Koch, and Y.-K. Chen, “Coherent Receivers for Practical Optical Communication Systems,” in Proc. OFC, page OThK4, 2007. [16] S. J. Savory, “Coherent Detection – Why is it Back?” in Proc. IEEE/LEOS Annual Meeting, Paper TUH1, 2007. [17] A. Hazneci and S. P. Voinigescu, “49-Gb/s, 7-tap transversal filter in 0.18 mm SiGe BiCMOS for backplane equalization,” IEEE CSICS, 2004, pp. 101–104. [18] H. Jiang and R. Saunders, “Advances in SiGe ICs for 40Gb/s Signal Equalization,” in Proc. OFC, page OTuE1, 2006. [19] B. Franz, D. Roesener, F. Buchali et al., “Adaptive Electronic Feed-Forward Equaliser and Decision Feedback Equaliser for the Mitigation of Chromatic Dispersion and PMD in 43 Gbit/s Optical Transmission Systems,” in Proc. ECOC, page We1.5.1, 2006. [20] S. Wada, R. Ohhira, T. Ito et al., “Compensation for PMD-Induced Time-Variant Waveform Distortions in 43-Gbit/s NRZ Transmission by Ultra-Wideband Electrical Equalizer Module,” in Proc. OFC, page OWE2, 2006. [21] M. Nakamura, K. Murata, and M. Tokumitsu, “Advances in 40 G Electronic Equalizers,” in Proc. OFC, page OThN6, 2007. [22] C. R. Doerr, S. Chandrasekhar, P. J. Winzer et al., “Simple Multi-Channel Optical Equalizer Mitigating Intersymbol Interference for 40-Gb/s Non Return-to-Zero Signals,” J. Lightw. Technol., 22(1), 249–256, 2004. [23] C. R. Doerr, P. J. Winzer, G. Raybon et al., “A Single-Chip Optical Equalizer Enabling 107-Gb/s Optical Non-Return-to-Zero Signal Generation,” in Proc. ECOC, page Th4.2.1, 2005. [24] P. J. Winzer, G. Raybon, C. R. Doerr et al., “107-Gb/s optical signal generation using electronic time-division multiplexing,” J. Lightw. Technol., 24, 3107–3113, 2006. [25] P. Hofmann, E. E. Basch, S. Gringeri et al., “DWDM Long Haul Network Deployment for the Verizon GNI Nationwide Network,” in Proc. OFC, page OTuP5, 2005. [26] Govind P. Agrawal, “Nonlinear Fiber Optics,” San Diego: Elsevier Science & Technology, 4th edition, 2006. [27] A. H. Gnauck, “Advanced Amplitude- and Phase Coded Formats for 40-Gb/s Fiber Transmission,” in Proc. IEEE/LEOS Annual Meeting, page WR1, 2004.

2. Advanced Optical Modulation Formats

83

[28] R. J. Essiambre et al., “Capacity Limits of Fiber Optic Communication Systems, in Proc. OFC, 2008. [29] H. Kogelnik, “On Optical Communication: Reflections and Perspectives,” in Proc. ECOC, page Mo1.1.1, 2004. [30] K. Schuh and E. Lach, “Recent advances in ultrahigh bit rate ETDM transmission systems,” J. Lightw. Technol., 24, 4455–4467, 2006. [31] F. Derr, “Optical QPSK transmission system with novel digital receiver concept,” IEE Electron. Lett., 27, 2177–2179, 1991. [32] J. Conradi, “Bandwidth-efficient modulation formats for digital fiber transmission systems,” In Optical Fiber Telecommunications IV B (I. Kaminow and T. Li eds), 2002, Academic Press, pp. 862–901. [33] S. Bigo, “Multiterabit DWDM terrestrial transmission with bandwidth-limiting optical filtering,” IEEE J. Sel. Top. Quantum Electron., 10(2), 329–340, 2004. [34] S. Bigo, G. Charlet, and E. Corbel, “What has Hybrid Phase/Intensity Encoding Brought to 40 Gbit/s Ultralong-Haul Systems,” in Proc. ECOC, page Th2.5.1, 2004. [35] P. J. Winzer and R.-J. Essiambre, “Advanced Optical Modulation Formats,” in Proc. ECOC, page Th2.6.1, 2003. [36] P. J. Winzer and R.-J. Essiambre, “System Trade-Offs and Optical Modulation Formats,” in Proc. OAA, page OTuC4, 2004. [37] P. J. Winzer, “Optical transmitters, receivers, and noise,” in Wiley Encyclopedia of Telecommunications (J. G. Proakis, ed), John Wiley & Sons, pp. 1824–1840. online: http:// www.mrw.interscience.wiley.com/eot/eot_sample_fs.html. [38] P. J. Winzer and R. J. Essiambre, “Advanced optical modulation formats,” in Proc. IEEE, 94, pp. 952–985, 2006. [39] G. P. Agrawal, “Lightwave Technology: Telecommunication Systems,” John Wiley and Sons, 2005. [40] J. G. Proakis, “Digital Communications,” Mc Graw Hill, 4th edition, 2001. [41] E. A. Lee and D. G. Messerschmitt, “Digital Communication,” Kluwer Academic Press, 1994. [42] B. Sklar, “Digital Communications,” Prentice Hall, 2th edition, 2001. [43] B. A. E. Saleh and M. C. Teich, “Fundamentals of Photonics,” John Wiley & Sons, 1991. [44] G. P. Agrawal, “Fiber-Optic Communication Systems,” John Wiley & Sons, 3rd edition, 2002. [45] C. E. Shannon, “A mathematical theory of communication,” The Bell System Technical Journal, 27, 379–423 and 623–656, 1948. [46] “Interfaces for the optical transport network (OTN),” Standard ITU-T G.709, 2003. [47] D. O. Caplan and W. A. Atia, “A Quantum Limited Optically-Matched Communication Link,” in Proc. OFC, page MM2, 2001. [48] P. J. Winzer, M. Pfennigbauer, M. M. Strasser, and W. R. Leeb, “Optimum filter bandwidths for optically preamplified RZ and NRZ receivers” J. Lightw. Technol., 19(9), 1263–1273, 2001. [49] P. J. Winzer and R.-J. Essiambre, “Optical Receiver Design Trade-Offs,” in Proc. OFC, page ThG1, 2003. [50] M. Pfennigbauer, M. Pauer, M. M. Strasser, and P. J. Winzer, “Dependence of optically preamplified receiver sensitivity on optical and electrical filter bandwidths measurement and simulation,” IEEE Photon. Technol. Lett., 14, 831–833, 2002. [51] G.-K. Chang, J. Yu, Y.-K. Yeo et al., “Enabling technologies for next-generation optical packetswitching networks” in Proc. IEEE, 94, 2006, pp. 892–910. [52] A. J. Phillips, R. A. Cryan, and J. M. Senior, “Novel Laser Intersatellite Communication System Employing Optically Preamplified PPM Receivers,” IEE Proc. Commun., 142, 1995, pp. 15–20. [53] D. O. Caplan, B. S. Robinson, R. J. Murphy, and M. L. Stevens, “Demonstration of 2.5-Gslot/s Optically-Preamplified M-PPM with 4 Photons/Bit Receiver Sensitivity,” in Proc. OFC, page PDP32, 2005. [54] J. Sinsky, A. Adamiecki, L. Buhl et al., “107-Gbit/s Opto-Electronic Receiver with Hybrid Integrated Photodetector and Demultiplexer,” in Proc. OFC, page PDP30, 2007.

84

Peter J. Winzer and Rene´-Jean Essiambre

[55] B. Mikkelsen, G. Raybon, R.-J. Essiambre et al., “320-Gb/s single-channel pseudolinear transmission over 200 km of nonzero-dispersion fiber,” IEEE Photon. Technol. Lett., 12, 1400–1402, 2000. [56] A. H. Gnauck, G. Raybon, P. G. Bernasconi et al., “1-Tb/s (6 170.6 Gb/s) transmission over 2000km NZDF using OTDM and RZ-DPSK format,” IEEE Photon. Technol. Lett., 15, 1618–1620, 2003. [57] M. Nakazawa, T. Yamamoto, and K. R. Tamura, “1.28 Tbit/s-70 km OTDM transmission using third- and fourth-order simultaneous dispersion compensation with a phase modulator,” IEE Electron. Lett., 36, 2027–2029, 2000. [58] H.-G. Weber, R. Ludwig, S. Ferber et al., “Ultrahigh-speed OTDM-transmission technology,” J. Lightw. Technol., 24, 4616–4627, 2006. [59] H.-J. Thiele and M. Nebeling editors, “Coarse Wavelength Division Multiplexing: Technologies and Applications,” CRC Press, 2007. [60] L. J. Cimini, “Analysis and simulation of a digital mobile channel using OFDM,” IEEE Trans. Commun., 33, 665–675, 1985. [61] A. J. Lowery, L. B. Du, and J. Armstrong, “Performance of optical OFDM in ultralong-haul WDM lightwave systems,” J. Lightw. Technol., 25, 131–138, 2007. [62] W. Shieh and C. Athaudage, “Coherent optical orthogonal frequency division multiplexing,” IEE Electron. Lett., 42, 587–589, 2006. [63] A. D. Ellis, F. G. C. Gunning, and T. Healy, “Coherent WDM: The Achievement of High Information Spectral Density Through Phase Control within the Transmitter,” in Proc. OFC, page OThR4, 2006. [64] T. Kawanishi, S. Sakamoto, and M. Izutsu, “High-speed control of lightwave amplitude, phase, and frequency by use of electrooptic effect,” IEEE J. Sel. Top. Quantum Electron., 13, 79–91, 2007. [65] T. Kawanishi, T. Sakamoto, S. Shinada et al., “LiNbO3 high-speed optical FSK modulator”, IEE Electron. Lett., 40, 691–692, 2004. [66] T. Sakamoto, T. Kawanishi, and M. Izutsu, “Continuous-phase frequency-shift keying with external modulation,” IEEE J. Sel. Top. Quantum Electron., 12, 589–595, 2006. [67] J. Mo, Y. J. Wen, Y. Dong et al., “Optical Minimum-Shift Keying Format and its Dispersion Tolerance.” in Proc. OFC, page JThB12, 2006. [68] S. Betti, G. De Marchis, and E. Iannone, “Polarization modulated direct detection optical transmission systems,” J. Lightw. Technol., 10(12), 1985–1997, 1992. [69] A. S. Siddiqui, S. G. Edirisinghe, J. J. Lepley et al., “Dispersion-tolerant transmission using a duobinary polarization-shift keying transmission scheme,” IEEE Photon. Technol. Lett., 14(2), 158–160, 2002. [70] S. Benedetto and P. Poggiolini, “Theory of polarization shift keying modulation,” IEEE Trans. Inf. Theory, 40, 708–721, 1992. [71] J. J. Lepley, J. G. Ellison, S. G. Edirisinghe et al., “Excess penalty impairments of polarization shift keying transmission format in presence of polarization mode dispersion,” IEE Electron. Lett., 36(8), 736–737, 2000. [72] A. Hodzic, B. Konrad, and K. Petermann, “Improvement of system performance in N 40-Gb/s WDM transmission using alternate polarizations,” IEEE Photon. Technol. Lett., 15, 153–155, 2003. [73] C. Xie, I. Kang, A. H. Gnauck et al., “Suppression of intrachannel nonlinear effects with alternatepolarization formats,” J. Lightw. Technol., 22(3), 806–812, 2004. [74] A. H. Gnauck, P. J. Winzer, L. L. Buhl et al., “12.3-Tb/s C-Band DQPSK Transmission at 3.2 b/s/ Hz Spectral Efficiency,” in Proc. ECOC, page Th4.1.2, 2006. [75] A. H. Gnauck, G. Charlet, P. Tran et al., “25.6-Tb/s CþL-band transmission of polarizationmultiplexed RZ-DQPSK signals,” in Proc. OFC, page PDP19, 2007. [76] L. E. Nelson and H. Kogelnik, “Coherent crosstalk impairments in polarization multiplexed transmission due to polarization mode dispersion,” Opt. Express, 7(10), 350–361, 2000. [77] Paul R. Prucnal, editor, “Optical Code Division Multiple Access: Fundamentals And Applications,” CRC Press, 2006. [78] H. G. Unger, “Planar Optical Waveguides and Fibers,” Clarendon Press: Oxford Engineering Science Series, 1977.

2. Advanced Optical Modulation Formats

85

[79] J. A. Buck, “Fundamentals of Optical Fibers,” John Wiley & Sons, 1995. [80] X. Shen, J. M. Kahn, and M. A. Horowitz, “Compensation for multimode fiber dispersion by adaptive optics,” Opt. Lett., 30, 2985–2987, 2005. [81] J. Kahn, “Compensating Multimode Fiber Dispersion using Adaptive Optics,” in Proc. OFC, page OTuL1, 2007. [82] H. R. Stuart, “Dispersive multiplexing in multimode optical fiber,” Science, 289(5477), 281–283, 2000. [83] S. Schoellmann, S. Soneff, and W. Rosenkranz, “10.7 Gb/s over 300 m GI-MMF using a 2 2 MIMO System Based on Mode Group Diversity Multiplexing,” in Proc. OFC, page OTuL2, 2007. [84] R. Gitlin, J. F. Hayes, and S. B. Weinstein, “Data Communications Principles,” Plenum Press, 1992. [85] G. Kramer, A. Ashikhmin, A. J. van Wijngaarden, and X. Wei, “Spectral efficiency of coded phase shift keying for fiber-optic communication,” J. Lightw. Technol., 21, 2438–2445, 2003. [86] M. Daikoku, I. Morita, H. Taga et al., “100 Gbit/s DQPSK Transmission Experiment Without OTDM for 100 G Ethernet Transport,” in Proc. OFC, page PDP36, 2006. [87] P. J. Winzer, G. Raybon, C. R. Doerr et al., “2000-km WDM Transmission of 10 107-Gb/s RZDQPSK,” in Proc. ECOC, 2006. [88] A. Sano, H. Masuda, Y. Kisaka, et al., “14-Tb/s (140 111-Gb/s PDM/WDM) CSRZ-DQPSK Transmission over 160 km using 7-THz Bandwidth Extended L-Band EDFA,” in Proc. ECOC, page Th4.1.1, 2006. [89] S. Walklin and J. Conradi, “Multilevel signaling for increasing the reach of 10-Gb/s lightwave systems,” J. Lightw. Technol., 17(11), 2235–2248, 1999. [90] R. A. Griffin and A. C. Carter, “Optical Differential Quadrature Phase Shift Key (oDQPSK) for High-Capacity Optical Transmission,” in Proc. OFC, page WX6, 2002. [91] R. A. Griffin, R. I. Johnstone, R. G. Walker et al., “10 Gb/s Optical Differential Quadrature Phase Shift Key (DQPSK) Transmission Using GaAs/AlGaAs Integration,” in Proc. OFC, page FD6, 2002. [92] M. Ohm, “Optical 8-DPSK and receiver with direct detection and multilevel electrical signals,” IEEE/LEOS Workshop on Advanced Modulation Formats, 2004, pp. 45–46. [93] M. Ohm and J. Speidel, “Quaternary optical ASK-DPSK and receivers with direct detection,” IEEE Photon. Technol. Lett., 15(1), 159–161, 2003. [94] S. Hayase, N. Kikuchi, K. Sekine, and S. Sasaki, “Proposal of 8-state per symbol (binary ASK and QPSK) 30-Gbit/s optical Modulation/Demodulation scheme,” in Proc. ECOC, page Th2.6.4, 2003. [95] K. Sekine, N. Kikuchi, S. Sasaki et al., “Proposal and Demonstration of 10-Gsymbol/s 16-ary (40 Gbit/s) Optical Modulation/Demodulation Scheme,” in Proc. ECOC, page We3.4.5, 2004. [96] T. Tokle, B. Zsigri, C. Peucheret, and P. Jeppesen, “Generation and transmission of 160 Gbit/s polarisation multiplexed RZ-DBPSK-ASK signal,” IEE Electron. Lett., 41, 433–435, 2005. [97] J. Hongo, K. Kasai, M. Yoshida, and M. Nakazawa, “1-Gsymbol/s 64-QAM coherent optical transmission over 150 km,” IEEE Photon. Technol. Lett., 19, 638–640, 2007. [98] K. A. S. Immink, “Codes for Mass Data Storage Systems,” Shannon Foundation Publishers, 2nd edition, 2004. [99] E. Forestieri and G. Prati, “Novel optical line codes tolerant to fiber chromatic dispersion,” J. Lightw. Technol., 19, 1675–1684, 2001. [100] B. Vasic, V. S. Rao, I. B. Djordjevic et al., “Ghost-pulse reduction in 40-Gb/s systems using line coding,” IEEE Photon. Technol. Lett., 16(7), 1784–1786, 2004. [101] D. Penninckx, M. Chbat, L. Pierre, and J.-P. Thiery, “The phase-shaped binary transmission (PSBT): A new technique to transmit far beyond the chromatic dispersion limit,” IEEE Photon. Technol. Lett., 9(2), 259–261, 1997. [102] J. B. Stark, J. E. Mazo, and R. Laroia, “Phased amplitude-shift signaling (PASS) codes: Increasing the spectral efficiency of DWDM transmission,” in Proc. ECOC, 1998, pp. 373–374. [103] C. N. Georghiades, “Modulation and coding for throughput-efficient optical systems,” IEEE Trans. Inform. Theory, 40, 1313–1326, 1994.

86

Peter J. Winzer and Rene´-Jean Essiambre

[104] S. Benedetto, G. Olmo, and P. Poggiolini, “Trellis coded polarization shift keying modulation for digital optical communications,” IEEE Trans. Inf. Theory, 43, 1591–1602, 1995. [105] H. Bulow, G. Thielecke, and F. Buchali, “Optical Trellis-Coded Modulation (oTCM),” in Proc. OFC, page WM5, 2004. [106] S. Randel, B. Konrad, A. Hodzˇic´, and K. Petermann, “Influence of bitwise phase changes on the performance of 160 Gbit/s transmission systems,” in Proc. ECOC, page P3.31, 2002. [107] D. M. Gill A. H. Gnauck et al., “/2 alternate-phase On/Off keyed 42.7 Gb/s long-haul transmission over 1980 km of standard single-mode fiber,” IEEE Photon. Technol. Lett., 16, 906–908, 2004. [108] J. Martensson, J. Li, A. Berntson et al., “Suppression of intra-channel four-wave mixing by phase modulation at one quarter of bit rate,” IEE Electron. Lett., 38, 1463–1465, 2002. [109] M. Forzati, A. Berntson, and J. Martensson, “IFWM suppression using APRZ with optimized phase-modulation parameters,” IEEE Photon. Technol. Lett., 16, 2368–2370, 2004. [110] S. Appathurai, V. Mikhailov, R. I. Killey, and P. Bayvel, “Investigation of the optimum alternatephase RZ modulation format and its effectiveness in the suppression of intrachannel nonlinear distortion in 40-Gbit/s transmission over standard single-mode fiber,” IEEE J. Select. Topics Quantum Electron., 10, 239–249, 2004. [111] M. Forzati, A. Berntson, J. Martensson, and R. J. Davies, “Performance analysis of single-MZM APRZ transmitters,” J. Lightw. Technol., 24, 2006–2014, 2006. [112] Y. Su, L. Moller, C. Xie et al., “Ultra high-speed data signals with alternating and pairwise alternating optical phases,” J. Lightw. Technol., 23, 26–31, 2005. [113] Y. Miyamoto, K. Yonenaga, A. Hirano et al., “Duobinary Carrier-Suppressed Return-to-Zero Format and its Application to 100-GHz Spaced 8 43-Gb/s DWDM Unrepeatered Transmission over 163 km,” in Proc. OFC, page TuU4, 2001. [114] Y. Miyamoto, A. Hirano, S. Kuwahara et al., “S-band 3 120-km DSF Transmission of 8 42.7Gbit/s DWDM Duobinary-Carrier- Suppressed RZ Signals Generated by Novel Wideband PM/ AM Conversion,” in Proc. OAA, page PD6, 2001. [115] A. Hirano and Y. Miyamoto, “Novel Modulation Formats in Ultra-High-Speed Optical Transmission Systems and their Applications,” in Proc. OFC, page ThM1, 2004. [116] K. S. Cheng and J. Conradi, “Reduction of pulse-to-pulse interaction using alternative RZ formats in 40-Gb/s systems,” IEEE Photon. Technol. Lett., 14, 98–100, 2002. [117] P. J. Winzer, A. H. Gnauck, G. Raybon et al., “40-Gb/s alternate-mark-inversion return-to-zero (RZ-AMI) transmission over 2000 km,” IEEE Photon. Technol. Lett., 15(5), 766–768, 2003. [118] A. H. Gnauck, X. Liu, X. Wei et al., “Comparison of modulation formats for 42.7-Gb/s singlechannel transmission through 1980 km of SSMF,” IEEE Photon. Technol. Lett., 16, 909–911, 2004. [119] J. Martensson, A. Berntson, A. Djupsjobacka et al., “Phase Modulation Schemes for Improving Intra-Channel Nonlinear Tolerance in 40 Gbit/s Transmission,” in Proc. OFC, Paper FE5, 2003. [120] A. Matsuura, K. Yonenaga, Y. Miyamoto, and H. Toba, “High-speed transmission based on optical modified duobinary signals,” IEE Electron. Lett., 35(9), 736–737, 1999. [121] P. Kabal and S. Pasupathy, “Partial-response signaling,” IEEE Trans. Commun., 23(9), 921–934, 1975. [122] X. Liu, X. Wei, A. H. Gnauck, C. Xu, and L. K. Wickham, “Suppression of intrachannel fourwave-mixing induced ghost pulses in high-speed transmissions by phase inversion between adjacent marker blocks,” Opt. Lett., 27, 1177–1179, 2002. [123] N. Alic´ and Y. Fainman, “Data-dependent phase coding for suppression of ghost pulses in optical fibers,” IEEE Photon. Technol. Lett., 16, 1212–1214, 2004. [124] L. K. Wickham, R.-J. Essiambre, A. H. Gnauck et al., “Bit pattern length dependence of intrachannel nonlinearities in pseudolinear transmission,” IEEE Photon. Technol. Lett., 16, 1591–1593, 2004. [125] R.-J. Essiambre, G. Raybon, and B. Mikkelsen, “Pseudo-linear transmission of high-speed TDM signals: 40 and 160 Gb/s.” In Optical Fiber Telecommunications IV, (I. Kaminow and T. Li, eds), Academic Press, 2002, pp. 232–304.

2. Advanced Optical Modulation Formats

87

[126] E. G. Shapiro, M. P. Fedoruk, S. K. Turitsyn, and A. Shafarenko, “Reduction of nonlinear intrachannel effects by channel asymmetry in transmission lines with strong bit overlapping,” IEEE Photon. Technol. Lett., 15, 1473–1475, 2003. [127] I. B. Djordjevic, S. K. Chilappagari, and B. Vasic, “Suppression of intrachannel nonlinear effects using pseudoternary constrained codes,” J. Lightw. Technol., 24, 769–774, 2006. [128] V. Pechenkin and F. R. Kschischang, “Constrained coding for quasi-linear optical data transmission systems,” J. Lightw. Technol., 24, 4895–4902, 2006. [129] T. L. Koch, “Laser sources for amplified and WDM lightwave systems.” In Optical Fiber Telecommunications III, (I. P. Kaminow and T. L. Koch, eds), Academic Press, 1997, pp. 115–162. [130] D. A. Ackermann, J. E. Johnson, L. J. P. Ketelsen et al., “Telecommunication lasers.” In Optical Fiber Telecommunications IV, (I. Kaminow and T. Li, eds), Academic Press, 2002, pp. 587–665. [131] K. Sato, S. Kuwahara, Y. Miyamoto, and N. Shimizu, “40 Gb/s direct modulation of distributed feedback laser for very-short-reach optical links.” IEE Electron. Lett., 38(15), 816– 817, 2002. [132] B. Wedding, W. Pohlmann, H. Gross, and O. Thalau, “43 Gbit/s transmission over 210 km SMF with a directly modulated laser diode,” in Proc. ECOC, page Mo4.3.7, 2003. [133] I. Tomkos, D. Chowdhury, J. Conradi et al., “Demonstration of negative dispersion fibers for DWDM metropolitan area networks” IEEE J. Select. Topics Quantum Electron., 7, 439–460, 2001. [134] B. Wedding, “New method for optical transmission beyond dispersion limit” IEE Electron. Lett., 28(14), 1298–1300, 1992. [135] B. Wedding, B. Franz, and B. Junginger, “10-Gb/s optical transmission up to 253 km via standard single-mode fiber using the method of dispersion-supported transmission,” J. Lightw. Technol., 12(10), 1720–1727, 1994. [136] S. Chandrasekhar, C. R. Doerr, L. L. Buhl et al., “Flexible Transport at 10 Gb/s from 0 to 675 km (11500 ps/nm) using a Chirp-Managed Laser and no DCF and a Dynamically Adjustable Dispersion-Compensating Receiver,” in Proc. OFC, page PDP30, 2005. [137] Y. Yu, R. Lewen, S. Irmscher et al., “80 Gb/s ETDM Transmitter with a Traveling-wave Electroabsorption Modulator,” in Proc. ECOC, page OWE1, 2005. [138] D. Walker, H. Sun, C. Laperle et al., “960-km transmission over G.652 fiber at 10 Gb/s with a laser/electroabsorption modulator and no optical dispersion compensation,” IEEE Photon. Technol. Lett., 17, 2751–2753, 2005. [139] C. R. Doerr, L. Zhang, P. J. Winzer et al., “Compact high-speed InP DQPSK modulator,” IEEE Photon. Technol. Lett., 19, 1184–1186, 2007. [140] A. Ougazzaden, C. W. Lentz, T. G. B. Mason et al., “40 Gb/s Tandem Electro-Absorption Modulator,” in Proc. OFC, page PD14, 2001. [141] D. McGhan, M. O’Sullivan, C. Bontu, and K. Roberts, “Electronic Dispersion Compensation,” in Proc. OFC, page OWK1, 2006. [142] S. K. Korotky, J. J. Veselka, C. T. Kemmerer et al., “High-speed, Low Power Optical Modulator with Adjustable Chirp Parameter,” Topial Meeting on Integrated Photonics Research, page TuG2, 1991. [143] A. H. Gnauck, S. K. Korotky, J. J. Veselka et al., “Dispersion penalty reduction using an optical modulator with adjustable chirp,” IEEE Photon. Technol. Lett., 3(10), 916–918, 1991. [144] H. Kim and A. H. Gnauck, “Chirp characteristics of dual-drive Mach-Zehnder modulator with a finite DC extinction ratio.” IEEE Photon. Technol. Lett., 14(3), 298–300, 2002. [145] P. J. Winzer, C. Dorrer, R.-J. Essiambre, and I. Kang, “Chirped return-to-zero modulation by imbalanced pulse carver driving signals,” IEEE Photon. Technol. Lett., 16(5), 1379–1381, 2004. [146] F. Heismann, S. K. Korotky, and J. J. Veselka, “Lithium niobate integrated optics: Selected contemporary devices and system applications,” In Optical Fiber Telecommunications III, (I. P. Kaminow and T. L. Koch, eds), Academic Press, 1997, pp. 377–462. [147] A. Mahapatra and E. J. Murphy, “Electrooptic modulators,” In Optical Fiber Telecommunications IV, (I. Kaminow and T. Li, eds), Academic Press, 2002, pp. 258–294.

88

Peter J. Winzer and Rene´-Jean Essiambre

[148] R. A. Griffin, R. G. Walker, and R. I. Johnstone, “Integrated devices for advanced modulation formats,” IEEE/LEOS Workshop on Advanced Modulation Formats, 2004, pp. 39–40. [149] M. Erman, P. Jarry, R. Gamonal et al., “Mach-Zehnder modulators and optical switches on III-V semiconductors” J. Lightw. Technol., 6(6), 837–846, 1988. [150] S. Akiyama, H. Itoh, T. Takeuch, et al., “Mach-Zehnder modulator driven by 1.2 V single electrical signal,” IEE Electron. Lett., 41(1), 40–41, 2005. [151] R. A. Griffin, B. Pugh, J. Fraser et al., “Compact, High Power, MQW InP Mach-Zehnder Transmitters with Full-Band Tunability for 10 Gb/s DWDM,” in Proc. ECOC, page Th2.6.2, 2005. [152] R. A. Griffin, A. Tipper, and I. Betty, “Performance of MQW InP Mach-Zehnder Modulators for Advanced Modulation Formats,” in Proc. OFC, page OTuL5, 2005. [153] K. Tsuzuki, H. Kikuchi, E. Yamada et al., “1.3-Vpp push-pull drive InP Mach-Zehnder Modulator Module for 40 Gbit/s Operation,” in Proc. ECOC, page Th2.6.3, 2005. [154] B. Culshaw and M. F. G. Wilson, “Integrated optic frequency shifter modulator.” IEE Electron. Lett., 17, 135–136, 1981. [155] M. Itsuzu, S. Shikama, and T. Sueta, “Integrated optical SSB modulator/frequency shifter.” IEE J. Quantum Electron., QE-17, 2225–2227, 1981. [156] Y. Muramoto, K. Yoshino, S. Kodama et al., 100 and 160 Gbit/s operation of uni-travellingcarrier photodiode module. IEE Electron. Lett., 40, 378–380, 2004. [157] A. Beling, H.-G. Bach, G. G. Mekonnen et al., “High-speed miniaturized photodiode and parallel-fed traveling-wave photodetectors based on inp.” IEEE J. Select. Topics Quantum Electron., 13, 15–21, 2007. [158] J. H. Sinsky, A. Adamiecki, C. A. Burrus et al., A “40-Gb/s integrated balanced optical front end and RZ-DPSK performance,” IEEE Photon. Technol. Lett., 8, 1135–1137, 2003. [159] K. Sato et al., “Record highest sensitivity of 28 dBm at 10 Gb/s achieved by newly developed extremely compact superlattice-APD module with TIA-IC,” in Proc. OFC, page FB11, 2002. [160] J. C. Campbell, “Recent advances in telecommunications avalanche photodiodes.” J. Lightw. Technol., 25, 109–121, 2006. [161] C. R. S. Fludger, T. Duthel, T. Wuth, and C. Schulien, “Uncompensated transmission of 86 Gbit/s polarisation multiplexed RZ-QPSK over 100 km of NDSF employing coherent equalisation,” in Proc. ECOC, page Th4.3.3, 2006. [162] C. R. S. Fludger, T. Duthel, D. van den Borne et al., “10 111 Gbit/s, 50 GHz spaced, POLMUX-RZ-DQPSK transmission over 2375 km employing coherent equalisation,” in Proc. OFC, page PDP22, 2007. [163] C. Laperle, B. Villeneuve, Z. Zhang et al., “Wave-Length Division Multiplexing (WDM) and Polarization Mode Dispersion (PMD) Performance of a Coherent 40 Gbit/s Dual-Polarization Quadrature Phase Shift Keying (DP-QPSK) Transceiver,” in Proc. OFC, page PDP16, 2007. [164] B. Mikkelsen, C. Rasmussen, P. Mamysnev, and F. Liu, “Partial DPSK with excellent filter tolerance and OSNR sensitivity,” IEE Electron. Lett., 42, 1363–1364, 2006. [165] C. Malouin, J. Bennike, and T. Schmidt, “DPSK receiver design optical filtering considerations,” in Proc. OFC, page OThK1, 2007. [166] S. Calabro, D. van den Borne, S. L. Jansen et al., “Improved Detection of Differential Phase Shift Keying Through Multi-Symbol Phase estimation,” in Proc. ECOC, page We4.P118, 2005. [167] X. Liu, “Receiver Sensitivity Improvement in Optical DQPSK and DQPSK/ASK Through DataAided Multi-Symbol phase Estimation,” in Proc. ECOC, page We2.5.6, 2006. [168] S. Bigo, “Modelling of WDM Terrestrial and Submarine Links for the Design of WDM Networks.,” in Proc. OFC, page OThD1, 2006. [169] S. Shimotsu, S. Oikawa, T. Saitou et al., “Single side-band modulation performance of a LiNbO3 integrated modulator consisting of four-phase modulator waveguides.” IEEE Photon. Technol. Lett., 13(4), 364–366, 2001. [170] P. M. Watts, V. Mikhailov, M. Glick et al., “Single sideband optical signal generation and chromatic dispersion compensation using digital filters.” IEE Electron. Lett., 40(15), 958–960, 2004. [171] W. Idler, S. Bigo, Y. Frignac et al., “Vestigial Side Band Demultiplexing for Ultra High Capacity (0.64 bit/s/Hz) Transmission of 128 40 Gb/s Channels,” in Proc. OFC, page MM3, 2001.

2. Advanced Optical Modulation Formats

89

[172] S. Bigo, Y. Frignac, G. Charlet et al., “10.2 Tbit/s (256 42.7 Gbit/s PDM/WDM) Transmission over 100 km TeralightTM Fiber with 1.28 bit/s/Hz Spectral Efficiency,” in Proc, OFC, page PD25, 2001. [173] Y. Frignac, G. Charlet, W. Idler et al., “Transmission of 256 Wavelength-Division and PolarizationDivision-Multiplexed Channels at 42.7 Gb/s (10.2 Tb/s capacity) over 3 100 km of TeraLightTM fiber,” in Proc. OFC, page FC5, 2002. [174] S. L. Jansen, R. H. Derksen, C. Schubert et al., “107-Gb/s Full-ETDM Transmission over Field Installed Fiber using Vestigial Sideband Modulation,” in Proc. OFC, page OWE3, 2007. [175] K. Schuh, E. Lach, B. Junginger et al., “8 107 Gbit/s Serial Binary NRZ/VSB Transmission over 480 km SSMF with 1 bit/s/Hz Spectral Efficiency and without Optical Equalizer,” in Proc. ECOC, Paper 2.3.1, 2007. [176] T. Tsuritani, A. Agata, I. Morita et al., “Performance Comparison Between DSB and VSB Signals in 20 Gb/s based Ultra-Long-Haul WDM Systems,” in Proc. OFC, page MM5, 2001. [177] P. Mamyshev, B. Mikkelsen, F. Liu et al., “Spectrally Efficient Pseudo Duo-Binary Modulation Formats for High-Speed Optical Data Transmission,” in Proc. Conf. on Lasers and Electrooptics, 2002. [178] A. Agarwal, S. Chandrasekhar, and R.-J. Essiambre, “42.7 Gb/s CSRZ-VSB for Spectrally Efficient Meshed Networks,” in Proc. ECOC, page We3.4.4, 2004. [179] A. Lender, “The duobinary technique for high-speed data transmission,” IEEE Trans. Commun. Electron., 82, 214–218, 1963. [180] A. Lender, “Correlative digital communication techniques,” IEEE Trans. Commun., 12(4), 128–135, 1964. [181] A. J. Price and N. LeMercier, “Reduced bandwidth optical digital intensity modulation with improved chromatic dispersion tolerance,” IEE Electron. Lett., 31(1), 58–59, 1995. [182] K. Yonenaga, S. Kuwano, S. Norimatsu, and N. Shibata, “Optical duobinary transmission system with no receiver sensitivity degradation,” IEE Electron. Lett., 31(4), 302–304, 1995. [183] T. Ono, Y. Yano, K. Fukuchi et al., “Characteristics of optical duobinary signals in Terabit/s capacity, high-spectral efficiency WDM systems,” J. Lightw. Technol., 16(5), 788–797, 1998. [184] S. Walklin and J. Conradi, “On the relationship between chromatic dispersion and transmitter filter response in duobinary optical communication systems.” IEEE Photon. Technol. Lett., 9(7), 1005–1007, 1997. [see also: Comments by D. Penninckx, pp. 902–903, 1998]. [185] H. Kim, G. Lee, H. Lee et al., “On the use of 2.5-Gb/s Mach-Zehnder modulators to generate 10-Gb/s optical duobinary signals.” IEEE Photon. Technol. Lett., 16(11), 2577–2579, 2004. [186] D. M. Gill, A. H. Gnauck, X. Liu et al., “42.7-Gb/s cost-effective duobinary optical transmitter using a commercial 10-Gb/s Mach-Zehnder modulator with optical filtering.” IEEE Photon. Technol. Lett., 17(4), 917–919, 2005. [187] P. J. Winzer, S. Chandrasekhar, C. R. Doerr et al., “42.7-Gb/s Modulation with a Compact InP Mach-Zehnder Transmitter,” in Proc. ECOC, page We1.6.1, 2006. [188] P. J. Winzer, G. Raybon, and M. Duelk, “107-Gb/s optical ETDM Transmitter for 100 G Ethernet Transport,” in Proc. ECOC, page Th4.1.1, 2005. [189] D. Penninckx, H. Bissessur, P. Brindel et al., “Optical Differential Phase Shift Keying (DPSK) Direct Detection Considered as a Duobinary Signal,” in Proc. ECOC, page We.P.40, 2001. [190] D. Penninckx, L. Pierre, J.-P. Thiery et al., “Relation between spectrum bandwidth and the effects of chromatic dispersion in optical transmissions,” IEE Electron. Lett., 32(11), 1023–1024, 1996. [191] W. Kaiser, M. Wichers, T. Wuth et al., “SPM limit of duobinary transmission,” in Proc. ECOC, page We7.2.2, 2000. [192] L. Boivin and G. J. Pendock, “Receiver Sensitivity for Optically Amplified RZ Signals with Arbitrary Duty Cycle.,” in Proc. OAA, page ThB4, 1999. [193] P. J. Winzer and A. Kalmar, “Sensitivity enhancement of optical receivers by impulsive coding,” J. Lightw. Technol., 17(2), 171–177, 1999.

90

Peter J. Winzer and Rene´-Jean Essiambre

[194] W. Idler, A. Klekamp, R. Dischler et al., “System Performance and Tolerances of 43 Gb/s ASK and DPSK Modulation Formats,” in Proc. ECOC, page Th2.6.3, 2003. [195] N. M. Froberg, G. Raybon, U. Koren et al., “Generation of 2.5 Gbit/s soliton data stream with an integrated laser-modulator transmitter,” IEE Electron. Lett., 30(22), 1880–1881, 1994. [196] Y. Kao, A. Leven, Y. Baeyens et al., “10 Gb/s Soliton Generation for ULH Transmission using a Wideband GaAs pHemt Amplifier,” in Proc. OFC, page FF6, 2003. [197] X. Liu and Y.-H. Kao, “Generation of RZ-DPSK using a Single Mach-Zehnder Modulator and Novel Driver Electronics,” in Proc. ECOC, page We3.4.2, 2004. [198] M. Suzuki, H. Tanaka, K. Utaka et al., “Transform-limited 14 ps optical pulse generation with 15 GHz repetition rate by InGaAsP electroabsorption modulator,” IEE Electron. Lett., 28(11), 1007–1008, 1992. [199] Y. Miyamoto, A. Hirano, K. Yonenaga et al., “320-Gbit/s (8 40 Gbit/s) WDM transmission over 367-km with 120-km repeater spacing using carrier-suppressed return-to-zero format,” IEE Electron. Lett., 35(23), 2041–2042, 1999. [200] A. Hirano, Y. Miyamoto, and K. Yonenaga, “40 Gbit/s L-band transmission experiment using SPM-tolerant carrier-suppressed RZ format,” IEE Electron. Lett., 35(25), 2213–2215, 1999. [201] A. Hodzˇic´, B. Konrad, and K. Petermann, “Alternative modulation formats in n 40 Gb/s WDM standard fiber RZ-transmission systems,” J. Lightw. Technol., 20, 598–607, 2002. [202] A. Agarwal, S. Banerjee, D. F. Grosz et al., “Ultra-high-capacity long-haul 40-Gb/s WDM transmission with 0.8-b/s/Hz spectral efficiency by means of strong optical filtering,” IEEE Photon. Technol. Lett., 15, 470–472, 2003. [203] G. Raybon, A. Agarwal, S. Chandrasekhar, and R.-J. Essiambre, “Transmission of 42.7-Gb/s VSB-CSRZ over 1600 km and Four OADM Nodes with a Spectral Efficiency of 0.8-bit/s/Hz,” in Proc. ECOC, page Mo3.2.7, 2005. [204] G. Raybon, “Performance of Advanced Modulation Formats in Optically-Routed Networks,” in Proc. OFC, 2006. [205] N. S. Bergano, “Undersea communication systems.” In Optical Fiber Telecommunications IV (I. Kaminow and T. Li, eds), Academic Press, 2002, pp. 154–197. [206] C. R. Menyuk, G. M. Carter, W. L. Kath, and R.-M. Mu, “Dispersion-managed solitons and chirped return to zero: What is the difference.” In Optical Fiber Telecommunications IV (I. Kaminov and T. Li, eds), Academic Press, 2002, pp. 305–328. [207] R. Ohhira, D. Ogasahara, and T. Ono, “Novel RZ Signal Format with alternate-Chirp for Suppression of Nonlinear Degradation in 40 Gb/s based WDM,” in Proc. OFC, page WM2, 2001. [208] E. A. Golovchenko, N. S. Bergano, C. R. Davidson, and A. N. Pilipetskii, “Modeling vs Experiments of 1610 Gbit/s WDM Chirped RZ Pulse Transmission Over 7500 km,” in Proc. OFC, page ThQ3, 1999. [209] B. Bakhshi, M. Vaa, E. A. Golovchenko et al., “Comparison of CRZ, RZ and NRZ Modulation Formats in a 64 12.3 Gb/s WDM Transmission Experiment Over 9000 km,” in Proc. OFC, page WF4, 2001. [210] N. S. Bergano, C. R. Davidson, and F. Heismann, “Bit-synchronous polarisation and phase modulation scheme for improving the performance of optical amplifier transmission systems,” IEE Electron. Lett., 32, 52–54, 1996. [211] E. A. Golovchenko, A. N. Pilipetskii, N. S. Bergano et al., “Modeling of transoceanic fiberoptic WDM communication systems.” IEEE J. Select. Topics Quantum Electron., 6, 337–347, 2000. [212] N. S. Bergano, “Wavelength division multiplexing in long-haul transoceanic transmission systems.” J. Lightw. Technol., 23, 4125–4139, 2005. [213] R. A. Griffin, R. G. Walker, R. I. Johnstone et al., “Integrated 10 Gb/s Chirped Return-to-Zero Transmitter using GaAs-AlGaAs Modulators,” in Proc. OFC, page PD15, 2001. [214] A. H. Gnauck and P. J. Winzer, “Optical phase-shift-keyed transmission.” J. Lightw. Technol., 23(1), 115–130, 2005. [215] M. Rice, S. Tretter, and P. Mathys, “On differentially encoded m-sequences.” IEEE Trans. Commun., 49, 421–424, 2001.

2. Advanced Optical Modulation Formats

91

[216] P. A. Humblet and M. Azizoglu, “On the bit error rate of lightwave systems with optical amplifiers.” J. Lightw. Technol., 9, 1576–1582, 1991. [217] P. J. Winzer, D. Stahl, and H. Kim, “Impact of Pulse Carver Chirp on RZ-DPSK Receiver Performance,” in Proc, ECOC, page We3.5.6, 2003. [218] T. Chikama, S. Watanabe, T. Naito et al., “Modulation and demodulation techniques in optical heterodyne PSK transmission systems,” J. Lightw. Technol., 8, 309–321, 1990. [219] T. Miyazaki and F. Kubota, “2-bit Per Symbol Modulation/Demodulation by DPSK over InverseRZ Optical Pulses,” in Proc. IEEE/LEOS Annual Meeting, page CThBB2, 2004. [220] T. Miyazaki and F. Kubota, “Superposition of DQPSK over inverse-RZ for 3-bit/symbol modulation demodulation.” IEEE Photon. Technol. Lett., 16, 2643–2645, 2004. [221] M. Zitelli, “Optical phase and intensity modulation using dark pulses.” IEEE Photon. Technol. Lett., 16, 1972–1974, 2004. [222] H. Kim and R.-J. Essiambre, “Transmission of 8 20 Gb/s DQPSK signals over 310-km SMF with 0.8-b/s/Hz spectral efficiency,” IEEE Photon. Technol. Lett., 15(5), 769–771, 2003. [223] J.-X.- Cai, D. G. Foursa, L. Liu et al., “RZ-DPSK field trial over 13 100 km of installed nonslope-matched submarine fibers,” J. Lightw. Technol., 23, 95–103, 2005. [224] H. Kim, “Cross-phase-modulation-induced nonlinear phase noise in WDM direct-detection DPSK systems,” J. Lightw. Technol., 21, 1770–1774, 2003. [225] P. J. Winzer and H. Kim, “Degradations in balanced DPSK receivers,” IEEE Photon. Technol. Lett., 15(9), 1282–1284, 2003. [226] P. J. Winzer, S. Chandrasekhar, and H. Kim, “Impact of filtering on RZ-DPSK reception.” IEEE Photon. Technol. Lett., 15(6), 840–842, 2003. [227] H. Kim and P. J. Winzer, “Nonlinear Phase Noise in Phase-Coded Transmission,” in Proc. OFC, page OThO3, 2005. [228] X. Gu and L. C. Blank, “10 Gbit/s unrepeated three-level optical transmission over 100 km of standard fibre.” IEEE Electron. Lett., 29, 2209–2211, 1993. [229] G. May, A. Solheim, and J. Conradi, “Extended 10 Gb/s fiber transmission distance at 1538 nm using a duobinary receiver,” IEEE Photon. Technol. Lett., 6, 648–650, 1994. [230] J. H. Sinsky, M. Duelk, and A. Adamiecki, “High-speed electrical backplane transmission using duobinary signaling,” IEEE Trans. Micro. Theo. Techn., 53(1), 152–160, 2005. [231] N. Alic, G. C. Papen, R. E. Saperstein et al., “Experimental demonstration of 10 gb/s NRZ Extended Dispersion Limited Reach over 600 km-SMF Link Without Optical Dispersion Compensation,” in Proc. OFC, page OWB7, 2006. [232] T. Tokle, C. R. Davidson, M. Nissov et al., “6500 km transmission of RZ-DQPSK WDM signals,” IEE Electron. Lett., 40(7), 444–445, 2004. [233] M. Ohm and T. Freckmann, “Comparison of different DQPSK transmitters with NRZ and RZ impulse shaping,” IEEE/LEOS Workshop on Advanced Modulation Formats, page ThB2, 2004. [234] A. H. Gnauck, P. J. Winzer, S. Chandrasekhar, and C. Dorrer, “Spectrally Efficient (0.8 b/s/Hz) 1-Tb/s (25 42.7 Gb/s) RZ-DQPSK Transmission over 28 100-km SSMF Spans with 7 Optical Add/Drops,” in Proc.” OFC, page Th4.4.1, 2004. [235] M. Cavallari, C. S. R. Fludger, and P. J. Anslow, “Electronic Signal Processing for Differential Phase Modulation Formats,” in Proc. OFC, page TuG2, 2004. [236] N. Yoshikane and I. Morita, “1.14 b/s/Hz spectrally efficient 50 85.4-Gb/s transmission over 300 km using copolarized RZ-DQPSK signals,” J. Lightw. Technol., 23(1), 108–114, 2005. [237] H. Kim and P. J. Winzer, “Robustness to laser frequency offset in direct-detection DPSK and DQPSK systems.” J. Lightwave Technol., 21(9), 1887–1891, 2003. [238] C. R. Doerr, D. M. Gill, A. H. Gnauck et al., “Simultaneous reception of both quadratures of 40-Gb/s DQPSK using a simple monolithic demodulator,” in Proc. OFC, page PDP12, 2005. [239] X. Liu, S. Chandrasekhar, A. H. Gnauck et al., “DSP-enabled compensation of demodulator phase error and sensitivity improvement in direct-detection 40-Gb/s DQPSK,” in Proc. ECOC, page Th4.4.5, 2006.

92

Peter J. Winzer and Rene´-Jean Essiambre

[240] P. J. Winzer, G. Raybon, S. Chandrasekhar et al., “10 107-Gb/s NRZ-DQPSK transmission at 1.0 b/s/Hz over 12 100 km including 6 optical routing nodes,” in Proc. OFC, page PDP24, 2007. [241] J. Wang and J. M. Kahn, “Impact of chromatic and polarization-mode dispersions on DPSK systems using interferometric demodulation and direct detection,” J. Lightw. Technol., 22(2), 362–371, 2004. [242] A. H. Gnauck, J. Sinsky, P. J. Winzer, and S. Chandrasekhar, “Linear Microwave-Domain Dispersion Compensation of 10-Gb/s Signals using Heterodyne Detection,” in Proc. OFC, page PDP31, 2005. [243] D. van den Borne, S. L. Jansen, G. D. Khoe et al., “A Comparison between Multi-Level Modulation Formats: 21.4-Gbit/s RZ-DQPSK and POLMUX-RZ-DPSK,” in Proc.” OFC, page OThR2, 2006. [244] T. Tokle, C. R. Davidson, M. Nissov et al., “Transmission of RZ-DQPSK over 6500 km with 0.66 bit/s/Hz Spectral Efficiency.” IEEE/LEOS Workshop on Advanced Modulation Formats, page ThA2, 2004. [245] G. Charlet, H. Mardoyan, P. Tran et al., “Nonlinear Interactions Between 10 Gb/s NRZ Channels and 40 Gb/s Channels with RZ-DQPSK or PSBT format, over Low-Dispersion Fiber,” in Proc. European Conf. on ECOC, page Mo3.2.6, 2006. [246] B. C. Collings and L. Boivin, “Nonlinear polarization evolution induced by cross-phase modulation and its impact on transmission systems,” IEEE Photon. Technol. Lett., 12, 1582–1584, 2000. [247] D. van den Borne, N. E. Hecker-Denschlag, G. D. Khoe, and H. de Waard, “Cross Phase Modulation Induced Depolarization Penalties in 2 10 Gbit/s Polarization-Multiplexed Transmission,” in Proc. ECOC, page Mo4.5.5, 2004. [248] G. Charlet, J. Renaudier, M. Salsi et al., “Efficient Mitigation of Fiber Impairments in an UltraLong Haul Transmission of 40 Gbit/s Polarization-Multiplexed Data, by Digital Processing in a Coherent Receiver,” in Proc. OFC, page PDP17, 2007. [249] N. S. Bergano, “Undersea amplified lightwave systems design.” In Optical Fiber Telecommunications III A (I. P. Kaminov and T. L. Koch eds), Chapter 10, Academic Press, 1997. [250] P. J. Winzer, M. Pfennigbauer, and R.-J. Essiambre, “Coherent crosstalk in ultra-dense WDM systems.” J. Lightw. Technol., 23(4), 1734–1744, 2005. [251] M. Pfennigbauer and P. J. Winzer, “Choice of MUX/DEMUX Filter Characteristics for NRZ, RZ, and CSRZ DWDM systems,” J. Lightw. Technol., 24, 1689–1696, 2005. submitted. [252] G. Raybon, S. Chandrasekhar, A. H. Gnauck et al., “Experimental Investigation of Long-Haul Transport at 42.7 Gb/s Through Concatenated Optical Add/Drop Nodes,” in Proc. OFC, page ThE4, 2004. [253] G. Raybon, S. Chandrasekhar, A. Agarwal et al., “Limitations of Optical Add/Drop Filtering on 42.7-Gb/s Transmission with 50-GHz Channel Spacing,” in Proc. ECOC, page Mo4.5.1, 2004. [254] S. Chandrasekhar, “Dispersion Tolerant Alternative 10-Gb/s Transmitters and Implications for WDM Optical Networking,” in Proc. OFC, 2006. [255] G. Bosco and P. Poggiolini, “Long-distance effectiveness of MLSE IMDD receivers,” IEEE Photon. Technol. Lett., 18, 1037–1039, 2006. [256] D. Castagnozzi, “Digital Signal Processing and Electronic Equalization (EE) of ISI,” in Proc. OFC, page WM6, 2004. [257] P. J. Winzer, R.-J. Essiambre, and S. Chandrasekhar, “Dispersion-Tolerant Optical Communication Systems,” in Proc. ECOC, page We2.4.1, 2004. [258] H. Kogelnik, R. M. Jopson, and L. E. Nelson, “Polarization-mode dispersion” In Optial Fiber Telecommunication (I. Kaminow and T. Li eds), Academic Press, 2002, pp. 725–861. [259] R. M. Jopson, L. E. Nelson, G. J. Pendock, and A. H. Gnauck, “Polarization-Mode Dispersion Impairment in Return-to-zero and Non Return-to-zero Systems,” in Proc. OFC, page WE3, 1999. [260] P. J. Winzer, H. Kogelnik, C. H. Kim et al., “Receiver impact on first-order PMD outage,” IEEE Photon. Technol. Lett., 15(10), 1482–1484, 2003. [261] C. Xie, S. Shen, and L. Moller, “Effects of transmitter imperfection on polarization mode dispersion impairments,” IEEE Photon. Technol. Lett., 15(4), 614–616, 2003.

2. Advanced Optical Modulation Formats

93

[262] C. H. Kim, H. Kim, R. M. Jopson, and P. J. Winzer, “Dependence of PMD Penalties on Decision Point and Receiver Bandwidth,” in Proc. OFC, page TuI4, 2002. [263] M. Brodsky, N. J. Frigo, M. Boroditsky, and M. Tur, “Polarization mode dispersion of installed fibers,” J. Lightw. Technol., 24, 4584–4599, 2006. [264] H. F. Haunstein, W. Sauer-Greff, A. Dittrich et al., “Principles for electronic equalization of polarization-mode dispersion,” J. Lightw. Technol., 22(4), 1169–1182, 2004. [265] F. Buchali and H. Bulow, “Adaptive PMD compensation by electrical and optical techniques,” J. Lightw. Technol., 22(4), 1116–1126, 2004. [266] B. Wedding and C. N. Haslach, “Enhanced PMD Mitigation by Polarization Scrambling and Forward Error Correction,” in Proc. OFC, page WAA1, 2001. [267] X. Liu, “All-channel Pmd Mitigation using Distributed Fast Polarization Scrambling in WDM Systems with FEC,” in Proc. OFC, page OMH4, 2007. [268] L. F. Mollenauer and J. P. Gordon, “Solitons in Optical Fibers: Fundamentals and Applications,” Academic Press, 2006, Section 3.3. [269] T. Mizuochi, K. Ishida, T. Kobayashi et al., “A comparative study of DPSK and OOK WDM transmission over transoceanic distances and their performance degradations due to nonlinear phase noise,” J. Lightw. Technol., 21, 1933–1943, 2003. [270] J. P. Gordon and L. F. Mollenauer, “Phase noise in photonic communications systems using linear amplifiers,” Opt. Lett., 15(23), 1351–1353, 1990. [271] A. Mecozzi, “Limits to long-haul coherent transmission set by the Kerr nonlinearity and noise of the in-line amplifiers,” J. Lightw. Technol., 12, 1993–2000, 1994. [272] H. Kim and A. H. Gnauck, “Experimental investigation of the performance limitation of DPSK Systems due to nonlinear phase noise,” IEEE Photon. Technol. Lett., 15, 320–322, 2003. [273] M. Ohm, R.-J. Essiambre, and P. J. Winzer, “Nonlinear Phase Noise and Distortion in 42.7-Gbit/s RZ-DPSK Systems,” in Proc. ECOC, page Tu1.2.1, 2005. [274] M. M. El Said, J. Sitch, and M. I. Elmasry, “An electrically pre-equalized 10-Gb/s duobinary transmission system,” J. Lightw. Technol., 23(1), 388–400, 2005. [275] P. J. Winzer and R. J. Essiambre, “Electronic Pre-Distortion for Advanced Modulation Formats,” in Proc. ECOC, page Tu4.2.2, 2005. [276] R.-J. Essiambre and P. J. Winzer, “Fibre Nonlinearities in Electronicaly Pre-Distorted Transmission,” in Proc. ECOC, page Tu3.2.2, 2005. [277] R. I. Killey, P. M. Watts, V. Mikhailov et al., “Electronic dispersion compensation by signal predistortion using digital processing and a dual-drive Mach-Zehnder modulator,” IEEE Photon. Technol. Lett., 17, 714–716, 2005. [278] K. Roberts, L. Strawczynski, C. Li et al., “Electronic precompensation of optical nonlinearity,” IEEE Photon. Technol. Lett., 18, 403–405, 2006. [279] R.-J. Essiambre and P. J. Winzer, “Impact of Fiber Nonlinearities on Advanced Modulation Formats using Electronic Pre-Distortion,” in Proc. OFC, page OWB1, 2006. [280] R.-J. Essiambre, P. J. Winzer, X. Q. Wang et al., “Electronic predistortion and fiber nonlinearity,” IEEE Photon. Technol. Lett., 18, 1804–1806, 2006.

This page intentionally left blank

3 Coherent optical communication systems Kazuro Kikuchi Department of Frontier Informatics, University of Tokyo, Kashiwa, Chiba, Japan

3.1 INTRODUCTION 3.1.1 Coherent Optical Communications Twenty Years Ago The research and development in optical fiber communication systems started around the first half of the 1970s. Such systems used intensity modulation of semiconductor lasers, and the transmitted optical signal intensity was detected by a photodiode, which was regarded as a square-law detector. This combination of the transmitter and receiver is called the intensity modulation/direct detection (IMDD) scheme, which has been commonly employed in current optical communication systems. However, coherent optical receivers were studied extensively in the 1980s [1]. Coherent receivers could linearly down-convert the whole optical signal to a baseband electrical signal by using heterodyne or homodyne detection, and had the following advantages against direct detection: (1) The shot-noise limited receiver sensitivity can be achieved with a sufficient local oscillator (LO) power. The LO gives us a signal gain, whereas the LO shot noise overwhelms the thermal noise of the receiver; thus we can achieve the shot-noise limited receiver sensitivity. (2) The frequency resolution at the intermediate frequency (IF) or baseband stage is so high that we can separate closely spaced wavelength-division multiplexed (WDM) channels at the electrical stage. (3) The ability of phase detection can improve the receiver sensitivity compared with the IMDD system. This is due to the fact that the distance between symbols, which are expressed as phasors on the complex plane, is extended by the use of the phase information. Optical Fiber Telecommunications V B: Systems and Networks Copyright  2008, Elsevier Inc. All rights reserved. ISBN: 978-0-12-374172-1

95

96

Kazuro Kikuchi

(4) The multilevel modulation format such as quadrature phase-shift keying (QPSK) can be introduced into optical communications by using phase modulation. A number of research groups challenged optical transmission experiments with coherent receivers. However, the invention of erbium-doped fiber amplifiers (EDFAs) made the shot-noise limited receiver sensitivity of the coherent receiver less significant. This is because the signal-to-noise ratio (SNR) of the signal transmitted through the amplifier chain is determined from the accumulated amplified spontaneous emission (ASE) rather than the shot noise. In addition, even in unrepeated transmission systems, the EDFA used as a low-noise preamplifier eliminated the need for the coherent receiver with superior sensitivity. Technical difficulties inherent in coherent receivers could not also be disregarded. The heterodyne receiver requires an IF, which should be higher than the signal bit rate, and the maximum bit rate of the heterodyne receiver is always less than the half of that the square-law detector can achieve. In contrast, the homodyne receiver is essentially a baseband receiver; however, the complexity in stable locking of the carrier phase drift has prevented its practical applications. From these reasons, further research and development activities in coherent optical communications have almost been interrupted for nearly 20 years. In contrast, rapid progress in EDFA technologies entirely changed the direction of R&D in optical communications. The EDFA-based system started to take benefit from WDM techniques to increase the transmission capacity of a single fiber. The hardware required for WDM networks became widely deployed owing to its relative simplicity and the relatively low cost associated with optical amplifier repeaters where multiple WDM channels could be amplified all at once. The WDM technique marked the beginning of a new era in the history of optical communication systems and brought forth 1000 times increase in the transmission capacity during the 1990s.

3.1.2 Rebirth of Coherent Optical Communications With the transmission-capacity increase in WDM systems, coherent technologies have restarted to attract a large interest over the recent years. The motivation lies in finding methods of meeting the ever-increasing bandwidth demand with multilevel modulation formats based on coherent technologies [2]. The first step of the revival of coherent optical communications research was ignited with the QPSK modulation/demodulation experiment featuring optical in-phase and quadrature (IQ) modulation and optical delay detection [3]. In such a scheme, one symbol carries two bits by using the four-point constellation on the complex plane; therefore, we can double the bit rate, while keeping the symbol rate, or maintain the bit rate even with the halved spectral width. The optical IQ modulation has been realized with Mach–Zehnder type push– pull modulators in parallel [4]. The IQ components of the optical carrier can be

3. Coherent Optical Communication Systems

97

modulated independently, enabling any kind of modulation formats. IQ modulators integrated on LiNbO3 substrates are now commercially available. The optical delay detector is composed of an optical one-bit delay line and a double-balanced photodiode. With such a receiver, we can compare the phase of the transmitted signal with that of the previous bit and restore the data, which is differentially precoded at the transmitter. Using two such receivers in parallel, we can demodulate the signal IQ components separately without the need for LO. A number of long-distance WDM QPSK transmission experiments have been reported recently, based on IQ modulation and differential IQ demodulation. The next stage has opened with high-speed digital signal processing (DSP) [5]. The recent development of high-speed digital integrated circuits has offered the possibility of treating the electrical signal in a DSP core and retrieving the IQ components of the complex amplitude of the optical carrier in a very stable manner. The authors have demonstrated a phase-diversity homodyne receiver for demodulating the 20-Gbit/s QPSK signal. Since the carrier phase is recovered after homodyne detection by means of DSP, this type of receiver has now commonly been called the “digital coherent receiver.” While an optical phase-locked loop (PLL) that locks the LO phase to the signal phase is still difficult to achieve, DSP circuits are becoming increasingly faster and provide us with simple and efficient means for estimating the carrier phase [6]. Any kind of multilevel modulation formats can be introduced by using the coherent receiver [7]. While the spectral efficiency of binary modulation formats is limited to 1 bit/s/Hz/polarization, which is called the Nyquist limit, modulation formats with M bits of information per symbol can achieve the spectral efficiency up to M bit/s/Hz/polarization. Although optical delay detection has been employed to demodulate the QPSK signal (M = 2) [3], further increase in multiplicity is hardly achieved with such a scheme. Another and probably more important advantage of the digital coherent receiver is the post signal-processing function. The IQ demodulation by our receiver is the entirely linear process; therefore, all the information on the complex amplitude of the transmitted optical signal is preserved even after detection, and conventional signal processing functions acting on the optical carrier, such as optical filtering and dispersion compensation [8,9], can be performed at the electrical stage after detection. Post signal-processing based on DSP has been employed in direct detection systems for adaptive equalization and maximum likelihood estimation (MLE) of the distorted signal, and has become a new trend of optical communications [10]. In contrast, the digital coherent receiver enables more sophisticated signal processing on the homodyne-detected complex amplitude because the optical phase information is not lost even after detection. For the moment, DSP at 10 Gsymbol/s in the digital coherent receiver has been performed still offline; however, the real-time operation has already been demonstrated at a few Gsymbol/s [11], and dedicated integrated circuits will make the real-time operation over 10 Gsymbol/s possible in the near future. The combination of coherent detection and DSP is thus expected to become a part of the next

Kazuro Kikuchi

98

generation of optical communication systems and provide new capabilities that were not possible without the detection of the phase of the optical signal.

3.1.3 Organization of this Chapter This chapter reviews the results of recent developments on the digital coherent receiver. After historical perspective given in Section 3.1, Section 3.2 provides the principle of coherent detection. Section 3.3 discusses the concept of the digital coherent receiver, and the basic performance of the digital coherent receiver is evaluated in Section 3.4. Section 3.5 deals with postprocessing functions, which enable compensation for transmission impairments at the receiver. Finally, a summary is provided in Section 3.6.

3.2 PRINCIPLE OF COHERENT DETECTION In this section, we describe the principle of operation of coherent detection. Especially, we show that with the homodyne receiver, which comprises phase and polarization diversities, we can restore the full information on the optical complex amplitude, namely the amplitude, the phase, and the state of polarization (SOP).

3.2.1 Coherent Detection The fundamental concept behind coherent detection is to take the product of electric fields of the modulated signal light and the continuous-wave LO. The optical signal coming from the transmitter can be described by Es ðtÞ = As ðtÞ expðj!s tÞ;

(3.1)

where As(t) is the complex amplitude and !s is the angular frequency. Similarly, the field of the LO can be described by ELO ðtÞ = ALO expðj!LO tÞ;

(3.2)

where ALO is the complex amplitude and !LO is the angular frequency of the LO. We note here that the complex amplitudes As and ALO are related to the power of the optical fields by Ps = jAsj2=2 and PLO = jALOj2=2, where Ps and PLO are the power of the signal and LO, respectively. Balanced detection is usually introduced into the coherent receiver as a mean to suppress the DC component and maximize the signal photocurrent. Figure 3.1 shows the block diagram of such a receiver. The concept resides in using a 3-dB

3. Coherent Optical Communication Systems

99 Balanced receiver

PC

Es

E1 I1 (t ) I2 (t ) EL0

I (t )

E2

LO laser Figure 3.1 Configuration of the coherent receiver.

optical coupler that adds a 180 phase shift to either the signal field or the LO field between the two output ports. When the signal and LO are copolarized, the electric fields incident on the upper and lower photodiodes are given as 1 E1 = pffiffiffi ðEs þ ELO Þ; 2 1 E2 = pffiffiffi ðEs  ELO Þ; 2

(3.3)

(3.4)

and the output photocurrents are written as 



As ðtÞ expðj!s tÞ þ ALO ðtÞ expðj!LO tÞ ms p ffiffi ffi ; I1 ðtÞ = R Re 2

 pffiffiffiffiffiffiffiffiffiffiffiffiffi R = Ps þ PLO þ 2 Ps PLO cos !IF t þ sig ðtÞ  LO ðtÞ 2 

As ðtÞ expðj!s tÞ  ALO ðtÞ expðj!LO tÞ ms pffiffiffi ; I2 ðtÞ = R Re 2

 pffiffiffiffiffiffiffiffiffiffiffiffiffi R = Ps þ PLO  2 Ps PLO cos !IF t þ sig ðtÞ  LO ðtÞ 2

(3.5)

(3.6)

where ms represents the mean square with respect to the optical frequencies, Re the real part, !IF the IF given by !IF = !s – !LO, and sig(t) and LO(t) are phases of the transmitted signal and LO, respectively. R is the responsitivity of the photodiode and is given as R=

e ; ! s h

(3.7)

where  h stands for the Planck’s constant, e is the electron charge, and  is the quantum efficiency of the photodiode. We neglect the sum frequency component

Kazuro Kikuchi

100

since it will be averaged out to zero due to the limited bandwidth of the photodiode. The balanced detector output is then given as IðtÞ = I1 ðtÞ  I2 ðtÞ = 2R

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

 Ps ðtÞPLO cos !IF t þ sig ðtÞ  LO ðtÞ :

(3.8)

PLO is always constant and LO(t) includes only the phase noise that varies with time.

3.2.2 Heterodyne and Homodyne Receivers Heterodyne detection refers to the case where j!IFj >> !b=2, with !b being the modulation bandwidth of the optical carrier determined by the bit rate. In such a case, Eqn (3.8) shows that the electric field of the signal light is down-converted to the IF signal including the amplitude information and the phase information. If we consider the PSK modulation format, Ps is constant and sig(t) = s(t) þ sn(t), where s(t) is the phase modulation and sn(t) the phase noise. The receiver output is given as pffiffiffiffiffiffiffiffiffiffiffiffiffi IðtÞ = 2R Ps PLO cosf!IF t þ s ðtÞ þ n ðtÞg;

(3.9)

and we can determine the complex amplitude on exp( j!IFt) from Eqn (3.9) as pffiffiffiffiffiffiffiffiffiffiffiffiffi Ic ðtÞ = 2R Ps PLO exp jfs ðtÞ þ n ðtÞg;

(3.10)

which is equivalent to the complex amplitude of the optical signal except for the phase noise increase. In this case, although the total phase noise n(t) = sn(t) – LO(t) might vary with time, electrical synchronous demodulation can be used to estimate the phase noise and decode the symbol s(t). Homodyne detection refers to the case where !IF = 0. The photodiode current then becomes

 pffiffiffiffiffiffiffiffiffiffiffiffiffi IðtÞ = 2R Ps PLO cos sig ðtÞ  LO ðtÞ :

(3.11)

The LO phase LO(t) must track the transmitter phase noise sn(t) such that n(t) = 0 to decode the symbol s(t) correctly. This function is realized by the optical phase-locked loop (OPLL); however, in practice, the implementation of such a loop is not simple and adds to the complexity of homodyne detection. In addition, Eqn (3.11) only gives the cosine component (in other words, the in-phase component with respect to the LO phase), and the sine component (the quadrature component) cannot be detected. Therefore, this type of homodyne receivers is not able to extract the full information of the signal complex amplitude.

3. Coherent Optical Communication Systems

101

3.2.3 Phase-diversity Homodyne Receiver If we prepare another LO, whose phase is shifted by 90, in the homodyne receiver, we can detect both IQ components of the signal light. This function is achieved by a 90 optical hybrid (see Section 3.4 for the more detailed optical circuit configuration). As shown in Figure 3.2, using the 90 optical hybrid, we can obtain four outputs E1, E2, E3, and E4 from the two inputs Es and ELO as E1 =

1 ðEs þ ELO Þ; 2

(3.12)

E2 =

1 ðEs  ELO Þ; 2

(3.13)

1 E3 = ðEs þ jELO Þ; 2

(3.14)

1 E4 = ðEs  j ELO Þ: 2

(3.15)

Output photocurrents from balanced photodetectors are then given as

 pffiffiffiffiffiffiffiffiffiffiffiffiffi II ðtÞ = II1 ðtÞ  II2 ðtÞ = R Ps PLO cos sig ðtÞ  LO ðtÞ ;

 pffiffiffiffiffiffiffiffiffiffiffiffiffi IQ ðtÞ = IQ1 ðtÞ  IQ2 ðtÞ = R Ps PLO sin sig ðtÞ  LO ðtÞ :

(3.16) (3.17)

In the case of PSK modulation, these are written as

pffiffiffiffiffiffiffiffiffiffiffiffiffi II ðtÞ = R Ps PLO cosfs ðtÞ þ n ðtÞg;

(3.18) PD1

E1

PC

II 1 (t )

Es 90° Optical hybrid

E2

II 2 (t )

E3

IQ 1 (t )

E L0 LO laser

IQ 2 (t ) E4 PD2

Figure 3.2 Configuration of the phase-diversity homodyne receiver.

II (t )

IQ (t )

Kazuro Kikuchi

102

pffiffiffiffiffiffiffiffiffiffiffiffiffi IQ ðtÞ = R Ps PLO sinfs ðtÞ þ n ðtÞg:

(3.19)

Using Eqns (3.18) and (3.19), we can restore the complex amplitude as pffiffiffiffiffiffiffiffiffiffiffiffiffi Ic ðtÞ = II ðtÞ þ jIQ ðtÞ = R Ps PLO expfj ðs ðtÞ þ n ðtÞg;

(3.20)

which is equivalent to the complex amplitude of the optical signal except for the phase noise increase. The receiver thus leads to the recovery of both the sine and cosine components. It is possible to estimate the phase noise n(t) varying with time and restore the phase information s(t) through DSP on the homodyne detected signal given by Eqn (3.20). In addition, even the frequency offset can be allowed if j!IFjb

where Im(x) is the mth-order modified Bessel function of the first kind. The optimal decision threshold satisfies I0 ð4SNRb

pffiffiffiffiffiffiffi Vth Þ expð2SNRb Þ = 1

(4.10)

4. Self-Coherent Optical Transport Systems

141

The optimal BER for direct-detection OOK can thus be numerically calculated by solving Eqns (4.8–4.10). For direct-detection OOK without polarization filtering, we have [4] BEROOK

2p =

 pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffii 1h 1  Q2 2 SNRb ; 2 SNRb Vth 2 1 þ expð2SNRb Vth Þ ð1 þ 2SNRb Vth Þ; 2 2

(4.11)

2

where Q2 ða; bÞ = Qða; bÞ þ ba eða þb Þ=2 I1 ðabÞ, and the optimal decision threshold can be numerically calculated. Figure 4.6 shows the BER performances of BPSK, DBPSK, and OOK with and without polarization filtering. Table 4.1 summarizes the receivers sensitivities at BER = 103 and 109 for the above signal formats. With polarization filtering at the receiver, the receiver sensitivity of DBPSK is slightly worse than that of the OBPSK: about 0.45 dB at BER = 109. This penalty is often referred to as the differential-detection penalty. The differential-detection penalty increases for DPSK formats having more than 2 symbols per bit, as will be discussed later. Without polarization filtering, the receiver sensitivity of DBPSK is further degraded by approximately 0.4 dB at BER = 109. Two commonly used modulation/detection configurations are direct-detection DBPSK and OOK that do not have the polarization filtering. The DBPSK outperforms the OOK by about 2.7 dB at BER = 109: this is commonly referred to as the 3-dB advantage of DBPSK over OOK in optical receiver sensitivity. DBPSK with balanced detection is also found to have 6 dB higher tolerance to coherent crosstalk than OOK [31]. –3

log10(BER)

–4

–5 –6

BPSK DBPSK, with PF DBPSK, no PF OOK, with PF OOK, no PF

–7 –8 –9

6

7

8

9

10

11

12

13

14

15

16

17

SNRb (dB) Figure 4.6 The optimal BER performances of BPSK, DBPSK, and OOK with and without polarization filtering (PF) (this figure may be seen in color on the included CD-ROM).

Xiang Liu et al.

142 Table 4.1

Receiver sensitivity comparison among different binary signal formats*. DBPSK BPSK 9

BER = 10

PPB SNRb OSNR(10 G)

BER = 103

PPB SNRb OSNR(10 G)

OOK

With PF

No PF

With PF

No PF

18 12.5 dB 8.5 dB

20 13 dB 9 dB

22 13.4 dB 9.4 dB

38.3 15.8 dB 11.8 dB

40.7 16.1 dB 12.1 dB

4.8 6.8 dB 2.8 dB

6.2 8 dB 4 dB

7.2 8.6 dB 4.6 dB

11.2 10.5 dB 6.5 dB

12.7 11 dB 7 dB

* Ideal Optical preamplifier with NF = 3 dB is assumed. OSNR(10 G) is the required OSNR values for a 10 Gb/s signal. PF: polarization filtering.

Receiver Impairments in Direct-Detection DBPSK One significant impairment in DBPSK receiver performance is the phase error due to nonexact phase matching between the two arms of the ODI. The phase error usually comes from temperature-induced phase mismatch or laser frequency offset from the passband center of the ODI. With the consideration of the phase error, we have for direct-detection DBPSK with polarization filtering [4] 2 2 1 BERDBPSK 1p = Qða; bÞ  eða þb Þ=2 I0 ðabÞ; 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi a = 2SNRb jsinðe =2Þj; b = 2SNRb jcosðe =2Þj;

(4.12)

where e is the phase error, and I0(x) is the zeroth-order modified Bessel function of the first kind, and Q(a, b) is the first-order Marcum Q-function. For direct-detection DBPSK without polarization filtering at the receiver, we have BERDBPSK

2p

2 2 1 = Qða; bÞ  eða þb Þ=2 I0 ðabÞ 2 1 ða2 þb2 Þ=2 b a þ e a  b I1 ðabÞ; 8

(4.13)

where the last term on the right-hand side of the equation represents the error contribution from the noise component whose polarization is orthogonal to that of the signal. When the phase error becomes zero, Eqns (4.12) and (4.13) reduces to Eqns (4.6) and (4.7), respectively. Figure 4.7 shows the OSNR penalty as a function of the phase error for direct-detection DBPSK. The phase error due to laser frequency offset, fe, can be expressed as e = 2p

fe ; FSR

(4.14)

4. Self-Coherent Optical Transport Systems

143

5 BER = 10–9 BER = 10–3

SNR Penalty (dB)

4

3

2

1

0

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Phase error (rad.) Figure 4.7 The OSNR penalty as a function of the phase error for direct-detection DBPSK without polarization filtering at BER = 109 and BER = 103 (this figure may be seen in color on the included CD-ROM).

where FSR is the free spectral range of the ODI. Usually, the FSR of the ODI equals the bit rate of the DBPSK signal, and thus the phase error resulting from a same laser frequency offset or uncertainty decreases as the bit rate of the DBPSK signal increases. Note that any laser has a finite linewidth or frequency jitter, which causes random phase errors with a given distribution, e.g., Lorentzian distribution, and degrades the performance of DPSK. The linewidth of commonly used DFB lasers is on the order of a few MHz, which causes negligible penalty for DPSK signals with multi-gigabits-per-second bit rate. However, laser frequency drift due to temperature and aging may cause sufficiently large frequency offset (e.g., >1 GHz) to cause severe degradation of the DPSK performance. In addition, the phase error due to a certain temperature change of the ODI is proportional to the length difference between the two arms of the ODI, which is inversely proportional to the signal bit rate. So the higher the signal rate, the higher the tolerance of the direct-detection DBPSK to laser frequency uncertainties and temperature changes at the receiver. This makes DBPSK particularly suitable for high-speed optical transmissions. Another receiver imperfection is the nonexact 1-bit delay of the ODI. DBPSK is quite tolerant to the delay mismatch. Negligible OSNR penalty was observed when an ODI with a FSR of 50 GHz (or a delay of 20 ps) was used for demodulating a 42.7 Gb/s DBPSK signal (with 23.4 ps bit period) [26]. Under severe optical filtering, partial-bit delay may actually provide better overall signal performance [32]. Other receiver imperfections such as the arrival time mismatch and amplitude mismatch between the two electrical signals detected by the balanced detector and receiver also contribute to the degradation of the DBPSK receiver sensitivity. Generally, the associated penalties become negligible when the time and amplitude mismatches are less than 5% of the bit period and the nominal amplitude,

Xiang Liu et al.

144

respectively [3]. Since the amplitude mismatch is related to the finite commonmode rejection of the balanced detector, so the required common-mode rejection ratio is desired to be higher than about 13 dB.

4.3.2 Direct-Detection DQPSK Receiver Configuration of Direct-Detection DQPSK Figure 4.8 shows the schematic of a direct-detection DQPSK receiver, consisting of two orthogonal ODIs that demodulate the in-phase (I) and quadrature (Q) tributaries of the signal, two balanced detectors, and a CDR unit with a binary decision circuitry. The I and Q decision variables can be expressed as p

 ðt  TÞ; uI ðtÞ = Re½e j 4 ERX ðtÞ ERX p

 ðt  TÞ; uQ ðtÞ = Im½e j 4 ERX ðtÞ ERX

(4.15)

where ERX is the normalized optical field of the DQPSK signal entering the receiver, and Re[x] and Im[x] are, respectively, the real and imaginary parts of complex variable x. Note that the DQPSK demodulator can be simplified by using a single optical delay interferometer followed by an optical hybrid [23] or star coupler [33]. For the optical encoding and detection configurations shown in Figure 4.3(a) and Figure 4.8, the relations among the precoder outputs and the original data are summarized in Table 4.2 [4]. The precoder outputs at the kth bit, pk and qk, can be obtained by performing the following logic operations on the original data at the kth bit, ak and bk, and the previous outputs of the precoder, pk1 and qk1, according to the following equation: pk = ak bk pk1 þ ak bk qk1 þ ak bk qk1 þ ak bk pk1 qk = ak bk qk1 þ ak bk pk1 þ ak bk pk1 þ ak bk qk1 ODI-1

(4.16)

BD-1 –

I

+π/4 DQPSK signal

ODI-2

CDR circuitry

BD-2 –

–π/4

Q

Figure 4.8 Schematic of a direct-detection DQPSK receiver (this figure may be seen in color on the included CD-ROM).

4. Self-Coherent Optical Transport Systems

145

Table 4.2 The relations among the precoder outputs and the original data for a typical DQPSK transmitter and receiver configuration [4].

Original data (akbk)

Optical phase difference

First precoder output, pk

Second precoder output, qk

(00) (01) (10) (11)

 /2 1.5 0

pk1 qk1 qk1 pk1

qk1 pk1 pk1 qk1

Optimal Receiver Sensitivity of Direct-Detection DQPSK Under the matched optical filer assumption, the optimal BER performance of direct-detection DQPSK with polarization filtering can be written as [4]. 1 2 2 BERDQPSK 1p = Qða; bÞ  eða þb Þ=2 I0 ðabÞ; 2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffi pffiffiffi with a = 2SNRb ð1  1= 2Þ and b = 2SNRb ð1 þ 1= 2Þ:

(4.17)

For direct-detection DQPSK without polarization filtering, we have BERDQPSK

2p

1 ða2 þb2 Þ=2 e

I0 ðabÞ 2   2 2 1 þ eða þb Þ=2 ba  ab I1 ðabÞ; 8

= Qða; bÞ 

(4.18)

where a and b are the same as those given in Eqn. (4.17). As a comparison, the optimal BER performance of coherent detection (homodyne or heterodyne) QPSK is [29] pffiffiffiffiffiffiffiffiffiffiffiffiffi 1 SNRb : BERQPSK = erfc 2

(4.19)

Figure 4.9 shows the BER performances of QPSK and DQPSK without polarization filtering, as compared to those of BPSK and DBPSK. QPSK performs the same as the BPSK, while the DQPSK (without polarization filtering) exhibits a differential-detection penalty of 2.4 dB over the BER range from 103 to 109. Comparing DQPSK with DBPSK, the relative sensitivity penalty of DQPSK is 1.5 dB at BER = 109 and 0.8 dB at BER = 103. These performance differences were confirmed experimentally [34]. In optical transmission systems with modern forward error correction (FEC) codes, the raw BER of interest

Xiang Liu et al.

146 –3 BPSK, QPSK DBPSK, no PF DQPSK, no PF

log10(BER)

–4

–5 –6 –7 –8 –9

6

8

10

12

14

16

SNRb (dB) Figure 4.9 Optimal BER performances of QPSK and DQPSK without polarization filtering, as compared to those of BPSK and DBPSK (this figure may be seen in color on the included CD-ROM).

is in the range around 103, and the receiver sensitivity disadvantage of the direct-detection DQPSK as compared to the DBPSK is about 1 dB. Receiver Impairments in Direct-Detection DQPSK Similar to DBPSK, DQPSK also suffers from implementation imperfections. Particularly, DQPSK is very sensitive to phase errors. The BER of an DQPSK signal in the presence of a phase error, e , is   = 12 BERDQPSK ðaþ ; bþ Þ þ BERDQPSK ða ; b Þ ; qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi   ffi with a = 2SNRb 1  cos p4 e ; and qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi   ffi b = 2SNRb 1 þ cos p4 e ;

BERDQPSK

e

(4.20)

where BERDQSPK can be obtained from Eqn (4.17) or Eqn (4.18) depending on whether polarization filtering is applied. Figure 4.10 shows the OSNR penalty (at BER = 109) of a 40-Gb/s DQPSK as a function of the laser frequency offset, as compared to that of a 40-Gb/s DBPSK signal. DQPSK requires approximately 5.5 times tighter laser frequency accuracy [35].

4.3.3 Direct-Detection m-ary DPSK Higher spectral efficiency can be obtained by using m-ary DPSK formats with more phase levels. For direct detection of m-ary DPSK, a straightforward but

4. Self-Coherent Optical Transport Systems

147

5 DBPSK DQPSK

SNR penalty (dB)

4

3

2

1

0

0

1

2

3

4

5

Frequency offset (GHz) Figure 4.10 The OSNR penalty (at BER = 1E-9) as a function of the laser frequency offset in 40-Gb/s DQPSK and DBPSK systems using 1-bit ODIs. No polarization filtering is used (this figure may be seen in color on the included CD-ROM).

cumbersome demodulation scheme is to use m/2 ODIs with the following different phase offsets (between the two arms of each ODI). DODI ðnÞ =

½1  2ðn  1Þp ; m

n = 1; 2; : : :; m=2:

(4.21)

For example, the demodulation of an 8-ary DPSK signal thus requires four ODIs with phase offsets of /8, –/8, –3/8, and –5/8 between the interfering arms. Following the demodulation, m/2 pairs of balanced detectors are needed to obtain m/2 decision variables, u[DODI(n)], where n = 1, 2,. . ., m/2. Generalizing the analysis on 8-ary DPSK to m-ary DPSK, we can recover the log2(m) data tributaries of an m-ary DPSK signal by [36] h p i h  p p i c1 =cI ¼ u >0 ; c2 ¼cQ ¼ u  >0 ; m 2 h p m p  i h  p p i c3 ¼ u þ >0 u  >0 ;::: 4   m 4  m     (4.22) 3 7 m/21 p >0 clog2 ðmÞ ¼ u p >0 u p >0 ;::: u m m m          3 p 7 p m/21 p  >0 ::: u p >0 :

u p >0 u m 2 m 2 m 2 Since there are correlations among the decision variables obtained by the m/2 ODIs, the first two decision variables are sufficient to derive all the other decision variables with the help of analog signal processing [37] or DSP [36] that will be discussed in detail in a later section on digital self-coherent receivers. Note that an

Xiang Liu et al.

148

appropriate precoder is needed to ensure the recovered data tributaries are the original ones. The receiver sensitivity of an m-ary PSK signal can be straightforwardly estimated from the minimum Euclidean distance between the two closest symbols in its constellation diagram (with a normalized average signal power). The receiver sensitivity penalty (in dB) of m-ary PSK with respect to coherent-detection BPSK can be approximated as [4] h  p  pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffii

log2 ðmÞ DSNRbPSK ðmÞ  20 log10 sin m

(4.23)

For m-ary DPSK signals, the errors are dominated by differential phase noise. We define a minimum differential phase distance for m-ary DPSK as dDDPSK ðmÞ =

pffiffiffi 2p 2p 1 ;

pffiffiffi = m m 2

(4.24)

pffiffiffi where pffiffiffi the factor of 2 comes from the fact that the differential-phase deviation is 2 times as large as the absolute phase deviation, assuming that the optical noise induced phase fluctuations in adjacent signal symbols are independent Gaussian random variables. Under the assumption of low BER where the minimum differential-phase distance determines the BER performance, we can then approximate the receiver sensitivity penalty (in dB) of m-ary DPSK with respect to coherentdetection BPSK as DSNRbDPSK ðmÞ

"

# dDDPSK ðmÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  20 log10

log2 ðmÞ 2  pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  = 20 log10 pffiffiffi log2 ðmÞ : 2m

(4.25)

Figure 4.11 shows the receiver sensitivity penalties of m-ary PSK and DPSK as compared to BPSK as a function of the number of phase states. The sensitivity penalties become ‡ 3 dB for m ‡ 8. In addition, the differential-detection penalty approaches 3 dB as m increases. This suggests that m-ary DPSK formats with m ‡ 8 are only viable in systems with sufficiently high OSNR.

4.3.4 Direct-Detection of Hybrid DPSK and PAM Combining DPSK with PAM provides another means to carry more bits per symbol. Since DPSK acts on signal phase and PAM acts on signal amplitude, they can be independently received. The DPSK portion of a DPSK/PAM signal

4. Self-Coherent Optical Transport Systems

149

16

m-ary PSK m-ary DPSK

14

SNR penalty (dB)

12 10 8

8ary-DPSK/PAM4

6

8ary-DPSK/PAM2

4

DQPSK/PAM2

2 0

4

8

12

16

20

24

28

32

m Figure 4.11 Receiver sensitivity penalties of m-ary PSK and DPSK as compared to BPSK as a function of the number of phase states. The symbols are for DPSK/PAM formats (this figure may be seen in color on the included CD-ROM).

can be received by the direct-detection schemes described previously, and the PAM portion can be received by a binary [16] or multilevel decision circuitry [18] depending the number of the PAM levels. For direct-detection DPSK/PAM, the ideal receiver sensitivity is optimized when the Euclidean distances between two adjacent PAM levels in its constellation diagram are equal, and they are also equal to the minimum differential-phase distance in the DPSK modulation. The optimal locations of the m p symbols of a DPSK/PAM signal with m DPSK phase states and p PAM levels are thus pffiffiffi     2pq 2pn ;

1þ sðn; qÞ = c exp j m m n = 0; 1; : : :; m  1;

q = 0; 1; : : :; p  1;

(4.26)

where c is a normalization constant. The receiver sensitivity penalty of directdetection DPSK/PAM as compared to coherent-detection BPSK can be estimated as DPSK=PAM

DSNRb

ðm; pÞ

8 ! " ) # p1  < d DPSK ðmÞ 2 X 2 1 D DPSK 1 þ q dD

ðmÞ

log2 ðm pÞ  10 log10 : 2 q=0 = 10 log10

(

) pffiffiffi 1 p2 p2 2p

log2 ðm pÞ :

1 þ ðp  1Þ þ ðp  1Þð2p  1Þ 2 2m2 3m m 

ð4:27Þ

150

Xiang Liu et al.

For DQPSK/PAM2 [14], the theoretical receiver sensitivity penalty as compared to BPSK with the same bit rate) is 4.7 dB, about 1.7 dB smaller than 8-ary DPSK which is also a 3-bit/symbol format. At 4-bit/symbol and 5-bit/symbol, the hybrid DPSK/PAM offers higher optimal receiver sensitivities than m-ary DPSK, as shown in Figure 4.11. However, DPSK/PAM usually suffers more transmitter/receiver bandwidth limitation induced ISI and nonlinear penalty than m-ary DPSK which may be regarded as a constant-amplitude format. Note that in principle, one can improve the receiver sensitivity of hybrid formats based on DPSK and PAM (or ASK) by arranging the signal constellation such that more DPSK phase states are assigned at high power regime than at low power regime, similar to conventional quadrature amplitude modulation (QAM) schemes commonly used in digital communication [29]. However, sophisticated transmitter and receiver designs are needed.

4.4 DIGITAL SELF-COHERENT OPTICAL RECEIVER With recent advances in high-speed electronic circuits and DSP, optical receiver technologies are evolving to provide unprecedented capabilities that may substantially impact optical transport systems. Combining square-law detection with ADC and DSP, maximum likelihood sequence estimation (MLSE) was used to improve receiver tolerance to chromatic dispersion [38]. However, square-law detection loses optical signal phase information, and limits the dispersion equalization capability. A coherent optical receiver is capable of recovering signal phase and amplitude, but traditionally requires a sophisticated optical phase-locked loop for its OLO. Recently, sampled coherent optical receivers [39–42] based on ADC and DSP have been proposed to omit the OLO and allow advanced signal processing for optical field reconstruction and subsequent electronic compensation of dispersion (EDC) [42–45] and polarization mode dispersion (EPMDC) [43, 44]. Realtime demonstration of sampled coherent receiver has also been reported [46–48]. As described in the Section 4.3, signals modulated with DPSK, such as DBPSK and DQPSK, can be received with direct-detection that does not require an OLO. Enhancing differential direct-detection with ADC and DSP, a new type of optical receiver, referred to as DSCOR in which a digital representation of received signal phase waveform is obtained, has recently emerged [49–51]. A phase reference approaching the accuracy of an ideal OLO can be digitally extracted in a DSCOR via data-aided MSPE [49] to improve its sensitivity to approach that of a sampled coherent receiver. The received signal optical field can also be digitally reconstructed in a DSCOR with the additional recovery of the signal amplitude, making digital compensation of dispersion possible [50, 51]. DSCOR is also naturally suitable for receiving high spectral-efficiency m-ary DPSK signals [17]. In this section, we present the concept and principle of DSCOR. Particularly, we describe data-aided MSPE for sensitivity improvement, electronic demodulator error compensation (EDEC), simplification in multilevel DPSK detection, and reconstruction of signal optical field, with

4. Self-Coherent Optical Transport Systems

151

which digital compensation of transmission impairments such as dispersion and nonlinear phase noise dispersion could be realized. A polarization-diversity version of the DSCOR allowing the reception of polarization-multiplexed self-coherent signals will also be discussed.

4.4.1 Receiver Architecture of DSCOR The schematic of DSCOR is shown in Figure 4.12. The optical complexity of the DSCOR is similar to that of a conventional direct-detection DQPSK receiver. The received signal, r(t) = |r(t)|exp[ j (t)], is first split into two branches that are connected to a pair of ODIs with orthogonal phase offsets  and  – /2, where  is an arbitrary phase value. Note that the phase orthogonality is assumed to be guaranteed, e.g., via the design reported in Ref. [33]. This simplifies the control of the pair of ODIs to a single phase control. The delay in each of the ODI, , is set to be approximately T/sps, where T is the signal symbol period and sps is the number of samples per symbol of the ADCs that convert the two detected analog signal waveforms, referred to as the I and Q components, to digitized waveforms uI (t) and uQ(t), which follow uðtÞ = uI ðtÞ þ j uQ ðtÞ = e j rðtÞ r  ðt  Þ:

(4.28)

The optical phase difference between adjacent sampling locations can be obtained from   e j ½ðtÞðtÞ = uðtÞej uðtÞej :

(4.29)

With the differential phase information being available, a digital representation of the received signal field can be obtained by

+

τ

–

DSP Unit

uI ADC

OA OF Optional intensity detection branch

ODI-2

ADC

BD-2

θ – π/2

+

τ

–

ADC

Signal processing/ data recovery

Signal

BD-1

θ

Field reconstruction

ODI-1

uQ

Figure 4.12 Schematic of DSCOR architecture based on orthogonal differential direct-detection and DSP. OA: optical preamplifier; OF: optical filter; ODI: optical delay interferometer; BD: balanced detector; ADC: analog-to-digital converter (this figure may be seen in color on the included CD-ROM).

Xiang Liu et al.

152

rðt0 þ n Þ = jrðt0 þ n Þje jðt0 Þ

n Y

e j Dðt0 þm Þ ;

(4.30)

m=1

where t0 is an arbitrary reference time, (t0) is a reference phase that can be set to 0, and the amplitude of the receiver signal can be obtained by an additional intensity detection branch [50] or approximated as below [51] jrðt0 þ n Þj  juðt0 þ n  Þ uðt0 þ n  þ Þj1=4 :

(4.31)

When the ISI caused by distortions such as dispersion and PMD is reasonably small, synchronous sampling with sps = 1 may be used and the signal amplitude may be approximated as a constant. Note that DSCOR can be designed to be polarization-independent to readily receive a single-polarization signal, while sampled coherent receiver requires polarization alignment between the signal and the OLO or polarization diversity. Once the received optical signal field is digitally available, advanced signal processing techniques, similar to those used in sampled coherent receivers, may be applied in DSCOR to mitigate transmission impairments, as to be discussed in Section 4.5. In the special case with sps = 1, the delay in the orthogonal ODI pair equals to the symbol period, and the I and Q decision variables for an m-ary DPSK signal can be directly obtained by setting  = /m (as discussed earlier). Any demodulator phase error e =   =m can be compensated by using the following simple EDEC process: uðtÞ ! eje uðtÞ:

(4.32)

Figure 4.13 shows the eye diagrams for the detected I and Q components of a 40-Gb/s DQPSK signal at different demodulator phase errors [49]. With e = p=8; the eyes are heavily distorted. With e = p=2; the eye of the I component is completely inverted, corresponding to a BER of 0.5. Figure 4.14 shows the measured BER φe = 0

φe = π/8

φe = π/4

φe = π/2

Figure 4.13 Measured I (upper) and Q (lower) eye diagrams of a 40-Gb/s DPQSK signal at different demodulator phase errors. Timescale is 20 ps/div [49] (this figure may be seen in color on the included CD-ROM).

4. Self-Coherent Optical Transport Systems

153

5 × 10–1

10–1

w/EDEC

BER

w/o EDEC

10–2

10–3

10–4

0.1

0

0.2

0.3

0.4

0.5

θe /π Figure 4.14 Measured BER of the 40-Gb/s DQPSK signal as a function of demodulator phase error. The optical power before the preamplifier is –38.5 dBm, corresponding to an OSNR of 16 dB [49] (this figure may be seen in color on the included CD-ROM).

as a function of the demodulator phase error with the EDEC. The open symbols show the calculated BERs without the EDEC. The EDEC effectively compensates for these demodulator errors with 40 Gb/s Optical Receivers,” in Proc. OFC 2004, Paper TuM4. [47] H.-G. Bach, A. Beling, R. Kunkel et al., “Ultra-high Bandwidth Optical Detectors (100 GHz) and Monolithic Photoreceivers for 80 Gbit/s NRZ/RZ Transmission Systems,” in Proc. ECOC 2004, Paper Tu3.1.2, 2004. [48] A. Beling, H.-G. Bach, G. G. Mekonnen et al., “Miniaturized waveguide-integrated p-i-n photodetector with 120-GHz bandwidth and high responsivity,” IEEE Photon. Technol. Lett., 17(10), 2152–2154, October 2005. [49] A. Beling, J. C. Campbell, H.-G. Bach et al., “High-Power Parallel-Fed Travelling Wave Photodetector Module for >100 GHz Applications,” in Proc. OFC 2007, Post-deadline paper PDP29, 2007. [50] Y. Suzuki, Z. Yamazaki, Y. Amamiya et al., “120-Gb/s Multiplexing and 110-Gb/s Demultiplexing ICs,” IEEE J. Solid-St. Cir., 39(12), 2397–2401, December 2004. [51] K. Ishii, H. Nosaka, K. Sano et al., “High-bit-rate low-power decision circuit using InP-InGaAs HBT technology,” IEEE J. Solid-St. Circ., 40(7), 1583–1588, July 2005. [52] I. D. Phillips, A. D. Ellis, T. Widdowson et al., “100 Gbit/s optical clock recovery using an electrical phase locked loop consisting of commercially available components,” IEE Electron. Lett., 36(7), 650–652, March 2000. [53] U. Du¨mler, M. Mo¨ller, A. Bielik et al., “86 Gbit/s SiGe receiver module with high sensitivity for 160 86 Gbit/s DWDM system,” IEE Electron. Lett., 42(1), 21–22, January 2006. [54] C. Schubert, R. H. Derksen, M. Mo¨ller et al., “107 Gbit/s Transmission Using An Integrated ETDM Receiver,” in Proc. ECOC 2006, Paper Tu1.5.5, 2006. [55] R.-E. Makon, R. Driad, K. Schneider et al., “80 Gbit/s Monolithically Integrated Clock and Data Recovery Circuit With 1:2 DEMUX using InP-based DHBTs,” in CISC 2005 digest, 2005, pp. 268–271. [56] J. H. Sinsky, A. L. Adamiecki, L. Buhl et al., “107-Gbit/s Opto-Electronic Receiver with Hybrid Integrated Photodetector and Demultiplexer,” in Proc. OFC 2007, Post-deadline paper PDP30, 2007. [57] G. Raybon, P. J. Winzer, and C. R. Doerr, “10 107-Gbit/s Electronically Multiplexed and Optically Equalized NRZ Transmission over 400 km,” in Proc. OFC 2006, Post-deadline paper PDP32, 2006. [58] P. J. Winzer, G. Raybon, and C. R. Doerr, “10 107 Gb/s Electronically Multiplexed NRZ Transmission at 0.7 bits/s/Hz over 1000 km Non-Zero Dispersion Fiber,” in Proc. ECOC 2006, Invited paper Tu1.5.3, 2006. [59] R. H. Derksen, G. Lehmann, C.-J. Weiske et al., “Integrated 100 Gbit/s ETDM Receiver in a Transmission Experiment over 480 km DMF,” in Proc. OFC 2006, Post-deadline paper PDP37, 2006. [60] K. Schuh, B. Junginger, E. Lach, and G. Veith, “1 Tbit/s (10 107 Gbit/s ETDM) Serial NRZ Transmission over 480 km SSMF,” in Proc. OFC 2007, Post-deadline paper PDP23, 2007. [61] K. Schuh, E. Lach, B. Junginger et al., “8 107 Gbit/s Serial Binary NRZ/VSB Transmission over 480 km SSMF with 1 bit/s/Hz Spectral Efficiency and without Optical Equalizer,” in Proc. ECOC 2007, Invited paper Mo2.3.1, 2007. [62] M. Daikoku, I. Morita, H. Taga et al., “100 Gbit/s DQPSK Transmission Experiment without OTDM for 100 G Ethernet Transport,” in Proc. OFC 2006, Post-deadline paper PDP36, 2006.

200

Karsten Schuh and Eugen Lach

[63] A. Sano, H. Masuda, Y. Kisaka et al., “14-Tb/s (140 111 Gb/s PDM/WDM) CSRZ DQPSK Transmission over 160 km using 7 THz Bandwidth Extended L-band EDFAs,” in Proc. ECOC 2006, Post-deadline paper Th4.1.1, 2006. [64] H. Masuda, A. Sano, T. Kobayashi et al., “20.4-Tb/s (204 111 Gb/s) Transmission over 240 km,” in Proc. OFC 2007, Post-deadline paper PDP20, 2007. [65] P. J. Winzer, G. Raybon, C. R. Doerr et al., “2000km-WDM Transmission of 10 107 Gb/s RZDQPSK,” in Proc. ECOC 2006, Post-deadline paper Th4.1.3, 2006. [66] P. J. Winzer, G. Raybon, S. Chandrasehkar et al., “10 107-Gb/s NRZ-DQPSK Transmission at 1.0 b/s/Hz over 12 100 km Including 6 Optical Routing Nodes,” in Proc. OFC 2007, Postdeadline paper PDP24, 2007. [67] C. R. S. Fludger, T. Duthel, D. van den Borne et al., “10 111-Gb/s, 50 GHz Spaced, POLMUXRZ-DQPSK Transmission over 2375 km Employing Coherent Equalization,” in Proc. OFC 2007, Post-deadline paper PDP22, 2007. [68] M. Duelk and M. Zirngibl, “100 Gigabit Ethernet–Applications, Features,” High-Speed Networking Workshop: The Terabits Challenge in conjunction with Infocom 2006. [69] M. Duelk, “Next-Generation 100 G Ethernet,” in Proc. ECOC 2005, Invited paper Tu3.1.2, 2005. [70] A. Kirsta¨dter, “Carrier-Grade Ethernet for Core Networks,” in Proc. OFC 2007, Paper OWK4, 2007. [71] J. Spencer, M. Du¨ser, A. Zapata et al., “Design Considerations for 100-GIGABIT Metro Ethernet (100 GbME),” in Proc. ECOC 2003, Th1.4.3, 2003.

6 Ultra-high-speed OTDM transmission technology Hans-Georg Weber and Reinhold Ludwig Fraunhofer Institute for Telecommunications, Heinrich-HertzInstitut, Einsteinufer 37, Berlin, Germany

6.1 INTRODUCTION The expected increase of transmission capacity in optical fiber networks presents us with many technology challenges [1]. One of these challenges is an optimized combination of wavelength-division multiplexing (WDM) and time-division multiplexing (TDM). TDM may be realized by electrical multiplexing (ETDM) or by optical multiplexing (OTDM) to a high-speed data signal, i.e., by electrical signal processing or optical signal processing. Presently, the first 40 Gb/s systems based on ETDM have been installed and in laboratories the first 100 Gb/s ETDMexperiments have been performed, e.g., Ref. [2]. In contrast, at the same data rates, OTDM transmission experiments have been carried out more than 10 years earlier. For instance, the first 100 Gb/s OTDM transmission experiment over a 36 km fiber link was reported already in 1993 [3]. Since then, OTDM transmission technology has been made a lot of progress towards much higher bit rates and much longer transmission links, as has been described in several review articles [4–6] and will be discussed in this chapter with special emphasis on the most recent developments. Recently, OTDM transmission technology succeeded in the transmission of a TDM data rate of 160 Gb/s over a record fiber length of 4320 km [7] and of a TDM data rate of 2.4 Tb/s over a fiber link length of 160 km [8]. The past has seen that OTDM will be replaced by ETDM as soon as electrical signal processing becomes available at the required TDM data rate. Therefore, OTDM transmission technology is often considered to be an interim technique with which to investigate the feasibility of ultra-high-speed data transmission in fibers. Today, the ultimate limits of ETDM technology are not known. An ETDM data rate of 160 Gb/s is very likely in the future. However, the performance of the ETDM terminal equipment may eventually be worse than the OTDM terminal Optical Fiber Telecommunications V B: Systems and Networks Copyright  2008, Elsevier Inc. All rights reserved. ISBN: 978-0-12-374172-1

201

202

Hans-Georg Weber and Reinhold Ludwig

equipment. OTDM receivers perform better than ETDM receivers already at data rates of 80 Gb/s. In this chapter we will show that the OTDM terminal equipment for 160 Gb/s TDM data transmission provides already very stable operation conditions. These results suggest that 160 Gb/s OTDM transmission systems can be used in deployed systems and can be operated error-free for years. ETDM technology must work hard to compete with these systems for instance as regards receiver sensitivity. Moreover, it is not sure that ETDM terminal equipment for 160 Gb/s will be less expensive and less energy consuming than the corresponding OTDM terminal equipment. The ultimate limits of OTDM transmission technology are not given by the terminal equipment but by the transmission properties of the fiber link including all repeater or amplifier stages. With higher TDM bit rate data transmission in fiber is stronger affected by chromatic dispersion (CD), polarization-mode dispersion (PMD), fiber nonlinearity, and the limited bandwidth of repeaters or amplifiers in the transmission link. This is independent of the signal processing in the terminal equipment whether it is based on ETDM or OTDM technology. At present therefore, a main task of OTDM technology is to explore the ultimate capacity for fiber transmission in a single wavelength channel. The most challenging view as regards OTDM technology is that optical networks will evolve into “photonic networks,” in which ultra-fast optical signals of any bit rate and modulation format will be transmitted and processed from end to end without optical–electrical–optical (O/E/O) conversion. This “photonic network” is a target for the distant future, and it presents us with the present challenge of investigating and developing high-speed optical signal processing and exploring the ultimate capacity for fiber transmission in a single wavelength channel. OTDM transmission technology is described in this chapter as follows: Section 6.2 gives a general description of an OTDM system, followed by a discussion of the OTDM transmitter in Section 6.3, of the OTDM receiver in Section 6.4 and of the fiber transmission line in Section 6.5. Transmission experiments are described in Section 6.6 starting with a review on 160 Gb/s transmission experiments in Section 6.6.1, on transmission at data rates beyond 160 Gb/s in Section 6.6.2, and on OTDM/WDM transmission experiments in Section 6.6.3. Then follows a detailed description of two 160 Gb/s transmission experiments: transmission with long-term stability in Section 6.6.4 and transmission over a record fiber length of 4320 km in Section 6.6.5. In Section 6.6.6 we report on transmission experiments at the TDM data rates of 1.28 and 2.56 Tb/s. Finally, Section 6.7 summarizes our conclusions on the present state of OTDM technology.

6.2 OTDM TRANSMISSION SYSTEM Figure 6.1 (upper part) is a schematic depiction of a 160 Gb/s OTDM transmission system as an example. On the transmitter side, the essential component is an optical pulse source. The repetition frequency of the generated pulse train is

6. Ultra-High-Speed OTDM Transmission Technology Fiber link

160 Gb/s OTDM transmitter

203 160 Gb/s OTDM receiver 40 Gb/s receiver

MOD MOD Pulse source

DEMUX MOD

160 Gb/s

40 GHz

40 Gb/s receiver 40 Gb/s receiver 40 Gb/s receiver

MOD

4 × 40 Gb/s

4 × 40 Gb/s Simplified laboratory system

40 Gb/s transmitter

40

160 Gb/s MUX

160 40 Gb/s DEMUX

40 Gb/s receiver

Figure 6.1 Schematic view of a 160 Gb/s OTDM transmission system (upper part of figure) and of a simplified laboratory system (lower part of figure).

assumed to be 40 GHz. In general, the repetition frequency depends on the base data rate (or on the symbol rate, see Section 6.3.2), at which the subsequent modulators (MOD) are driven by an electrical data signal (base rate signal). The 40 GHz optical pulse train is coupled into four optical branches, in which four 40 Gb/s non-return-to-zero (NRZ) electrical data signals drive four modulators (MOD) and generate four 40 Gb/s optical return-to-zero (RZ) data signals. The modulation formats include on–off keying (OOK) and differential phase shift keying (DPSK). The setup in Figure 6.1 can also be used for the modulation format differential quadrature phase shift keying (DQPSK). However, in this case one has to take into account that the bit rate (expressed in bit per second, b/s) is twice the symbol rate (expressed in baud, for details see Section 6.3.2). The four optical data signals (TDM channels) are combined to a multiplexed 160 Gb/s optical data signal by bit-interleaving with use of an appropriate delay in the four optical branches. Multiplexing (MUX) can be such that all bits of the multiplexed data signal have the same polarization (single polarization (SP) signal, SP multiplexing) or adjacent bits have alternating (orthogonal) polarization (AP multiplexing, AP-signal). On the receiver side, the essential component is an optical demultiplexer (DEMUX), which separates the four base rate data signals (TDM channels) again for subsequent detection and electrical signal processing. The transmission link requires in general compensation for CD and PMD, which both depend on the type of singlemode fiber (SMF) used in the transmission system. The DEMUX shown in Figure 6.1 comprises two parts, an optical gate and a clock recovery device. The optical gate is a fast switch with a switching time that is shorter than the bit period (6.25 ps for 160 Gb/s) of the multiplexed data signal.

204

Hans-Georg Weber and Reinhold Ludwig

The clock recovery device provides the timing signal for the optical gate. In Sections 6.3, 6.4, and 6.5 we discuss the OTDM transmitter, the OTDM receiver, and the fiber transmission line in more detail. Laboratory systems are frequently simplified as follows (Figure 6.1, lower part): on the transmitter side, only one modulator is used and combined with the pulse source for a 40 Gb/s optical transmitter. The generated optical data signal is then multiplexed by a fiber delay line multiplexer (MUX) to a 160 Gb/s data signal using either SP or AP multiplexing. On the receiver side, the DEMUX selects only one 40 Gb/s TDM channel, which is detected by one 40 Gb/s optoelectronic receiver at a given time. In a proper experiment all TDM channels are measured successively in this way. In this chapter we discuss point-to-point transmission as shown in Figure 6.1. We do not consider network components and subsystems such as add–drop multiplexers, e.g., Refs [9–12], wavelength converters, e.g., Refs [13–16], modulation format converters, e.g., Ref. [17], optical regenerators, e.g., Ref. [18] and optical sampling systems, e.g., Refs [19, 20].

6.3 OTDM TRANSMITTER 6.3.1 Pulse Sources The pulse source is a key component in an OTDM transmitter. The pulse source must provide a well-controlled repetition frequency and wavelength, a pulse width significantly shorter than the bit period of the multiplexed data signal, a timing jitter much less than the pulse width, low-amplitude noise, and a high extinction ratio. For instance, typical values for stable 160 Gb/s (bit period 6.25 ps) OOK transmission are RMS jitter Crystal frequency contour -Outgoing k vector --> Outgoing vg vector --> Crystal interface Conserved k line

2

kz

1 0 –1 –2 –3

–4

–3

–2

–1

0

1

2

3

4

kx 3 2

κy a

1 0 –1 –2 –3 –3

–2

–1

0

1

2

3

κx a Figure 20.14 (a) Wavevector construction for superprism refraction at the surface of a 2D hexagonal photonic crystal. A wave in free space (blue arrow) is refracted into a wave in the crystal lying on the red iso-frequency contour. The refraction direction is determined by conservation of the tangential component of momentum. (b) Map of the resolving power of a superprism (this figure may be seen in color on the included CD-ROM).

vectorial Maxwell’s equations to be solved in the Fourier domain. The MTL [39, 40] approach is equivalent to RCWA, except that MTL uses summations over individual modes to represent the electromagnetic fields. This method has the advantage of describing the fields in terms of an equivalent transmission-line network, which yields physical insight for various applications using different mechanisms. Another benefit of the MTL method is that it eliminates possible stability problems, which may be encountered in other numerical schemes, such as the transfer matrix method.

Robert Scarmozzino

822

The geometry is assumed to consist of homogenous incident and outgoing regions, with an arbitrary, but periodic structure sandwiched between them. For typical scattering problems, we want to calculate the reflected and transmitted light waves from the incident field. The structure is sliced along the primary direction, z, into “layers.” Within each layer the structure is assumed independent of z, and periodic in the transverse directions x and y. According to Bloch’s Theorem, the transverse field components in a periodic layer can be expressed as X X i 2 pxþ 2 qy X   ax; m; p; q fm eim z þ gm eim z ; e x y Ex = eiðkx;0 xþky;0 yÞ Ey = e

iðkx;0 xþky;0 yÞ

Hx = e

iðkx;0 xþky;0 yÞ

p

q

m

p

q

m

p

q

m

p

q

m

X X i 2 pxþ 2 qy X   ay; m; p; q fm eim z þ gm eim z ; e x y

X X i 2 pxþ 2 qy X   e x y bx; m; p; q fm eim z  gm eim z ;

X X i 2 pxþ 2 qy X   iðkx;0 xþky;0 yÞ by; m; p; q fm eim z  gm eim z : Hy = e e x y Substituting the above into Maxwell’s equations results in an eigenvalue problem for the modes: Ax = x; where = 2m is the eigenvalue to be calculated, and the a and b coefficients in the field representations are derived from the corresponding eigenvectors. A solution for the entire structure can be obtained using well-known transmission-line methods to solve the required boundary-value problem. The scattering characteristics can then be obtained in the following systematic fashion. The modal voltage and current amplitude for jth layer are vðmjÞ = fmð jÞ eim z þ gðmjÞ eim z ; iðmjÞ = fmð jÞ eim z  gðmjÞ eim z :

The superscript j indicates that the various quantities are for a specific ( jth) layer. Imposing the required boundary condition (at zjjþ1 ) between the jth and ( j þ 1)th layers result in X  ax; m; p; q ð jÞ m

ay; m; p; q

m

by; m; p; q

X  bx; m; p; q ð jÞ

vðmjÞ ðzjjþ1 Þ = iðmjÞ ðzjjþ1 Þ =

X  ax; m; p; q ð jþ1Þ m

ay; m; p; q

m

by; m; p; q

X  bx; m; p; q ð jþ1Þ

vðmjþ1Þ ðzjjþ1 Þ;

ð jþ1Þ jþ1 im ðzj Þ:

This approach not only facilitates physical insights into the problem but also provides important numerical advantages [39, 40].

20. Simulation Tools for Devices, Systems, and Networks

823

Applications RCWA can be used to determine the reflection and transmission, from a periodic structure, of a plane wave, as a function of wavelength and incident angle. Numerous applications can benefit from this technique including outcoupling from surface gratings, diffractive optical element design, and metrology. Figure 20.15 shows several reflected orders of a blazed surface grating at a particular wavelength. Each order is peaked at a different angle. The simulation can also be stepped through wavelength at a particular angle, to produce the optical spectrum. Figure 20.16 shows the field pattern of the –1 reflected order which is peaked at an angle of 20 . The periodicity is along x in this example. Reflected orders from blazed grating Diffraction efficiency (a.u.)

1.0

–2R –1R

0.8

0R 1R

0.6

2R

0.4 0.2 0.0 –80 –60 –40 –20

0

20

40

60

80

Launch angle (degree) Figure 20.15 Diffraction efficiency of five reflected orders of a blazed surface grating (this figure may be seen in color on the included CD-ROM).

Field reflected from blazed grating 4.0 0.2

Z (μm)

0.0 –0.2 –0.4 –0.6 –0.8 –1.0 0

1

2

3

0.0

X (μm) Figure 20.16 Field of the –1 reflected order (20 from normal) of a blazed grating (shown outlined in white and inverted, at top of figure) (this figure may be seen in color on the included CD-ROM).

Robert Scarmozzino

824 Contour map of index profile 0.5

2.298

0.4

Z (μm)

0.3 0.2 0.1 0.0 –0.1 –0.8 –0.6 –0.4 –0.2

0.0

0.2

0.4

0.6

0.8

1.0

X (μm) Figure 20.17 Surface filter with a very sharp resonance that is both a function of incident angle and wavelength (this figure may be seen in color on the included CD-ROM).

Figure 20.17 depicts several periods of a resonant grating filter (RGF). These exploit a resonance condition that arises when incident light gets trapped in the waveguide via evanescent coupling. When this light is out-coupling again, it interferes destructively with the incoming light, producing very sharp resonances for narrow ranges of incident angle and wavelength. Hence only a few layers are needed for narrow band reflection. The angular dependence also provides a mechanism for tunability. The response of this filter is shown in Figure 20.18 as a contour plot of angle and wavelength. The first inset shows a cut along wavelength taken at 40 and the other inset shows a cut along angle taken at 1.55 mm. Zero-order reflection vs angle vs wavelength 50

1.0

Launch angle (degree)

At 40°

At 1.55 μm 40

30 1.50

1.52

1.54

1.56

1.58

1.60

–1.0

Wavelength (μm) Figure 20.18 Filter response showing the linear variation of the peak with angle and wavelength along with insets showing a single angle (40 ) and a single wavelength (1.55 mm) (this figure may be seen in color on the included CD-ROM).

20. Simulation Tools for Devices, Systems, and Networks

825

20.2.2 Active Optical Device Modeling Active optical device modeling requires the simultaneous calculation of the optical field and the gain medium. For semiconductor-based active optical devices such as lasers, amplifiers, and modulators, the gain medium is pumped by carrier injection, requiring the solution of the electronic transport equations, and in some cases, the thermal transport equation as well. Various approaches are present in the literature, but generally fall into two broad categories; those that solve the coupled rate equations in time only and those that solve the transport equations on a spatial mesh. Additionally, there is a traveling wave approach which falls somewhere in between, accounting for time and one spatial dimension. Background The standard rate-equation approach typically includes the following three equations: the carrier density Eqn (20.16), the photon density Eqn (20.17), and the phase of the cavity mode Eqn (20.18) [41–46]. dN i I =  Rspon ðNÞ  Rnr ðNÞ  GðN; SÞS; dt qV dS S =  þ spon Rspon ðNÞ þ GðN; SÞS; dt    d  1 "S = GðN; SÞ  þ : dt 2  

(20.16) (20.17) (20.18)

Here, I is the injection current, N is the carrier density, S is the photon density,  is the optical phase,  is the current injection efficiency, q is the electron charge, V is the cavity volume,  is the photon lifetime, G is the confinement factor,  is the spontaneous emission coupling coefficient,  is the linewidth enhancement factor, and  is a gain saturation parameter. Also, R represents the spontaneous and nonradiative recombination and G is the optical gain, both of which depend on carrier density and temperature. R and G have various analytic forms depending on which physical processes are being modeled. Standard methods for solving sets of coupled first-order differential equations, such as the fourth-order Runge–Kutta method, are typically employed to resolve the above system. The second approach, which is attempts to include the spatial variations of the carriers and fields, resolves the physical equations at the nodes of a spatial mesh, and is hence much more computationally intensive [49–51].  þ   NA þ p  n = 0; r  "stat r’ þ q ND @n þ r  jn þ U = 0; @t

(20.19) (20.20)

Robert Scarmozzino

826

@p þ r  jp þ U = 0; @t

@Sm; ! @t

r2 Em þ "opt k02 Em = 0;   1 Sm; ! þ Rspon = Gm;!  m; ! : n;!

(20.21) (20.22) (20.23)

The carrier dynamics are governed by the Poisson equation (20.19) and the continuity equations for electrons (20.20) and holes (20.21). Here,  is the electrostatic potential, n and p are the electron and hole densities, respectively, and ND and NA are the ionized donor and acceptor concentrations, respectively. U represents the total generation and recombination, both radiative and nonradiative (such as Auger and Shockley–Read–Hall (SRH) recombination), while j is the current density. The optical equations are split into two parts; the first being a spatial part (20.22) from which the modal eigenvalues and field patterns, E, are determined. The second is the photon rate equation (20.23), which connects the photon population to the carrier densities via the modal gain, G, and spontaneous recombination Rspon. In most modern devices, there is the additional complication of quantum wells, which are used engineer the gain spectrum. These necessitate the solution of a multiband Schroedinger equation to determine the carrier distributions in energy and space, as well as their transition rates. The most common numerical approach for this is the KP method [52–57], which is a perturbation technique, typically employing a basis of four or eight bands. This system of equations must be solved self-consistently; usually by iterating between a fully coupled Newton–Raphson solution of the transport (20.19–20.21) and photon rate equation (20.23), and independent solutions of the modes (20.22) and gain (KP method). Such an approach is inherently unstable and a variety of techniques are required to insure a convergent simulation. Applications A broad range of devices may be simulated by the methods described in the previous section; Fabry–Perot (FP) edge-emitting lasers, VCSELs, distributed feedback lasers (DFBs), semiconductor optical amplifiers (SOAs), LEDs, electroabsorptive (EA) modulators, and photodetectors (PDs) to name a few. The particular approach used depends on the information desired. For example, if the system performance (discussed in subsequent sections of this chapter) of the device is required, the rate-equation approach is better suited. However, if a particular geometry or material composition must be optimized to improve the design, then a spatial simulation is necessary. While empirical models for spatial hole burning, self-heating, and mode competition, can be found in the rate equation approach, a spatial simulation can give a clear picture of these effects and how they impact device performance. Additionally, numerous parameters can be calculated directly from the simulation results, such as modal confinement factors, nearfield and farfield patterns, energy bands, carrier densities, current fluxes, and

20. Simulation Tools for Devices, Systems, and Networks

827

lattice temperature profiles. The following examples are spatial simulations of an FP laser and a VCSEL, and will be used to demonstrate the type of information that can be obtained from the spatial mesh approach, and its importance to device design. Example 1: Fabry–Perot Edge-Emitting Laser Diode. An FP laser is typically formed with a single or multiple quantum well active region surrounded by a separate confinement layer for optical confinement in the vertical (Y) direction. An etched ridge structure is used for optical confinement in the lateral direction (X). Finally, the device is cleaved front and back (Z), creating semiconductor–air interfaces (30% reflectivity) which form an FP cavity (shown schematically in Figure 20.19). Figure 20.19 also shows the nonuniform simulation mesh that digitizes the structure. Note that the symmetry of the problem has been exploited for computational efficiency, and only the right half of the structure is simulated. Nonuniform meshing is critical to resolve all the layers of the device, which can vary in thickness by orders of magnitude, especially if quantum wells are present. The simulation begins with a cold cavity mode calculation and the calculation of the material gain, both shown in Figure 20.20. In the case of FP lasers, the (longitudinal) mode density is so high that the lasing wavelength is assumed to track the gain peak (about 980 nm for this example). In other types of lasers, such as VCSELs and DFBs the cavity has better mode selectivity, and must be carefully designed so that the fundamental resonance collocates closely with the peak of the gain. Next, the transport can be simulated, giving rise to the L–I–V curve shown in Figure 20.21. From this information, the lasing threshold (10 mA) and slope efficiency of the device can be determined. If temperature is included in the simulation, the L–I curve can begin to roll-off due to self-heating effects in the active region that cause the gain to saturate. The dynamic performance of the Z

Nonuniform mesh (621 nodes, 544 polygons)

Vertical direction (μm)

Y

X

2.1 1.8 1.5 1.2 0.9 0.6 0.3 0

1

2

3

4

5

6

7

Horizontal direction (μm)

Figure 20.19 Schematic of a ridge (FP) laser (shown left), and nonuniform simulation mesh (shown right) superimposed on the device structure (only right half is simulated). The active region, with single quantum well, is shown in the inset (with and without the mesh) (this figure may be seen in color on the included CD-ROM).

Robert Scarmozzino

828 Mode[0,0] (N eff = 3.422298)

1.0

2.1

Legend: N = 1.0e – 12 cm–2

1.8

N = 1.3e – 12 cm–2

Gain (cm–1)

Y direction (μm)

Material gain

×104 0

1.5 1.2 0.9

N = 1.67e + 12 cm–2

–1

N = 2.0e + 12 cm–2

–2

0.6 0.3 –3 –6

–4

–2

0

2

4

0.0

6

0.6

0.7

0.8

0.9

1.0

1.1

Wavelength

X direction (μm)

Figure 20.20 Fundamental mode (left) along with an outline of the structure (white), and material gain spectra (right) at several carrier densities, as generated by the KP method, for a single InGaAs/AlGaAs quantum well (this figure may be seen in color on the included CD-ROM).

Light and voltage vs current

1 10

0

0 0

10

20

30 Current (mA)

40

50

60

–3 dB

Light (mW)

20

Transient response

20.153

0

Response (dB)

Voltage (V) {blue}

30

Optical power (mW) {green}

Frequency response

2

20.152

0

200

400

600

800

1000

Time (ps)

–10

–20 0

10 20

30 40 50 60 70 80 Frequency (GHz)

90 100

Figure 20.21 L–I–V curve for this device (left), and frequency response (right) at a bias of about 20 mA. The transient response of the light output to a small injected square pulse (right) is shown in the inset (this figure may be seen in color on the included CD-ROM).

device can also be obtained from this type of simulation. For example, a small signal response can be obtained by injecting a small current pulse into the bias contact of the device and observing the resulting perturbation in the optical output power. Figure 20.21 also depicts both the time- and frequency-domain responses, revealing a –3dB cutoff frequency in the 15–20 GHz range, for this device. Example 2: Vertical Cavity Surface-Emitting Laser. A VCSEL has several advantages over FP lasers. It emits from the surface of the chip, taking up less real estate, and has a mode that couples better to fibers. Also, the cavity mode selectivity ensures better stability of the lasing wavelength. However, a commensurate disadvantage is the increased thermal sensitivity, caused by temperature detuning between the lasing wavelength and the gain peak. Hence, many factors must be considered in the proper design of a VCSEL, making the use of simulation critical. In the following example, a cylindrical oxide aperture VCSEL is discussed. Shown schematically in Figure 20.22, this device has several GaAs/Al0.2Ga0.8As

20. Simulation Tools for Devices, Systems, and Networks

829

Active region

R

Mode intensity (×1e–16) (blue) index profile (green)

Fundamental cavity mode and index profile Y

4

QWs Top facet

Substrate

3

2

1

0 0

1

2

3

4

5

6

7

8

9

10

Y direction (μm) Figure 20.22 Schematic of a cylindrical VCSEL (left), and simulated mode and index profiles (right) (this figure may be seen in color on the included CD-ROM).

quantum wells centered in a lambda cavity that is terminated by top (32 periods) and bottom (36 periods) DBR mirrors. Also, near the center of the cavity is a thin oxide layer with small diameter aperture that is responsible for confining modes and carriers toward the center of the cylinder, where their overlap will be optimal. The vertical profile of the mode is also shown in Figure 20.22, along with the index profile. The inset shows an expanded view of the quantum wells and their overlap with the central intensity peak. Device operation is strongly dependent not only on how the mode overlaps with the quantum wells, but also on how it intersects with the oxide aperture. Figure 20.23 shows the transmission spectrum of the cavity superimposed with the material gain, at several temperatures and a constant sheet carrier density of 2 1012 cm–2. Clearly, the gain peak undergoes a red shift with increasing temperature, that changes its relative position with respect to the lasing resonance, which remains fixed. This detuning results in a roll-over of Thermal roll-over

Light and voltage vs current 3

7

Gain peak

6

0 T = 300 K T = 312 K T = 325 K

Mode resonance

2

5 4 3

1

2 1

–1

0 0.75 0.78 0.81 0.84 0.87 0.90 0.93 0.96

Wavelength (μm)

Optical power (mW)

8

1

Voltage (V)

Transmission (a.u.) Material gain (×1e – 3 cm–1)

Transmission spectra and material gain

0 0

10

20

Current (mA)

Figure 20.23 Transmission spectrum of the cavity shown superimposed with the material gain at several temperatures (left), and the L–I–V curve depicting the effects of thermal roll-over (right) (this figure may be seen in color on the included CD-ROM).

830

Robert Scarmozzino

the L–I curve, also shown in Figure 20.23, and can actually cause the device to turn off at higher bias currents. A caveat with active device simulation on a spatial mesh is that it requires accurate knowledge of many material parameters, and in the end, may still deviate from experimental results due to the un-simulated portions of the problem (i.e., those not within the simulation domain). For example, parasitic effects due to the laser die or package, or even due to the measurement apparatus itself, may cause discrepancies between simulation and experiment. The power of these simulations comes from their ability to show the performance trends of the device, as its material or geometric parameters are varied.

20.3 OPTICAL COMMUNICATION SYSTEM SIMULATION In this section we explore system-level modeling. In optical communication system modeling, the information is represented by optical, electrical, and binary signals and is propagated from a transmitter to a receiver through various network components. Each component is modeled with equations representing the underlying physics to capture the way it processes its input signal to produce an output signal. Network components include optical fiber, DFB and VCSEL lasers, EDFA and SOA amplifiers, electrical and optical filters, photodetectors, electrical amplifiers, and many others. The optical fiber is the medium in which the signal propagates for distances of up to thousands of kilometers. The fiber type divides optical communication systems into two main categories, single-mode and multimode. In single-mode systems the small fiber core diameter enables the propagation of only two degenerate modes, the X and Y polarizations. These systems ensure the highest bandwidth and reach. In multi-mode systems a larger fiber core enables the propagation of multiple modes, each characterized by its own loss, propagation constant, and spatial profile. These systems have lower bandwidth, but also advantage in terms of cost. In this section, we describe some of the available methods for single- and multi-mode system simulation, with particular emphasis on optical fiber modeling. We also present some of the methods available for estimating system performance in terms of bit error rate (BER) and the related Q parameter. The following discussion is by no means exhaustive, and the interested reader can find further material in the references section.

20.3.1 Single-Mode Systems As outlined above, single-mode systems incorporate a wide range of network components. Simulation requires modeling of each of these elements, thereby allowing a designer to assess the impact of transmitters, transmission media, and receivers on an optical link’s overall performance. For example, a designer may

20. Simulation Tools for Devices, Systems, and Networks

831

wish to compare the advantages and disadvantages of direct versus external modulation in the transmitter, or the impact of different optical amplification schemes on system performance. Arguably the fiber itself plays the largest role in determining a single-mode link’s performance. Linear effects such as loss, dispersion, and polarization mode dispersion (PMD) can lead to pulse spreading which ultimately undermines the optical signal-to-noise ratio (OSNR) at the receiver. Nonlinear behaviors such as four-wave mixing (FWM), self-phase modulation (SPM), cross phase modulation (XPM), stimulated Brillouin scattering (SBS), stimulated Raman scattering (SRS), modulation instability, and parametric gain can further degrade a signal’s transmission due to crosstalk and additional distortion effects. In this section, we discuss the system-level modeling of singlemode fiber, and present an example of single-mode system simulation. The propagation of an optical pulse in a single-mode fiber is governed by the nonlinear Schro¨dinger equation, which can be written as @A @A i @ 2 A  þ A = i jAj2 A: þ 1 þ 2 @z @t 2 @t2 2

(20.24)

In Eqn (20.24), A is the complex field envelope, z is the distance,  1 is inversely proportional to the group velocity g, 2 accounts for the group velocity dispersion (GVD),  accounts for the effects of fiber loss, and is the nonlinear coefficient. Equation (20.24) can be modified to include the effects of third-order dispersion, amplification, amplified spontaneous emission noise, and PMD to represent a more comprehensive model of single-mode optical fiber transmission [58–63]. Only the HE11 mode propagation is relevant in single-mode optical fiber systems. However, due to the circular symmetry of the fiber, there are two orthogonal HE11 solutions. Ideally, the two HE11 modes should be degenerate, but fiber birefringence can break the degeneracy. Therefore, it is important to account for the fact that the fiber propagation is bimodal, and that both linear and nonlinear phenomena shuffle optical power between the two modes in complicated ways. This mode mixing of the propagating light can impact certain effects that need a high degree of coherence to develop. Linear and nonlinear birefringence, PMD, FWM, and sideband instability (SI) are a few of the phenomena that require a bimodal or vector signal representation to be properly investigated. To include bimodal effects, the mode amplitude in the spectral domain can be written as A = Að ; zÞ =

"

Ax ð ; zÞ Ay ð ; zÞ

#

;

(20.25)

where Ax ( , z) and Ay ( , z) are the components of the mode amplitude vector that are defined with respect to a linear ð^ x; y^Þ polarization basis. The frequency

refers to the complex envelope representation [58], i.e., = ! – !0, where !0 is the nominal optical transmission frequency.

Robert Scarmozzino

832

The Split-Step Method The propagation equation (Eqn (20.24)) is a second-order nonlinear partial differential equation that does not generally have an analytical solution. In almost all cases, numerical approaches have to be employed. The split-step Fourier method is the most commonly used numerical scheme that is employed for solving Eqn (20.24). This method is considerably faster than most finite difference methods due to the use of FFT algorithms [64]. It is also a very convenient method due to its simplicity and flexibility in dealing with higher order dispersion, Raman effects, and filtering [58]. All commercial optical system simulation tools use the split-step method to perform the integration of the fiber propagation equation. The form of such an equation is as follows: @Aðt; zÞ = fL þ N gAðt; zÞ; @z

(20.26)

where A(t,z) is the optical field, L is the linear operator responsible for dispersion and other linear effects, and N is the nonlinear operator that accounts for the Kerr effect and other nonlinear effects like SRS or pulse self-steepening. The split-step integration algorithm works by separately applying L and N to A(t,z) over small spans of fiber Dz. The distribution of the step sizes along the fiber is an important factor in determining the efficiency of the split-step method. A number of step-size selection criteria, mostly based on physical intuition, have been used to optimize the split-step method. A computational cost for a given global accuracy serves as the figure of merit for each criterion. Some of the more commonly used methods are discussed in Ref. [65]. The split-step method is based on an approximate solution that assumes that over a small distance Dz, the dispersive and nonlinear effects can be considered to act independently. For this two-step approach, the solution of Eqn (20.26) can be expressed as Aðz þ DzÞ = expðDzLÞ expðDzN ÞAðz; tÞ:

(20.27)

A more commonly used solution, referred to as symmetric split-step, includes the effect of nonlinearity in the middle of the segment rather than at the segment boundary. For the symmetric split-step, the solution to Eqn (20.26) is expressed as     

  Dz Dz Dz L exp DzN A z þ ; t exp L : Aðz þ DzÞ  exp 2 2 2

(20.28)

Since the linear and nonlinear operators do not commute in general, both solutions (20.27) and (20.28) are only approximations to the exact solution. Using the Baker–Hausdorff formula [66], it can be shown that the symmetric

20. Simulation Tools for Devices, Systems, and Networks

833

split-step method improves the leading-order error term from second-order to third-order in the step size Dz when compared to solution (20.27). The frequency-domain split-step (FDSS) and the time-domain split-step (TDSS) methods are the two popular approaches that are used to numerically solve Eqn (20.28). These two methods differ in the way L is calculated. The nonlinear operator N is always calculated in the time domain by both methods. The techniques are briefly discussed in the next two subsections. Frequency-Domain Split-Step FDSS calculates L in the frequency domain [58], by multiplying the FFT of A[n], the signal in sampled time, with the FFT of h[n], the impulse response in sampled time. The resulting array is converted back to the time domain with an inverse FFT, as expressed in the following formula: FDSS ) A0L ½n = A½n  h½n = FFT1 ðFFTðA½nÞ FFTðh½nÞÞ:

(20.29)

It is well known from signal processing and simulation theory that FDSS as calculated above does not implement a linear convolution product, but a circular convolution, as indicated by the  symbol. The circular convolution can create a signal fold-over error effect, or aliasing, in the resulting array A0 L[n], which may contain less samples than needed to compute the actual convolution product AL[n]. The use of the FFT algorithm implies periodic boundary conditions. In cases where the pulse energy spreads rapidly, it may be difficult to avoid hitting the window boundary. This fundamental issue with the FDSS method can lead to numerical instabilities as the energy reaching one edge of the window automatically reenters the window from the other edge. Initially, it may appear that this approach introduces unphysical correlations due to the wrapping of signals and noise. In fact, these correlations may have only a very minor effect and can always be essentially eliminated by using longer bit patterns. Moreover, periodic boundary conditions are important for WDM simulations using FDSS since initially coincident pulses in different channels quickly separate due to walk-off. If the pulses did not re-enter the time window, interchannel interactions would be grossly underestimated leading to overoptimistic predictions for cross talk, gain tilt, and XPM effects. However, the wrapping of the signals and noise is intrinsic to the method, thus it is difficult to avoid when the effect is undesirable and extreme care must be exercised to ensure that the FDSS method is used properly. Time-Domain Split-Step Being a linear operator, L is fully characterized by its impulse response h(t) and the mathematically correct way to compute its effect on A(t,z) is via a convolution

Robert Scarmozzino

834

product in time. TDSS calculates L in the time domain by calculating the convolution product in sampled time [67], which can be written as TDSS ) AL ½n = A½n  h½n =

1 X

k = 1

A½kh½n  k:

(20.30)

The effective implementation of TDSS is more difficult than FDSS to achieve. In the TDSS method, the linear operator L can be either implemented by means of infinite impulse response (IIR) filters [67] or finite impulse response (FIR) filters. The FIR implementation can be achieved by automatically truncating the theoretically infinite h(t) impulse response of L and calculating the FIR filter with a sophisticated technique that ensures full control on the overall end-to-end error. One of the best implementations of the FIR filter algorithm in terms of complexity is the famous overlap-and-add algorithm [68] that takes advantage of FFTs while correcting the aliasing problems that affect FDSS. To filter M signal samples of A[n], the computational effort of L at each Dz (i.e., for each filter) is on the order of M[log2(K) þ 2], where K is the number of samples of the impulse response h(t) of the FIR filter which implements L. The drawback of this FIR filter implementation is the need to segment, block-process, and then streamline the data. For comparison, the computational complexity of FDSS for each L is on the order of M[log2(M) þ 1], which can be considerably larger. The intrinsic signal and noise wrapping that could possibly lead to numerical errors in the FDSS method do not occur in the TDSS method. TDSS imposes no constraints whatsoever on the shape and spectra of the optical signals. Furthermore, TDSS requires no tricky workarounds since it guarantees the absence of intrinsic errors. TDSS computes a true convolution in the time domain and therefore channels are free to shift in time. It should be noted that distant channels in a WDM system shift with respect to each other to an extent that they only partially overlap. Thus to preserve accuracy, the nonoverlapping bits are cut off at the end and only the bits that have always beaten with all channels are preserved in an efficient TDSS method. 40 Gbit/s NRZ, Duobinary, DPSK, and CSRZ-DPSK Single-Mode System Simulation Example The split-step methods described above can be used to simulate the propagation in optical fiber of single- and dual-polarization signals described in terms of amplitude and phase information sampled over the simulation time span. The analog nature of the signal representation makes the methods suitable for study of the performance of different modulation formats at a given bitrate. In the following we show the simulation results of four different modulation formats in ultrahigh spectrally efficient WDM systems. Besides standard nonreturn-to-zero (NRZ) intensity modulation, we analyze duobinary, DPSK, and

20. Simulation Tools for Devices, Systems, and Networks

835

CSRZ-DPSK formats, and optimize the system dispersion map for each of them. These examples were used to obtain results presented in [69]. The transmitter and receiver layout of the four analyzed modulation formats is shown in Figure 20.24. The different architectures, based on seven-channel WDM transmission, were simulated using a commercial optical communication system design tool. Each transmitter uses the ad hoc setup obtained after optimization of each modulation format under analysis. The channel spacing is set to 50 GHz and the bit rate to 42.65 Gbit/s (40 Gbit/s þ typical FEC overhead). The transmission line is a Laser

M-Z Mod

NRZ

PRBS gen

(...)

Electrical filter

Electrical filter

0 Vx Squared driver

PRBS gen

Laser

Duobinary precoder

M-Z Mod

–Vx

z

Duobinary

(...)z

Electrical filter

Electrical filter

Vx

Squared driver

M-Z Mod

Laser

DPSK

T

PRBS gen

(...)z (...)z

diff precoder AMZ –½Vx

Electrical filter

+½Vx

Squared driver

Electrical filter

CSRZ-DPSK Laser +Vz

M-Z Mod

M-Z Mod

f = Rb / 2 –Vz

–½Vx +½Vx Squared driver diff precoder PRBS gen

Electrical filter

T

(...)z (...)z

AMZ Electrical filter

Figure 20.24 From top to bottom: NRZ, duobinary, DPSK, and CSRZ-DPSK transmitter and receiver layout [69].

Robert Scarmozzino

836

five-span periodically amplified fiber link based on nonzero dispersion-shifted fiber (NZDSF) with dispersion parameter D = 4.46 ps/nm/km, dispersion slope S = 0.09 ps/nm2/km, and loss dB = 0.22 dB/km. All values are referred to the wavelength of the WDM comb’s center channel. The length of each span is 100 km, and a variable attenuator is used after the fiber spans to set the overall span loss to a fixed value tot. Dispersion management is obtained via a Dispersion compensation unit (DCU) inserted between two amplification stages after the attenuator. The noise figure F of both the erbium-doped fiber amplifiers (EDFAs) is equal to 6.5 dB (nsp = 2.23). A further DCU is placed before the receiver to adjust the overall link dispersion. Like the transmitter structure, the receiver is configured as defined in the back-to-back optimization phase. For each modulation format, the OSNR at the input of the receiver is set to give a target BER of 10–6 in the back-to-back WDM condition. This target OSNR (a different value for each modulation format) can be achieved by choosing the proper transmitted power, depending on the overall span loss tot. The resilience of each modulation format to variation of the dispersion was tested by varying the amount of dispersion Din-line and DRX (measured in ps/nm) introduced by the in-line DCU and the receiver DCU, respectively. The Q surface at the receiver vs in-line and predetection dispersion is shown in Figure 20.25. The size of the Q interval is 5.5, 1.5, 1.2, and 1 dB, respectively, for NRZ, duobinary, Q[dB]

Q[dB]

NRZ

Duobinary

13.6 13.4

12

13.2 11 13 12

10

11 10

12.8

13.5

12.6

13 9

9

12.5

8 7

80 60

6 5 –100

13

14

8

20 –80

0 –60

–40

7

–20 –20

0

20

40

60

80

–40 –60 –80 100 –100

12.4 80

12 60

40

TotalResidue

11.5 –100

12.2

40 20 –80

12

0 –60

–40

–20 –20

0

20

6

40

60

80

–40 –60 –80 100 –100

11.8

TotalResidue

InLineResidue

InLineResidue

Q[dB]

Q[dB]

CSRZ-DPSK

11.2

DPSK

11.2

11 11 10.8 10.8

11.6 11.4 11.2 11

10.6

10.6

11.2 11 10.4

10.8

10.8 10.6 10.4

10.4

10.6 10.4

10.2

10.2

10.2 80

10 9.8 –100

11.6 11.4

60

10.2

40 20 –80

0 –60

–40

0

20

40

60

InLineResidue

80

–40 –60 –80 100 –100

9.8

TotalResidue

60 40

10

20 10

–20 –20

80

10 9.8 –100

–80

0 –60

–40

–20 –20

0

20

40

60

80

100

–40 –60 –80 –100

9.8

TotalResidue

InLineResidue

Figure 20.25 Clockwise from top-left: Q surface at the receiver vs. in-line and predetection dispersion for NRZ, duobinary, DPSK, and CSRZ-DPSK modulation formats (this figure may be seen in color on the included CD-ROM).

20. Simulation Tools for Devices, Systems, and Networks

837

DPSK, and CSRZ-DPSK modulation formats. The simulation results show that advanced modulation formats have greater dispersion resilience than the standard NRZ intensity modulation scheme.

20.3.2 Multi-mode Systems Recent advances and standards efforts [70] in Ethernet-based multigigabit optical data links have revived interest in design optimization of optical communication systems that use multi-mode fibers and multi-mode light sources. Some of the typical multi-mode design objectives, among others, include optimization of system bandwidth, compliance with standards requirements, evaluation of chromatic and modal dispersion, and calculation of differential mode delay (DMD) and encircled flux (EF), all of which can be studied via suitable multi-mode system simulation tools. Unlike single-mode systems, multi-mode designs require analyses of both temporal and spatial effects [71–73]. For example, it is relatively easy to analyze coupling between a single-mode fiber and a single-mode laser. In contrast, analysis of coupling between a multi-mode fiber and a multi-mode laser requires knowledge of the spatial distributions of both the laser’s optical field profiles and the fiber’s transverse mode profiles. Misalignment between the fiber and laser also affects the analysis. Furthermore, as shown in Figure 20.26, an optical pulse launched into a multi-mode fiber simultaneously excites multiple fiber modes, each having a different group velocity [74–77]. This causes each mode to arrive at the end of the fiber at different times, thereby creating a spreading of the pulse, also known as modal dispersion. Below we elaborate on these issues with particular emphasis on the multi-mode fiber, and then present an example of a multi-mode link simulation.

Representation of an Optical Signal For single-mode simulations, it is sufficient to consider only the time-varying amplitude of a particular mode. Multi-mode systems, however, with their complicated mode couplings between components, demand that each mode’s transverse

Figure 20.26 Modal dispersion in a multi-mode fiber.

Robert Scarmozzino

838

shape be explicitly accounted for within the optical field representation. Hence, we include transverse mode profiles for both polarizations as part of the optical signal. The complete representation of an optical field for a single mode, including unique shapes for the X and Y polarizations, is

 EðtÞ = ^x  Ax ðtÞ

x ðx; yÞe

j2fo t

  þ c:c: þ ^y  Ay ðtÞ

y ðx; yÞe

j2fo t

 þ c:c: ;

(20.31)

where is a scaling factor, fo is the optical carrier frequency, t is time, x is the X coordinate, y is the Y coordinate, and c.c. denotes complex conjugate. The field amplitudes for the X and Y polarizations are Ax(t) and Ay(t), respectively, while the normalized transverse mode profiles for the X and Y polarizations are cx(x, y) and cy(x, y), respectively. The profiles are normalized to ensure that the full optical power is carried inside the amplitude functions. In most cases, the mode profiles for the X and Y polarizations will be identical, but for generality, they can be specified separately. These mode shapes include donut and spot modes, Laguerre–Gaussian (LG) and Hermite–Gaussian modes, and linearly polarized (LP) fiber modes. These profiles will generally vary depending on the component from which they are generated. For example, VCSELs can often be characterized by LG modes. Typically, the multi-mode output of a given component can be treated at a single wavelength with mutually orthogonal transverse modes. Propagation of Light in a Multi-mode Fiber As seen earlier in Figure 20.26, the modal dispersion causes different fiber modes to arrive at the fiber output with different delays. Typically, a multi-mode fiber would support hundreds of transverse modes at a particular wavelength, thereby making the analysis of modal delays a very involved one. There are various approaches for these calculations. One approach is to use the BPM to determine the transverse optical modes of a multi-mode fiber and their respective propagation constants [78–82]. A second approach is to assume a parabolic refractive index profile for which analytical expressions exist for fiber modes and delays [83, 84]. Of course, this assumption would not permit studies on the effects of index perturbations. A third approach determines the modes of a fiber by solving the radial Helmholtz equation subject to a radially varying index profile n(r). The individual propagation constants are determined by using Maxwell’s equations to match boundary conditions, and then solving the resulting characteristic equation. The advantage of this approach is that the index profile can be an arbitrary function. When performing system-level analysis of modal delays, a common approximation is the so-called degenerate mode group (DMG) formulation. This technique is based on the observation that modal delay values tend to fall into clusters; the DMG formulation takes advantage of this fact by analyzing the clusters rather than the individual delays themselves. The well-known Wentzel-Kramers-Brillouin

20. Simulation Tools for Devices, Systems, and Networks

839

(WKB) approximation [72, 73] of electromagnetic theory is based on the scalar Helmholtz equation  d2 1d l2 2 2 2 þ k n ðr Þ    2 = 0 þ r dr2 r dr

(20.32)

with assumed solutions of the form ðr Þ =

0

exp½ikSðr Þ;

(20.33)

where r is the radial distance from the center of the fiber core, k is the wave number, n(r) is the multi-mode fiber’s radially varying index profile,  is the modal propagation constant, and l is the given azimuthal index for which solutions clm(r) for each radial index m are found. Furthermore, S(r) can be expanded in powers of 1/k: 1 Sðr Þ = S0 ðr Þ þ S1 ðr Þ þ    k

(20.34)

The fundamental WKB assumption is that if the index varies only slightly over the distance of a wavelength, terms of higher order in Eqn (20.34) can be neglected [73], leading to reasonable approximations for c(r). In Refs [72] and [73], it is shown that under the WKB approximation, the modal delays for the parabolic index fiber can be approximated by WKB =

  Ln1 n1 Neff ; þ 2c Neff n1

(20.35)

where L is the fiber length, n1 is the peak index at the fiber center, c is the speed of light, and Neff is the effective index. An Example of the System-Level Impact of Modal Delays Consider the layout for a typical multi-mode system simulated using a commercial tool [85] as shown in Figure 20.27. From left to right, the components are a PRBS pattern generator, a laser driver, a VCSEL, a coupler, a multi-mode fiber, a photoreceiver, and an eye diagram analyzer. The VCSEL is modulated by a 10-Gb/s NRZ data stream and produces an 820-nm Gaussian beam with a 5-mm waist. The coupler is set to provide an offset of 25 mm between the VCSEL and the multi-mode fiber axis. After the simulation is over, the multi-mode fiber model produces a plot of the relative power coupled into each mode as a function of the modal delay as shown in Figure 20.28, where the received eye diagram is also shown. Reducing the lateral offset to

Robert Scarmozzino

840 Δy

VCL

ΔZ

01101

M

Δx 1.0

30

Signal (real) (V)

1.0

20

0.8

y (μm)

10

0.6

0

0.4

–10

0.2

–20

0.0 0

2

4

6

8

10

–30

12

0

× 10–9

10

20

30

40

50

0.0

x (μm)

Time (s)

5-μm Waist Gaussian beam

10-Gb/s square-wave modulation

× 10–5 0.05

10

0.04

Signal (V)

Power coupling coefficient

Figure 20.27 A simple multi-mode link (this figure may be seen in color on the included CD-ROM).

0.03 0.02 0.01

9 8 7

0.00 1413

1414

1415

0

× 10–9

1

2

× 10–10

Modal delay

Time (s)

×10–5

0.05

30 0.04

28

0.03

Signal (V)

Power coupling coefficient

Figure 20.28 Modal delay plot (left) and received eye (right) for a 25-mm laser-fiber offset.

0.02 0.01 0.00

26 24 22 20

1413

1414

18

1415 –9

0

1

×10 Modal delay

2

×10–10 Time (s)

Figure 20.29 Modal delay plot (left) and received eye (right) for a 5-mm laser-fiber offset.

20. Simulation Tools for Devices, Systems, and Networks

841

5 mm results in more tightly clustered delay values, and hence a better received eye diagram, as shown in Figure 20.29.

20.3.3 Performance Estimation Introduction The simulation of an optical communication system requires not only implementation of detailed models for the link subsystems and components, but also handling of stochastic physical processes and accurate techniques for the evaluation of the system performance. The stochastic physical processes taking place in an optical communication system are the following: • noise [ASE (amplified spontaneous emission) thermal, shot, phase noise, etc.] • statistics of systems parameters (length, dispersion value, PMD, etc.) • statistics of digital sources (random digital sequences) Focusing our attention on binary digital systems, the most commonly used performance parameters are • BER • Q-factor, defined as Q=

m1  m0 ; 1 þ 0

where m1, 1 (m0, 0) are the mean and the standard deviation of the received signal at the sampling instant when a logical “1” (“0”) is transmitted. In a laboratory or field experiment, BER can be easily and accurately measured using commercial BER testers, which are based on direct error counting by comparison of the input and output digital streams. Otherwise, BER evaluation is usually a very difficult task in a software simulation environment for the following reasons: • the reference BER for an optical communication system is very low (10–9 and lower), • the signal at the output of the link is usually effected by intersymbol interference, • the number of simulated bits is limited with respect to the available CPU time. The most commonly used techniques in communications system simulations are direct error counting, Monte Carlo methods, and semianalytical techniques. A detailed analysis of these techniques may be found in Ref. [86]. Direct error

Robert Scarmozzino

842

counting is seldom used in the simulation of optical systems. In fact, a BER close to 10–9 requires [86, Figure 2] the simulation of at least 1010 bits to achieve a reasonable accuracy, and would thus need a prohibitively large CPU time even for extremely simple link setups. Monte Carlo Method In the Monte Carlo approach, all the stochastic processes are simulated by extracting a random variable from a random number generator. All noise sources are simulated by extracting a noise sample, adding it to each signal sample and propagating them together. The system parameters are simulated by extracting random numbers once at the beginning of the simulation. The digital signal sources are simulated through pseudo-random sequence generators. The system performance is then estimated calculating a few moments—typically mean and variance—of the received signal. In the Monte Carlo approach the noise is propagated in the same way as the signal, reflecting physical reality. In case nonlinearities are relevant, the method is accurate since signal and noise can mix together nonlinearly. Moreover for very complex transmission systems, Monte Carlo is often the only available general purpose approach because the noise statistics at the receiver can only be handled with a Gaussian approximation. Simulated Number of Bits and Confidence Interval The estimate of the system performance over a finite and typically small time frame using the Monte Carlo method is affected by statistical fluctuations of the results. A very common parameter for the estimation of system performance by simulation is the well-known Q-factor, defined as Q=

m1  m0 ; 1 þ 0

(20.36)

The BER is then often evaluated as   1 Q BER = erfc pffiffiffi 2 2

(20.37)

This last relation is exact only when:

• noise at the receiver is Gaussian, • ISI is negligible. There is usually a widespread perception among the users of simulation tools of the scope and consequences of these two assumptions. On the contrary, it is much

20. Simulation Tools for Devices, Systems, and Networks

843

less understood that there is an intrinsic uncertainty in the evaluation of the Q-factor whenever it is measured over a finite number of bits. To investigate this point, let us suppose that the above assumptions are exactly met, i.e., that the signal is not significantly affected by intersymbol interference and that the noise is Gaussian. Let us start by focusing on the estimate m* of the mean of a Gaussian-distributed variable of mean m and standard deviation  . The estimate is evaluated over N samples of the variable as m =

N 1X si ; N i=1

(20.38)

where si are the N samples of the variable. From standard estimation theory [87], it can be demonstrated that the estimate m* has an uncertainty whose standard deviation is given by  dev½m  = pffiffiffiffi : N

(20.39)

Similarly, the estimate of the variance 2* has an uncertainty whose standard deviation is given by [87]: rffiffiffiffi 2 2  : dev½  = N 2

(20.40)

Starting from these results, and assuming that we are simulating Ntot bits and that we have an equal number of “1” and “0” symbols, it can be shown that the uncertainty in the estimation of the Q-factor has a standard deviation which is approximately given by Q dev½Q  ffi pffiffiffiffiffiffiffiffiffiffi : 2Ntot

(20.41)

As commonly accepted, we will now assume as a “confidence interval” of the estimate a range given by two standard deviations, which, for Gaussian variables, accounts for a 95% probability. We can interpret the confidence interval as follows: the estimate Q* has a 95% probability to be in the range: ½Q  2 dev½Q ; Q þ 2 dev½Q :

(20.42)

In Figures 20.30 and 20.31, we report this range for a nominal Q = 6 (15.56 dB) and Q = 8.2 (18.2 dB), corresponding to the two commonly used reference BERs of 10–9 and 10–16, respectively.

Robert Scarmozzino

844 18

Upper Q value Central Q value Lower Q value

17.5 17

Q (dB)

16.5 16 15.5 15 14.5 14 13.5

0

100

200

300

400

500

600

700

800

900

1000

Total number of simulated bits Figure 20.30 Q-factor estimate range (95% confidence interval) for a nominal Q = 6 (15.56 dB), corresponding to BER = 10–9 (this figure may be seen in color on the included CD-ROM).

20.5

Upper Q value Central Q value Lower Q value

20 19.5

Q (dB)

19 18.5 18 17.5 17 16.5 16

0

100

200

300

400

500

600

700

800

900

1000

Total number of simulated bits Figure 20.31 Q-factor estimate range (95% confidence interval) for a nominal Q = 8.2 (18.2 dB), corresponding to BER = 10–16 (this figure may be seen in color on the included CD-ROM).

It is interesting to note that, in both cases, approximately 500 bits are required to have Q-factor confidence intervals of 1 dB. Moreover, it should be noted that, even though this is often done in a simulation environment, the estimation over less than 100 bits leads to an exceedingly high confidence interval, greater than 2 dB.

20. Simulation Tools for Devices, Systems, and Networks

845

In order to asses the “error” in the evaluation of the Q-factor over a finite number of bits, we consider the two extremes of the 95% confidence interval given by

Qmin = Q  2 dev½Q ; Qmax = Q þ 2 dev½Q :

(20.43)

The estimate range can be expressed as range = Qmax  Qmin

(20.44)

and is given by 0

rffiffiffiffiffiffiffiffi1 2 B1 þ C N B tot C rffiffiffiffiffiffiffiffi range = 20 log10 B C: @ 2 A 1 Ntot

(20.45)

This is a very useful formula, which shows that the Q-factor range is independent from the Q-factor nominal value and is only dependent on the total number of bits Ntot. The following table reports this range in dB vs the number of simulated bits.

Number of simulated bits Ntot 32 64 128 256 512 1024

Q-factor uncertainty range (dB) 4.43 3.10 2.18 1.53 1.09 0.77

An even more useful insight into this description of the estimation accuracy can be obtained by Figures 20.32 and 20.33, where the BER range (evaluated from the Q-factor in the Gaussian approximation) is plotted vs the number of simulated bits for the two nominal BERs equal to 10–9 and 10–16. As it can be shown by direct comparison, the confidence interval at a BER around 10–9 is given by roughly 1 order of magnitude for Ntot = 512. We believe that this is an acceptable range for most applications and we thus advise the user to simulate at least 512 bits when even the accurate indication of the BER value is in the 10–9 range. The situation is worst at lower bit rates, as it can be seen from the graphs at BER = 10–16. In fact for a given Ntot, the range in terms of BER increases while decreasing the reference BER. This is due to the fact that the BER vs Q-function is

Robert Scarmozzino

846 1e–06

Upper BER value Central BER value Lower BER value

1e–07 1e–08

BER

1e–09 1e–10 1e–11 1e–12 1e–13 1e–14

0

100

200

300

400

500

600

700

800

900

1000

Total number of simulated bits Figure 20.32 BER estimate range (95% confidence interval) for a nominal BER = 10–9 (this figure may be seen in color on the included CD-ROM).

1e–10

Upper BER value Central BER value Lower BER value

1e–12 1e–14

BER

1e–16 1e–18 1e–20 1e–22 1e–24 1e–26

0

100

200

300

400

500

600

700

800

900

1000

Total number of simulated bits Figure 20.33 BER estimate range (95% confidence interval) for a nominal BER = 10–16 (this figure may be seen in color on the included CD-ROM).

steeper for smaller BER. Thus for a fixed range in term of Q-factor correspond to a BER range that increases for the decreasing reference BER, as shown in Figure 20.34. For example when simulating 512 bits at BER = 10–16, the range is more than 2 orders of magnitude.

BER

20. Simulation Tools for Devices, Systems, and Networks

847

–5 –6 –7 –8 –9 –10 –11 –12 –13 –14 –15 –16 –17 –18 –19 –20 12

13

14

15

16

17

18

19

Q (dB) Figure 20.34 BER vs Q-factor (this figure may be seen in color on the included CD-ROM).

Semianalytical Techniques In the semianalytical techniques, the stochastic processes related to the noise are taken into account analytically. The signal is simulated without noise, the noise is propagated analytically, and explicit analytic formulas are used to evaluate the BER. Several versions of semianalytical techniques are available. The common requirement to all these techniques is the exact knowledge of the noise distribution at the receiver. Whenever this is true, the semianalytical techniques are very powerful and provide extremely accurate results even when a few bits are simulated (see Ref. [86, Section VI]). Moreover, since the noise statistics do not have to be recovered from the signal, the semianalytical techniques seamlessly handle systems with strong ISI. Karhunen–Loe`ve Technique The exact probability density functions (PDFs) of the received signal after pre- and postdetection optical and electrical filtering can be efficiently evaluated by using the method described in Ref. [86], based on the Karhunen–Loe`ve decomposition of signal and noise in the frequency domain. It can be shown (see Refs [88, 89]) that the received signal after optical and electrical filtering can be written in the form: vðtÞ =

X i

jbi ðtÞ þ i ðtÞj2 i ;

(20.46)

Robert Scarmozzino

848

where bi(t) and i(t) are the coefficients of the series expansion of the Fourier transform S( f ) and NASE( f ) of the noiseless received signal s(t) and of the ASE noise random process nASE(t), respectively: bn ðtÞ = vn ðtÞ =

Z

Z

Sð f Þ exp½ j2ft n ð f Þdf ;

(20.47)

NASE ð f Þ exp½ j2ft n ð f Þdf :

{m(t)} is a set of orthornormal functions satisfying the following eigenvalue integral equation: Z

m ð f2 ÞK ð f1 ; f2 Þdf2 = m m ð f1 Þ:

(20.48)

For a direct detection receiver: KDD ð f1 ; f2 Þ = RGHo ð f1 ÞHL ð f1  f2 ÞHo ðf2 Þ;

(20.49)

where R and G are the photodiode responsivity and avalanche gain, respectively, and HO and HL are the optical and electrical filter transfer functions, respectively. Similar expressions can be obtained for differential receivers. Using numerical integration algorithms (see Ref. [90, Chapter 4]), the solution of Eqn (20.48) can be reduced to the eigenvalue and eigenvector problem for a Hermitian matrix: Z

m ð f2 ÞK ð f1 ; f2 Þdf2 

fMAX Z

m ð f2 ÞK ð f1 ; f2 Þdf2 

fMAX

2k X i=0

m ðxi ÞK ð f1 ; xi Þ;

(20.50)

  where xi = fMAX ðikÞ . The integration interval [–fMAX, fMAX] must be chosen in k a way that the integrand evaluated for f > fMAX is sufficiently small and does not affect the result. Equation (20.48) can thus be rewritten as 2k X i=0

m ðxi ÞK ðxn ; xi Þ = m m ðxn Þ;

n; m = 0; . . . ; 2k;

(20.51)

i.e., in matrix form: K  j = t  j:

(20.52)

with Ki,j = K(xi,xj), ’ij = ’j(xi),  ij =  i i,j. Since K is Hermitian, the problem Eqn (20.53) is straightforward to solve using the Jacobi routines in Ref. [90, Chapter 11].

20. Simulation Tools for Devices, Systems, and Networks

849

Since nASE(t) is a white random process, it can be shown that the noise expansion coefficients ni(t) are statistically independent Gaussian random variables with zero mean and variance N0 i. Thus Eqn (20.46) represents a weighted sum of nonzero mean-squared Gaussian random variables, whose moment generating function (MGF) is equal to [91]:

hðt; zÞ =

"

zjbi j2 i exp  nY val 1 1 þ zi N0 i=0

#

ð1 þ zi N0 ÞM

(20.53)

with M = 1 and 2 for single- and dual-polarization representations, respectively. If additive electrical noise sources, which are statistically independent from the optical ASE noise, are also present (like shot or thermal noise), the overall MGF of the decision variable can be evaluated by multiplying Eqn (20.53) by the MGF of a Gaussian random process with zero mean and variance 2(t) = (sh(t))2 þ (th)2, i.e.,  1 hel ðt; zÞ = exp  2 ðtÞ z2 : 2

(20.54)

Using the steepest descent approximation method [92], the probability of error can be evaluated as P0e; k ðsth ; tÞ =

Z1

sth

Pie; k ðsth ; tÞ =

Zsth

1

exp½ðt; z0 Þ PDFk ðs; tÞ ds  pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; 200 ðt; z0 Þ

(20.55)

exp½ðt; z1 Þ PDFk ðs; tÞ ds  pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 200 ðt; z1 Þ

with (t,z) defined as ðt; zÞ = ln

   exp½zsth  hðt; zÞ exp  12 2 ðtÞz2 : j zj

(20.56)

00 is the second derivative of  with respect to z, and z0 and z1 are, respectively, the positive and negative roots of the equation: #0 ðt; zÞ =

@ðt; zÞ = 0: @z

(20.57)

Equation (20.57) can be efficiently solved using the bisection method described in Ref. [90, Chapter 9].

Robert Scarmozzino

850

Finally the overall BER at a given decision threshold sth and sampling instant ts is evaluated as BERðsth ; ts Þ =

N 1X P0;1 ðsth ; ts þ kT Þ: N k = 1 e;k

(20.58)

20.4 NETWORK MODELING 20.4.1 Introduction In this section, we discuss optical network modeling, which allows for the study of high-level optical network behaviors under various conditions. The networks under study are generally comprised of a WDM-based optical layer along with a service operator network and traditional transport network technologies like Synchronized Optical NETwork (SONET) and synchronized digital hierarchy (SDH). A variety of optical network architectures are available in the optical layer, and generally incorporate optical terminal multiplexers (OTM), Optical Add and Drop Multiplexers (OADM) and Optical Cross Connects (OXCs). Furthermore, the transparent nature of the optical network supports a wide range of physical and logical topologies, as well as different routing and protection mechanisms. Examples of some network modeling applications include network dimensioning, protection and restoration mechanisms, and protocol behaviors and throughputs. Below we discuss a small subset of the topics related to optical network modeling. For further information, the interested reader is referred to the references section.

20.4.2 Network Dimensioning Network dimensioning finds out the resources required to implement an optical network. The allocation of resources includes fiber, wavelengths, and WDM equipment required to build the physical topology to satisfy the given set of traffic demands. In the network planning process, traffic demands are the forecasted traffic demands. Network planning can be done either with static traffic or dynamic traffic. With static traffic planning, the traffic is known in advance, the planning can be done for a single period or multiple periods. With dynamic traffic planning, the traffic is uncertain. The traffic forecasts are given in terms of statistical data. For example, it can be given as the level of traffic churn or number of hubs in the network or number of nodes in the optical ring. In this case the designed network should be flexible enough to accommodate the traffic dynamism and minimize the network cost [101–103].

20. Simulation Tools for Devices, Systems, and Networks

851

Network design/planning software helps to achieve the design goals quickly and to do a systematic planning process [99, 100]. During the network planning process, the planner needs to collect an initial data set, which will be used to plan the network later. This initial data set includes • Fiber connectivity between the locations: in the ideal case, physical connections between all locations are need. However, creating full mesh connectivity with independent fiber routes between locations is very expensive. So it is required to consider physical connectivity limitations to determine the possible permitted links during the network planning process. • Traffic demands (logical level demands): traffic demand specifies the number of wavelength channels required between two nodes. These wavelength channels are routed through the fiber links in the network in such a way that it minimizes the network cost. Each traffic demand has parameters such as (a) (b) (c) (d) (e)

Start and end nodes. Bandwidth. Quantity. Protection/restoration requirements. Routing diversity requirements. • Network elements: there are three types of technologies available today. WDM terminal multiplexers and demultiplexers (OTM), OADM and OXC. In each of these elements, there are constraints due to available technology at the time of planning. For example, there is a limited number of optical add and drop channels in an OADM, or number of EDFAs that can cascade along an optical path. Also in the planning process one should consider technologies that will mature in the near future. Input parameters required for the network elements include (a) Number of wavelengths available per fiber. (b) OADM capacity—number of add–drop channels. (c) OXC capacity—number of optical cross connections it supports. (d) Power budget (span loss) for the EDFA. (e) Number of cascaded EDFAs possible before regeneration. • Routing and survivability schemes: the routing of the optical path should take into consideration the client layer requirements. For example, the client layer may require several channels between two nodes with a guarantee of carrying all channels of the given demand together along with the same path. This is called grouped routing of set of optical channels. Another very important aspect of optical network planning is the planning of optical protection and restoration. Parameters used in the routing and survivability schemes are (1) Routing metric—shortest path, minimum hops, minimum cost, maximum availability.

852

Robert Scarmozzino

(2) Diversity requirements—node diverse, line diverse and nondiverse. (3) Client layer protection requirements—1 þ 1 protection, 1:1 protection, optical ring protection schemes like optical channel layer protection, Optical multiplexer layer protection, optical multiplexer section shared protection (OMS-SPRing). (4) Restoration requirements—end-to-end path restoration, link restoration, and adjacent node restoration. Let us consider the design of WDM rings with known traffic demands. There are three types of popular WDM rings: • WDM rings with dedicated protection: this is also called path-protected ring. These can be unidirectional (UPSR) or bidirectional (BPSR). Every optical channel is protected from fiber cuts by another dedicated protection channel, which runs on the opposite side of the ring. • WDM rings with shared protection: these rings are always bidirectional, although in some cases it is advantageous to route some optical demands unidirectionally. In case of a fiber break on one section of the ring, enough spare transmission capacity is provisioned on the rest of the ring to restore the optical demands interrupted on the broken section. An example of this type of ring is the OMS-SPRing. • WDM rings with no protection: protection may be carried out in client layers in this case, using capacity on the same ring or elsewhere in the network, but the WDM ring does not need to reserve any wavelengths for protection purposes. Consider the capacity planning for the WDM ring with dedicated protection, the working and protection channels can share the same wavelength since it travels in the opposite direction of the ring. So each optical demand can be assigned one wavelength. Consider a ring with N nodes and full mesh connectivity with one wavelength demand between each node. This means there are N(N – 1)/2 demands present in this case and we need N(N – 1)/2 wavelengths to satisfy the entire demands.

20.4.3 Discrete Event Simulations for Optical Networks Apart from the network planning/design, network modeling also deals with the study of optical network protocols, protection mechanisms and quality of service. In this section, we will see discrete event simulations and a few optical network applications using discrete event simulation models. A popular methodology to generate discrete event systems is Petri Nets [111–117]. Petri Nets are an efficient, visual approach to model the dynamic behavior of queues and protocols. The basic Petri Nets formalism consists of a

20. Simulation Tools for Devices, Systems, and Networks

853

very small set of symbols [117–119]. This makes the approach easy to learn. Concepts like parallelism and synchronization are expressed with a few icons. Petri Nets are defined in terms of three symbols. Place (represented by a circle) to model states, conditions, events, and queues. Transition (represented by a rectangle) to model state transitions and activities. Arc (represented by oriented lines) to define cause–effect relationships by connecting places and transitions. Arcs going from places to a transition define the transition’s input places; whereas arcs going from a transition to places define the transition’s output places. A place is often an input to a transition and an output from other transitions at the same time. Current state (true—conditions or occurring events) is represented by drawing a dot within the suitable place. Such a dot is called a token. Petri Nets evolve by moving tokens from an initial state defined by a set of initial tokens until no more tokens can be moved according to the following simple rule: “When all the input places of a transition contain at least one token the transition can fire. The effect of a transition firing is to remove tokens from some places and to create tokens in some other places. When a transition fires, it fetches one token from each of its input places and releases a new token in each of its output places.” Example: Protection Scheme in Packet Transmission We would like to evaluate the protection switching mechanism of a WDM ring. For example consider a WDM ring with four nodes. In normal working conditions, the traffic will flow in the main (normal) ring, in case of a fiber cut, the traffic will flow in the protection ring. There is a supervisory link, which is used to transmit messages between the nodes. The objective is to study the series of events and messages required to invoke a protection mechanism in a WDM ring and predict the robustness of the protection protocol. To simulate such a scenario, we need four objects, the node, the working link, the protection link and the supervisory link. Top-level view of the ring is given in Figure 20.35. Both normal and protection rings are unidirectional, while the supervisor link transmits in both directions. The packets are generated from each node and circulated in the ring. Each packet carries a value to model the carrier signal status: TRUE means that there is a carrier signal and the link is working properly. FALSE means absence of the carrier signal. If a FALSE value is detected, then the receiver node sends a special message to the adjacent nodes informing of the detection of a failure and starts switching the transmission to the protection link. The Node Object The node object is made of two states: NORMAL and PROTECTION. The node is initially in the normal state. The node has a TX section, RX section, Fault management section and a restoration management section as shown in Figure 20.36. Each section behavior is described below. The TX section generates packets at a particular rate, which is user defined. For example, it generates one packet every 0.01 time units. When the state is

Robert Scarmozzino

854 NORMAL_12:LINK N1:NODE

N2:NODE

PROTECTION_12:LINK SLINK_12:SLINK

NORMAL_43:LINK

SLINK_23:SLINK

PROTECTION_43:LINK

PROTECTION_23:LINK

SLINK_41:SLINK

NORMAL_23:LINK

SLINK_34:SLINK

N4:NODE

N3:NODE

PROTECTION_34:LINK NORMAL_34:LINK

Figure 20.35 The top-level view.

NORMAL:NUL(1) RX_Fault_Signal, 2

RESTORE:NUL Restore_Management, 1

Fault_Management, 2 FAULT:NUL TX

RX

PROTECTION:NUL RX_Restore_Signal, 1

Figure 20.36 Main view of node object.

NORMAL, the packet is transmitted on the normal ring, and when the state is PROTECTION it will go on the protection ring. The details of the TX section are given in Figure 20.37. When the RX section as shown in Figure 20.38 receives a packet with a FALSE value, meaning that the carrier signal is lost, it detects the fault on the link and immediately activates a procedure at higher priority (2) managed by the Fault_Management section. Any packet received on the working link when the node is not in normal state is then discarded, and these are the packets that the node lost. When a FAULT is detected by the Fault_Management section as shown in Figure 20.39, it switches the mode from NORMAL to PROTECTION. A GO_PROTECTED signal is sent to the adjacent nodes, predecessor and follower,

20. Simulation Tools for Devices, Systems, and Networks

855

TX_WORKING

Transmit on Normal Links NORMAL_TX:SIGNAL

This is the packets generator GENERATE_PACKET

NORMAL:NUL(1) PACKET:SIGNAL

Node states PROTECTION:NUL

GEN_ENABLE:NUL(1)

PROTECTION_TX:SIGNAL Transmit on Protection Links

TX_PROTECTED

Figure 20.37 The TX section.

DISCARD_PACKET

Receive on Normal Links

Packets received before protection switching completed

SIGNAL_OK, 2 NORMAL_RX:SIGAL LOST_CARRIER, 1

FAULT:NUL

NORMAL:NUL(1)

Receive on Protection Links

PROTECTION_RX:SIGNAL

RECEIVE_PACKET_P

Figure 20.38 The RX section.

FAULT:NUL

NORMAL:NUL(1)

FAULT_DISCARD

DETECT_FAULT, 1 WAIT_PROTECTION_ACK:NUL SWITCH_TO_PROTECTION PROTECTION:NUL SIGNAL_TO_PREDECESSOR:MSG

SIGNAL_FROM_PREDECESSOR:MSG

SIGNAL_TO_FOLLOWER:MSG

SIGNAL_FROM_FOLLOWER:MSG

Figure 20.39 The fault_management section.

through the supervisor LINK and then the Fault_Management subnet waits to get an ACK signal from both adjacent nodes. When the ACK signals are received, the node switches to PROTECTION State, which disables transmission on the normal link and switches transmission of

Robert Scarmozzino

856 RX protection signal from predecessor node

RX protection signal from follower node

SIGNAL_FROM_FOLLOWER:MSG

FAULT:NUL

SIGNAL_FROM_PREDECESSOR:MSG

RX_FAULT_ACK_FOLLOWER

RX_FAULT_ACK_PREDECESSOR SIGNAL_TO_PREDECESSOR:MSG

SIGNAL_TO_FOLLOWER:MSG

Figure 20.40 RX_Fault_signal.

packets on the protection link, as described in the packet generation TX subnet of Figure 20.37. While in the protection mode, a state simply receives packets, without examining the carrier presence, as it is assumed, in this simple example, that the protection link is always working. When the RX_Fault_Signal section (Figure 20.40) receives the GO_PROTECTION signal from an adjacent node, the ACK signal is sent back to that node and an internal fault generated, which is managed as described in the fault management section Figure 20.39. Restoration To restore normal mode of operations, the node must receive a signal in place RESTORE, which activates the Restore Management section. This section works the same way the Fault Management section does, but this time it switches the nodes from PROTECTION mode to NORMAL. The restore action will be taken inside the Restore Management section (Figure 20.41). If in protection mode, the restore command make the node exit the protection state, inform the adjacent nodes of the changes, and when both predecessor and follower nodes send back an ACK message, the node switches eventually in the normal state, so that packets will be sent through the normal link again, as described in the TX section. When the Restore Management section receives the restore command, it sends an ACK message as reply; an internal restore is generated to be managed as RESTORE:NUL

PROTECTION:NUGO_RESTORE, 1

DISCARD_RESTORE

WAITING_R_ACK:NUL

SIGNAL_TO_PREDECESSOR:MSG

SIGNAL_TO_FOLLOWER:MSG

SWITCH_NORMAL_LINK

SIGNAL_FROM_PREDECESSOR:MSG

SIGNAL_FROM_FOLLOWER:MSG

Figure 20.41 Restore_Management section.

NORMAL:NUL(1)

20. Simulation Tools for Devices, Systems, and Networks

857

RX restore signal from predecessor node

RX restore signal from follower node

SIGNAL_FROM_PREDECESSOR:MSG

SIGNAL_FROM_FOLLOWER:MSG

RESTORE:NUL

RX_RESTORE_ACK_PREDECESSOR

RX_RESTORE_ACK-FOLLOWER

SIGNAL_TO_PREDECESSOR:MSG

SIGNAL_TO_FOLLOWER:MSG

Figure 20.42 RX_Restore_Signal section.

described in the previous view, and to communicate to all the adjacent nodes to switch to the normal link (Figure 20.42). The Link Object Each single link is unidirectional; the link object propagates packets with a propagation delay, for example 0.005 time unit. The link object is having two states: WORKING and BROKEN, where WORKING is the initial state. The user can make any link working or broken by a toggle switch (Figure 20.43). The Supervisor Link Object The fourth object is the supervisor link, which propagates supervisor signals among nodes. All messages simply pass through the supervisor link with a propagation delay, for example at 0.005 time units (Figure 20.44). Simulation The WDM ring shown in Figure 20.35 is simulated under various parameter values. Figure 20.45 shows one of the simulation results obtained. In this graph, time is marked in the X-axis and discrete events (fault detection and switch to protection) are marked in the Y-axis. At time t = 1.0 a link fault is simulated PACKET_IN:SIGNAL

MARK_TRUE

INT_PACKET:SIGNAL

PACKET_OUT:SIGNAL

PROPAGATE WORKING:NUL(1)

BREAK, 1

ON_OFF_TOGGLE

Models Link Delay

RESTORE, 1

BROKEN:NUL

MARK_FALSE

Figure 20.43 The link object.

Robert Scarmozzino

858 Supervisor Link SIGNAL_FROM_PREDECESSOR:MSG

SIGNAL_TO_PREDECESSOR:MSG

SEND_SIGNAL_TO_FOLLOWER

SIGNAL_TO_FOLLOWER:MSG

SEND_SIGNAL_TO_PREDECESSOR SIGNAL_FROM_FOLLOWER:MSG

Figure 20.44 The supervisory link.

Time 1

1.02

1.04

1.06

1.08

\N1\DETECT_FAULT \N1\SWITCH_TO_PROTECTION \N2\DETECT_FAULT \N2\SWITCH_TO_PROTECTION \N3\DETECT_FAULT \N3\SWITCH_TO_PROTECTION \N4\DETECT_FAULT \N4\SWITCH_TO_PROTECTION

Figure 20.45 Simulation results (this figure may be seen in color on the included CD-ROM).

by the user. Since the link propagation delay time in this simulation run is 0.005 time units, N2 (Node 2) detects the fault at time 1.005. Both N1 and N3 are then informed of the fault event, at time 1.01 (\N1\DETECT_FAULT and \N3\DETECT_FAULT), after the propagation delay time of the supervisor link. At time 1.015 N2 receives both ACKs from N1 and N3 and switches to protection mode (\N2\SWITCH_TO_PROTECTION). At the same time N4 is informed of the fault (\N4\DETECT_FAULT). At time 1.02 N1 and N3 receive their ACK and switch to protection mode (\N1\SWITCH_TO_PROTECTION and \N3\SWITCH_TO_PROTECTION). At time 1.025 N4 receives ACK from N1 and N3 and finally switches to protection mode (\N4\SWITCH_TO_ PROTECTION). At this point, all nodes are transmitting and receiving packets into the protection link. From this graph it is easy to understand the system behavior, and to predict the performance of the protection and restoration protocols and their robustness.

ACKNOWLEDGMENTS The author would like to acknowledge the contributions and assistance of several colleagues at RSoft: Evan Heller, Michael J. Steel, Mayank Bahl, and Mingming Jiang (device/component portion); Enrico Ghillino, Jigesh Patel, Dwight Richards, and Pablo Mena (system portion); Anil Ramapanicker (network portion).

20. Simulation Tools for Devices, Systems, and Networks

859

REFERENCES [1] M. D. Feit and J. A. Fleck, “Light propagation in graded-index optical fibers,”, Appl. Opt., 17, 3990, 1978. [2] See references in D. Yevick, “A guide to electric field propagation techniques for guided-wave optics,” J. Opt. Quantum Electron., 26, S185, 1994. [3] R. Scarmozzino, A. Gopinath, R. Pregla, and S. Helfert, “Numerical techniques for modeling guided-wave photonic devices,” J. Sel. Top. Quantum Electron., 6, 150, 2000. [4] K. S. Yee, “Numerical solution of initial boundary value problems involving Maxwell’s equations in isotropic media,” IEEE Trans. Antennas Propagat., AP-14, 302, 1966. [5] Tavlove, “Computational electrodynamics: The finite-difference time-domain method,” Artech House, Norwood, MA, 1995. [6] M. Bahl, N. C. Panoiu, R. M. Osgood, Jr, “Nonlinear optical effects due to 1D Kerr defects in a 2D photonic crystal,” Phys. Rev. E, 056604, 2003. [7] N. Panoiu, M. Bahl, and R. Osgood, Jr, Opt. Express., 12–18, 1605. [8] J. Jin, The Finite Element Method In Electromagnetics, 2nd edition, New York, John Wiley & Sons, 2002. [9] K. Saitoh and M. Koshiba, “Full-vectorial imaginary-distance beam propagation method based on a finite element scheme: application to photonic crystal fibers,” IEEE J. Quantum Electron., 38(7), 927, 2002. [10] M. Koshiba, S. Maruyama, and K. Hirayama, “A vector finite element method with the highorder mixed-interpolation-type triangular elements for optical waveguiding problems,” IEEE J. Lightw. Technol., 12(3), 495, 1994. [11] D. Schulz, C. Glingener, M. Bludszuweit, and E. Voges, “Mixed finite element beam propagation method,” IEEE J. Lightw. Technol., 16(7), 1336, 1998. [12] Y. Tsuji, M. Koshiba, and T. Shiraishi, “Finite element beam propagation method for threedimensional optical waveguide structures,” IEEE J. Lightw. Technol, 15(9), 1728, 1997. [13] M. Koshiba and Y. Tsuji, “Curvilinear hybrid edge/nodal elements with triangular shape for guided-wave problems,” IEEE J. Lightw. Technol, 18(5), 737, 2000. [14] Y. Tsuji and M. Koshiba, “Guided-mode and leaky-mode analysis by imaginary distance beam propagation method based on finite element scheme,” IEEE J. Lightw. Technol., 18(4), 618, 2000. [15] F. Peterson, “Vector Finite Element Formulation for Scattering from Two-Dimensional Heterogeneous Bodies,” IEEE Trans. Antennas and Propagat., 43(3), 357, 1994. [16] Greenwood and J. Jin, “A Novel Efficient Algorithm for Scattering from a Complex BOR Using Mixed Finite Elements and Cylindrical PML,” IEEE Trans. Antennas and Propagat., 47(4), 620, 1999. [17] J.-F. Lee, M. Wilkins, and R. Mittra, “Finite-element analysis of axisymmetric cavity resonator using a hybrid edge element technique,” IEEE Trans. on Microwave Theory and Techniques, 41(11), 1981, 1993. [18] J. C. Ne´de´lec, “Mixed finite elements in R3,” Numerische Mathematik, 35, 315, 1980. [19] J. Z. Huang, R. Scarmozzino, G. Nagy et al., “Realization of a compact and single-mode optical passive polarization converter,” IEEE Photon. Technol. Lett., 12(3), 317, 2000. [20] E. Yablonovitch, “Inhibited spontaneous emission in solid-state physics and electronics,” Phys. Rev. Lett., 58, 2059, 1987. [21] S. John, “Strong localization of photons in certain disordered dielectric superlattices,” Phys. Rev. Lett., 58, 2486, 1987. [23] N. W. Ashcroft and N. D. Mermin, Solid State Physics, Philadelphia, Saunders College, 1976. [22] E. Yablonovitch, “Photonic crystals: What’s in a name?,” Opt. Photon. News, 18, 12, 2007. [24] K. M. Ho, C. T. Chan, and C. M. Soukoulis, “Existence of a photonic gap in periodic dielectric structures,” Phys. Rev. Lett., 65, 3152, 1990.

860

Robert Scarmozzino

[25] R. D. Meade, A. M. Rappe, K. D. Brommer et al., “Accurate theoretical analysis of photonic band-gap materials,” Phys. Rev. B, 48, 8434, 1993. [26] S. Johnson and J. D. Joannopoulous, “Block-iterative frequency-domain methods for Maxwell’s equations in a planewave basis,” Opt. Express, 8, 173, 2001. [27] K. Sakoda, Optical Properties of Photonic Crystals, Springer, 2001. [28] K. Busch and S. John, “Photonic band gap formation in certain self-organizing systems,” Phys. Rev. E, 58, 3896, 1998. [29] G. J. Pearce, T. D. Hedley, and D. M. Bird, “Adaptive curvilinear coordinates in a plane-wave solution of Maxwell’s equations in photonic crystals,” Phys. Rev. B, 71, 195108, 2005. [30] J. A. West, C. M. Smith, N. F. Borrelli et al., “Surface modes in air-core photonic band-gap fibers,” Opt. Express, 12, 1485, 2004. [31] S. G. Johnson, P. R. Villeneuve, S. Fan, and J. D. Joannopoulos, “Linear waveguides in photoniccrystal slabs,” Phys. Rev. B, 62, 8212, 2000. [32] M. J. Steel, “Reflection symmetry and mode transversality in microstructured fibers,” Opt. Express., 12, 1497, 2004. [33] H. Kosaka et al., “Superprism phenomena in photonic crystals,” Phys. Rev. B, 58, R10096, 1998. [34] T. Baba and T. Matsumoto, “Resolution of photonic crystal superprism,” Appl. Phys. Lett., 81, 909, 2002. [35] M. J. Steel et al., “Analytic properties of photonic crystal superprism parameters,” Phys. Rev. E, 71, 056608, 2005. [36] R. Petit, Electromagnetic Theory of Gratings, Berlin, Springer-Verlag, 1980, p. 174. [37] M. G. Moharam, T. K. Gaylord, “Rigorous coupled-wave analysis of metallic surface-relief gratings,” J. Opt. Soc. Am. A, 3, 1780, 1986. [38] Lifeng Li, “New formulation of the Fourier modal method for crossed surface-relief gratings,” J. Opt. Soc. Am. A, 14, 2758, 1997. [39] Mingming Jiang, Theodor Tamir, and Shuzhang Zhang , “Modal theory of diffraction by multilayered gratings containing dielectric and metallic components,” J. Opt. Soc. Am. A, 18, 807, 2001. [40] Chung-Hsiang Lin, K. Ming Leung, and Theodor Tamir, “Modal transmission-line theory of three-dimensional periodic structures with arbitrary lattice configurations,” J. Opt. Soc. Am. A, 19, 2005, 2002. [41] P. V. Mena, J. J. Morikuni, and K. W. Wyatt, “Compact Representations of Mode Overlap for Circuit-Level VCSEL Models,” in Proc. IEEE/LEOS Annual Meeting Conf., 2000, p. 234. [42] J. W. Scott, R. S. Geels, S. W. Corzine, and L. A. Coldren, “Modeling temperature effects and spatial hole burning to optimize vertical-cavity surface-emitting laser performance,” IEEE J. Quantum Electron., 29, 1295, 1993. [43] N. Bewtra, D. A. Suda, G. L. Tan et al., “Modeling of quantum-well lasers with electro-optothermal interaction,” IEEE J. Sel. Top. Quantum Electron., 1, 331, 1995. [44] G. P. Agrawal and N. K. Dutta, Semiconductor Lasers, 2nd edition, New York, Van Nostrand Reinhold, 1993. [45] M. X. Jungo, D. Erni, and W. Ba¨chtold, “VISTAS: A comprehensive system-oriented spatiotemporal VCSEL model,” IEEE J. Sel. Top. in Quantum Electron., 9, 939, 2003. [46] L. A. Coldren and S. W. Corzine, Diode Lasers and Photonic Integrated Circuits, New York, John Wiley and Sons, Inc., 1995. [47] Matt Grupen and Karl Hess, “Simulation of carrier transport and nonlinearities in quantum-well laser diodes,” IEEE J. Quantum Electron., 34(1), 120, 1998. [48] B. Grote et al., “Quantum-Well Laser Diodes: Temperature and Many-Body Effects,” in Advanced Simulation and Analysis of Optoelectronic Devices (J. Piprek ed.), 2004. [49] B. Grote et al., “Integration of microscopic gain modeling into a commercial laser simulation environment,” in Physics and Simulation of Optoelectronic Devices XI Proc. SPIE, (M. Osinski, H. Amano, and P. Blood, eds), 4986, 2003, p. 413. [50] J. Piprek, Semiconductor Optoelectronic Devices—Introduction to Physics and Simulation, New York, Academic Press, 2003.

20. Simulation Tools for Devices, Systems, and Networks

861

[51] J. Piprek, Optoelectronic Devices – Advance Simulation and Analysis, New York, Springer, 2005. [52] Paul von Allman, “Derivation of the effective-mass equation for a superlattice: A perturbational approach,” Phys. Rev. B, 46(23), 15377, 1992-I. [53] Paul von Allman, “Conduction subbands in GaAs/AlxGa1–xAs quantum well: Comparing different kp models,” Phys. Rev. B, 46(23), 15382, 1992-I. [54] J. Hader, et al., “Quantitative prediction of semiconductor laser characteristics based on low intensity photoluminescence measurements,” IEEE Photon. Technol. Lett., 14(6), 762, 2002. [55] J. Hader, J. V. Moloney, and S. W. Koch, “Microscopic theory of gain, absorption, and refractive index in semiconductor laser materials – influence of conduction-band nonparabolicity and coulomb-induced intersubband coupling,” IEEE J. Quantum Electron., 35(12), 1878, 1999. [56] W. W. Chow, S. W. Koch, Semiconductor-Laser fundamentals, Physics of the Gain Materials, Berlin, Springer, 1999. [57] W. W. Chow, S. W. Koch, and M. Sargent III, Semiconductor-Laser Physics, Springer, Berlin (1994). [58] G. P. Agrawal, Nonlinear Fiber Optics, 2nd edition, London, UK, Academic Press, 1995. [59] Hasegawa and Y. Kodama, Solitons in Optical Communications Clarendon, Oxford, UK, 1995. [60] R. Menyuk, “Application of multiple-length-scale methods to the study of optical fiber transmission,” J. Eng. Math., 36, 113–136, 1999. [61] G. J. Foschini and C. D. Poole, “Statistical theory of polarization dispersion in single mode fibers,” J. Lightw. Technol., 9(11), 1439–1456, 1991. [62] F. Curti, B. Daino, G. De Marchis, and F. Matera, “Statistical treatment of the evolution of the principal states of polarization in single-mode fiber,” J. Lightw. Technol., 8(8), 1162–1166, 1990. [63] P. K. A. Wai and C. R. Menyuk, “Polarization mode dispersion, decorrelation, and diffusion in optical fibers with randomly varying birefringence”, J. Lightw. Technol., 14(2), 148–157, 1996. [64] J. W. Cooley and J. W. Tukey, Math. Comput., 19, 297, 1965. [65] Oleg V. Sinkin, Ronald Holzlohner, John Zweck, and Curtis Menyuk, “Optimization of the splitstep Fourier method in modeling optical-fiber communications systems,” J. Lightw. Technol., 21(1), January 2003. [66] G. H. Weiss and A. A. Maradudin, J. Math. Phys., 10, 129, 1976. [67] Andrea Carena, Vittorio Curri, Roberto Gaudino et al., “A time-domain optical transmission system simulation package accounting for nonlinear and polarization-related effects in fiber,” J. Sel. Areas Commun., 15(4), May 1997. [68] M. C. Jeruchim, P. Balaban, and K. S. Shanmugan, “Simulation of Communication Systems,” New York: Plenum, 1992. [69] G. Bosco, A. Carena, V. Curri et al., “Modulation formats suitable for ultra high spectral efficient WDM systems,” IEEE J. Sel. Top. Quantum Electron., 10(2), March/April 2004. [70] IEEE Standard 802.3aeTM-2002 (Amendment to IEEE Std 802.3, 2002 Edition). [71] D. Krumbholz, E. Brinkmeyer, and E. G. Neumann, “Core/cladding power distribution, propagation constant, and group delay: Simple relation for power-law graded-index fibers,” J. Opt. Soc. Am., 70(2), 179–183, 1980. [72] D. Gloge and E. A. J. Marcatili, “Multimode theory of graded-core fibers,” Bell Sys. Tech. J., 52(9), 1563–1579, 1973. [73] M. J. Adams, Introduction to Optical Waveguides. New York, NY, John Wiley & Sons, Inc., 1981. [74] J. Ritger and S. E. Golowich, “A simplified model of modal dispersion in multimode fiber,” NIST Symposium on Optical Fiber Measurements, 2002, pp. 95–99. [75] D. Marcuse, Principles of Optical Fiber Measurements, New York, NY, Academic Press, 1981. [76] R. Mickelson, Guided Wave Optics, New York, NY, Van Nostrand Reinhold, 1993. [77] J. B. Schlager and D. L. Franzen, NIST Symposium on Optical Fiber Measurements, September 1998, pp. 127–130. [78] J. Morikuni, P. Mena, B. K. Whitlock, and R. Scarmozzino, “Simulation of Modal Delays for 10 Gb/s Multimode Fiber Applications,” in Proc. IEEE/LEOS Annual Meeting Conf., Glasgow, Scotland, 2002, pp. 903–904.

862

Robert Scarmozzino

[79] R. Scarmozzino, A. Gopinath, R. Pregla, and S. Helfert, “Numerical techniques for modeling guided-wave photonic devices,” IEEE J. Sel. Top. in Quantum Electron., 6(1), 150–162, 2000. [80] Z. Huang, B. K. Whitlock, and R. Scarmozzino, “CAD optimizes component design at many levels,” WDM Solutions, August 2000, pp. 41–48. [81] D. Yevick and B. Hermansson, “New formulations of the matrix beam propagation method: Application to rib waveguides,” IEEE J. Quantum Electron., 25(2), 221–229, 1989. [82] S. Jungling and J. C. Chen, “A study and optimization of eigenmode calculations using the imaginary-distance beam-propagation method,” IEEE J. Quantum Electron., 30(9), 2098–2105, 1994. [83] G. R. Hadley and R. E. Smith, “Full-vector waveguide modeling using an iterative finitedifference method with transparent boundary conditions,” IEEE J. Lightw. Technol., 13(3), 465–469, 1995. [84] J. Gowar, Optical Communication Systems, New York, NY, Prentice Hall, 1984. [85] J. Morikuni, P. Mena, B. K. Whitlock, and R. Scarmozzino, “Multimode System Simulation as an Alternative to Spreadsheet Analysis for the Study of Gb/s Optical Communication Systems,” in Proc. IEEE/LEOS Annual Meeting Conf., paper MO4, Tucson, Arizona, October, 2003, p. 27. [86] M. C. Jeruchim, “Techniques for estimating the bit error rate in the simulation of digital communication systems,” IEEE J. Sel. Areas in Commun., SAC-2, (1), January 1984. [87] A. Papoulis, Probability, Random Variables and Stochastic Processes, New York, McGraw Hill, 1991. [88] J. S. Lee and C. S. Shim, “Bit error rate analysis of optically preamplified receivers using an eigenfunction expansion method in optical frequency domain,” IEEE J. Lightw. Technol., 12, 1224–1229, 1994. [89] A. H. Gnauck and P. J. Winzer, “Optical phase-shift-keyed transmission,” IEEE J. Lightw. Technol., 23(1), 115–130, January 2005. [90] W. H. Press, S. A. Teukolsky, and W. T. Vetterling, Numerical recipes in C, Cambridge University Press, 1992. [91] G. Bosco and P. Poggiolini, “On the joint effect of receiver impairments on direct-detection DQPSK systems,” IEEE J. Lightw. Technol., 24(3), 1323–1333, 2006. [92] C. W. Helstrom, “Distribution of the filtered output of a quadratic rectifier computed by numerical contour integration,” IEEE Trans. on Inform. Theory., IT-32(4), July 1986, 450–463. [93] B. E. A. Saleh and M. C. Teich, Fundamentals of Photonics, New York, NY, John Wiley and Sons, 1991. [94] M. C. Jeruchim, P. Balaban, and K. S. Shanmugan, Simulation of Communication Systems, New York, Plenum Publishing Corporation, 1992. [95] J. Wang and K. Peterman, “Small signal analysis for dispersive optical fibers communication systems,” IEEE J. Lightw. Technol., 10(1), 96–100, January 1992. [96] G. Bosco and R. Gaudino, “Towards New Semi-Analytical Techniques for BER Estimation in Optical System Simulation,” in Proc. NFOEC 2000, Denver (CO), Tuesday Sess. E1, USA, August, 2000. [97] L. Kazovsky, S. Benedetto and A. Willner, Optical Fiber Communication Systems, Artech House, 1996. [98] A. M. Mathai, S. B. Prevost, Quadratic Forms in Random Variables, New York, Marcel Dekker, 1992, Chapter 3. [99] MetroWAND Ring user guide. [100] MetroWAND Mesh user guide. [101] V. Poudyal, R. H. Cardwell, O. J. Wasem et al., “The effects of demand levels and number of wavelengths on the economics of metropolitan WDM,” in Proc. NFOEC, 2001, 2001, pp. 1133–1144. [102] R. H. Cardwell, O. J. Wasem, V. Poudyal et al., “Planning for interoperable and reliable metropolitan DWDM ring networks to achieve minimum cost-of-ownership,” in Proc. NFOEC’99, Chicago.

20. Simulation Tools for Devices, Systems, and Networks

863

[103] M. Ajmone Marsan, A. Bianco, E. Leonardi et al., “All-optical WDM multi-rings with differentiated QoS,” IEEE Commun. Mag., 37(2), 58–66, 1999. [104] E. Modiano, “WDM-based packet networks,” IEEE Commun. Mag., 37(3), 130–35, 1999. [105] K. K. Ramakrishnan, G. Hjalmtysson, and J. E. Van der Merwe, “The role of signaling in quality of service enabled networks,” IEEE Commun. Mag., 37(6), 124–32, 1999. [106] M. Ajmone Marsan et al., “SR3: A bandwidth-reservation MAC protocol for multimedia application over all-optical WDM multi-rings,” in Proc. INFOCOM 1997, Kobe, Japan, pp. 761, April, 1997. [107] C. K. Chan, L. K. Chen, and K. W. Cheung, “A fast channel-tunable optical transmitter for ultrahigh-speed all-optical time-division multiaccess networks,” IEE JSAC 1996, 14(5), 1052–56, 1996. [108] Cidon, and Y. Ofek, “MetaRing – a full-duplex ring with fairness and spatial reuse,” IEEE Trans. Commun., 41(1), 110–120, 1993. [109] P. Labourdette, G. W. Hart, and A. S. Acampora, “Branch-exchange sequences for reconfiguration of lightwave networks,” IEEE Trans. Commun., 42(10), 2822–2832, October 1994. [110] G. Ellinas and T. E. Stern, “Automatic protection switching for link failures in optical networks with bi-directional links,” in Proc. GLOBECOM 1996, London, UK, November 1996. [111] M. Baldassari and G. Bruno, “PROTOB: An object-oriented methodology for developing discrete dynamic systems,” Comput. Languages J. 16(1), 39–63, 1991 (reprinted in High-level Petri Nets, pp. 624–48, Springer-Verlag, 1991). [112] M. Baldassari and G. Bruno, “An environment for object-oriented conceptual programming based on PROT nets,” Advances in Petri Nets, Springer-Verlag, 1988. [113] Artifex 4.2 Reference Manual, by ARTIS Software Corporation, Torino, May 1999. [114] Artifex 4.2 Installation Guide & Getting Started, by ARTIS Software Corporation, Torino, May 1999. [115] The Artifex Language, by ARTIS Software Corporation, Torino, August 1997. [116] T. Murata, “Petri Nets: Properties, Analysis and Applications,” in Proc. IEEE 1989, 77(4), 1989, pp. 541–580. [117] K. Jensen, Colored Petri Nets, Volume 1: Basic Concepts, Springer-Verlag, Berlin and Heidelberg GmbH & Co. KG, 1992. [118] K. Jensen, Colored Petri Nets, Volume 2: Analysis Methods, Springer-Verlag, Berlin and Heidelberg GmbH & Co. KG, 1994. [119] K. Jensen, Colored Petri Nets, Volume 3: Practical Use, Springer-Verlag, Berlin and Heidelberg GmbH & Co. KG 1997.

This page intentionally left blank

Index to Volumes VA and VB

1 1 WSS, A:317 1 2 MMI coupler, A:271, 273 1 2 thermo-optic switch, A:393 1 4 wavelength-selective switch (WSS), A:723, 727 1 9 WSS, A:317 waveguide layout, A:317 1 K WSS, A:317 1 N2 using two-axis micromirror array, A:720, 721 100-Gb/s transmitter and receiver PIC, A:352 10GBASE-LRM, A:699 1480-nm pumps/14xx-nm high-power lasers, A:122–124 2 2 MEMS wavelength add/drop switch, A:721 2 2 MMI coupler, A:273 2 2 star coupler, A:273, 297 mutual coupling between waveguides, A:300 20 Gb/s dual XFP transceiver, A:416–417 2D MEMS switches, A:714 2R (reamplification and reshaping), A:760 2R regenerators, B:722 3D MEMS switches, A:716–719 Fujitsu’s 80 80 OXC, A:718 Lucent’s optical system layout for OXC, A:718 structure of, A:715 3D photonic crystals, B:8 3R (reamplification, reshaping, and retiming), A:760 3R regeneration, A:782–784 3rd harmonic generation, B:11 40-Gb/s WDM transceiver, A:417 integrated driver and MZ modulator, A:418 receiver performance, A:419

8-channel CWDM (de)multiplexer, A:304 8/9 factor, A:656 9xx-nm MM pump lasers, classification of heat transport classification, A:125–126 optical classification, A:126–127

A access-layer networks, B:514 acknowledgment pulse, B:780, 784 acoustooptic tunable filters, B:253 active optical device modeling, B:825 adaptive equalization techniques, B:122, 157 adaptive filter, A:679 Adaptive unconstrained routing, B:649 add/drop filters, B:505 multiplexers, B:306, 487, 524, 635 Address resolution protocol, B:393 adiabatic directional coupler, A:274 advanced modulation formats, B:12, 23, 28, 52, 64, 70, 74, 75, 837 ‘‘Advanced Research and Development Activity’’ (ARDA) agency, A:32 aerial cable terminal, B:424 aerospace applications, limitations, A:375 alignment accuracy, B:135 all-optical signal processing devices, A:11, 760, 761, 766, 778, 818 future prospects nanowire-based integrated alloptical regenerator in chG, A:819 nonlinear-induced phaseshift vs length for a pulse, A:819 alternate chirp return-to-zero, B:58 alternate mark inversion, B:41 alternate phase, B:41

865

866 alternate polarization modulation, B:41 signal, B:207, 602 American National Standards Institute, B:415 amplification scheme, B:71 amplified spontaneous emission (ASE), A:6, 57, 58, 63, 114, 127, 132, 154, 158, 159, 231, 393, 505, 662, 769, 854 amplified spontaneous emission, B:6, 25, 30, 96, 137, 235, 310, 714, 753, 796, 831, 841 amplifier cascades, B:498 chip, B:188, 189 control, B:243, 325, 326, 331 design, B:18, 568 gain, B:236, 242, 253, 326, 330, 331, 753 noise, B:247 spacing, B:73, 274, 496 amplitude distortion, B:651 impairments, B:426, 428 implementations, B:153 measurement techniques, B:245 modulated pilot tone, B:256 phase modulation (chirp), 44 power spectrum, B:245 shift keying, B:34, 123, 132 signal processing, B:147 analog micromirror, A:723, 724 two-axis for WSS, A:724 analog transport system vs digital transport system, A:344 analog-to-digital conversion, B:106, 131 anisotropic etching, A:736 antenna remoting, B:758 APD, see avalanche photodiodes (APDs) Apollo Cablevision, B:459 application-specific integrated circuits, B:165 ARDA, see ‘‘Advanced Research and Development Activity’’ (ARDA) agency arrayed waveguide grating (AWG), A:273, 279–311, B:6, 311, 661, 697, 807 aberrations, A:299–301

Index athermal (temperature-insensitive), A:305–307 channels’ peak transmissions shift, A:306 silicone-filled, A:306 specifications and athermal condition, A:307 band de/multiplexer using, A:303 chirped, A:309–311 component in WDM systems, A:394 conventional vs silicone-filled athermal, A:307 creating flat spectral-response, A:288 crosstalk, A:284–286 demultiplexing properties of 32-ch 10-GHz, A:285 32-ch 50-GHz, A:284 32-ch 100-GHz, A:289 32-ch 100-GHz, parabolic horns, A:289 400-ch 25-GHz, A:285 sinusoidal phase fluctuations, A:286 dense InP, A:349 diffraction order, A:281 dispersion, A:286–287 characteristics of three kinds of, A:287 flat spectral-response, A:287–290 flattening passbands, techniques, A:288 see also arrayed waveguide grating (AWG), periodic stationary imaging gaussian spectral response, A:283–284 grating efficiency, improving, A:300, 301 grating order of zero, see arrayed waveguide lens (AWL) high spectral sampling, A:297–299 sampling theorem, A:298–299 use, A:299 interference condition, A:281 parabola-type flat, A:289 dispersion characteristics of 32-ch 100-GHz, A:290, 291 phase retardations, A:290 passband, A:283 and achieving multiple zero-loss maxima, A:290–291 ‘‘flat’’ passbands, A:287

Index periodic stationary imaging, A:290 achieving multiple zero-loss maxima, A:290 possible ways of laying out the MZI–AWG, A:295 second star coupler transmissivity, A:300 spatial range of, A:282 tandem configuration, A:308–309 three-arm interferometer input, A:296–297 two-arm-interferometer input, A:293–296 creating two zero-loss maxima per FSR, A:293–294 waveguides, A:279 in first slab, A:279 path-length difference, A:282 in second slab, A:280 wavelength for fixed light output position, A:282 wavelength spacing, A:282 arrayed waveguide grating (AWG) demultiplexer, A:297 16-Channel, consisting of three-arm interferometer, A:298 arrayed waveguide lens (AWL), A:301 Artificial defects, A:470–471 ASE noise, signal impairment, A:769 ASE, see amplified spontaneous emission (ASE) asymmetric directional coupler, A:273, 274 asymmetric Fabry–Perot modulator, B:751 asymmetric-bandwidth interleavers, B:168 asynchronous amplitude histogram, B:280 asynchronous digital subscriber line, B:457 asynchronous histogram monitoring techniques, B:280 asynchronous optical packets, switching of, B:631 asynchronous transfer mode, B:416, 479, 516 adaptation layer, B:523 AT&T, B:16, 293, 410, 415, 443, 448, 451, 454, 461, 465, 466, 468, 511, 526, 528, 530, 549 long lines business unit, B:530 Project Lightspeed, B:443

867 athermal (temperature-insensitive) AWGs, A:305–307 Atlantic and the Pacific Oceans, B:564 Attenuation mechanisms, PCF, A:495–498 absorption/scattering, A:494–495 bend loss, A:495–496 confinement loss, A:498 Autocorrelation and first-order PMD, A:633 automatic dispersion compensation, B:213 automatic protection switch, B:487 automatic switched optical network, B:615, 642 avalanche photodiodes (APDs), A:835, B:5, 6, 46 low-excess noise APDs, see multiplication regions, low-excess noise APDs performance of InGaAs APDs as single-photon detectors, A:835 separate absorption/charge/multiplication APDs, A:245–247 InP/InGaAsP/InGaAs SACM APDs, A:246 ‘‘long-wavelength’’ photodetectors, A:245 SACM structure, ‘‘C’’ (charge layer), A:246 single-photon avalanche detectors, A:256–258 Geiger-mode operation, A:256 quantum cryptography/quantum key distribution, A:256 single-photon-counting avalanche photodetector (SPAD), A:256 average effective refractive index, B:713 AWG multiplexer distributed feedback (DFB) lasers spectrum with, A:353 AWG, see arrayed waveguide grating (AWG) AWG-based switch fabrics, B:727 AWL, see arrayed waveguide lens (AWL)

B back-to-back bit error rate, B:113 configuration, B:225, 582, 603, 605 measurements, B:126, 224

868 back-to-back (continued ) performance, B:67, 215 sensitivity, B:38, 53, 54, 154 backbone routers, B:531, 546 Baker–Hausdorff formula, B:832 balanced detection, B:47, 61, 72, 73, 100, 138, 141, 144, 147, 168, 170, 218, 225, 633, 759 balanced driving see push-pull; operation band edge engineering, A:456 broad area coherent laser action, A:473–476 lattice point control and polarization mode control, A:476 resonance action at band edge, A:473 band engineering, A:456, 476–477 group velocity of light, control of, A:477–478 light propagation, angular control of, A:478–479 band gap/defect engineering, A:456–457 three-dimensional photonic crystals, A:467 fabrication technology, A:467–468 light propagation, control of, A:472–473 spontaneous emission, control of, A:468–472 two-dimensional photonic crystal slabs, A:456–457 heterostructures, effect of introducing, A:459–463 high-Q nanocavities, A:463–466 line/point-defect systems, control of light by combined, A:457–459 nonlinear and/or active functionality, introduction of, A:466–467 band multiplexer, A:303 5-band 8-skip-0 multiplexer, A:304 bandwidth, B:26, 200, 454, 753, 769, 851 limitations, B:193, 705, 720 -on-demand, B:302, 304, 549, 556 -time trading, B:781 -times-distance, B:445, 446, 458, 461, 468 banyan networks, B:773, 774, 778 BASE, see broad-area single-emitter (BASE)

Index Baud, B:29, 165, 166 beam propagation method (BPM), A:282 central wavelength, A:282 finite-difference (FD-BPM), A:299, 300 Fourier transform (FT-BPM), A:299–300 intensity distribution, double-peaked, A:288 mutual coupling between waveguides, A:300 shorter wavelength component, A:282 beam propagation method, B:805 beam-steering mirrors, B:316 Bell company, B:451, 462 Bell system, B:522, 530 Bellcore, B:522, 530 bending losses, A:387 guide configurations, A:387 BER measurement of QD laser module, A:37 Bernoulli traffic, B:785, 786 BERTS, see bit-error rate test set (BERTS) bidirectional line-switched rings, B:523 binary differential phase shift keying, B:58 binary electronic drive signals, B:62 binary optical intensity eye diagram, B:54 binary phase shift keyed, B:757 binary tree-like configuration, B:794 bipolar feedback signal, B:211 birefringence, A:275, 494, 607–609 birefringence’s derivative, A:612 bit error rate, B:26, 113, 133, 180, 238, 385, 494, 533, 567, 571, 582, 596, 600–602, 604, 804, 830 measurements, B:65, 215, 219, 224, 676 tester, B:219 bit-error rate test set (BERTS), A:204 BitTorrent, B:457 blind equalization, B:122 estimator, B:160 block length, B:164 blocker, see 1 1 WSS blocking/nonblocking, B:653 Boltzmann’s constant, B:742 boundary-value problem, B:822

Index BPM, see beam propagation method (BPM) Bragg diffraction, A:473 Bragg gratings, A:327, 512–513, B:253, 586 Bragg reflection, A:464 Bragg wavelength, B:632 branch stations, B:569 Bridge Protocol Data Units, B:362 bridging, B:346, 358, 362, 568, 604 Brillouin gain coefficient, B:746 spectra, B:427, 428 Brillouin scattering, A:507–509 backward, A:507 forward, A:508–509, B:831 broad-area single-emitter (BASE), A:127 broadband access, B:417 asynchronous histogram, B:276 deployment, B:409, 615 digital cross-connect system, B:525 over power line, B:466 storm, B:362, 365 broadbanding techniques principles of, A:682–683 RC broadbanding, A:683–687 series shunt-peaked broadbanding, A:694–697 series-peaked broadbanding, A:692–693 shunt-peaked broadbanding, A:687–692 Broglie wavelength, B:3 bubble-and-bust, B:2, 3 buffer delay, B:709 memory, B:656 parameters, B:710 technologies, comparison of, B:719 buffering, B:226, 536, 629, 630, 637, 643, 650, 651, 665, 667–679, 671, 677, 683, 695, 696, 698, 699, 701, 707, 710, 715, 729, 733, 734, 771, 772, 777, 778, 781, 792, 793 electronic messages, B:792 mechanisms, B:630 space, B:781

869 bufferless photonic switching nodes, B:778 built-in optical buffer, B:649 buried oxide (BOX) layer, A:384 burst assembly, B:648 burst header cell, B:645 burst mode operation, B:385 receivers, B:170, 359 burst scheduling, B:651 butt-joint regrowth, A:345–346, 348 butterfly network see banyan networks

C cabinet installation, B:425 cabinet size reduction, B:425 cable management technology, B:422 cable modem, B:406, 414, 440, 460, 461, 466 cable television, B:17, 305, 378, 403, 739, 748, 758 cables with PF polymer-based POFs, A:599 canonical format indicator, B:366 capability index (Cpk), A:356 capacity upgrade, B:13, 131, 171, 596 capital expenditure, B:482, 484, 506, 507 capital investment, B:418 card failure, B:544 carrier craft personnel, B:423 carrier density, B:749, 825, 829 carrier phase-estimation method, B:102 carrier suppressed return-to-zero, B:40, 56, 57, 597 carrier-grade telecommunication equipment, B:65 see also Ethernet carrier-to-noise ratio, B:426, 744 Cartesian grid, B:34 cascaded filtering, B:320 cascaded optical injection locking (COIL), A:153, 158 cascaded-ring-optical-waveguide filters, A:397, 398 catastrophic optic mirror damage (COMD), A:119 catastrophic problems, B:244

870 CD see chromatic dispersion cellular-mobile telephony, B:449, 454 chalcogenide fiber-based wavelength conversion, A:789–793 BER vs received optical power, A:791 demonstrating XPM-based wavelength conversion, A:790 principle of XPM wavelength conversion, A:789 required calculated pump intensity, A:793 spectra of As2Se3 fiber output, A:791 chalcogenide fibers high quality fabrication processes, A:763 nonlinear material characteristics, A:764 PLD development, A:763 and waveguide devices, A:763 chalcogenide waveguides, A:793–796 BERs of converted pulses, A:795 cross phase modulation-based, A:794 channel conditioning, B:308 channel demultiplexing, B:123, 776 channel filtering effects, B:500 channel frequencies, B:125 channel identification, B:243 channel power attenuation, B:330 monitoring scheme, B:252 channel spacing, B:52, 53, 71, 123, 124, 126, 127, 195, 218, 234, 238, 247, 322, 390, 391, 479, 484, 492, 493, 495, 499, 500, 596, 604, 605, 675, 835 channel spectra, B:169, 564, 596, 604, 605 channel telemetry, B:309 channel-degrading effects, B:233, 236, 285 channel-to-channel cross-polarization, B:605 chip cost vs assembly plus testing, A:345 chirp, B:4, 5, 42–45, 51, 54, 56, 58, 59, 135, 205, 235, 245, 247, 264, 265, 266, 277, 319, 493, 651, 653 effects, B:264 fluctuation, B:265 managed laser, B:42 multiprocessors, B:765 parameter, B:264

Index Chirped AWG, A:309 interleave Chirping, A:310–311 WSS using two and AWL, A:311 linear chirping, A:310 chirped fiber Bragg gratings, B:181, 195 chirped return-to-zero, B:58, 596 chromatic dispersion, B:2, 15, 24, 31, 37, 42, 54, 63, 68–70, 72, 77, 110, 118, 120, 121, 123, 131, 133, 134, 137, 150, 155, 165, 166, 168–170, 180–182, 193, 202, 203, 208, 212–214, 223, 225, 234, 235, 243, 245, 247, 256, 257, 260, 261, 263–271, 273, 275, 281–284, 348, 359, 365, 366, 379, 380, 384, 391, 414, 565, 566, 571–573, 576–581, 593, 596, 601–603, 616, 630, 682, 775, 816, 818, 828, 836, 844, 846, 847 compensation of, B:212 chromatic dispersion, transmission impairments, A:800, 808 circuit-switched layer, B:522, 530, 536, 551 telephone network, B:19, 542 cladding-pumped fiber, A:556 class of service, B:545, 648 cleaving/splicing, PCF, A:500 client layer protection requirements, B:852 client optoelectronic, B:485 client service interfaces, B:491 clock distribution circuits, B:187 clock modulation, B:211, 224 clock recovery, B:108, 138, 192, 193, 203, 204, 208, 210, 211, 216, 217, 219, 224–226, 267, 280, 282, 348, 356, 660, 673, 681, 781, 788 circuit, B:108, 216 device, B:203, 204, 211, 216, 225, 226 clock-phase margin, B:192 CMOS integration and integrated silicon photonics examples, A:416–420 microwave OE oscillators, A:420–421 microwave spectrum analyzer, A:421 SOI transistor, A:414–416 CMOS technology, state of art in, A:700 CMOS, see complementary metal oxide semiconductor (CMOS)

Index CMOS-like gate dielectric, charge accumulation, A:404 coarse WDM, A:304, B:478, 516 coaxial cable, B:349, 410, 413, 419, 551, 617 code division multiplexing, B:656 coherent crosstalk, B:131, 170, 241 coherent demodulation, B:12, 28, 46, 69, 116 see also delay demodulation coherent detection, A:705–706, B:13, 23, 24, 26, 33, 49, 52, 63, 64, 76, 78, 79, 97, 98, 118, 121, 131, 137, 145, 148, 149, 154, 195, 196, 268, 272, 606, 743, 745 performance of, B:145 principle of, B:98 coherent optical communications, B:95, 96 coherent receiver, B:46, 47, 48, 64, 96–99, 102, 121, 132, 147, 150, 152, 162, 166, 170, 171, 319, 745 see also digital coherent receiver coherently connected interferometers (FIR-based TODCs), A:329 two-arm interferometers, A:329–330 multiple controls, A:329, 330 single controls, A:329–330 COIL, see cascaded optical injection locking (COIL) coincidence-counting rate biphoton amplitude, A:844 coincidence-to-accidental ratio (CAR), A:865 collective blindness, new dawn, A:25–26 coherent three-dimensional clusters, A:26 ‘‘Frank–van der Merwe’’ growth mode, A:26 collimators, B:111 collision avoidance, B:695 colorless drop, B:313 combiner, see optical coupler COMD, see catastrophic optic mirror damage (COMD) commercial packet switches, B:20 committed information rate, B:378 common gate cascode shunt-peaked amplifier, A:690

871 communication engineering techniques, B:23, 28 communication symbols, B:28 community antenna television, B:440 Compact resonant ring modulator, A:409 competitive local exchange carrier, B:448, 524 complementary metal oxide semiconductor (CMOS), A:381, B:166 composite triple beat, B:426 consent decree, B:519 constant bit rate, B:549 constant modulus algorithm, B:161 see also Godard algorithm constant-amplitude format, B:150 constellation diagram, B:46, 121, 133, 148, 159 constrained coding, B:39–41 continuous-time DFE using analog feedback, A:698 control devices inter-signal, A:312 large channel count ROADMs, A:316–317 reconfigurable optical add/drop multiplexers, A:313–316 wavelength-selective switches, A:317–318 intra-signal, A:318 optical chromatic dispersion compensators, A:325–335 optical equalizers, A:322–325 temporal pulse waveform shapers, A:319–322 coplanar waveguide (CPW), A:243 core segment, B:555 core-layer networks, B:529 correlative coding, B:41, 53, 57 cost effective-related technologies, B:502 cost reduction, B:13, 131, 171, 186, 401, 421, 422, 431 cost-effective technologies, B:502 cost-effectiveness of silicon, B:20 counter-pumped amplifiers, B:333 coupled mode theory, A:458 coupled nonlinear Schrodinger equations, A:652–655

872 coupled resonator optical waveguides (CROW), A:444 coupled-resonator waveguides, B:713 coupling, B:8, 9, 36, 44, 191, 246, 316, 324, 326, 328, 330, 332, 338, 416, 421, 422, 428, 504, 505, 671, 683, 752, 757, 807, 814, 819, 824, 825, 837 waveguides, B:422 CPW, see coplanar waveguide (CPW) Crank–Nicholson scheme, B:807 cross phase modulation, B:52, 60, 64, 72, 79, 123, 124, 126, 127, 137, 162, 166, 167, 170, 171, 209, 211, 241, 273, 274, 566, 567, 572, 578, 831, 833 cross-phase modulation (XPM), A:537, 657, 761 crossbar, B:778, 782, 790 crosstalk, B:33, 253, 321, 353, 651, 653, 728 crystal and waveguide lasers, B:748 current injection, B:825 current-controlled oscillator, B:749 customer premises equipment, B:521 cybersphere, B:457 cyclic redundancy check value, B:359

D data data data data

burst transmission, B:627 communication channel, B:645 communication network, B:645 modulators (MEMS devices), A:743–745 Data Over Cable Interface Specification, B:414 data signal processing, B:13 data stream intensity autocorrelation, B:268 data traffic and evolution of the optical network, B:19 data transmission upgrading to 10Gb/s, A:672 Data Vortex see topology data warehousing, B:481 data-aided MSPE scheme, B:164 datacom applications with POFs, A:602 DC properties of long-wavelength, QW lasers

Index DC measurements of optical gain/ refractive index change/LEF, A:56–58 E-TEK Dynamics LOFI, A:57 Hakki–Paoli method, A:57 DC-measured experimental data/ theory, comparison, A:59–63 Auger recombination process, A:61 LEFs of InAlGaAs and InGaAsP systems, comparison of ASE spectra, A:64 optical gain/refractive index change/ LEF, theory of non-Markovian gain model, A:55 DDF, see dispersion-decreasing fiber (DDF) decision feedback equalizer (DFE), A:675–676, B:214 better performance than LE, A:675 designing of, A:676 low SNR, A:675 structure of, A:675 decision flip-flops, B:191 decisive break-throughs, A:26–27 defect budgeting, A:361 defect density and functionality per chip, A:361–363 position of PIC technology, A:363 deflection, B:651, 725, 774, 775, 777, 792–796 nodes, B:783 routing, B:725, 777, 792, 793 signal, B:774, 775, 793, 795, 796 degenerate mode group, B:838 degenerate photon pairs for quantum logic in telecom band photon-pair generation in optical fiber, A:863–869 CAR plotted as function of single counts/pulse, A:865 pump preparation and CPS scheme, A:864 single counts and coincidence counts, A:868 two-photon interference, A:867 delay demodulation, B:12, 46–49, 64 delay interferometer, B:46, 133, 144, 193, 194

Index delay line buffers, B:710 multiplexer, B:223 demodulation, B:28, 46, 48, 49, 58, 62, 69, 96, 97, 100, 106, 107, 115–117, 123, 137, 147, 601, 603 demodulator phase error, B:152, 153 demultiplexer, B:123, 160, 165, 183, 191–193, 203, 208, 210, 215–218, 224, 226, 314, 353, 415, 618, 671, 676, 699, 718, 724, 790, 851 demultiplexing, A:782, 792, B:26, 53, 123, 125, 137, 180, 183–185, 192, 208, 209, 210, 215–218, 222, 226, 256, 296, 314, 320, 415, 526, 527, 561, 604, 656, 676 applications, B:216 components, B:226 Dense InP AWG (25GHz)-based 64-channel O-CDMA encoder/decoder chip, A:349 dense wavelength division multiplexed, B:4, 119, 351 dense wavelength-division multiplexing (DWDM), A:4, 109, 116, 172 design of experiments (DOE), A:357 destination address, B:358, 360, 362, 375, 653, 682, 774, 792, 795, 796 detection techniques, B:18, 28, 745 DGD see differential group delay DGD changes vs outside temperature, A:618 dial-up modem, B:440 dielectric thin-film filters, B:253 differential binary phase-shift keying, B:134 differential delays, B:381 differential detection, B:58, 61, 116, 164, 598, 602 differential group delay (DGD), A:607, B:70, 160, 162, 168, 214, 219, 222, 223, 262–266, 268–272, 570, 571, 582–583, 603 differential modal dispersion (DMD), A:699 differential mode delay, B:837 differential phase-shift keying, B:33, 49, 50, 58, 62, 131, 203, 256, 269, 598, 625

873 differential quadrature phase-shift keying, B:184, 390 differential quadrature PSK (DQPSK) demodulator, A:278 differential-detection penalty, B:141, 145, 148, 153, 155 differential-phase distance, B:148 differentially coherent demodulation, B:46 diffraction order, A:281 diffractive spectrometers and spectral synthesis characteristics of, A:739–741 optical MEMS, A:740 optical system, A:739 digital coherent receiver, B:97, 98, 105–107, 111, 114, 123, 125, 127, 128, 132, 157, 158, 162, 164, 165 advantages of, B:123 digital communication receiver, B:30 digital cross-connect, B:623 digital equalization, B:157 digital homodyne receiver, B:123, 124 digital line module (DLM), A:366 digital micromirror devices, successful MEMS product, A:713 digital network, constructing, A:344–345 digital optical network (DON), A:344, 376 architecture, A:367–368 monolithically integrated TX and RX PICs, A:351 digital radio-frequency, B:12 digital ROADMs, A:368 data ingress/egress in, A:369–370 digital self-coherent receivers, B:131, 147, 150 digital signal processing, B:23, 48, 63, 64, 69, 70, 76, 78, 79, 97, 107, 131, 163, 196 digital subscriber line, B:410, 438, 515 multiplexer, B:462 digital TV, B:413 evolution of, B:449 digital video recorder, B:554 diode-pumped crystal, B:748 diodepumped solid-state lasers (DPSSL), A:108 direct-contact wafer bonding, A:346 direct-detection schemes, B:149

874 directional coupler, A:272 adiabatic, A:274 asymmetric, A:273, 274 curved, A:274, 275 multi-section, A:274 directional separability, B:306 directly coupled resonator configuration, A:432, 433, 437, 439 directly modulated lasers, B:42, 722, 757 Discovery Gate, B:381 discovery window, B:381 discrete event simulations, B:852 discrete multitone, B:415 dispersion compensation fibers, A:526–539 design examples, A:532–534 experimental results, A:534–536 fiber profile designs, A:531–532 future developments, A:538–539 mode field diameter/effective area, A:536–537 nonlinear properties, A:537 principle, A:527–531 multiple path interference (MPI), A:531 parameter k, A:529 PMD, see polarization mode dispersion (PMD) Rayleigh scattering, A:530 splicing loss, A:538 technologies, A:526–527 dispersion compensation, B:9, 10, 12, 13, 58, 71, 79, 97, 121, 165, 166, 181, 194, 195, 208, 213, 214, 219, 234, 236, 255, 261, 263, 296, 317, 318, 319, 322, 353, 390, 495, 505, 571, 706, 713, 716, 757 devices, B:571 fiber, B:181–183, 495, 581, 741 inaccuracies in, B:234 technique, B:213, 214 unit, B:836 dispersion compensators, B:8, 195, 213, 261, 356, 573, 625, 634, 714 all-pass filter, A:733 MEMS Gires–Tournois interferometer, A:734 dispersion equalization capability, B:150

Index dispersion management, B:319, 836 dispersion map, B:52, 54, 58, 60, 63, 64, 74–76, 79, 170, 244, 274, 317–320, 334–336, 339, 390, 577, 835 effect of, B:75 dispersion slope, B:58, 213, 214, 222, 223, 335, 576, 577, 579, 581, 582, 836 dispersion tolerance, B:55, 69, 70, 182, 319, 352, 388, 493, 495, 496 dispersion-decreasing fiber (DDF), A:548, 553–554 dispersion-limited propagation distance, B:51 dispersion-managed fiber cable systems, A:538 dispersion-managed fiber, B:213 dispersion-managed transmission, B:167 dispersion-related impairment, B:281 dispersion-shifted fiber, B:12, 213, 576 dispersive impairments monitoring, B:260 distortion, B:247 distributed feedback (DFB) lasers, A:347 L–I–V characteristics, A:363 typical light vs current/voltage (L–I–V), A:352 distributed feedback, B:62, 110, 388, 663, 749, 826 lasers, B:826 distribution network, B:784 DLM, see digital line module (DLM) DMD, see differential modal dispersion (DMD) donut-shaped far-field pattern, A:476 doped fiber amplifier, B:4, 6, 96, 235, 413, 561, 593, 748, 752, 836 double sideband, B:264 double-clad fibers, fiber lasers/amplifiers by OVD, A:555–572 concept, A:555–556 double-clad fiber 101, A:556–562 requirements, A:558 double-clad fibers, high-performance Yb-Doped Al/Ge counter-graded composition design/fabrication, A:568–570 reduced-overlap, A:569 introduction, A:565

Index SBS effect in optical fiber, A:565–566 SBS-managed fiber composition design, A:566–568 SBS-Managed LMA, characteristics of, A:570–572 double-gate switch ‘‘off ’’ state, A:313, 314 ‘‘on’’ state, A:313–314 double-gate thermooptic switches (TOSWs), A:313 double-stage polarization, B:181 DPSSL, see diodepumped solid-state lasers (DPSSL) DRAM (Dynamic Random Access Memory)/flash memory cells, A:23 driver amplifier, B:45, 183, 185–191, 193, 195 drop spectra, A:462 drop-and-continue function, B:305 Drude model, A:399 dual-band filter, transmission spectrum of measured with tunable source and OSA, A:834 dual-frequency pumps degenerate frequency polarizationentangled photon pairs produced by, A:869 duobinary, B:41, 45, 53, 54, 133, 256, 258, 357, 390, 495, 834, 836 DWDM, see dense wavelength-division multiplexing (DWDM) dynamic bandwidth services, B:549 dynamic channel equalizers (DCEs), A:720 dynamic gain equalization filter, A:301–303 arrangement with lower insertion loss, A:302 consisting of two AWGs connected by AWL, A:302 dynamic gain equalizers (DGEs), A:719 dynamic optical circuit switching, B:626 dynamic photonic crystals, A:443 creating, A:445 future prospects, A:450–451 structure used to stop light, A:447 dynamic photonic crystals, stopping light in, A:443 conditions for stopping light, A:444–445

875 numerical demonstration in photonic crystal, A:448–450 from tunable bandwidth filter to lightstopping system, A:447–448 tunable Fano resonance, A:445–447 tuning spectrum of light, A:443–444 dynamic random-access memory, B:710 dynamic spectral equalizer, A:720

E e-commerce, B:481 e-Field Domain Signal Processing, A:705–706 e-Health, B:614 E-Line, B:378 e-science applications, B:555 EAM, see electroabsorption modulator (EAM) EDC, see electronic dispersion compensation (EDC) EDFA, see erbium-doped fiber amplifier (EDFA) edge emitting lasers, B:4, 826 edge routers, B:649 EEQ, see electrical equalization (EEQ) effective-index microstructured optical fibers (EI-MOF), A:575–579 air-filling fraction, A:576 attenuation, A:579 basic concepts, A:575–576 dispersion, A:578–579 endlessly single-mode (ESM), A:577 small modal area/nonlinearity, A:577–578 V-number, A:576–577 EI-MOF, see effective-index microstructured optical fibers (EI-MOF) EIT, see Electromagnetically Induced Transparency (EIT) electrical amplifiers, B:758, 830 electrical equalization (EEQ), A:322 electrical multiplexing, B:14, 201 electrical phase shifters, B:192 electrical sampling systems, B:205 electrical signal processing, B:214 electrical spectrum analyzer, B:276

876 electrical switching fabric, B:502 electrical synchronous demodulation, B:100 electrical time-division multiplexing, B:179 electro-absorption devices, B:5 modulator, B:5, 42, 43, 186, 204, 388, 636, 663 electro-optic bandwidth, B:191 coefficient, B:188 effect, B:807 modulators, B:5 phase comparator, B:210 electroabsorption (EA), A:192 electroabsorption modulator (EAM), A:185, 189, 198, 208, 210, 348 small-signal frequency response, A:353 electroabsorption-modulated lasers (EML), A:347 electromagnetically induced transparency (EIT), A:445 electron charge, B:99, 825 electronic and optical properties, A:28–31 biexciton-binding energy, A:31 excitonic/biexcitonic recombination, A:31 GaAs-based QD technology, A:29 piezoelectric potentials, A:29 shape/composition, QDs, A:28 tuneability of the emission of (InGa)As/GaAs QDs, A:29, 30 electronic bottleneck, B:705, 706 electronic buffers, B:710, 718, 733 electronic compensation for 10-GB/S applications LAN: transmission over MMF, A:699 electronic equalization for 10GBASE-LRM, A:700–701 metro and Long-Haul transmission over SMF E-field domain signal processing, A:705–706 electronic compensation of PMD, A:704–705 electronic equalization to compensate chromatic dispersion, A:703–704 modeling of transmission channel and impairments, A:701–703

Index electronic demodulator error compensation, B:150 electronic demultiplexer circuits, B:192 electronic dispersion compensation (EDC), A:539, B:12, 158, 353, 356, 495 electronic equalization, B:107 and adaptation techniques, A:678–681 BER, adapting coefficients, A:679 implementation considerations, A:673 LMS adaptive transversal filter, A:680 low bit-error rate (BER) application, A:673 minimize BER, A:679 nonclassical channel, A:673 performance metrics for adaptive filter, A:679 synchronous histogram monitor, A:679 to compensate chromatic dispersion, A:703–704 compensation of PMD, A:704–705 for 10GBASE-LRM, A:700–701 electronic logic gates, B:787 electronic multiplexer, B:26, 32, 184, 185, 187–189, 191, 192, 415 electronic packet switch, B:642, 650, 656, 696–699, 701, 705, 720, 721, 725, 731, 732, 735 electronic polarization demultiplexing, B:157, 159 electronic predistortion, B:76, 78 electronic signal processing in optical networks adaptation techniques, A:678–681 data center applications, A:707 electronic compensation for 10-GB/S applications, A:699–707 electronic equalization decision feedback equalizer, A:675–676 linear equalizer, A:673–675 maximum A posteriori equalization, A:677–678 maximum likelihood sequence estimation, A:676–677 variants on theme, A:678

Index high-speed electronic implementation architectural considerations, A:697–699 broadbanding techniques, A:682–697 prospects and trends for nextgeneration systems, A:707–708 role of, A:671–672 storage area networks, A:706–707 electronic signal processing, B:7, 23, 24, 37, 180, 184, 185, 195 electronic signal-processing, adaptive, A:681, 701, 705, 708 high-speed, A:672, 681 receiver-based electronic dispersion compensation (EDC), A:672, 701 signal impairments, mitigate, A:671–672 upgrading data transmission to 10Gb/s over installed fiber, A:672 electronic switch fabrics, B:729 electronic time division multiplexing, B:13, 14, 32, 179, 180–183, 185–195, 197, 199, 201, 202, 210, 218, 224–226, 705, 706 electronic transmission mitigation, B:391 electronically predistorted optical waveforms, B:44 electrooptical effects in silicon, A:398 electrostatic actuation, A:715 embedded DRAM (eDRAM), B:718 EML, see electroabsorption-modulated lasers (EML) emulation, A:641–652 emulators, A:645–647 first-order, limitations, A:642–645 sources, A:647–650 with variable DGD sections, A:650–652 encircled flux, B:837 encode information, B:28 end-to-end connection, B:708 networked information, B:23, 26 optical data path, B:791 endlessly single mode (ESM), A:490, 577 entertainment video service, B:411 equalization filter coefficients, B:161

877 equalizers, A:322 erbium-doped fiber amplifier (EDFA), A:107, 116, 124, 135, 161, 170, 343, B:1, 2, 4, 13, 96, 113, 117, 125, 169, 180, 195, 200, 215, 216, 219, 221–223, 225, 235, 241, 248, 253, 257, 263, 265, 294, 325, 326, 330–334, 338, 388, 389, 413, 496, 497, 561, 564, 565, 572, 584–593, 790, 830, 836, 851 gain spectrum, B:338 systems, B:331, 591 technologies, B:96 error correction, B:564, 582, 583, 706 coding, B:28, 38, 39 error correlation information, B:277 error distribution, B:182 error-free operation, B:217 performance, B:672, 679 ESM, see endlessly single mode (ESM) Ethernet, B:14–16, 19, 132, 169, 179, 196, 200, 225, 345, 346–353, 355, 357–363, 365–383, 385–387, 389–397, 399, 406, 417, 464, 477, 480, 481, 483–486, 490, 491, 493, 502, 506, 507, 513, 515–518, 520, 526, 528, 529, 532, 543, 550, 553, 620, 623, 624, 638, 642, 682, 703, 837 carrier, B:16, 169, 369, 638 frame format, B:358 layer spanning tree protocol, B:543 layering, B:347, 379 passive optical network, B:379, 387, 703 physical layer, B:351 service models, B:378 switches, B:486, 517, 528, 553, 623 transport, B:169, 179, 370, 528, 529 virtual circuits, B:378 Euclidean distance, B:148, 149 European Commission, B:452, 466, 615 evanescent coupler, see directional coupler evanescent coupling, B:824 excess burst rate, B:378 excess information rate, B:378 explicit telemetry, B:309 extinction ratio, B:43, 56, 59, 67, 493, 651, 663

878 extra link capacity, B:545 eye diagram, B:51, 54, 55, 57, 60, 63, 152, 191, 193, 194, 219, 243, 245, 276, 277, 280–283, 581, 582, 594, 663, 676, 839 eye monitoring techniques, B:279

F fabric failure, B:544 fabrication technology, A:467–468 Fabry–Perot, B:253, 352, 440, 681, 751, 826, 827 filters, B:253 lasers, B:352, 440, 827 factory automation, A:357 Fano-interference, A:445 configuration, A:432–433, 438 Faraday rotation, principle, A:350 Fast Ethernet lines, B:518 see also Ethernet fast mixing assumption, A:615 fast mixing model, A:616 fault management, B:243, 244, 278, 856 FAX see dial-up modem FBG, see Fiber Bragg Grating (FBG) FDTD, see finite-difference time-domain (FDTD) federal communications commission, B:17, 402, 403, 407, 408, 413, 453, 458, 468, 519, 528 feed-forward arrangement, B:710 pump, B:333 scheme, B:164 feedback loop, B:77, 78, 214, 497 fiber attenuation, B:246, 388, 591 Fiber Bragg Grating (FBG), A:112, 114, 116, 117, 121, 129, B:9, 181, 253, 661, 586, 632 Fiber Channel, A:706–707, B:485, 486 fiber coupling, B:316, 428 fiber lasers, B:747 fiber lasers/amplifiers, A:503 fiber nonlinearities, B:24, 26, 28, 38, 40–42, 52, 57, 58, 63–65, 69, 70, 73–76, 79, 123, 133, 166, 202, 208, 213, 245, 245, 248, 260, 268, 273–276, 325, 330, 338, 760

Index monitoring techniques, B:275 as source for correlated photons coincidence counting results, A:836 coincidence rates as function in single-photon rates, A:835 correlations between signal and idler photons, measuring, A:834–835 dual-band spectral filter based on double-grating spectrometer, A:834 fiber polarization controller, A:833 photon-counting measurements, A:833 photon-pair source, efficiency of, A:833 quantum efficiency vs dark-count probability, A:835–836 spectral filter with fiber Bragg gratings, implementation of, A:837 transmission curves, A:834 transmission efficiency in, A:834 two-pump photons, A:833 as source for entangled photons coincidence counts and single counts, A:856 degree of polarization entanglement, A:584, 857 efficient source for developing quantum communication technologies, A:858 measured values of S for four Bell states, A:857 measurement of polarization entanglement, A:855–858 nature of entanglement generated, A:851 observed polarization (in) dependence of parametric fluorescence, A:856 parametric fluorescence, source of quantum-correlated photon pairs, A:852 photon-counting modules, A:855 polarization correlations, A:855–856 polarization- entangled states, measurements, A:855 polarization entanglement, experimental setup and sinusoidal variations, A:853

Index polarization-entangled photon pairs in 1550-nm telecom band, A:851 quality of polarization entanglement, A:857 quantum efficiencies, detection of, A:855 signal and idler photon pairs, A:852 fiber optics applications communication, B:47, 50, 71 current status of, B:442 history of, B:445 fiber performance technology, B:426 fiber polarization controller, A:833 fiber refractive index, B:566, 567 fiber to the node, B:410 fiber transmission impairments, B:24 line, B:212 medium, advantages of, B:739 fiber-based access, B:403, 405, 408, 418, 446 broadband, B:16, 401, 403, 404, 410, 431, 438, 453, 469 optical signal processing, B:215, 217 solutions, B:403 fiber-drawing techniques, A:620 fiber-fed base stations, B:760 fiber-in-the-loop deployments, B:468 fiber-optic quantum information technologies degenerate photon pairs for quantum logic in telecom band Hong-Ou-Mandel interference, A:869–876 photon-pair generation in optical fiber, A:863–869 fiber nonlinearity as source for correlated photons, A:832–837, 851–858 high-fidelity entanglement with cooled fiber, A:585–863 quantum theory of four-wave mixing in optical fiber, A:838–851 fiber-to-the-building, B:438, 464 fiber-to-the-curb, B:16, 404, 408, 410, 413, 415, 417, 419–421, 429–431, 438, 442, 443, 458, 462

879 fiber-to-the-home, B:5, 16, 17, 251, 401, 404, 410, 413, 416, 417, 419, 421, 422, 426, 428, 430, 440–442 fiber-to-the-premises, B:410, 416, 418, 437–444, 446, 448, 452, 453, 461, 466–469, 613–615, 624 fiber’s birefringence, A:607 field programmable gate array, B:165, 493 field tests and PMD, A:616–621 figure of merit, A:334–335 for dispersion compensators, A:335 for TODC, A:334–335 filter characteristics, B:53, 67, 68, 320 detuning, B:277 offset, B:283 finite element method, B:812 finite impulse response, B:159, 834 finite linewidth, B:143 see also frequency jitter finite measurement bandwidths, A:617 finite-difference time-domain (FDTD), A:441, 442, B:809 switching dynamics, A:442 first-in-first-out buffer, B:711 first-order PMD, A:633 autocorrelation, A:634 penalty, A:629, 638 fixed filter based nodes, B:499 fixed frequency grid, B:183 fixed wavelength converters, B:725 ‘‘flat’’ passbands, A:287 flat-top filter construction theorem, A:293 flip-chip bonding, A:346 flip-chip, B:189, 422 forgetting factor, B:154 forward error correction, B:10, 12, 23, 28, 39, 60, 63, 65, 67, 125, 145, 167, 180, 205, 247, 306, 372, 385, 493, 564, 571, 582, 583, 596, 604 forward propagating slab waveguide, A:391 four-dimensional signal space, B:50 four-point constellation, B:96 four-stage distribution network, B:786, 787

880 four-wave mixing (FWM), A:760, 762, 833 four-photon scattering (FPS) process, A:838 photon–photon scattering process, A:838 signal and idler photons, A:838 four-wave mixing, B:40, 52, 166, 167, 209, 211, 215–217, 241, 253, 260, 273, 274, 567, 573, 578, 668, 673, 675, 676, 831 Fourier space, B:818 Fourier transforms, B:818 frame check sequence, B:359 frame relay, B:479, 481 France Telecom, B:458 Franz–Keldysh modulators, A:399 Franz-Keldysh effect (FK), A:195, 196, 213, B:5 free spectral range, B:194, 257 free-space optics, B:139 frequency drift, B:62, 113, 139, 247 frequency estimation, B:158, 162, 165 see also phase estimation frequency jitter, B:143 frequency offset compensation, B:163 frequency-domain explanation, B:54 split-step, B:833 frequency-encoding code division multiplexing (FE-CDM), A:320 frequency-shift keying, B:33, 133, 663 Fresnel reflections, B:246 Fujitsu’s 80 80 OXC, A:717 Full Service Access Network Group, B:415 full width at half maximum (FWHM), A:39, 114 full-duplex Ethernets, B:348 fusion splices, B:246 FWHM, see full width at half maximum (FWHM) FWM, see four-wave mixing (FWM)

G GaAs-based QD technology, A:29 gain error, B:330, 332, 333 gain peaking, B:189

Index gain ripple, B:308, 330, 338 impact of, B:330 gain-flattening filter, B:326, 584, 585 Galerkin method, B:812 gallium arsenide, B:44 Gas-Based nonlinear optics high-harmonic generation, A:510–511 induced transparency, electromagnetically, A:511 stimulated Raman scattering, A:510 Gate message, B:381 gate oxides, carrier accumulation on, A:403–404 gate switching energy, B:707 Gaussian approximation, B:842, 845 Gaussian AWG, A:283–284 wide passband, A:288 Gaussian distribution, B:115 Gaussian noise, B:30, 115, 259 Gaussian random variables, B:849 Gemini network interface, B:792 generalized multi-protocol label switching (GMPLS), A:367, B:347, 615, 642 generic framing procedure, B:373 germanium photodetectors, A:409–414 benefits, A:411–412 and photoreceivers for integrated silicon photonics, A:409–414 p-type Ge contacts, A:413 single heterojunction, A:413 waveguide (WPDs), A:412 GI-POF see graded-index plastic optical fiber (GI-POF) Gigabit Ethernet, B:10, 200, 346, 348, 349, 351, 353, 355, 357, 370, 372, 373, 385, 392, 394, 485, 521, 703 link, B:346, 392 physical layer, B:349 private line, B:521 see also Ethernet Gigabit interface convertor, B:346 Gires–Tournois etalons, A:326–327 Global Lambda Integrated Facility, B:617 global system for mobile communications (GSM), A:179 Godard algorithm, B:122 Google, B:389

Index Gordon-Mollenauer phase noise, B:162, 167 graded-index plastic optical fiber (GI-POF), A:597, 707 gradient-type optimization algorithm, B:161 grating coupler, A:389–390 backward scattering, A:391–392 fiber-coupling efficiency of, A:392 grating mirrors, reflectivity spectrum, A:348 Gray coding, B:35 grid computing, B:549, 555, 611, 613, 617, 620, 629 grid spacing, B:253 grooming, B:502, 518, 526, 528 Group Velocity Dispersion (GVD), A:494–495 hollow core, A:495 solid core, A:494–495 group-delay ripple, B:320 group-velocity dispersion, B:120, 495, 831 Groupe speciale mobile, B:756 grouped routing, B:851 GSM, see global system for mobile communications (GSM) guard time, B:656, 660, 670, 681, 775, 776, 781, 784, 786, 788 GVD, see group velocity dispersion (GVD)

H half mirror, B:112 half-rate decision, B:192 harmonic mixer, B:192 head-end switch, B:393 header error control, B:373 Hello message, B:376 Helmholtz equation, B:805, 806, 817, 838, 839 Hermite–Gaussian modes, B:838 Hermitian matrix, B:848 heterodyne efficiency, B:47 heterodyne receiver, B:13, 96, 100, 102, 105 see also homodyne receiver heterodyne signal, B:104, 746 heterojunction bipolar transistor, B:179 heterostructure photo-transistor, B:755

881 high speed optical modulators, recent developments EAMs, see traveling-wave EAMs external modulators, key performance data, A:186 high-speed modulation, see high-speed modulation, optical modulators intensity modulators based on absorption changes, A:195–197 Electroabsorption (EA), A:195 FK effect, A:195 IS Stark-shift modulator, A:197 Kramers–Kro¨nig relation, A:185 modulators based on phase changes and interference, A:193–195 LiNbO3 technology, A:195 linear Pockels effect, A:193 Mach–Zehnder configuration, A:194–195 novel types of modulators, see modulators, novel types on-off keying (OOK), A:184 principles and mechanisms of external optical modulation, A:185–186 quantum-confined Stark effects (QCSE), A:185 high-Q nanocavities, A:463–466 high-bit-rate digital subscriber line, B:443 high-capacity packet switches, B:696, 697 high-definition television, B:406, 413, 457, 461–463, 551, 552 high-fidelity entanglement with cooled fiber measurements of ratio between coincidence counts and accidentalcoincidence counts, A:861 photon-counting modules, A:860 two-photon interference, A:862 violation of Bell’s inequality, A:863 high-index-contrast waveguide, A:386 types and performance on SOI, A:384–388 high-performance optical channels, B:244 high-power photodetectors charge-compensated uni-traveling carrier photodiodes, A:239 saturation currents/RF output powers, A:239

882 high-power photodetectors (continued) high-power PDA, A:240 Partially Depleted Absorber (PDA) photodiodes, A:240 high-power waveguide photodiodes, A:242–244 coplanar waveguide (CPW), A:243 modified uni-traveling carrier photodiodes, A:240–242 modified UTC (MUTC), A:240 saturation, impact of physical mechanisms, A:234 thermal considerations Au bonding, advantage, A:244 transimpedance amplifier (TIA), A:233 uni-traveling carrier photodiodes, A:235–238 advantages, compared with PIN structure, A:235 optical preamplifiers, advantage, A:236 pseudo-random bit sequence (PRBS) signal, A:237 space-charge effect, A:238 high-quality video, B:553, 554 high-speed direct modulation of strained QW lasers InAlGaAs and InGaAsP systems/ comparison with theory, experimental results, A:66–69 microwave modulation experiment, A:66 small-signal modulation response, theory, A:65–66 K factor, A:65 high-speed edge emitters, tunnel injection resonances, A:43–44 high-speed electronic implementation architectural considerations, A:697–699 broadbanding techniques common source amplifier, A:684 compensated degenerative amplifier, A:686 implementation of adaptive, A:681 principles of, A:682–683 RC broadbanding, A:683–687 series shunt-peaked broadbanding, A:694–697

Index series-peaked, A:692–693 shunt-peaked, A:687–692 small signal, high-frequency equivalent circuit of amplifier, A:684, 695 small signal, high-frequency model of amplifier, A:684, 692 use of CMOS technology, A:682 electronic compensation for 10-GB/S applications, A:699–707 data center applications, A:707 electronic equalization for 10GBASE-LRM, A:700–701 LAN: transmission over MMF, A:699–701 metro and long-haul transmission over SMF, A:701–706 storage area networks, A:706–707 prospects and trends for nextgeneration systems, A:707–708 high-speed low-chirp semiconductor lasers, A:53 gain and differential gain, A:54 quantum dot lasers, see quantum dots (QD), lasers QW lasers, DC properties of long-wavelength, see DC properties of long-wavelength, QW lasers strained QW lasers, high-speed direct modulation, see high-speed direct modulation of strained QW lasers high-speed modulation, optical modulators chirp issues, modulators sinusoidal modulation, A:191 issues/principles/limitations, A:187–190 factors, A:187 flip-chip mounting, A:188 optoelectronic integrated circuits (OEICs), A:187 ‘‘traveling-wave,’’ A:190 high-speed nanophotonics device structure, A:37 directly modulated QD lasers, A:36–37

Index eye pattern/bit-error rate measurements, QD laser module, A:37–38 nonreturn-to-zero (NRZ), A:37 saturation of radio frequency (RF) amplifier, A:38 temperature-dependent BER measurements, A:38 high-Speed Ethernet, B:391, 411 photodiode, B:185, 191, 192, 270 transmission experiments, B:213 Higher Speed Study Group, B:179, 390 higher-layer protocol identifiers, B:389 highly nonlinear fibers (HNLF), A:761 hinge model, A:615–616, 620, 631–632 histogram techniques, B:277 hitless Operation, B:306 hollow-core PBGF, A:579–585 definition, A:581 dispersion, A:578–579 Bessel function, zeroth-order, A:582 five transverse mode intensity profiles, A:584 modal properties, A:584–585 origin, A:580–581 homodyne detection, B:13, 48, 95, 97, 100, 133 homodyne phase-diversity receiver, B:105, 112, 113, 115, 117 homodyne receiver, B:96–98, 100, 102, 106, 111, 123, 124, 127 optical circuit, B:111 Hong-Ou-Mandel interference with fibergenerated indistinguishable photons copolarized identical-photon source, 50/50 Sagnac-loop, A:871 dual-frequency copolarized pump, preparation of, A:872 experimental parameters, A:873 experimental results, A:874 experimental setup, A:872 fiber-based source of photon pairs at telecom-band wavelength near 1550 nm, A:869 four-wave mixing, A:869 photons produced by QS source, A:871

883 QS source and use in HOM experiment, A:873 QS source by measuring CAR, features of, A:872–875 quantum interference at beam splitter, A:870 real-life imperfections, A:875 theoretical simulation results of HOM dip, A:875 hubbed rings, B:302 hybrid communication network, B:226 hybrid integration platform, B:787 Hybrid Optical Packet Infrastructure, B:617 hybrid transmission, B:169 hybrid-fiber coaxial, B:16, 404, 460

I IEEE Higher Speed Study Group, B:179 III–V photonic integrated circuits, A:343–345, 376–377 future of OEO networks enabled by III–V VLSI, A:372–373 on chip amplifiers offer additional bandwidth, A:373–374 mobile applications for optical communications, A:375–376 power consumption and thermal bottleneck challenge, A:374–375 manufacturing advances for III-V fabrication implying scalability defect density and functionality per chip, A:361–363 design for manufacturability, A:356–358 in-line testing, A:358–359 yield management methodologies, A:359–361 manufacturing advances for III–V fabrication implying scalability, A:355–356 network architecture impact of LSI PICs component consolidation advantages, A:366–367 data ingress/egress in digital ROADM, A:369–370

884 III–V photonic integrated circuits (continued) DON architecture, A:367–368 network management advantages, A:370–372 Q-improvement cost, A:369 photonic material integration methods, A:345–346 small-scale integration large-scale III–V photonic integration, A:351–355 multi-channel interference-based active devices, A:348–351 single-channel multi-component chips, A:347–348 III–V semiconductor arena Franz–Keldysh modulators, A:399 Quantum-Confined Stark Effect (QCSE) modulators, A:399 impairment mitigation, B:18, 75, 196 impairment of the transmitted signal, B:308 impairment-aware first-fit algorithm, B:240 impairment-aware routing, B:237 impairment-unaware algorithms, B:240 implementation of MLSE, challenges, A:700 in-band OSNR monitoring, A:800 In-Fiber devices, A:501–502 in-line testing, A:358–359 in-phase and quadrature, B:96 incumbent local exchange carriers, B:419, 448 indium phosphide, B:44, 186 modulators, B:45 semiconductors, B:6 individual bulk devices, B:253 information spectral density, B:604 InGaAs/GaAs semiconductor, A:474 ingression paths, B:794 injection-locked lasers, B:5 inner ring, B:538 InP PLCs, A:275 geometry as source of birefringence, A:275 InP-based transmitter devices data capacity for, A:365 number of functions per chip for, A:364

Index InP/InGaAsP quantum well structure, A:468 InP/InGaAsP/InGaAs SACM APDs, A:246 input–output coupling, A:388–389 gratings, A:389–392 tapers, A:389, 390 insertion loss, B:654 ‘‘instantaneous frequency’’ of electric field, A:444 instruction-level parallelism, B:768 integrated silicon photonics chips, examples, A:416 20 Gb/s dual XFP transceiver, A:416–417 40-Gb/s WDM transceiver, A:417 integrated driver and MZ modulator, A:418 receiver performance, A:419 Integrated yield management (IYM) triangle, A:359 Intel, B:769 intelligent optical switch, B:528 intensity eye diagrams, B:50, 53, 57, 60 see also eye diagram intensity modulation, B:30, 60, 95, 133, 137, 625, 663, 739, 745, 749, 752, 834, 837 direct detection, B:95, 739 intensity-dependent refractive index coefficient, B:273 inter-nodal management, B:243 inter-symbol interference, B:31, 135, 349, 565, 582, 843 interchannel effects, B:166 interconnects, B:720, 721, 732 interframe gaps, B:373 Interior Gateway Protocols, B:532 intermediate frequency, B:48, 95, 743 intermodulation distortion (IMD), A:166, 168, B:426 international telecommunication union, B:347, 569 Internet Engineering Task Force, B:532 Internet Group Management Protocol, B:553 Internet lingo peer-to-peer, B:555

Index Internet Protocol, B:345, 346, 448, 480, 516, 695, 701 television, B:15, 19, 295, 345, 389, 391, 441, 448, 449, 457, 461, 462, 464, 529, 535, 553, 555 Internet service provider, B:389, 515, 516, 518, 523, 526, 528, 551–555 interoperability, B:620 interstage, B:585, 787 intersymbol interference (ISI), A:322 effects, A:672 intra-fiber device (cutting/joining), A:499–502 cleaving/splicing, A:500 in-fiber devices, A:501–502 mode transformers, A:500–501 intrachannel dispersion slope, B:581–583 intrachannel effects, B:167, 579, 580, 597 intrachannel four-wave mixing, B:40, 597, 600, 601 see also four-wave mixing intrachannel nonlinearities, B:41, 71, 73, 75 intradyne detection, B:63, 64 intradyne receiver, B:102, 164–166 inverse dispersion fiber, B:125 ISO standards organization, B:513 ITRS Semiconductor Roadmap, B:706

J J K WSS, A:317 Japanese Ministry of International Trade, B:458 JET signaling, B:646–648, 679 jitter, B:143, 204, 206, 218, 237, 283, 533, 554, 651, 653 Jones matrix, B:159, 160 Jones matrix formalism, A:541 Jones vector, A:654, 658 Joule heating, B:755 just-enough-time, B:645 just-in-time, B:645, 646

K Karhunen–LoSˇve Technique, B:847 ‘‘KEISOKU’’, A:82 Kerr effect, B:12, 24, 25, 70, 566, 567, 572

885 Kerr nonlinear material, A:438–439 transmitted vs input power, A:440 Kerr nonlinearities, A:499, 778, 789, 790, 794 Kerr-related nonlinear effects, PCF, A:504–507 correlated photon pairs, A:505–506 parametric amplifiers/oscillators, A:505 soliton self-frequency shift cancellation, A:507 supercontinuum generation, A:504–505 killer application, B:461, 551

L label switched paths, B:365, 541 labor-intensive changes, B:430 Laguerre–Gaussian modes, B:838 see also Hermite–Gaussian modes LAN, see local area networks (LAN) LAN/MAN Standard Committee, B:346 lanthanum-doped lead zirconium titanate, B:655 Large Channel Count ROADMs, A:316–317 large effective area core (LMA), A:543, 558, 562, 570–572 Large Hadron Collider, B:614 large-scale photonic integrated circuits (LS-PICs), A:351–352 100 Gb/s DWDM transmitter module, A:366 challenges to production, A:356 design, A:357 future, A:373 manufacturing, A:359–360, 362 network architecture impact of component consolidation advantages, A:366–367 data ingress/egress in digital ROADM, A:369–370 DON architecture, A:367–368 network management advantages, A:370–372 Q-improvement cost, A:369 output power distributions on 40-channel, A:364

886 large-scale photonic integrated circuits (LS-PICs) (continued) process capability (Cp)/capability index (Cpk), A:356 response vs frequency curves, A:353 upper control limit (UCL)/lower control limit (LCL), design, A:357 wafer processing, A:358 wafers yields vs chip size for time periods of production, A:362 and yield improvement, A:359–361 see also Integrated yield management (IYM) triangle laser chips, B:422 laser frequency offset, B:142, 143, 146 laser optical fiber interface (LOFI), A:57 laser rangefinders, B:423 latching/nonlatching, B:653 latency, B:21, 310, 374, 504, 533, 549, 550, 554, 646, 648–650, 664, 766–769, 772, 773, 775, 780–785, 787–789, 791, 792, 795, 796 lattice constants, A:461 Laudable characteristics, B:244 Layer, 2 functions, B:16, 358 Layer, 3 intelligence, B:507 LCOS, see liquid crystal-on-silicon (LCOS) LCOS-based 1 9 WSS, A:725 LE see linear equalizer legacy transport networks, B:371 LFF, see low-fill factor (LFF) light emitter, A:468 substance, A:469 light transmission , input port to output port, A:314 light-emitting diode, B:811 Lightguide Cross Connect, B:520 lightpath labeling, B:309 lightwave systems, historic evolution of, B:26 line coding, B:23, 24, 28, 38, 39, 53, 72, 348, 349, 355, 370, 373 line defect, A:457, 458 line-coded modulation, B:53 line/point-defect systems, A:457, 458 linear equalization, B:582

Index linear equalizer, A:673–675 designing transversal filter adaptation algorithm for tap weights, A:675, 676–678 analog vs digital implementation, A:675 number of taps, A:674–675 removal of ISI, linear filter, A:673, 676 structure of, A:674 linear optical quantum computing (LOQC), A:863 linear refractive index of silica, B:273 linearly polarized electromagnetic wave, A:443 linewidth enhancement factor (LEF), A:54 link aggregation group, B:374 link capacity adjustment scheme, B:347 link state advertisements, B:540 link utilization, B:782 liquid crystal-on-silicon (LCOS), A:723 features of, A:723–724 lithium niobate, B:5, 44, 187, 206, 726, 748, 751 LMA, see large effective area core (LMA) loading coils, B:10 Local Access and Transport Area, B:519 local area networks (LAN), A:668, B:236, 345 local fault, B:357 ‘‘local field’’ model, A:247 local loop unbundling, B:453 local oscillator power, B:95 localization of fiber failure, B:250 LOFI, see laser optical fiber interface (LOFI) logically inverted data pattern, B:61 long-distance monopolies, B:452 optical communications, B:565 long-haul applications, B:51, 166 long-wavelength VCSELS, A:83–87 loop timing, B:385 Lorentzian distribution, B:143 loss of signal, B:357, 533 low-density parity check, B:353 low-dispersion, B:2

Index low-fill factor (LFF), A:132 low-pass filtering effects, B:190 lower sideband, B:265 Lucent’s optical system layout for OXC, A:718

M M M star coupler, ideal, A:297 Mach–Zehnder delay interferometer (MZDI), A:276, 277, 382, 401, 402 demodulators, A:278 interleavers, A:279 plot of optical power from, A:295 possible ways of laying out the MZI–AWG, A:295 switches, usage of SAG, A:348 Mach-Zehnder characteristic, B:194 delay line interferometer, B:265 demodulator, B:599 electro-optic amplitude, B:598 electro-optic modulator, B:594 intensity, B:663 interferometer, B:193, 206, 209, 256, 598, 602, 634, 750 modulator, A:405, B:42, 43, 126, 134, 184, 206, 594, 758, 807 based on CMOS gate configuration, A:406 lumped-element, A:408 traveling-wave, A:408 push–pull modulators, B:96 Management Information Base, B:386 Manakov equation, A:656 Manakov-PMD equation, A:655–656 MAP, see Maximum A Posteriori Equalization (MAP) Marcum Q-function, B:140, 142 master-oscillator power-amplifier (MOPA), A:112 master-planned communities, B:441 matched filter, B:30, 31 matrix coefficients, B:162 matrix inversion, B:160 Maxichip, A:133

887 maximal-ratio combining, B:105, 109 Maximum A Posteriori Equalization (MAP), A:677–678 with BCJR algorithm, A:677–678 to improve system performance with FEC, A:678 maximum likelihood estimation, B:97 sequence estimation, B:23, 150, 182 maximum likelihood sequence estimation (MLSE), A:676–677 whitened matched filter (WMF), transversal filter, A:676–677 Maximum Refractive Index, A:488–489 maximum transport unit, B:682 Maxwell distribution, A:613, 614–615 Maxwell’s equations, B:804, 805, 809, 816, 820, 822 McIntyre’s local-field avalanche theory, A:247 MCVD, see modified chemical vapor deposition (MCVD) MDSMT, see modified dead-space multiplication theory (MDSMT) Mean time between failure, B:534 Mean time to repair, B:535 measured BER vs average optical power, A:700 media-independent interface, B:347, 348 medium access control, B:346 memory elements, B:20, 766, 767, 772 memory-less modulation, B:37 MEMS technologies and applications, emerging MEMS-tunable microdisk/microring resonators, A:747–748 lightconnect’s diffractive MEMS VOA, A:746 photonic crystals with MEMS actuators, A:748–749 MEMS, see Microelectromechanical system (MEMS) metalorganic chemical vapor deposition (MOCVD), A:27, 29, 35, 82, 83, 87, 88 metalorganic vapor phase epitaxy (MOVPE), A:346

888 metro and long-haul transmission over SMF, A:701–707 e-field domain signal processing, A:705–706 electronic compensation of PMD, A:704–705 electronic equalization to compensate chromatic dispersion, A:703–704 modeling of transmission channel and impairments, A:701–703 Metro Ethernet Forum, B:347 services, B:481 metro layer networks, B:518 metro networks, B:17, 196, 307, 484, 487, 495, 496, 498, 502, 505, 514, 524, 548 architectures, evolution of, B:482 geography of, B:505 modulation formats for, B:493 metro nodes, B:496 metro optical transport convergence, B:484 metro segment, B:552 metropolitan area networks, B:346, 483, 511 micro-ring and micro-disk resonators, A:349 fabricated eight-channel active, A:350 vertically coupled to I/O bus lines, A:350 microelectro mechanical systems, B:10, 249, 312 microelectromechanical system (MEMS), A:83, 89, 99 emerging MEMS technologies and applications MEMS-tunable microdisk/microring resonators, A:747–748 photonic crystals with MEMS actuators, A:748–749 optical switches and crossconnects 3D MEMS switches, A:716–719 two-dimensional MEMS switches, A:714–716 other optical MEMS devices data modulators, A:743–745 variable attenuators, A:745–747 tunable lasers, A:741–743 wavelength-selective mems components diffractive spectrometers and spectral synthesis, A:739–741

Index dispersion compensators, A:733–734 reconfigurable optical add–drop multiplexers, A:721–722 spectral equalizers, A:719–721 spectral intensity filters, A:730–733 transform spectrometers, A:734–738 wavelength-selective crossconnects, A:728–730 wavelength-selective switches, A:722–728 microoptoelectromechanical systems (MOEMS), A:713 microstructured optical fibers (MOF) effective-index microstructured optical fibers (EI-MOF), A:575–579 attenuation, A:579 basic concepts, A:575–576 dispersion, A:578–579 endless single mode/large effective area/coupling, A:576–577 small modal area/nonlinearity, A:577–578 hollow-core PBGF, A:579–585 waveguiding, A:573–575 microwave monolithic integrated circuit, B:745 microwave OE oscillators, A:420–421 spectrum analyzer, A:421 Miller multiplication of transistor, A:685 misalignment, B:837 mitigation, B:70 MLSE, see multiple likelihood sequence estimation (MLSE) MMF, see multimode fiber (MMF) MMI, see multimode interference (MMI) mobile applications for optical communications, A:375–376 MOCVD, see metalorganic chemical vapor deposition (MOCVD) modal transmission line theory, B:820 mode division multiplexing, B:36 mode transformers, A:500–501 mode-locked fiber lasers, B:204 mode-locked laser diodes, B:204 mode-locked pulse train, frequency spectrum of, A:321 mode-locked QD lasers

Index device structure, A:39 hybrid mode-locking, A:41–42 passive mode-locking, A:39–40 full width at half maximum (FWHM), A:39 modes see sub-pulses modified chemical vapor deposition (MCVD), A:491, 539, 562–564 modified dead-space multiplication theory (MDSMT), A:254 modified uni-traveling carrier photodiodes (MUTC), A:240 modular deployment, B:306 modulation format, B:12, 13, 18, 24, 27, 28–30, 34, 36, 37, 41, 42, 46, 49, 50, 52, 55–58, 61, 62, 64, 65, 67–76, 97, 100, 108, 132–135, 161, 169, 171, 181, 183, 184, 194, 195, 202, 203, 215–217, 222, 223, 226, 237–239, 256, 257, 259, 260, 264, 272, 309, 321, 390, 413, 415, 493–495, 569, 572, 583, 593, 596–600, 604–606, 660, 663, 834–837 auxiliary polarization, B:40 binary, B:28, 37, 63, 70, 97, 185 coded, B:28, 39, 41, 53 duobinary, B:41, 45, 53, 495 response, B:748, 750 tone techniques, B:253 waveforms, B:28–30, 33, 36, 46, 59 with memory, B:37, 38 modulator devices, A:404–409 reducing modulators size, A:407 variable attenuator waveguide, A:404 modulator, B:134, 183, 206, 258, 662 bias, B:44, 56, 190 extinction ratios, B:67 modulators, novel types Intersubband Electroabsorption Modulators, A:208–211 effects, A:209 negative chirp/high-optical transmission, A:209 Silicon-Based Modulators, A:211–213 complementary metal–oxide– semiconductor (CMOS), A:212 G–L /Ge intervalley scattering time, A:213

889 MOEMS, see microoptoelectromechanical systems (MOEMS) molecular beam epitaxy (MBE), A:37, 44, 82, 87, 110, 252 moment generating function, B:849 monitoring physical-layer impairments, B:245 monolithic integration, B:684 Monte Carlo methods, B:841, 842 Moore’s law, B:735, 765 MOPA, see master-oscillator poweramplifier (MOPA) Mosaic browser, B:404 MPEG-2 and, B:4 compression, B:462 multi-Gbaud rates, B:64 multi-mode fiber model, B:757, 839 multi-mode interference devices, B:807 multi-mode Systems, B:837 multi-processor chips, B:20 multi-protocol label switching, B:365, 513, 643 multi-section directional coupler, A:274 multi-service platform, B:528 multichannel entertainment digital video, B:535 multicore architectures, emergence of, B:797 multifamily dwelling units, B:464 multifiber wavelength-selective switch, B:296 multilayer network stack, B:537 multilevel coding, B:40, 123 multilevel decision circuitry, B:149 multilevel modulation, B:13, 24, 37, 38, 40, 78, 96, 97, 226 multimode fiber (MMF), A:672 multimode fiber-coupled 9xx nm pump lasers active-cooled pump diode packages beam symmetrization coupling optics, A:132–133 Maxichip, A:133 ‘‘wallplug efficiency,’’ A:133 BA pump diode laser broad-area single-emitter (BASE), A:127 frequency stabilization, A:129 slow-axis BPP,causes for NA degradation, A:127

Index

890 multimode fiber-coupled 9xx nm pump lasers (continued) classification of 9xx-nm mm pump lasers heat transport classification, A:125–126 optical classification, A:126–127 combined power, A:133 passive-cooled 9xx-nm MM pumps beam symmetrization coupling optics, A:131–132 direct coupling, A:129–131 low-fill factor (LFF), A:132 simple lens coupling, A:131 Single Emitter Array Laser (SEAL), A:132, 133 pulse operation, A:133 multimode interference (MMI), A:112, 232, 271, 393, 730 multiparty gaming, B:480 multipath interference, B:322 multiple access interference, B:633 multiple input multiple-output, B:36, 161 multiple likelihood sequence estimation (MLSE), A:626 multiple service operators, B:378, 413 multiple-quantum-well, B:749 multiple-wavelength transmission schemes, B:767 multiplexing, B:20, 26, 28, 29, 31–33, 36, 47, 49, 53, 64, 78, 119, 131, 132, 180, 183, 185, 186, 193, 195, 201, 203, 204, 206, 207, 208, 215, 217, 218, 222, 223, 226, 238, 242, 247, 251, 253, 294, 296, 297, 304, 311–313, 320, 322, 346, 347, 353, 368, 372, 415, 417, 444, 478, 484, 485, 490, 491, 498–500, 514, 517, 522, 526, 527, 604, 617, 626, 629, 635, 642, 644, 656, 661, 665, 669, 695, 803 see also demultiplexing multiplication regions, low-excess noise APDs AlInAs, A:250–252 active region, definition, A:251 heterojunction, A:252–256 electrons gain energy, A:254 ‘‘initial-energy effect,’’ A:254 MDSMT, A:254

thin, A:248–250 ‘‘dead space,’’ A:248 noise reduction in thin APDs, A:250 ‘‘quasi-ballistic,’’ A:249 multipoint-to-multipoint, B:359, 378, 380, 481 multiport switch, B:500, 501 multisource agreement, B:346 multistage topologies, B:773, 782, 788 multisymbol phase estimation, B:49, 132 MUX see multiplexing

N nanocavity with various lattice point shifts, A:465 Napster, B:468 narrowband digital cross-connect system, B:522 National LambdaRail, B:617 national research and educational networks, B:611 National Television System Committee, B:411 Negative Core-Cladding Index Difference all-solid structures, A:493 higher refractive index glass, A:492–493 hollow-core silica/air, A:492 low leakage guidance, A:493 surface states, core-cladding boundary, A:493 network architectures, B:297, 410, 482 automation, B:503 component failures, B:522, 537 diameter, B:553, 767, 769, 789 dimensioning, B:850 element, B:293 evolution, B:511, 549 fault management, B:244 management, B:2, 234, 239, 305, 310, 324, 336, 339, 366, 440, 445, 483, 556, 766 modeling, B:850 network interface, B:620 -on-chip, B:797 optimization problem, B:547

Index optimization process, B:547 premise equipment, B:528 restoration, B:537 scaling, B:308 segments, B:513 survivability, B:487 switching, B:623 NOBEL project, B:615 node object, B:853 node-sequenced channel, B:339 node-to-node demand connection, B:489 noise spectral density, B:137 non-adiabatic parabolic horn, A:288 non-Gaussian beat noise, B:61 non-return-to-zero, B:35, 50–56, 58–63, 69–72, 76, 108, 123, 135, 168–170, 184–187, 189, 190, 194, 195, 200, 203, 256, 258, 261, 265, 267, 269–272, 282, 322, 493, 495, 594, 596, 599, 834–837, 839 on-off keying, B:256, 493 phase-shift keying, B:108 non-zero dispersion-shifted fiber, B:195, 213, 255 nonlinear distortion, B:24, 60, 76, 77, 121 nonlinear effect, B:2, 10, 64, 73, 137, 159, 162, 166, 167, 181, 235, 241, 243, 260, 268, 272–274, 276, 317, 495, 566, 567, 572, 573, 577, 580, 586, 593, 594, 596, 597, 600–602, 605, 625, 743, 745, 755, 759, 832 nonlinear fibers, A:546–555, 765 designs, A:548–555 core refractive index change /core radius, A:549 W-profile, A:551–555 requirements, A:547–548 effective interaction length, A:547 figure of merit, A:547–548 nonlinear efficiency, A:547–548 nonlinear impairments, B:24, 170, 578, 589, 590, 597 nonlinear optical loop mirror, B:209, 222 nonlinear optical materials, A:760, 766 bismuth glass, A:760, 765, 789 chalcogenide glasses (ChG), A:762–764 silicon, A:763–764, 794

891 nonlinear optical methods of generating tunable optical delays and buffering, A:811–818 Nonlinear optical signal processing, A:760 Nonlinear optics in communications future prospects, A:818–820 optical delays and buffers, A:811–818 optical performance monitoring, A:800–811 optical regeneration, A:769–771 optical signal regeneration through phase sensitive techniques, A:784–787 3R regeneration, A:782–784 2R regeneration in integrated waveguides, A:778–782 2R regeneration through SPM in chalcogenide fiber, A:771–777 optical switching, A:796–800 optical-phase conjugation, A:787–789 parametric amplification, A:766–769 phase-matched vs nonphase-matched processes, A:761–762 platforms, A:762–766 wavelength conversion chalcogenide fiber-based wavelength conversion, A:789–793 wavelength conversion in chalcogenide waveguides, A:793–796 nonlinear phase fluctuation, B:127, 276 noise compensation, B:157, 162, 274 noise, B:60, 72, 73, 131, 151, 154, 157, 162, 164, 167, 274, 276, 601 rotation, B:274 nonlinear polarization scattering, B:137, 259 nonlinear refractive index ((Kerr effect), A:760 nonlinear signal-noise interactions, B:72 nonlinear taper, B:808 nonlinear transmission, B:41, 52, 54–58, 60, 63, 64, 71, 79, 135, 154, 155, 274 nonlinearity compensation, B:79 nonperiodic problems, B:819 nonreturn-to-zero (NRZ), A:37, 42, 45, 95, 198, 204, 237, 324

892 nonstatic nature of optical networks, B:234 nonzero dispersion, B:2, 41, 72, 255, 576, 836 nonzero dispersionshifted fibers (NZDSF), A:525–526, 533–534 normalization constant, B:149 Nyquist limit, B:97, 104, 114 NZDSF, see nonzero dispersionshifted fibers (NZDSF)

O OEICs, see optoelectronic integrated circuits (OEICs) OMM’s 16 16 switch, A:716 on-off keying (OOK), A:184 B:13, 30, 123, 131, 203, 211, 214–218, 256, 493, 594, 599, 601, 663 OOK, see on-off keying (OOK) open shortest path first, B:532 open systems interconnection, B:513 operation, administration, and maintenance, B:347 operational costs, B:482–485, 506, 507 operational savings, B:420 OPM, see optical performance monitoring (OPM) optical add and drop multiplexers, B:15, 253, 295, 487, 498, 850 optical amplifier, B:3, 25, 26, 30, 30, 65, 73, 74, 96, 138, 170, 182, 183, 209, 211, 217, 241, 242, 247, 248, 260, 274, 294, 310, 314, 479, 484, 495, 496, 505, 533, 565–269, 580, 625, 654, 739, 742, 743, 745, 752, 753, 757, 760 optical bandpass filter, B:54, 219, 223, 257 optical buffers, B:665 optical burst switching, B:170, 556, 615, 627, 649, 650, 655, 679, 701 optical channel configuration, B:101 monitor, B:248, 249 switching, B:642 optical chromatic dispersion compensators, A:325–335 tunable ODC, A:325 optical clock recovery, B:216, 219

Index optical code division multiple access, B:36, 617, 625, 632, 633 optical communication systems, fibers speciality in dispersion compensation, A:526–538 double-clad fibers for fiber lasers, A:555–572 nonlinear fibers, A:546–555 single polarization fibers, A:539–546 optical communication technologies, B:26 optical coupler, A:271 optical crossconnect (OXC), A:716, B:615, 642, 645, 647, 850 optical data modulator, B:185 optical delay interferometer, B:133 optical delays and buffers, A:811–818 counterphased pump modulation, impairment, A:816 dual-pump parametric delay, A:816 effect of pump synchronization, A:816 nonlinear transmission, A:813 one-pump parametric delay, A:816 phase dithering in one-pump device, A:816 system performance of tunable dual-pump parametric delay, A:818 transmission spectrum of FBG, A:812–813 tunable delay with power or strain, A:815 two-pump counterphased delay line, A:817 Optical demodulators, A:278 optical demultiplexer, B:123, 203 optical dispersion compensation, B:71 optical duobinary, B:53, 54, 62, 139 optical equalizers (OEQ), A:322–325, B:181, 182, 190, 193–195 ineffectiveness, A:324 symmetric and asymmetric ISI impulse responses, A:324 two-tap, A:323–324 band-limited photodiode and its equalization, A:324 107-Gb/s NRZ bandwidth before and after equalization, A:324 using serial arrangement of MZIs, A:328 using finite-impulse response (FIR) filters, A:325

Index optical fibers polarization effects in, A:539–542 polarization-maintaining, A:542–544 single-polarization, A:544–546 optical field vectors, B:29 optical field, change in, A:295 optical filters, A:275, B:30, 31, 46, 52, 54, 62, 68, 69, 97, 119, 123, 131, 133, 139, 143, 166, 195, 211, 223, 235, 245, 256, 265, 267, 268, 284, 295, 320, 479, 496, 499, 506, 605, 661, 776, 830 designs example, A:294 Fourier filter-based interleaver, A:294 nonlinear phase response, A:276 represented in frequency domain by Fourier transform, A:292 in time domain by impulse response, A:292 zero-loss maxima per FSR multiple, A:293, 294 with one, A:293, 294 see also flat-top filter construction theorem optical frequency, B:268, 272, 712, 713 optical gates, B:209 optical header, B:630, 644, 657, 658, 660–665, 678, 684 processing technologies, B:658 swapping, B:678 technique, B:663 optical hybrid, B:101, 103, 144 optical impairments, B:239, 244 optical index, B:570, 805 Optical injection locking (OIL), A:145 optical injection phase-locked loop (OIPLL), A:179, B:757 optical intensity, B:30, 34, 46, 50, 51, 54, 59, 60, 135, 273, 274, 664 profile, A:288 optical interconnection networks, B:766, 771, 777 Optical Internet Forum, B:347 Optical Internetworking Forum, B:620 Optical isolators, A:350 TE mode waveguide, A:351

893 optical label encoding, B:31 switching, B:19, 643, 678, 682 optical line terminal, B:251, 379 optical local oscillator, B:133 optical memory, B:637 optical MEMS devices, other data modulators, A:743–745 etalon modulator, A:744 etalon variable attenuator, A:744 microdisk resonator, A:747 optical MEMS, developments and fabrication technologies of, A:730–732, 748–749 Optical Mesh Service, B:549 optical metropolitan networks, evolution of, B:477 optical microelectromechanical systems (MEMS), A:713 applications of, A:713 developments of, A:714 see also microoptoelectromechanical systems (MOEMS) optical modulation amplitude, photodetector capacitance, A:410–411 optical modulation formats, B:12, 23, 26, 27, 39, 42, 45, 49, 132, 134 overview of, B:133 optical modulator, B:217, 309, 757 optical monitoring, B:623, 636 optical multiplexing, B:14, 36, 201 optical network unit (ONU), A:172, 174, 176, B:250, 379, 460, 464, 633 optical networking, B:617 optical noise, B:31, 137, 138, 148, 164, 166, 245, 253, 254, 309, 330, 593 optical nonlinearity, A:760 optical packet switching, B:170, 615, 617, 623, 624, 627–633, 635–638, 642, 644, 646, 649–651, 654–658, 660, 662–665, 667, 669–671, 675–677, 679, 684, 696, 697, 699, 701, 712, 715, 720, 725, 733, 775, 776 optical packet synchronizer, B:665 optical parametric amplification (OPA), A:761, B:260

894 optical passbands, A:287 optical performance monitoring (OPM), A:760, 800–811 advantages and disadvantages of phase-matched and nonphase-matched processes, A:800, 804, 811 configuration of, A:800 DGD monitor, A:808 measuring in-band OSNR, A:804 nonlinear optical loop mirror (NOLM), A:806 nonlinear power transfer curve of OPA-based in-band OSNR monitor, A:805 optical power spectrum of signal measured on OSA, A:810 optical spectrum analyzer methods, A:803 optical spectrum at output of OPA, A:805 OSNR monitoring curves of NOLM-based device, A:807 for OPA-based in-band OSNR monitor, A:805 principle of device operation, A:801 principle of residual dispersion monitoring, A:800, 801 provides signal quality and identify network fault, A:800 spectral monitoring signal vs 40-Gb/s RZ receiver BER penalty, A:802 optical performance monitoring, B:15, 233 optical phase conjugation, role in signal processing, A:787 optical phase, B:11, 26, 43–46, 57, 59, 72, 97, 100, 111, 150, 151, 154, 170, 207, 208, 210, 211, 316, 746, 825 conjugation, B:11, 26 diversity homodyne receiver, B:111 lock loop, B:97, 100, 150, 746 optical power loss, B:246 optical preamplification, B:105, 113, 114, 753 optical protection mechanisms, B:568 optical pulses, B:50, 56, 181, 204, 205, 208, 214, 572 compressor, B:215

Index optical regeneration, A:769–771 ASE noise, signal impairment, A:769 experimental configuration, A:772 limiting signal degradation, using 2R and 3R regenerators, A:769–770 optical signal regeneration through phase sensitive techniques generic 2R, higher order 2R processing, excessive idler broadening, A:786 power transfer functions, A:785 two-pump PA operating in small-signal, large-signal and transfer characteristics, A:786 optical-phase conjugation, A:787–789 all-optical networking layer, A:787 feasibility of two-pump FPP in all-optical networks, A:788 principle of noise compression, A:770 3R regeneration, A:782–784 BER improvement, optimizing parameters, A:784 capability to retime signal timing jitter, A:782 operating principle of, A:782–783 power transfer function, A:784 structural design and principle of, A:782–783 2R regeneration in integrated waveguides, A:778–782 autocorrelation of input and output pulses, A:781 capable of operating T/s bit rates, A:782 power transfer function, resulting in OSNR reduction and BER improvement, A:782 structure of, A:776 testing integrated waveguide 2R regenerator, A:780 transmission spectrum, A:779 unfiltered pulse spectra, A:780 2R regeneration through SPM in chalcogenide fiber, A:771–777 better performance, A:779 calculated Q-factor, A:776 input and output eye diagrams, A:776 optimized Q-factor improvement vs FOM, A:777

Index output pulse spectra of the chG, A:773 power transfer curves, A:771, 773–774 principle of operation, A:771, 772 wavelength conversion advantage of, A:789 chalcogenide fiber-based wavelength conversion, A:789–793 chalcogenide waveguides, A:793–796 optical regenerators, B:204, 520, 721, 723 optical resonators, B:717 optical return loss, B:246 optical ring resonator, B:717 optical routers, B:683 optical sampling technique, B:205 optical sensors, A:514 optical signal processing based on VCSEL technologies, A:94–97 optical signal to noise ratio, B:42, 65, 131, 181, 234, 319, 566, 572, 587, 594, 600, 636, 653, 796, 831 optical signal, representation of, B:837 processing, B:4, 11, 201, 202, 204, 208, 214, 215, 217, 225, 697 optical signal-to-noise ratio (OSNR) penalty, A:623, 703–705 optical spectral filters, B:498 optical spectrum analyzer (OSA), A:150 optical splitter splits, A:271 1 2 couplers, A:271–272 2 2 couplers, A:272–275 optical subcarrier monitoring, B:255 optical supervisory channel, B:242 optical switch fabrics, B:724 optical switches and crossconnects 2D MEMS switch, A:714–716 optical switching, A:796–800, B:19, 209, 239, 322, 506, 507, 620, 636, 637, 643–645, 651, 654, 655, 665, 669, 679, 684, 725, 731, 772, 773, 778, 789 devices, B:772 OC-768 Switching in HNLF parametric device, A:800 signal band of pump operating in normal and anomalous dispersion, A:797 two-pump parametric switching, A:798 optical terminal multiplexers, B:850

895 optical time division multiplexing, B:14, 32, 56, 119, 120, 180, 191, 193–195, 201–209, 211, 213, 215–219, 221–223, 225–227, 229, 231, 249, 617, 664, 705 transmitter, B:14, 195, 201–204, 222, 225 optical time-domain reflectometry (OTDR), conventional, A:176 optical transmission link capacitybandwidth product, growth in, A:760 optical transmitter, B:421 optical transponder, B:132, 520, 533, 536 optical transport network, B:483, 487 evolution of, B:132 optical wavelength domain reflectometry, B:250 optical-electrical-optical conversion, B:202, 615 Optical-electrical-optical (OEO) consumption vs 2.5G (Sonet/SDH), A:371 optical-optical-optical, B:615 optical-to-electrical conversion, B:245, 695, 731 optically transparent mesh transmission, B:334, 336 optimal receiver sensitivity, B:139, 145 optimal signal-to-noise ratio, B:42, 52, 55, 56, 59, 63–70, 72–76, 125, 126, 131, 134, 136–138, 142, 143, 146, 148, 154, 155, 166, 168, 169, 181, 190, 235, 239, 243, 245, 248, 249, 253–260, 263, 266, 274, 281–284, 319, 321–324, 335, 336, 390, 494, 495, 566, 582, 587, 600, 603, 653, 831, 836 monitoring, B:253, 255, 257–260, 266 penalty, B:68–70, 76, 142, 143, 146, 166, 181, 582, 603 performance, B:63, 169 optimization algorithms, B:505 optoelectronic bandwidth, B:188 optoelectronic integrated circuits (OEICs), A:187 optoelectronic oscillator, A:420 optoelectronics, B:426, 440, 461, 491, 501 base rate receiver, B:211 buffer, B:678 components, B:26, 37, 188, 734 conversion stages, B:569

896 optoelectronics (continued) demodulation, B:46 integrated circuits, B:754 integration, B:64 switch fabrics, B:697, 733 switching, B:653 Optoplex’s high-isolation, B:249 orthogonal codes, B:36 orthogonal frequency division multiplexing, B:32 Orthogonal modulation header technique, B:663 orthogonal polarization, B:30–33, 46, 47, 157, 159, 161, 183, 254, 259, 262, 263 orthogonal transverse modes, B:838 oscillator laser, B:164, 745 voltage-controlled, B:192, 193, 210 oscilloscope digital sampling, B:196 digital storage, B:170 electrical sampling, B:191, 205 Osmosis, B:789, 790, 791 outage map, A:628 outer ring, B:538, 539 Outside plant cabinet technology, B:424 outside vapor deposition (OVD), A:539, 552, 562–565 OVD, see outside vapor deposition (OVD) overlay switching, B:454 OXC, see optical crossconnect (OXC)

P P–N Junction in depletion, A:399–402 in forward bias, A:402–403 p-type doping, B:4 P2P (E-Line), B:378 packaged photodiodes, B:421 packaged switch, A:716 packet buffering, B:643 packet dropping, B:664 packet loss, B:378, 379, 533, 536, 554, 667, 669 packet switch architectures, B:697 packet switching nodes, B:21, 774 packet synchronization processing, B:657

Index packet transmission, protection scheme in, B:853 parabolic horn, width, A:288–289 parabolic refractive index profile, B:838 Parabolic waveguide horns, A:288 straight multimode waveguide, A:291 paradigm changes in semiconductor physics and technology, A:27–28 parallel loss elements, B:253 parallelism and synchronization, B:853 parametric amplification (PA), A:760, 766–769 amplified spontaneous emission, A:767 effect of pump spectral tuning, A:767–768 modulational instability, A:767 operating principle, A:767 parametric gain, A:768 parametric fluorescence of FWM, A:833 parametric gain, A:766, 768 parametric processor, functions of, A:760, 762 parasitic effects, B:830 partial response, B:53 Partially Depleted Absorber (PDA) photodiodes, A:240 partially depleted-absorber photodiode, B:755 passband ripple, B:320 passive device modeling methods, B:805 passive optical network, B:246, 379, 624, 703 passive optics, B:787 passive waveguide devices/resonators filters, A:393–395 highly compact splitters, A:393 photonic crystal resonators, A:395 ring resonators, A:395–397 splitters and couplers, A:393 passive/active components for POFs, A:599–601 path adjustments, B:784 iterations, B:785, 786, 788 path-protected ring, B:852 path-routing hand offs, B:236 pattern-dependent performance, B:188 PBG, see photonic band gap (PBG)

Index PBGF, see photonic band-gap fibers (PBGFs) PCF, see Photonic crystal fibers (PCF) PDCD, see polarization-dependent chromatic dispersion (PDCD) peer-to-peer, B:345, 479, 480, 555 per-channel dispersion compensation, B:318 management, B:317, 319 per-channel equalization, B:311, 312 per-channel symbol rate, B:23, 24, 71 perfectly matched layer, B:813 performance analysis and metrics, B:781 estimation, B:841 evaluation, B:785 functions, B:535 monitoring techniques, B:243 periodic dispersion maps, B:317 permanent virtual circuit, B:523 Petri Nets, B:852, 853 PF polymer POFS, connectors for, A:599–600 PHASARs (phased arrays), see arrayed waveguide grating (AWG) phase demodulator, B:224 phase difference, B:44, 48, 112, 116, 133, 139, 163, 164, 192, 206, 210–212, 264, 272, 599 phase distortions, B:59 phase error, B:142, 143, 146, 153, 163 phase estimation, B:49, 102, 110, 113–117, 132, 158, 162–165 algorithms, B:162 phase modulation, B:34, 38, 42, 44, 45, 47, 49, 58, 59, 60, 100, 109, 110, 118, 134–136, 164, 184, 204, 205, 214, 223, 276, 572, 596, 597, 744–746, 759, 831 phase noise, B:76, 100, 102, 107, 110, 111, 115, 118, 127, 162, 164, 167, 206, 274, 748, 841 tolerance, B:115 phase portrait, B:282, 283 phase shaped binary transmission, B:41, 53, 190 phase wrapping, B:163 phase-diversity homodyne receiver, B:97, 101, 102

897 phase-encoded data, B:163 phase-locked loop, B:97, 192, 210 phase-matched vs nonphase-matched processes, A:757–759 phase-modulated data signal, B:211 phase-modulated formats, B:60, 64 characteristic of, B:60 phase-sensitive detection, B:267 phase-shift keyed (PSK), A:278 phase-shift keying, B:34, 96, 108, 123, 133, 161, 269, 593, 601 formats, B:599 phased amplitude-shift signaling, B:53 phasor diagrams, B:167, 274 ‘‘phonon bottleneck,’’ A:25 photo sharing, B:480 photocurrents, B:104, 108, 112, 113 photodetector capacitance, A:410 optical modulation amplitude as function of, A:411 photodetector, B:5, 7, 46, 47, 54, 206, 211, 212, 219, 253, 264, 268, 349, 636, 745, 753–755, 758, 760, 787, 826, 830 photodetectors, advances in avalanche, see avalanche photodiodes high-power, see high-power photodetectors waveguide, see waveguide photodiodes photodiode, B:5, 6, 46, 95, 97, 99, 100, 105, 112, 113, 138, 180, 184, 191–193, 205, 210, 259, 270, 325, 351, 421, 422, 561, 565, 566, 582, 598, 741–746, 754–756, 848 responsivity, B:742, 746, 848 photon density, B:825 electron resonance, B:748 lifetime, B:4, 825 photonic band gap (PBG), A:485, 490 photonic band-gap fibers (PBGFs), A:573, 575, 579–581 Photonic crystal add–drop filter, A:395 photonic crystal cladding characteristics, A:487–490 maximum refractive index, A:488–489 photonic band gaps, A:490 transverse effective wavelength, A:489

898 photonic crystal fibers (PCF), A:525–526, 761 applications, A:502–515 Bragg gratings, A:512–513 Brillouin scattering, A:507–509 fiber lasers/amplifiers, A:503 gas-based nonlinear optics, A:509–511 high power/energy transmission, A:502–503 Kerr-related nonlinear effects, A:504–507 laser tweezers in hollow-core PCF, A:513–514 optical sensors, A:514 telecommunications, A:511–512 design approach, A:487 fabrication techniques, A:485–487 guidance, characteristics, A:490–498 attenuation mechanisms, A:495–498 birefringence, A:494 classification, A:490 Group Velocity Dispersion, A:494–495 Kerr Nonlinearities, A:499 Negative Core-Cladding Index Difference, A:492–493 Positive Core-Cladding Index Difference, A:490–492 intra-fiber devices (cutting/joining ), A:499–502 cleaving/splicing, A:500 in-fiber devices, A:501–502 mode transformers, A:500–501 photonic crystal cladding, characteristics, A:487–490 maximum refractive index, A:488–489 photonic band gaps, A:490 transverse effective wavelength, A:489 photonic crystal resonators, A:395, 440, 441 photonic crystal theory: temporal coupledmode formalism, A:433–434, 451 stopping light in dynamic photonic crystals, A:445 conditions for stopping light, A:447–448

Index future prospects, A:450–451 numerical demonstration in photonic crystal, A:448–450 from tunable bandwidth filter to light-stopping system, A:447–448 tunable Fano resonance, A:445–447 tuning spectrum of light, A:443–444 temporal coupled-mode theory for optical resonators, A:432–438 to predict optical switching, A:438–442 photonic crystal, A:456–457, 461, 478 light-stopping process in, A:449 numerical demonstration in, A:448–450 point and line defect states in, A:432 switches, A:438 Photonic double heterostructure, A:466 photonic heterostructures, A:456 photonic integrated circuit (PIC) transmitter (TX), A:347 photonic integrated circuit (PIC), A:347, 354, 365 future, A:373 100-Gb/s transmitter and receiver, A:352 Q measurement vs wavelength, A:355 ultralong-haul transmission, A:354–355 using hybrid Raman/EDFA amplifier, A:355 see also planar lightwave circuits (PLCs) photonic material integration methods butt-joint regrowth, A:345–346, 348 direct-contact wafer bonding, A:346 quantum well intermixing (QWI), A:346, 348 selective area growth (SAG), A:346, 348 photonic, B:6–9, 20, 21, 202, 226, 296, 310, 339, 483, 507, 532, 615, 617, 632, 634, 636, 637, 668, 706, 713, 755, 766, 772, 774, 776–778, 781, 782, 797, 798, 807, 811, 814, 816, 817, 820 band, B:8, 811, 816, 817, 820 bottleneck, B:705, 706 cross-connect, B:296, 304, 311, 312, 532

Index crystal, B:7–9, 636, 637, 668, 713, 811, 814, 817 integrated circuit, B:6, 7, 778, 807 per bit, B:137 photoreceiver, B:46, 191, 192, 839 physical coding sublayer, B:348 physical layer, B:346, 513 development, B:348 fault management, B:243 measurement parameters, B:245 optical encryption, B:170 physical medium attachment, B:348 dependent, B:348 PIC-based all-optical circuits, A:761, 777 pilot-tone-based monitoring technique, B:256 pin semiconductor structures, B:43 planar holographic bragg reflectors, A:311–312 4-Channel CWDM multiplexer using, A:312 Planar lightwave circuits (PLCs), A:269, B:6, 139, 181, 206, 207, 217, 223, 253, 296, 655 and birefringence, A:275 PDW shift, A:275 parameter ‘‘delta,’’ A:270 polarization conversion, A:276 surface and oscillating electric field, A:275 wavelength shift and refractive index, A:305 planar lightwave circuits in fiber-optic communications, A:269–270, 336 basic waveguide theory and materials index contrast, A:270 optical couplers/splitters, A:271–275 polarization effects, A:275–277 inter-signal control devices, A:312 large channel count ROADMs, A:316–317 reconfigurable optical add/drop multiplexers, A:313–316 wavelength-selective switches, A:317–318 intra-signal control devices, A:318 optical chromatic dispersion compensators, A:325–335

899 optical equalizers, A:322–325 temporal pulse waveform shapers, A:319–322 passive optical filtering, demodulating, and demultiplexing devices, A:277 arrayed waveguide gratings, A:279–311 Mach–Zehnder interferometers, A:277–279 planar holographic bragg reflectors, A:311–312 plane wave expansion, B:816 plasmonic VCSELS, A:91–93 plastic optical fiber (POF), A:672, 707 datacom applications, A:602 development, A:595–598 in attenuation of, A:595–596 in high-speed transmission, A:597–598 enabling 10-Gb/s transmission, A:672 passive/active components, A:599–601 connectors/termination for PF Polymer, A:601 PMMA, connectors/termination methods, A:599–600 transceivers for, A:601 varieties, A:598–599 product-specifications of, A:598 typical construction of PMMA, A:598–599 plastic optical fiber, B:9 platform developments, A:762–766 advances in, A:765–766 chalcogenide glasses, nonlinear material for all-optical devices, A:762 parameters for silicon and ChG with nonlinear FOM, A:760 properties of chalcogenide, A:762–763 properties of SM selenide-based chG and bismuth oxide, A:764 PLC, see silicon planar lightwave circuits (PLCs) pluggable transmitters, next-generation, B:493 PMD emulator (PMDE), A:642, 644–644 PMD source (PMDS), A:642 cautions about, A:650

900 PMD Taylor expansion, A:632–633 validity of, A:633–634 PMD, see polarization mode dispersion (PMD) PMF, see polarization maintaining fibers (PMF) PMMA, see polymethylmethacrylate (PMMA) PMMA-Based POFs, connectors for, A:599–600 POF, see Plastic optical fiber (POF) point defect, A:457, 458 quality factors of, A:458–459 Point of Interface, B:526 point-of-presence, B:519 point-to-multipoint, B:348, 379, 512 point-to-point connections, B:512, 553, 641 emulation, B:382, 384 Ethernet development, B:16, 347 links, B:325, 569, 767, 771, 804 transmission, B:204, 225, 316 WDM systems, B:524, 532 Poisson, B:534, 805, 826 polarization alignment, B:108, 114, 152 polarization beam splitter, B:46, 103, 112, 157, 183, 223, 269 polarization control, B:9, 47, 125, 182–184, 582, 746 polarization controller, B:125, 182–184, 582 polarization conversion, A:276 polarization crosstalk, A:276 polarization demultiplexer, B:165, 183, 184, 224 polarization demultiplexing, B:137, 157, 165, 166, 222 polarization dependence, B:102, 111, 389, 651, 757 polarization diversity, B:33, 47, 102, 104, 105, 111, 112, 114, 151, 152, 155, 162, 746 polarization effects, optical fibers, A:539–542 birefringence B, A:540 Jones matrix formalism, A:541 in telecommunication, A:541

Index polarization entangled photon-pairs with compact counterpropagating scheme (CPS), creating, A:859 Ekert’s QKD protocol, implementing, A:863 polarization filtering, B:139–142, 145, 146 polarization fluctuation, B:114 polarization hole burning, B:263 polarization independent gate, B:210 polarization interleaving, B:32 polarization maintaining fibers (PMF), A:523, 525, 542–544 polarization management, B:33 polarization matching, B:746 polarization mode dispersion (PMD), A:531, 537–538, 605–606, 609–610 birefringence, A:607–608 emulation, A:641–642 emulators, A:645–647, 651 emulators with variable DGD sections, A:650–652 first-order emulation and its limitations, A:642–645 sources, A:647–650 environmental fluctuations, A:617 fiber models, A:612 fiber plant, elementary model of installed, A:614–616 field tests, survey of, A:616–621 high-order effects experimental and numerical studies, A:639–641 PMD Taylor expansion, A:632–637 practical consideration, A:637–639 notation convention, A:606 and optical nonlinearities, A:652 coupled nonlinear Schrodinger equations, A:652–655 effect of nonlinear polarization rotation on PMDCs, A:659–661 interactions between third-order nonlinearities and PMD, A:662 Manakov-PMD equation, A:655–656 nonlinear polarization rotation, A:656–659 polarization evolution and, A:609

Index propagation, A:606–607 properties of, in fibers, A:611–614 transmission impairments caused by the first-order, A:621 first-order PMD penalties from real receivers, A:624–627 PMD-induced outages, implication of the hinge model on, A:631–632 PMD-induced outages, probability of, A:627–631 Poole’s approximation, A:622–625 polarization mode dispersion, B:9, 10, 15, 24, 37, 63–65, 68, 70, 71, 77, 79, 123, 131, 133, 134, 137, 150, 152, 155, 159–162, 166–170, 180–182, 184, 196, 202, 203, 212, 214, 215, 219, 222, 224–226, 234, 235, 239, 243, 245, 255–257, 259, 262–266, 268–272, 279, 281–284, 309, 321, 346, 348–351, 356, 357, 383, 385, 390, 570–573, 582, 583, 593, 603, 654, 831, 841 compensation of, B:214 effects of, B:160 mitigation, B:167, 168, 219, 268 monitoring technique, B:270 polarization mode dispersion (PMD), transmission impairments, A:800, 808 polarization modes, A:544–546 polarization modulator, B:602, 603 polarization multiplexing, B:24, 33, 47, 63, 64, 78, 79, 123, 137, 183, 184, 195, 196, 226 polarization nulling method, B:254, 255 polarization orientation, B:165 polarization parameters, B:109 polarization related impairment, B:263 polarization scramblers, B:168, 182, 583 polarization sensitivity, B:114 polarization shift keying, B:30, 133 polarization-dependent chromatic dispersion (PDCD), A:641, 643 polarization-dependent loss (PDL), A:354 polarization-dependent wavelength (PDW), A:275 shift resulting in PMD, A:276 polarization-dependent wavelength shift (PDWS), A:354

901 polarization-dependent frequency, B:62, 154 gain, B:263 loss, B:160, 184, 255, 308, 321, 654 polarization-division multiplexing, B:131 Polarization-entangled degenerate photonpair generation in optical fiber CAR as function of single-channel photon-detection rate, A:866 measuring, A:866 plotted as function of single counts/ pulse, A:865 correlated/entangled photon pairs in single spatiotemporal mode, A:869 degenerate correlated-photon, A:865 determining single-count rate for optimum CAR value, A:866 experimental layout, A:864 polarization-independent behavior for single counts and coincidence counts, A:864 properties of photon pairs, A:867 two-photon interference (TPI) with visibility, A:867 Polarizationindependent (PI) SOAs, A:373 polymer modulators, B:751 polymethylmethacrylate (PMMA), A:595 POF Cords, A:598–599 Poole’s approximation, A:622–624 Poole’s formalism, A:639 Positive Core-Cladding Index Difference, A:490–492 multiple cores, fibers, A:492 ultralarge area single mode, A:491 potential dispersion monitoring, B:283 power amplifiers, B:329, 330 power consumption and thermal bottleneck challenge, A:374–375 power coupling, B:326, 328, 330, 336–339 power detection, B:268 power dissipation, B:798 power fading effect, B:264 power penalties, B:118, 121, 234, 672, 740 power splitters, B:305, 314, 316, 393 power splitting ratio, A:622

Index

902 power transients, B:247 predicted end of life, B:323 predistortion algorithms, B:77 prehistoric era, A:24 single QD, d-function, A:24 ‘‘phonon bottleneck,’’ A:25 threshold current density of laser, A:24 zero dimensional structures, A:24 ‘‘principal state transmission,’’ B:214, 222 Principal States of Polarization (PSP), A:607–608 Principle of dispersion compensating fibers, A:527–531 dispersion compensation, A:528–531 fiber dispersion, A:527–528 printed circuit boards, B:355 Private Branch Exchange, B:518 probability density functions, B:847 Process capability (Cp), A:356 Process monitor (PM) wafers, A:358 product sensitivity analysis (PSA), A:361 propagation delay, B:757, 780, 857, 858 propagation solution technique, B:809 protection switching, B:374 protocol data units, B:359 Protocol independent multicast, B:553 provider backbone bridge, B:369 pseudo-level modulation, B:40 pseudo-linear transmission regime, B:213 pseudo-multilevel format, B:40, 55, 57 pseudo-multilevel modulation, B:40, 41 pseudo-random binary sequence, B:59, 113, 116, 167, 207 pseudo-wire setup, B:550 Public-Switched Telephone Network, B:526 pull-tapes, B:423 pulse amplitude modulation, B:133, 353 pulse carver, B:56–59, 62, 180, 264, 599 pulse carving schemes, B:599 Pulse code modulation, B:522 pulse compression unit, B:223 pulse generation, B:135, 216, 274 pulse pattern generator, B:113 pulse position modulation, B:32, 663 pulse sources, B:204 pulse-to-pulse overlap, B:579, 580, 601 Pump diode lasers

high-radiance diode laser technologies, A:133–134 multimode fiber-coupled 9xx nm, see multimode fiber-coupled 9xx nm pump lasers 1480-nm pumps and 14xx-nm high-power, A:122–124 optical power supply, A:107–108 power photonics, A:108 ‘‘all fiber’’ telecom technology, A:108 DPSSLs, A:108 VCSELs, A:108 single-mode fiber 980-nm pumps, see single-mode fiber 980-nm pumps status/trends/opportunities ‘‘loss budget,’’ A:135–136 status, Raman amplification, A:135 trends and opportunities, A:136 telecom optical amplifiers, A:108 VCSEL pump and high-power diode lasers, A:134–135 pump-diode breaks, B:585 pump-to-pump interactions, B:591 push-pull configurations, B:635 modulator, B:96, 113 operation, B:44

Q Q-factor, B:243, 245, 248, 250, 263, 278, 281, 600–605, 653, 841–847 Q-tag, B:366, 369, 370 Q/Bit error rate monitoring techniques, B:278 Q factor, A:458–459 of ideal LE, A:673–674 of point-defect cavity, A:459, 464 Q-improvement cost, A:369 for optical and electronic performance, A:370 QCSE, see quantum-Confined Stark Effect (QCSE) quadratic impact, B:260 quadrature amplitude modulation, B:34, 110, 150, 392, 416 quadrature phase shift keying, B:38, 62, 96, 203, 583

Index Quadrature point, B:51 quadrature signal space, B:50, 78 quality of service (QoS), 18, 237, 238, 248, 253, 275, 346, 417, 482, 484, 485, 523, 524, 531, 533, 535–537, 545–547, 549, 615, 627, 628, 638, 645, 648, 657, 679, 852 quantum dots (QD) amplifier modules device structure, A:42 eye pattern and BER measurements, A:42–43 error-free modulation, A:42–43 external cavity laser (ECL), A:42–43 quantum dots (QD), A:23, 47, B:3, 4, 637 lasers optical gain and LEF p-doped and tunneling injection QDs, A:70–74 p-type doping and tunneling injection, A:74–76 limitations, A:75 utilization of, A:35–36 quantum efficiency, B:99, 105, 726, 754, 755 quantum entanglement, A:830–831, 851 incompatibility with realism, A:831 measurements of correlated quantities, A:832 measuring states of particles, A:830 quantum information processing (QIP), A:838 generation of entangled states and Bell’s inequalities, A:838 quantum measurement, process of, A:829 quantum mechanics (QM), A:829 quantum theory of four-wave mixing in optical fiber, A:838–851 applications of QIP, A:838 coincidence counting formula, A:848–849 coincidence-counting rate with Gaussian filters, calculation of, A:847–848 coincidence-counting result, A:839 coincidence-photon counting, A:839 coupled classical-wave equations for pump, A:840 dependence of photon-counting results, A:847, 848–851

903 dependence on ratio of pump bandwidth to filter bandwidth, A:851 experiment vs theory, A:850 Hamiltonian in quantum FWM process, A:841 interaction Hamiltonian, A:840–841 measurement techniques, coincidencephoton counting, A:839 negative effect of SRS, A:838 polarization entanglement, A:839 process of FWM, A:838 quantum efficiencies of detection in signal and idler channels, A:839 single-photon counting rate, formula, A:844, 846–848 state vector evolution, calculating, A:841–844 theoretical model, A:839–840 two-photon state, calculated by first-order perturbation theory, A:843 two-photon states, A:844–845, 850–851 Quantum well intermixing (QWI), A:346, 348 quantum wells (QWs), A:110 Quantum-Confined Stark Effect (QCSE), A:185, 196, 208, 210, 212 modulators, A:399, 400 quantum-confined Stark effect, B:749 quantum-mechanical, B:11 quasi-linear transmission, B:71

R radio frequency communication, B:23, 79 radio frequency tone, measurement of, B:263, 268 Raman gain coefficient, A:761, 763, 764, 792 Raman amplifiers, B:4, 325, 332, 333, 496, 580, 586–593 couplers, B:222 effects, B:592, 832 gain, B:11, 587, 588, 591, 592 interactions, B:333, 592, 593

904 Raman (continued) preamplifier, B:589 pump modules, B:221 scattering, B:273, 325, 330, 389, 586–588, 831 tilt errors, B:333 random access memories, B:772 random nonlinear phase shift, B:274 rapid spanning tree protocol, B:364 rate-equation approach, B:825, 826 Rayleigh backscattering, B:589 scattering, B:246 real-time optical Fourier transformation, B:272 Real-Time Protocol, B:516 Real-valued baseband, B:52 Rearrangeably nonblocking topologies, B:782 received signal r(t), A:669 receiver (RX) PICs, A:351, 366 determining PDL and PDWS, A:354 with integrated PI-SOA, A:374 key DC performance characteristics, A:354 receiver impairments, B:142, 146 receiver sensitivity enhancement, B:153 penalty, B:38, 148–150 sensitivity, B:113, 139, 145, 153 Receiver side equalization, B:495 receiver technology advances, B:421 Receiver-based electronic dispersion compensation (EDC), A:701 recirculating loop buffer, B: reconfigurable optical add–drop multiplexer (ROADM), A:313–316, 721–722 athermal 16-ch ROADM, A:315 40-channel reflective, A:316 64-channel reflective, A:317 designs, A:316–317 modular 80-channel, A:318 transmission spectra of athermal, A:315 waveguide of 16-channel optical, A:313

Index reconfigurable optical add/drop multiplexer, B:6, 11, 15, 24, 65, 68, 132, 168–170, 196, 293–297, 299, 300–317, 319–325, 327, 329, 331, 333–339, 341, 343, 499, 500, 502, 520, 521, 524, 526, 528, 529, 532, 533, 545, 548, 622, 623, 636, 642, 642 cascade penalties, B:319 layer, B:521, 522, 526, 545, 548 network layers, B:521 technologies, B:293, 543 transmission system design, B:316 Reed–Solomon codes, B:180 refractive index, A:460–461, B:9, 11, 24, 273, 274, 633, 712, 713, 715, 749, 806, 807, 818, 819, 838 of defect, A:466 regenerators, B:720 Regional Bell Operating Companies, B:468 relative group delay, measurement of, B:266 relative sensitivity penalty, B:145 remote antenna units, B:756 remote fault, B:357, 394 remote mammography, B:614 remote node, B:379 residual dispersion, B:125, 213, 222, 317, 571, 580, 581 resilient packet ring, B:484 resonance spectra, tuning of, A:466 resonant grating filter, B:824 resonant mode, amplitude of, A:435 resonant ring modulator, A:408 resonator buffers, B:717 restoration, B:537, 538, 542, 852, 856 Restore Management, B:856 return-to-zero, B:40, 55, 56, 58, 108, 135, 180, 203, 208, 493, 593, 594, 596, 635 dispersion maps, B:319 format, B:572 modulation, B:50, 55 phase-shift keying, B:108 quadrature phase shift keying, B:606 ridge waveguide, A:385 ring hopping, B:525 ring resonators, A:326, 395–397

Index configured add–drop filter, A:396 ring add–drop filters, A:396–397 ring-based network, B:489 ring-to-ring interconnections, B:489 Ritz method, B:812, 813 ROADM, see reconfigurable optical add–drop multiplexer (ROADM) round trip latency, B:787, 788 propagation time, B:779 time, B:77, 381 router common cards, B:536 routers, B:649, 683, 701, 707 routing latency, B:788 power, B:300–305 properties, B:300 technologies, B:19 row decoder, B:718 rule of thumb, B:71, 181, 582, 602, 669

S S-tag, B:369 Sagnac interferometer, B:209 sampled symbol values, B:35 sampling of the signal, B:108 sampling theorem, A:298–299 coefficient, A:299 SAN, see storage area networks (SAN) satellite carriers, B:553 satellite-based technologies, B:466 SBS, see stimulated Brillouin Scattering (SBS) scalability, B:369, 653, 675 scalability: chip size vs cost, A:363–365 Scalable Photonic Integrated Network, B:778 scalar electric field, B:806 scattering losses, A:388 scattering, B:567, 586–588, 783, 813 SEAL, see Single Emitter Array Laser (SEAL) second-order PMD (SOPMD), A:633, 640, 646 selective area growth (SAG), A:346, 348 self-coherent

905 modulation formats, B:132, 134, 169, 171 optical transport systems, B:171 receivers, B:157, 166 signals, B:131, 133, 137 Self-homodyne detection, B:133 self-managed network, B:235 self-phase modulation (SPM), A:657, 756 B:123, 136, 273, 317, 566, 567, 572, 637, 831 semianalytical techniques, B:847 semiconductor optical amplifier (SOA), A:348 and isolator, A:351 semiconductor quantum dots, future applications devices collective blindness, A:25–26 electronic/optical properties, A:28–31 nanophotonics, see high-speed nanophotonics prehistoric era, see prehistoric era QD amplifier modules, see quantum dots (QD) amplifier modules QD Lasers, see Mode-Locked QD lasers semiconductor devices, B:5, 44, 215 lasers, B:748–750, 752 optical amplifier, B:43, 170, 248, 388, 496, 635, 654, 715, 774, 826 sensitivity penalties, B:148 separate absorption and multiplication (SAM) APD structures, A:245 serial filters, B:253 serial-to-parallel demultiplexer, B:718 series-peaked amplifier, A:692 server consolidation, B:481 service level agreement, B:18, 347, 378, 536 Session Initiation Protocol, B:518 severe electrical low-pass filtering, B:53 SG-DBR laser, wavelength-tunable, A:348 integration with EAM and SOA, A:348 shared restoration, B:300, 542 shared-risk-link group, B:521, 537 SHEDS (super high-efficiency diode source) program, A:108, 114, 124 Shortest Path First Fit, B:649 shunt peaking, A:688

906 shunt-peaked amplifier utilizing common base cascode, A:689 shunt-peaked common source amplifier, A:688 Si-CMOS fabrication processes, B:188 side-coupled resonator configuration, A:432, 433, 437–438 nonlinear, A:440 sideband instability, B:831 signal amplitude, B:102, 134, 148, 150, 152, 159, 167 signal bandwidth, B:24, 37, 52, 58, 63, 247, 320, 322, 746, 753, 758 signal bit rate, B:96, 282 signal conditioner, B:356 signal distortions, B:26, 31, 41, 42, 54, 70, 75, 76, 79 signal launch power, B:26, 60, 64, 73, 76 signal photocurrent, B:98, 105 signal processing applications Bragg grating filters in chG, importance of, A:763 signal quality assessment, B:243 signal space, B:28–31, 34, 35, 37, 39, 50 signal-dependent noise, B:38 Signal-noise interaction, B:72 nonlinearities, B:79 signal-signal interactions, B:72 signal-spontaneous beat noise, B:753 signal-to-noise ratio, B:12, 15, 26, 26, 65, 67, 96, 104, 105, 113, 115, 116, 131, 137, 143, 236, 248, 254, 255, 278, 308, 309, 313, 392, 415, 498, 572, 580, 587, 593, 651, 653, 719, 728, 731, 743–746, 749, 753, 754, 758, 759, 774, 804 silica planar lightwave circuits, B:253 silica PLCs, A:275 strain as source of birefringence, A:275 silicon electronics, B:188 silicon germanium bipolar, B:166, 182, 186, 187 silicon microelectronics industry, R&D investments, A:382 silicon nanowires development of high quality fabrication processes, A:763

Index enhance nonlinear optical efficiencies, A:761 as material platform, A:763–764 mode field area reduction, A:761 silicon photonic waveguides, nonlinear effects, A:421–423 Raman amplification, A:423 optically pumped CW silicon laser, A:422 two-photon absorption, A:421 silicon photonics, A:381–382, B:20, 188, 636 achieving electrically pumped laser, approaches, A:423 epitaxial approach, A:424–425 extrinsic gain, A:424 hybrid wafer-bonded III–V epitaxial structures, A:423–424 active modulation silicon photonics carrier accumulation on gate oxides, A:403–404 modulation effects, A:397–399 modulator devices, A:404–409 P–N junction in depletion, A:399–402 P–N junction in forward bias, A:402–403 CMOS integration and examples of integrated silicon photonics chips, A:416–420 microwave OE oscillators, A:420–421 microwave spectrum analyzer, A:421 SOI transistor description, A:414–416 germanium photodetectors and photoreceivers, A:409–414 high-index-contrast waveguide types and performance on SOI, A:384–388 implemented on SOI, A:381–382 input–output coupling, A:388–389 gratings, A:389–392 tapers, A:389 integration, B:636 interleaver and demuxing performance, A:419

Index manufacture in CMOS process, A:417 nonlinear effects, A:421–423 passive waveguide devices and resonators filters, A:393–394 photonic crystal resonators, A:395 ring resonators, A:395–397 splitters and couplers, A:393 SOI wafer technology, A:383–384 toward silicon laser, A:423–425 trends and applications, A:425–426 waveguide, A:393 silicon planar lightwave circuits (PLCs), A:715 silicon processing technology, B:7 silicon strip, A:385 silicon–InP lasers and detectors, laser structure, A:349, 350 silicon-grating coupler, A:390 Silicon-On-Insulator (SOI), A:381 advantages, A:384 created by SMARTCUT technique, A:383 integration of photonics, A:414–415 loss mechanisms, A:387 requiring BOX, A:383–384 waveguide types, A:384 simplex/duplex PMMA POF cords, A:598–599 Single Emitter Array Laser (SEAL), A:132 single multiplexer circuit, B:187 single polarization fibers (SPF), A:544–546 single QDS, A:32–35 ARDA agency, A:32 ‘‘International Technology Road Map for Semiconductors,’’ A:33 nanoflash, obstacles, A:33 nanoflash memory cell based on QDs, future, A:34 purcell effect, A:32 ‘‘Quantum Information Science and Technology Road Map,’’ A:32 single-photon emitter (SPE), A:32 single sideband, B:52, 206, 265 measurement of, B:206 modulation, B:45, 52

907 single-channel fiber optic, B:13 optical systems, B:576 transmission, B:52, 73, 76, 125 single-ended detection, B:61, 72, 73 single-fiber capacity, B:295 single-mode fiber (SMF), A:37, 42, 82, 86, 161, 488, 490, 494–496, 498, 500–501, 506, 509, 672 single-mode fiber 980-nm pumps DWDM, A:109 failure rate, A:120–122 critical process parameters, A:120–122 materials for 980-nm pump diodes, A:109–110 InGaAs QW, A:109 low-loss large optical cavity (LOC), A:110, 114 optical beam, A:110–112 alpha-distributed feedback (DFB) lasers, A:112 broad-area (BA) diode lasers, A:112 mode filters, A:112 MOPA, A:112 narrow stripe technology, A:111–112 Fiber Bragg Grating (FBG), A:111–112 output power scaling amplified spontaneous emission (ASE), A:114 constant mirror reflectivity scaling, A:114 ‘‘constant power ratio scaling’’/ ‘‘constant photon lifetime scaling,’’ A:113 length scaling of laser diode, A:112–114 SHEDS (super high-efficiency diode source) program, A:114 vertical epitaxial structure, A:114–115 packaging, A:117 reliability catastrophic junction meltdown, A:118–119 catastrophic optic mirror damage (COMD), A:119 electron beam-induced current (EBIC) charge thermography, A:120

908 single-mode fiber 980-nm pumps (continued) failure modes, CMD/NRR/TAD, A:118 nonabsorbing mirror (NAM), A:119 NRR heating, A:119 TAD at facets, A:119–120 spectral stability, A:115–117 Fabry–Perot laser, A:116 single-mode fiber, B:49, 203, 255, 756 single-photon emitter (SPE), A:32 single-photon-counting avalanche photodetector (SPAD), A:256 single-polarization, B:9, 63, 64, 152, 157, 203 single-rack router, B:704 single-waveguide phase modulator, B:134, 136, 781 sinusoidal transmission, B:61 size, weight, and power (SWaP), A:375 slot efficiency, B:781, 785, 786, 788 slot waveguide, A:386 slow-light waveguide, B:712, 714 small form factor pluggable, B:346 small-scale integration, III-V photonic integrated circuit large-scale III–V photonic integration, A:351–355 multi-channel interference-based active devices, A:348 multi-channel ring resonator devices, A:349 novel devices and integration methods, A:349–351 single-channel multi-component chips, A:347–348 SMARTCUT technique (for SOI), A:383 processing sequence, A:384 SMF, see single-mode fiber (SMF) SOI CMOS process, A:409 SOI transistor, A:414–416 fabricated at Freescale Semiconductor, A:414 space division multiplexing, B:36, 789 space domain, B:650 space switching, B:670 SPAD, see single-photon-counting avalanche photodetector (SPAD)

Index spanning tree protocol, B:358, 362, 364, 374, 529, 543 spatial dispersion of focal position, A:281 of input-side position, A:282 spatial effects, B:837 specialty fibers active fibers, A:525 coupling fibers or bridge fibers, A:525 dispersion compensation fibers, A:525 high numerical aperture (NA) fiber, A:525 highly nonlinear optical fiber, A:525 photonic crystal fibers (PCFs), A:525 photo-sensitive fibers, A:525 polarization control fibers, A:525 spectral compactness, B:63 spectral density, B:65, 67, 137, 604 spectral efficiency, B:12, 13, 24, 97, 123, 124, 127, 131, 133, 136, 137, 146, 150, 155, 169, 183, 184, 195, 208, 217, 218, 226, 391 spectral intensity filters 4 4 WSXC, A:729 advantages and challenges, A:732 tunable Fabry–Perot (F–P), A:732 spectrally narrow formats, B:69 spectrum monitoring, B:268, 272, 593 speed digital signal processing, B:28, 97 SPF, see single polarization fibers (SPF) splicing attributes, B:428 splicing loss, A:538 split-step method, B:832–834 splitter size reduction, B:425 splitters and couplers, A:393 splitting of packets, B:699 splitting ratios, B:388 SPM, see self-phase modulation (SPM) spontaneous emission factor, B:138, 753 spontaneous-spontaneous beat noise, B:753 spoofing, B:486 spun fiber, A:613 spur-free dynamic range (SFDR), A:146, 152, 166, 168, 169, 178, 179 spurious free dynamic ranges, B:758 square-law detection, B:58, 95, 96, 150 square-shaped optical pulse, A:320–321 waveforms and frequency spectra, A:320

Index squelch tables, B:540 SRS, see stimulated Raman scattering (SRS) standard estimation theory, B:843 standard single-mode fiber, B:26, 169, 565 standardized terrestrial optical communications, B:28 star coupler, B:144, 808 star coupler, second transmissivity, A:300 star-shaped hub-and-spoke architecture, B:348 Stark effect, B:5, 749 start frame delimiter, B:359 state diagrams, B:39 state of polarization (SOP), A:607 steady-state channel power control, B:324 stimulated Brillouin Scattering (SBS), A:530, 548, 555, 565–568, B:273, 389, 426, 745, 831 stimulated Raman scattering (SRS), A:499, 510 Stochastic Differential Equation techniques, A:613 storage area networks (SAN), A:672, 706–707, B:196 straight-line phase modulator, B:59, 60 strick-sense nonblocking, B:651 strictly nonblocking topologies, B:782 sub-pulses, B:570 subcarrier multiplexing header technique, B:661 submarine applications, B:585 repeaters, B:568 Subscriber Loop Carrier, B:410 subtending rings, B:526 subwavelength granularity, B:622, 624, 642, 651, 684 superprism dispersion, B:820 surface emitting lasers, recent advances long-wavelength VCSELS, A:83–87 optical signal processing based on VCSEL technologies, A:94–97 optical Kerr effect, A:95 plasmonic vcsels, A:91–93 VCSEL-based slow light devices, A:97–99 surface-emitting laser (VCSEL), A:597

909 swept acousto-optic tunable filters, B:253 switch controller, B:630, 657 switch fabrics, B:724 technologies, B:730 switching nodes, B:21, 773–776, 778–780, 782, 783, 787, 792, 793, 796 speed, B:310 technologies, B:625, 651 voltage, B:44, 45, 206, 727 synchronization, B:58, 184, 187, 359, 385, 416, 643, 657, 660, 665–668, 678, 697, 757, 781, 788, 853 inaccuracies, B:781 synchronized digital hierarchy, B:32, 350, 850 Synchronizer architectures, B:667 synchronous optical network (SONET), B:32, 296, 306, 308, 310, 350–352, 355–357, 365, 370, 372, 374, 377, 378, 390, 392, 478, 482–484, 487–491, 493, 498, 502, 507, 513–518, 521–528, 530–532, 536, 538–540, 542–546, 548, 618, 620, 622, 623, 627, 637, 643, 646, 682, 850 cross-connect layers, B:532 interfaces, B:392, 524 rings, B:516, 518, 522–524, 526, 528, 531, 532, 542, 543, 545, 548 system penalties, B:68 systematic errors, B:335

T Taguchi method, A:357 tail-end switch, B:542 tandem AWG, A:308–309 10 GHz-spaced 1010-channel, A:308 demultiplexing properties of all the channels, A:309 tandem switch, B:526 tap coefficients, B:121, 122 delay, B:282 tapped delay line, A:322–323, 327 parallel arrangement, A:328–329 serial arrangement, A:327–328

910 Taylor series, A:632–637 TCP/IP, B:483, 617, 701, 707 Telco metro network, B:519, 548 telecom packaging technology, A:130 tell-and-wait protocols, B:646 temperature variations, B:70, 181 temporal coupled-mode theory for optical resonators, A:432–438 applied to photonic crystal structure, A:433 directly coupled resonator configuration, A:432, 437 Fano-interference configuration, A:433, 438 light from one port to other direct pathway, A:433 resonant pathway, A:433 perfect crystal with photonic band gap, A:431 physical implementation, A:440 side-coupled resonator configuration, A:432, 433, 437–438 nonlinear, A:440 using to predict optical switching, A:438–442 temporal pulse waveform shapers, A:319–322 experimental original spectra and auto-correlated pulse waveforms, A:321 square-shaped optical pulse, A:320–321 synthesized spectra and corresponding cross-correlated pulse waveforms, A:321 TeraFlops, B:769 ternary symbol alphabet, B:39 terrestrial cables, B:568, 569 terrestrial networks, B:295, 511, 564, 567–569, 581 terrestrial systems see undersea systems Thermo-optic device, dependence on temperature, A:398 thin-film filter technology, B:249, 421 components, B:421 third-order nonlinearities and PMD, A:662

Index three-dimensional photonic crystals, A:467 fabrication technology, A:467–468 light propagation, control of, A:472–473 spontaneous emission, control of, A:468–472 three-dimensional photonic crystals, B:8 signal space, B:30 three-modulator transmitter architecture, B:58 threshold crossing alert, B:533 TIA, see transimpedance amplifier (TIA) tilt error, B:333, 339 time delay units, B:654 time division multiplexing, B:17, 32, 179, 180, 201, 217, 347, 415, 426, 481, 484, 514, 656, 705 time domain header technique, B:660 time slot interchanger, B:667 time switch, B:665 Time-domain split-step, B:833 timing extraction devices, B:210 timing jitter, B:206, 245 TIR, see total internal reflection (TIR) titanium indiffused lithium niobate, B:748 tone-fading techniques, B:264 topology, B:21, 362–366, 376, 486, 487, 514, 521, 540, 541–543, 768, 769, 773, 777, 778, 782–785, 788, 790, 792–794, 796, 850 banyan, B:778, 782 Data Vortex, B:789, 792–796 deflection routing, B:792 hub-and-spoke, B:528 modifications, B:782 omega, B:778, 782–785 tree-like, B:543 total internal reflection (TIR), A:573 total internal reflection mirror, B:671 TPA, see two-photon absorption (TPA) traditional best-path algorithm, B:240 traditional first-fit algorithm, B:240 traffic congestion, B:537 traffic generating node separation, B:505 traffic generation modules, B:782

Index traffic matrix, B:524, 547, 555 training sequence, B:160 transceivers, POFs, A:601 IEC standards for, A:601 transform spectrometers advantages and challenges of MEMS, A:736 Fourier transform, A:735 grating, A:737 semitransparent detector, A:738 single-chip integrated, A:736 transient control, B:326, 331 transimpedance amplifier (TIA), A:233 transimpedance amplifier configuration, A:410 Transmission Control Protocol, B:669, 701 transmission distances, B:42, 43, 77, 213, 222, 234, 317, 325, 351–353, 388, 596 transmission experiments, B:123, 194, 215 transmission fiber, B:33, 102, 103, 213, 226, 274, 332, 333, 389, 495, 565, 570, 581, 582, 587–589, 592 transmission impairments, B:12, 13, 24, 38, 41, 63, 65, 67, 78, 98, 122, 123, 131–133, 152, 157, 166, 171, 221, 240, 325, 348, 565 compensation of, B:77, 78, 151, 157 impact of, B:571 origin of, B:122 transmission line methods, B:822 transmission of optical data signals, B:180 transmission penalty, B:127, 133, 320 transmission spectra TO switches are ‘‘off,’’ A:314 TO switches are ‘‘on,’’ A:315 transmission speed, B:625 transmission system model, B:742, 743 transmission technologies, B:507, 513, 771 transmitted signal, B:97, 99, 108, 122, 237, 308, 319, 542 transmitter (TX) PIC, A:351, 355 10-channel working at elevated temperature, A:375 channels operating at 40 Gb/s, A:363 key performance metrics, A:352

911 transmitter chirp, B:268, 277, 493 transmitter side modulation formats, B:495 transoceanic distances, B:564, 565, 568, 572, 573, 601 transmissions, B:572 transpacific distance, B:577, 601, 604 transparency, B:237, 621 transparent bridging, B:362, 367 transponder, B:37, 38, 64, 132, 299, 300, 302, 303, 308, 314, 315, 353, 355, 377, 378, 495, 564, 605 transport network, B:32, 131, 132, 167, 168, 182, 239, 370–372, 374, 377, 480, 483, 487, 511, 529, 553, 554, 615, 618, 622, 850 transversal filters, B:122, 214 transverse effective wavelength, A:489 transverse electric (TE) polarizations, birefringence, A:275 transverse magnetic (TM) polarizations, birefringence, A:275 transverse mode, A:473 traveling-wave EAMs future development, A:207–208 high-efficiency modulators for 100 Gb/s and beyond, A:205–207 ridge-waveguide eam structure, A:198–200 beam-propagation method (BPM), A:199 TWEAM structures for increased impedance, A:200–205 benzocyclobutene (BCB) layer, A:202 bit-error rate test set (BERTS), A:204 frequency limitations, A:201 LiNbO3-based intensity Mach–Zehnder interferometer (MZI) modulators, A:200 travelling-wave amplifier, B:191 tributary adapter module (TAM), A:367 tunable bandwidth filter, A:447 band diagrams calculation, A:448 photonic bands for structure, A:447 two-cavity structure, A:446–447 Tunable laser technologies, B:492

Index

912 tunable lasers commercial uses, A:743 external cavity semiconductor diode lasers, A:741, 742 Littrow and Littman configuration, A:741 tunable ODC (TODC), A:325 AWG-based polymer thermooptic lens, A:334 polymer thermooptic lens, measured response, A:335 silica thermooptic lens, A:333 chirped bragg grating, A:327 coherently coupled MZIs, A:330 figure of merit, A:334–335 finite impulse response (FIR), A:325, 327–328 tapped delay lines, see tapped delay line Gire–Tournois etalons, A:327 gratings, A:331–334 arrayed waveguide grating, A:331– 334 diffraction grating, A:331 virtually imaged phased array (VIPA), A:331, 332 infinite impulse response (IIR), A:325, 326–327 Bragg Gratings, A:327 Gires–Tournois etalons, all-pass filters, A:326 ring resonators, all-pass filters, A:326 ring resonators, A:326 3-Stage, A:330 4-Stage, A:330, 331 4-Stage version realized in silica PLC, A:330, 331 transversal filter in InP waveguides, A:329 in silica waveguides, A:328 types of, A:325 tuneable polarization controller, B:183 tuneable wavelength converter, B:630, 665, 670, 672, 701 Tuning and switching functions, B:316 tunneling injection, B:4

TWEAM (traveling-wave EAMs), A:198, 200–206, 209, 210 twisted pair, B:10, 345, 349, 410, 413, 415, 460 two to four fiber pairs, B:564 two-dimensional photonic crystal slabs, A:456–457 heterostructures, effect of introducing, A:459–463 high-Q nanocavities, A:463–466 line/point-defect systems, control of light by combined, A:457–459 nonlinear and/or active functionality, introduction of, A:466–467 two-photon absorption (TPA), A:760, B:11, 268 TX optical sub-assemblies (TOSAs), A:347 typical fiber performance, A:555

U Ultra-compact 4-channel arrayedwaveguide grating filter, A:394 ultra-high-performance systems, B:2 Ultra-long-haul distance, B:601 transmission, B:181, 221ultra-low latency propagation, B:780 ultra-low-polarization-dependent fibers, B:2 ultrafast nonlinear interferometer, B:676 ultrahigh-speed laser modulation by injection locking applications analog optical communications, A:177–179 global system for mobile communications (GSM), A:179 metropolitan area networks, A:177 optical time-domain reflectometry (OTDR), conventional, A:176 passive optical networks, A:171–176 Rayleigh backscattering effects, A:175 wireless local area networks (WLANs), A:179

Index ultrahigh-speed laser modulation by injection locking (continued) Fabry–Perot (FP) laser, A:145 OIL, basic principle, A:146–151 optical spectrum analyzer (OSA), A:150 OIL VCSELs, modulation properties, A:151–160 advantest dynamic chirp measurement system, A:163 cascaded optical injection locking (COIL) structure, A:153 cavity resonance shift in wavelength, A:155 COIL, promising approach, A:160 erbium-doped fiber amplifier (EDFA), A:161 large signal modulation, A:160–163 small-signal modulation response, A:151–160 OIL VCSELs, nonlinearity/dynamic range experiments, A:166–168 intermodulation distortion (IMD), A:166 simulations, A:168–169 Optical injection locking (OIL), A:145 relative intensity noise of OIL VCSELs, A:169–171 RF link gain enhancement of OIL VCSELs, A:163–165 spur-free dynamic range (SFDR), A:146 vertical-cavity surface- emitting lasers (VCSELs), A:146 Ultrawave, B:220, 221 undersea cable, B:561, 564, 571, 581, 601, 606 undersea systems, B:565, 569 undirectional ethernet broadcast, B:392 uni-travelling carrier, B:191, 755 unidirectional link routing protocol, B:393 unidirectional path-switched rings, B:523 universal ethernet interface, B:346 Universalmobile telecommunication system, B:756 unshielded twisted pair, B:349 US Congress, B:451 US Telecommunications policy, B:520

913 User datagram protocol, B:725 user network interface, B:347, 620

V vapor-doping OVD process, A:562–563 advantage, A:563–565 variable attenuators MEMS data transceiver, A:745 waveguide, A:404 variable optical attenuator (VOA), A:276, 352, B:308 power transfer function of, A:353 variants on theme equalization architectures for better performance DFE with precursor canceller, A:678 fixed delay tree search, A:678 VCSEL-based slow light devices, A:97–99 VCSELs diodes, A:44–47, 108 vector modulation/demodulation architecture, B:106 Verizon, B:403, 408, 410, 416, 438, 442, 443, 448, 459, 461, 465–468 vertical-cavity surface-emitting lasers (VCSELs), A:146, B:4, 5, 668, 814, 827, 828, 839 very high-speed digital subscriber line, B:415 Very-long-baseline interferometry, B:614 vestigial sideband, B:52, 266 Video hub office, B:553 video-on-demand, B:413, 552–555 virtual channel identifier, B:523 virtual communities, B:480 virtual concatenation, B:347, 526 virtual private network, B:546, 613 Viterbi algorithm, A:676, B:164, 182 Voice over internet protocol, B:345, 455, 479, 516, 518, 551, 637 voice-telephony connections, B:454 VSB signals, B:52, 267

W wafer bonding (integration technique), A:349 wafer-bonded hybrid laser, A:423 CW light-current curve, A:424

914 ‘‘wallplug efficiency,’’ A:132 waveform distortion, B:25, 244, 274, 493, 566 waveguide single side cavity, transmission, A:445–446 transmission spectra of one- and two-cavity structures, A:446 two cavities side-coupled, A:445 transmission spectrum, A:446–447 waveguide-grating couplers, A:391 waveguide modulators, B:751 waveguide parameters, A:280 in first and second slab, A:281 path-length difference, A:282 waveguide photodiodes edge-coupled waveguide photodiodes, A:222–224 ‘‘double core,’’ A:223 evanescently coupled waveguide photodiodes, A:224–230 electro-optic sampling, A:228 enhancements/modifications, A:225 MMI, A:232 negative aspects, A:224 waveguides, AWG, A:279 in first slab, A:279 improve coupling efficiency to highly sampled output, A:300 layout and results from demultiplexer consisting of MZI, A:296 path-length difference, A:282 in second slab, A:280 wavelength add/drop, B:295 wavelength blocker, B:296, 500 wavelength coded tag, B:250 wavelength conversion, A:789–796, B:11, 204, 300, 499, 625, 632, 634–636, 638, 651, 654, 655, 663, 669, 671–673, 675, 676, 677, 682, 682, 683, 696, 696, 697, 701, 720–723, 725–727, 731, 734 wavelength division multiplexed, B:1, 2, 6, 7, 13, 14, 15, 17–19, 24, 28, 32, 33, 42, 53, 58, 63, 68, 71, 76, 95–97, 118, 125, 126, 132, 137, 162, 166–170, 181–184, 194–197, 200–202, 204,

Index 208, 215–218, 226, 239, 241–243, 247, 248, 250–252, 256, 263, 268, 273, 274, 284, 294, 295–297, 300, 304, 309, 312, 314, 316, 321, 322, 324–325, 329, 336, 337, 346, 351, 370, 379, 388–393, 430, 477, 481, 484–487, 491–493, 502, 503, 506, 507, 514, 516, 521, 522, 524, 531–533, 535, 536, 543–549, 551, 556, 561, 564–567, 572, 573, 576, 578, 579, 581, 594, 596, 597, 604, 615, 618, 622, 625, 633, 642, 643, 656, 662, 675, 676, 678, 699, 703, 706, 720, 749, 771, 772, 778, 789, 803, 833–836, 850–853, 857 wavelength division multiplexing (WDM), A:43, 82, 172, 270, 393, 432, 511, 524 CWDM, A:304 multiplexing channels, A:287 wavelength domain, B:650 wavelength filters, B:422, 506 wavelength independent modulation, B:45 wavelength independent power couplers, B:312 wavelength integration and control, surface emitting lasers, A:87–91 wavelength locked distributed feedback, B:62 wavelength management, B:132, 183, 196 wavelength monitoring techniques, B:248 wavelength multiplexing header technique, B:662 Wavelength, polarization, and fabrication (WPF) changes, A:272, 273 wavelength routing, B:335, 336 switching fabric, B:672 wavelength selective couplers, B:193 crossconnect, B:242 Wavelength-selective crossconnects (WSXC), A:728 4 4 WSXC, A:729 wavelength-selective MEMS components spectral equalizers, A:719–721 wavelength-selective switches (WSS), A:317–318

Index free space-based wavelength-selective switches 1 4 wavelength-selective switch, A:723 1 N2 using two-axis micromirror array, A:725 analog micromirror, A:722 free space-based wavelengthselective switches, A:722–726 LCOS-based 1 9 WSS, A:723 two-axis analog micromirror array for WSS, A:726 hybrid PLC-MEMS wavelengthselective switches, A:726–727 1 9 WSS with hybrid integration of PLC and MEMS, A:727 advantages of, A:726 monolithic wavelength-selective switches, A:727–728 1 4 WSS, A:722 wavelength services, B:521 wavelength spacing, A:282 wavelength speedup, B:780 wavelength striped message format, B:775 wavelength stripes, B:780 WDM see wavelength division multiplexed; wavelength division multiplexing (WDM) WDM technology, A:82 Wentzel-Kramers-Brillouin, B:838 whitened matched filter (WMF), A:676 Wi-Fi, B:236, 418, 466, 614 Wide area network (WAN), A:371 router-to-router bandwidth compared to, A:372 Wide Area Network, B:528 interface sublayer, B:352 ‘‘wide-band’’ (de)multiplexers, A:287 Wideband digital crossconnect system, B:522 WiMax, B:614, 754 wire waveguide, see silicon strip wireless-fidelity, B:418 wireless local area networks (WLANs), A:179, B:739, 756, 757

915 wireless-over-fiber systems, B:756 WLANs, see wireless local area networks (WLANs) WMF see whitened matched filter (WMF) World Wide Web, B:468, 617 world-wide web and Internet, A:372 WPF changes, see Wavelength, polarization, and fabrication (WPF) changes W-profile design, A:551–555 WSS, see wavelength-selective switches (WSS) WSXC see Wavelength-selective crossconnects (WSXC)

X XAUI interface, B:353, 356 xDSL, B:421, 462, 516, 551 XPM, see cross-phase modulation (XPM) XPolM, A:652, 657, 658, 659, 660, 661, 662

Y Yahoo, B:389 Y-branch coupler, A:271, 272 yield management methodologies, A:359–361 cluster analysis and negative binomial model, A:360 in-line analysis, A:361 limited yield analysis, A:360 refined yield analysis, A:360 Y-junction, B:776, 777 YouTube, B:457, 458, 468, 552

Z zero phase difference, B:211 zero transmission, B:59, 206 zone analysis, A:361 Zoomy communications, B:441

This page intentionally left blank