WILEY Broadband Optical Access Networks User Manual

BROADBAND OPTICAL
ACCESS NETWORKS

BROADBAND OPTICAL
ACCESS NETWORKS

LEONID G. KAZOVSKY
NING CHENG
WEI-TAO SHAW
DAVID GUTIERREZ
SHING-WA WONG

A JOHN WILEY & SONS, INC., PUBLICATION



Copyright

C

2011 by John Wiley & Sons, Inc. All rights reserved.

Published by John Wiley & Sons, Inc., Hoboken, New Jersey.
Published simultaneously in Canada.

No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or
by any means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except as
permitted under Sections 107 or 108 of the 1976 United States Copyright Act, without either the prior
written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to
the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, [phone],
fax [phone], or on the web at www.copyright.com. Requests to the Publisher for permission
should be addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken,
NJ 07030, [phone], fax [phone], or online at http://www.wiley.com/go/permission.

Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in
preparing this book, they make no representations or warranties with respect to the accuracy or
completeness of the contents of this book and specifically disclaim any implied warranties of
merchantability or fitness for a particular purpose. No warranty may be created or extended by sales
representatives or written sales materials. The advice and strategies contained herein may not be suitable
for your situation. You should consult with a professional where appropriate. Neither the publisher nor
author shall be liable for any loss of profit or any other commercial damages, including but not limited to
special, incidental, consequential, or other damages.

For general information on our other products and services or for technical support, please contact our
Customer Care Department within the United States at [phone], outside the United States at
[phone] or fax [phone].

Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may
not be available in electronic format. For more information about Wiley products, visit our web site at

www.wiley.com

Library of Congress Cataloging-in-Publication Data Is Available

Kazovsky, Leonid G.

Broadband optical access networks / Leonid G. Kazovsky, Ning Cheng, Wei-Tao Shaw, David

Gutierrez, Shing-Wa Wong.

Includes index.

ISBN 978-0-470-18235-2

Printed in Singapore

oBook ISBN: 978 0470 910931
ePDF ISBN: 978 0470 910924
ePub ISBN: 978 0470 922675

10987654321

CONTENTS

FOREWORD xi
PREFACE xiii
ACKNOWLEDGMENTS xv

1 BROADBAND ACCESS TECHNOLOGIES: AN OVERVIEW 1

1.1 Communication Networks / 2

1.2 Access Technologies / 4

1.2.1 Last-Mile Bottleneck / 4

1.2.2 Access Technologies Compared / 5

1.3 Digital Subscriber Line / 6

1.3.1 DSL Standards / 7

1.3.2 Modulation Methods / 8

1.3.3 Voice over DSL / 8

1.4 Hybrid Fiber Coax / 9

1.4.1 Cable Modem / 10

1.4.2 DOCSIS / 10

1.5 Optical Access Networks / 11

1.5.1 Passive Optical Networks / 11

1.5.2 PON Standard Development / 12

1.5.3 WDM PONs / 13

1.5.4 Other Types of Optical Access Networks / 15

v

vi CONTENTS

1.6 Broadband over Power Lines / 18

1.6.1 Power-Line Communications / 18

1.6.2 BPL Modem / 19

1.6.3 Challenges in BPL / 20

1.7 Wireless Access Technologies / 20

1.7.1 Wi-Fi Mesh Networks / 21

1.7.2 WiMAX Access Networks / 22

1.7.3 Cellular Networks / 24

1.7.4 Satellite Systems / 25

1.7.5 LMDS and MMDS Systems / 26

1.8 Broadband Services and Emerging Technologies / 28

1.8.1 Broadband Access Services / 29

1.8.2 Emerging Technologies / 30

1.9 Summary / 31
References / 33

2 OPTICAL COMMUNICATIONS: COMPONENTS

AND SYSTEMS 34

2.1 Optical Fibers / 35

2.1.1 Fiber Structure / 35

2.1.2 Fiber Mode / 38

2.1.3 Fiber Loss / 45

2.1.4 Fiber Dispersion / 47

2.1.5 Nonlinear Effects / 52

2.1.6 Light-Wave Propagation in Optical Fibers / 54

2.2 Optical Transmitters / 55

2.2.1 Semiconductor Lasers / 55

2.2.2 Optical Modulators / 66

2.2.3 Transmitter Design / 71

2.3 Optical Receivers / 72

2.3.1 Photodetectors / 73

2.3.2 Optical Receiver Design / 76

2.4 Optical Amplifiers / 78

2.4.1 Rare-Earth-Doped Fiber Amplifiers / 81

2.4.2 Semiconductor Optical Amplifiers / 83

2.4.3 Raman Amplifiers / 85

2.5 Passive Optical Components / 86

2.5.1 Directional Couplers / 86

2.5.2 Optical Filters / 89

CONTENTS vii

2.6 System Design and Analysis / 93

2.6.1 Receiver Sensitivity / 93

2.6.2 Power Budget / 98

2.6.3 Dispersion Limit / 99

2.7 Optical Transceiver Design for TDM PONs / 101

2.7.1 Burst-Mode Optical Transmission / 102

2.7.2 Colorless ONUs / 104

2.8 Summary / 106
References / 107

3 PASSIVE OPTICAL NETWORKS: ARCHITECTURES

AND PROTOCOLS 108

3.1 PON Architectures / 109

3.1.1 Network Dimensioning and Bandwidth / 110

3.1.2 Power Budget / 110

3.1.3 Burst-Mode Operation / 112

3.1.4 PON Packet Format and Encapsulation / 113

3.1.5 Dynamic Bandwidth Allocation, Ranging,

and Discovery / 114

3.1.6 Reliability and Security Concerns / 114

3.2 PON Standards History and Deployment / 115

3.2.1 Brief Developmental History / 115

3.2.2 FTTx Deployments / 116

3.3 Broadband PON / 117

3.3.1 BPON Architecture / 118

3.3.2 BPON Protocol and Service / 121

3.3.3 BPON Transmission Convergence Layer / 125

3.3.4 BPON Dynamic Bandwidth Allocation / 130

3.3.5 Other ITU-T G.983.x Recommendations / 133

3.4 Gigabit-Capable PON / 133

3.4.1 GPON Physical Medium–Dependent Layer / 134

3.4.2 GPON Transmission Convergence Layer / 137

3.4.3 Recent G.984 Series Standards, Revisions, and

Amendments / 142

3.5 Ethernet PON / 144

3.5.1 EPON Architecture / 144

3.5.2 EPON Point-to-Multipoint MAC Control / 147

3.5.3 Open Implementations in EPON / 152

3.5.4 Unresolved Security Weaknesses / 155

viii CONTENTS

3.6 IEEE 802.av-2009 10GEPON Standard / 156

3.6.1 10GEPON PMD Architecture / 156

3.6.2 10GEPON MAC Modifications / 158

3.6.3 10GEPON Coexistence Options / 161

3.7 Next-Generation Optical Access System Development
in the Standards / 162

3.7.1 FSAN NGA Road Map / 162

3.7.2 Energy Efficiency / 163

3.7.3 Other Worldwide Development / 164

3.8 Summary / 164
References / 165

4 NEXT-GENERATION BROADBAND OPTICAL

ACCESS NETWORKS 166

4.1 TDM-PON Evolution / 167

4.1.1 EPON Bandwidth Enhancements / 168

4.1.2 GPON Bandwidth Enhancements / 168

4.1.3 Line Rate Enhancements Research / 169

4.2 WDM-PON Components and Network Architectures / 172

4.2.1 Colorless ONUs / 173

4.2.2 Tunable Lasers and Receivers / 174

4.2.3 Spectrum-Sliced Broadband Light Sources / 176

4.2.4 Injection-Locked FP Lasers / 178

4.2.5 Centralized Light Sources with RSOAs / 179

4.2.6 Multimode Fiber / 181

4.3 Hybrid TDM/WDM-PON / 184

4.3.1 TDM-PON to WDM-PON Evolution / 184

4.3.2 Hybrid Tree Topology Evolution / 186

4.3.3 Tree to Ring Topology Evolution / 195

4.4 WDM-PON Protocols and Scheduling Algorithms / 202

4.4.1 MAC Protocols / 203

4.4.2 Scheduling Algorithms / 204

4.5 Summary / 211
References / 211

5 HYBRID OPTICAL WIRELESS ACCESS NETWORKS 216

5.1 Wireless Access Technologies / 217

5.1.1 IEEE 802.16 WiMAX / 217

5.1.2 Wireless Mesh Networks / 225

CONTENTS ix

5.2 Hybrid Optical–Wireless Access Network Architecture / 241

5.2.1 Leveraging TDM-PON for Smooth Upgrade of Hierarchical
Wireless Access Networks / 242

5.2.2 Upgrading Path / 244

5.2.3 Reconfigurable Optical Backhaul Architecture / 247

5.3 Integrated Routing Algorithm for Hybrid Access Networks / 258

5.3.1 Simulation Results and Performance Analysis / 260

5.4 Summary / 262
References / 263

INDEX 267

FOREWORD

Broadband optical access networks are crucial to the future development of the
Internet. The continuing evolution of high-capacity, low-latency optical access net­works will provide users with real-time high-bandwidth access to the Web essential
for such emerging trends as immersive video communications and ubiquitous cloud
computing. These ultrahigh-speed access networks must be built under challenging
economic and environmental imperatives to be “faster, cheaper, and greener.” T his
book presents in a clear and illustrative format the technical and scientific concepts
that are needed to accomplish the design of new broadband access networks upon
which users will surf the wave of the twenty-first-century Internet.

The book is coauthored by Professor Leonid Kazovsky and his graduate stu­dents. Professor Kazovsky is a recognized leader and authority in the field and has
a long and distinguished track record for making highly timely and significant re­search contributions within the general area of optical communication systems and
optical networks.He has contributed over thelast 40 years in the areas of wavelength­division-multiplexed(WDM) andcoherent transmission systems for the core network
as well as transmission systems and network architectures and technologies at the
metro and access levels. This book builds on Professor Kazovsky’s research con­ducted at Bellcore (where he worked in the 1980s), at Stanford University (where he
has worked since 1990), and at numerous European research organizations during
sabbaticals in the UK, the Netherlands, Italy, Denmark, and (most recently) Sweden.
This rich set of influences gives the book and its readers the benefits ofbroad exposure
to diverse research ideas and approaches.

Professor Kazovsky heads the Photonics and Networking Research Laboratory
at Stanford University. He and his team of researchers are focusing on broadband
optical access networks. They bring their ongoing research results to this unique

xi

xii FOREWORD

book, bridging fundamentalsof optical communicationandnetworking system design
with technology issues and current standards. Once that foundation is laid, the book
delves into currenthigh-capacity research issues, including evolution to WDMoptical
access, convergedhybrid optical/wireless accessnetworks,and implementation issues
of broadband optical access. Research ideas generated by Professor Kazovsky’s
research group havebeen widely adopted worldwide, includingin framework projects
of the European Union.

We strongly recommend this book, as it offers timely, accurate, authoritative,
and innovative information regarding broadband optical access network design and
implementation. We’re confident that you will enjoy readingthe book and learn much
while doing so.

Daniel Kilper and Peter Vetter

Alcatel-Lucent Bell Labs, Murray Hill, New Jersey

James F. Kelly

Google, Mountain View, California

Alan Willner

USC Viterbi School of Engineering, Los Angeles, California

Biswanath Mukherjee

University of California Davis, Davis, California

Anders Berntson, Gunnar Jacobsen, and Mikhail Popov

Acreo, Stockholm, Sweden

PREFACE

The roots of this book were planted about a decade ago. At that time, I became
increasingly convinced that wide-area and metropolitan-area networks, where much
of my group’s research has been centered at that time, were in good shape. Although
research in these fields was (and still is) needed, that’s not where the networking
bottleneck seemed to be. Rather, the bottleneck was (and still is in many places) in
the access networks, which choked users’ access to information and services. It was
clear to me that the long-term solution to that problem has to involve optical fiber
access networks.

That conviction led me to switch the focus of my group’s research to optical
access networks. In turn, that decision led to a decade of exciting exceptionally
interesting research into the many challenges facing modern access networks. These
challenges include rapidly increasing demands for larger bandwidthand better quality
of service,graceful evolution to more powerful solutionswithout completerebuilding
of existing infrastructure, enhancing network range and number of users, improving
access networks’ resilience, simplifying network architecture, finding better control
strategies, and solving the problem of fiber/wireless integration. All these problems
would have to be solved while maintaining the economic viability of access networks
so that operators would be prepared to make the necessary (and huge) investment in
fiber and other infrastructure.

Finding solutions fortheforegoing problems occupiedmostof my researchgroup’s
time and attention for much of the past decade. In the beginning of that decade (and
for a long time after that), my group, the Photonics and Networking Research Labo­ratory (PNRL) at Stanford University, was one of very few (or perhaps even the only)
university research group working on fiber access, as many other optical researchers
tended to discount optical accessissues as trivial. Although that made funding for our

xiii

xiv PREFACE

research difficult to find, that position allowed us to make many pioneering contribu­tions widelyused andcited today. Later, many other university and industrial research
groups entered the field, and several large-scale research efforts were organized, most
notably in Europe, where serious research into both passive optical networks (PONs)
and active optical networks (AONs) has been conducted over the last several years.
Notable European efforts in broadband fiber access include ICT ALPHA (architec­tures for flexible photonic home and access networks, focused on AON, PON, and
technoeconomics), ICT OASE (optical access seamless evolution, focused on PON,
technoeconomics, and business models) and ICT SARDANA (focused on PON and
optical metropolitan networks). These efforts resulted in extremely fast progress in
the field. It was gratifying to see many PNRL research results adopted, used, and
developed further by these (and other) efforts, especially in SARDANA.

Many of my colleagues working on optical access research encouraged me over
the past fewyears tointegrate results ofthe PNRLresearch on optical access networks
into a single volume and publish it to ensure the broadest possible dissemination of
our results. They feel that our results, when published in a single volume rather than
the current combination of conference and journal articles, will further stimulate new
research, plant new ideas, and lead to exciting new developments.

For a long while, I was reluctant to do so. The field of broadband fiber access
networks is exceptionally broad; in addition, it is still very young and is developing
and changing very fast. Thus, writing acomprehensive book on this subjectis (nearly)
impossible. Eventually, though, a stream of inquiries for additional information about
our research convinced me to change my mind, and my research students and myself
began the time-consuming process of writing our book.

Our goal was fairly modest: to summarize in one place the research results pro­duced by the PNRL over the past decade or so. The reader should keep this goal in
mind. We make no attempt to cover the entire field, just to provide a summary of our
research. Even that goal proved to be difficult to achieve, as we are continuing our
research as new technologies emerge, so our understanding of the field continues to
evolve with time. However, we trust that the reader will consider this book a useful
addition to his or her knowledge base of optical access networks.

Stanford University
Stanford, California

Leonid Kazovsky

ACKNOWLEDGMENTS

This book is based on research results obtained by our research group, the Photonics
and Networking Research Laboratory at Stanford University. Our research on broad­band fiber access networks, conducted over adecade orso, required a consistent effort
by a large group of exceptionally talented graduate students, postdocs, and visitors.
Some of these contributors are co-authors of the book, while others are working
in other organizations and on other projects and so were too busy to help with the
book-writing process. We are thankful to all of them, however.

Our research on broadband fiber access networks required a sizable team and a
substantial amount of experimental, theoretical, and simulation efforts. This would
be impossible without the generous and long-term support of our sponsors. We are
grateful to our sponsors, who trusted us with the necessary resources. Our main
sponsors in that area were, or are, the National Science Foundation under grants
0520291 and 0627085, KDDI Laboratories, Motorola, the Stanford Networking Re­search Center (no longer in existence), ST Microelectronics, ANDevices, Huawei,
Deutsche Telecom, and Alcatel-Lucent Bell Laboratories.

We also thank the many research visitors to our group (mainly postdocs or visiting
professors), who helped in a variety of ways, ranging from making research contribu­tions to our book, to providing suggestions and comments on its contents, to taking
part in one or more of our broadband access research projects. In particular, we are
grateful to Dr. Kyeong Soo (Joseph)Kim ofSwanseaUniversity; Professor Chunming
Qiao of SUNY Buffalo; Dr. Luca Valcarenghi of Scuola Superiore Sant’Anna , Italy;
Professor David Larrabeiti of Universidad Carlos III de Madrid, Madrid, Spain; and
Dr.Divanilson Campelo of University ofBrasilia, Brazil.Many others helped aswell;
unfortunately, a comprehensive list would be too long to include here.

xv

xvi ACKNOWLEDGMENTS

We are grateful to the challenging, exciting research environment at Stanford
University, where the lead author of this book has had the pleasure of working for the
past two decades. Without that environment,this book would never have materialized.

Last but not least, we would like to thank our many colleagues all over the world
for stimulatingdiscussions, fortheir friendship, and for their help. We are particularly
grateful to Prof. Vincent Chan, MIT; Prof. Alan Willner, USC; Drs. James Kelly and
Cedric Lam of Google, Inc.; Prof. Andrea Fumagali, University of Texas; Profs. Ben
Yoo and BiswanathMukherjee, University ofCalifornia, Davis; Profs. Djan Khoeand
Dr. Harm of the Technical University of Eindoven, the Netherlands; Prof. Giancarlo
Prati of the Scuola Superiore St. Anna, Pisa, Italy; Prof. Palle Jeppesen of the Danish
Technical University, Copenhagen, Denmark; Drs. Gunnar Jacobsen, Mikhail Popov,
and ClausLarsen of Acreo, Stockholm, Sweden; Dr. Shu Yamamoto of KDDI, Japan;
and Dr. Frank Effenburger of Huawei.

Stanford University
Stanford, California

Leonid Kazovsky

Ning Cheng

Wei-Tao Shaw

David Gutierrez

Shing-Wa Wong

CHAPTER 1

BROADBAND ACCESS TECHNOLOGIES: AN OVERVIEW

In past decades we witnessed the rapid development of global communication infras­tructure and theexplosivegrowthof the Internet,accompanied by ever-increasinguser
bandwidth demands and emerging multimedia applications. These dramatic changes
in technologies and market demands, combined with government deregulation and
fierce competitionamong data, telcom, and CATV operators, have scrambledthe con­ventional communication services and created new social and economic challenges
and opportunities in the new millennium. To meet those challenges and competi­tions, current service providers are striving to build new multimedia networks. The
most challenging part of current Internet development is the access network. As an
integrated part of global communication infrastructure, broadband access networks
connect millions of users to the Internet, providing various services, including inte­grated voice, data, and video. As bandwidth demands for multimedia applications
increase continuously, users require broadband and flexible access with higher band­width and lower cost. A variety of broadband access technologies are emerging to
meet those challenging demands. While broadband communication over power lines
and satellites is being developed to catch the market share, DSL (digital subscriber
line) and cable modem continue to evolve, allowing telecom and CATV companies
to provide high-speed access over copper wires. In the meantime, FTTx and wireless
networks have become a very promising access technologies. The convergence of
optical and wireless technologies could be the best solution for broadband and mo­bile access service in the future. As new technology continues to be developed, the
future access technology will be more flexible, faster, and cheaper. In this chapter

Broadband Optical Access Networks, First Edition. Leonid G. Kazovsky, Ning Cheng, Wei-Tao Shaw,
David Gutierrez, and Shing-Wa Wong.



C

2011 John Wiley & Sons, Inc. Published 2011 by John Wiley & Sons, Inc.

1

2 BROADBAND ACCESS TECHNOLOGIES: AN OVERVIEW

we discuss current access network scenarios and review current and emerging broad
access technologies, including DSL, cable modem, optical, and wireless solutions.

1.1 COMMUNICATION NETWORKS

Since the development oftelegraph and telephone networks in the nineteenth century,
communication networks have come a long way and evolved into a global infrastruc­ture. More than ever before, communications and information technologies pervade
every aspect of our lives: our homes, our workplaces, our schools, and even our
bodies. As part of the fundamental infrastructure of our global village, communi­cation networks has enabled many other developments—social, economic, cultural,
and political—and has changed significantly how people live, work, and interact.

Today’s global communication network is an extremely complicated system and
covers a very large geographic area, all over the world and even in outer space. Such
a complicated system is built and managed within a hierarchical structure, consisting
of local area, access area, metropolitan area, and wide area networks (as shown in
Figure 1.1). All the network layers cooperate to achieve the ultimate task: anyone,
anywhere, anytime, and any media communications.

Local Area Networks

Local area networks (LANs) mainly connect computers and
other electronicdevices (servers, printers,etc.) within an office, a single building, or a
few adjacent buildings. Therefore, the geographical coverage of LANs is very small,
spanning from a few meters to a few hundred meters. LANs are generally not a part
of public networks but are owned and operated by private organizations. Common

FIGURE 1.1 Hierarchical architecture of global communication infrastructure.

COMMUNICATION NETWORKS 3

topologies for LANs are bus, ring, star, or tree. The most popular LANs are parts of
the Ethernet, supporting a few hundred users with typical bit rates of 10 or 100 Mb/s.

Access Networks

The computers andother communication equipmentof a private
organization are usually connected to a public telecommunication networks through
access networks.Access networks bridge end users to service providers through twist
pairs (phone line), coaxial cables, or other leased lines (such as OC3 through optical
fiber). The typical distance covered by an access network is a few kilometers up to
20 km. For personal users, access networks use DSL or cable modem technology
with a transmission rate of a few megabits per second; for business users, networks
employ point-to-point fiber links with hundreds of megabits or gigabits per second.

Metropolitan Area Networks

Metropolitan area networks (MANs) aggregate the
traffic from access networks and transport the data at a higher speed. A typical area
covered by a MAN spans a metropolitan area or a small region in the countryside.
Its topology is usually a fiber ring connecting multiple central offices, where the
transmission data rate is typically 2.5 or 10 Gb/s.

Wide Area Networks

Wide area networks (WANs) carry a large amount of traffic
among cities, countries, and continents. MAN multiplexes traffic from LANs and
transports the aggregated traffic at a much higher data rate, typically tens of gigabits
per second or higher using wavelength-division multiplexing (WDM) technology
over optical fibers. Whereas a WAN covers the area of a nation or, in some cases,
multiple nations, a link or path through a MAN could be as long as a few thousand
kilometers. Beyond MANs, submarine links connect continents. Generally, the sub­marine systems are point-to-point links with a large capacity and an extremely long
path, from a few thousand up to 10,000 km. Because these links are designed for
ultralong distances and operate under the sea, the design requirements are much more
stringent than those of their terrestrial counterparts. Presently, submarine links are
deployed across the Pacific and Atlantic oceans. Some shorter submarine links are
also widely used in the Mediterranean, Asian Pacific, and African areas.

Service Convergence

Historically, communication networks provide mainly
three types of service: voice, data, and video (triple play). Voice conversation us­ing plain old telephony is a continuous 3.4-kHz analog signal carried by two-way,
point-to-point circuits with a very stringent delay requirement. The standard TV sig­nal is a continuous 6-MHz analog signal usually distributed with point-to-multipoint
broadcasting. Data transmission is typically bursty with varying bandwidth and de­lay requirements. Because the traffic characteristics of voice, data, and video and
their corresponding requirements as to quality of service (QoS) are fundamentally
different, three major types of networks were developed specifically to render these
services in a cost-effective manner: PSTN (public-switched telephone networks) for
voice conversation, HFC (hybrid fiber coax) networks for video distribution, and the
Internet for data transfer. Although HFC networks are optimized for video broad­casting, the inherent one-way communication is not suitable for bidirectional data or

4 BROADBAND ACCESS TECHNOLOGIES: AN OVERVIEW

voice. PSTN adopts circuit switching technology to carry information with specific
bandwidth or data rates, such as voice signals. However, circuit-switched networks
are not very efficient for carrying bursty data traffic. With packet switching, the
Internet can support bursty data transmission, but it is very difficult to meet stringent
delay requirements for certain applications. Therefore, no single network can satisfy
all the service requirements.

Emerging multimedia applications suchas videoon demand, e-learning, and inter­active gaming require simultaneous transmission of voice, data, and video. Driven by
user demands and stiff competition, service providers are moving toward a converged
network for multimedia applications, which will utilize Internet protocol (IP) tech­nologies to provide triple-play services. As VoIP (voice over IP) has been developed
in the past few years and more recently IP TV has become a mature technology,
all network services will converge into an IP-based service platform. Furthermore,
the integration of optical and wireless technologies will make quadruple play (voice,
data, video, and mobility) a reality in the near future.

1.2 ACCESS TECHNOLOGIES

Emerging multimedia applications continuously fuel the explosive growth of the
Internet and gradually pervade every area of our lives, from home to workplace. To
provide multimedia service to every home and every user, access networks are built
to connect end users to service providers. The link between service providers and
end users is often called the last mile by service providers, or from an end user’s
perspective, the first mile. Ideally, access networks should be a converged platform
capable of supporting a variety of applications and services. Through broadband
access networks, integrated voice, data, and video service are provided to end users.
However, the reality is that access networks are the weakest links in the current
Internet infrastructure. While national information highways (WANs and MANs)
have been developed in most parts of the globe, ramps and access routes to these
information highways (i.e.,the first/lastmile) are mostlybike lanes or at best,unpaved
roads, causing traffic congestion. Hence, pervasive broadband access should be a
national imperative for future Internet development. In this section we review current
access scenarios and discuss the last-mile bottleneck and its possible solutions.

1.2.1 Last-Mile Bottleneck

Due to advances in photonic technologies and worldwide deployment of optical
fibers, during the last decade the telecommunication industry has experienced an ex­traordinary increase in transmission capacity in core transport networks. Commercial
systems with 1-Tb/s transmission can easily be implemented in the field, and the
state-of-the-art fiber optical transmission technology has reached 10 Tb/s in a single
fiber. In the meanwhile, at the user end, the drastic improvement in the performance
of personal computers and consumer electronic devices has made possible expand­ing demands of multimedia services, such as video on demand, video conferencing,

ACCESS TECHNOLOGIES 5

TABLE 1.1 Multimedia Applications and Their Bandwidth Requirements

Application Bandwidth Latency Other Requirements

Voice over IP (VoIP) 64 kb/s 200 ms Protection
Videoconferencing 2 Mb/s 200 ms Protection
File sharing 3 Mb/s 1 s
SDTV 4.5 Mb/s/ch 10 s Multicasting
Interactive gaming 5 Mb/s 200 ms
Telemedicine 8 Mb/s 50 ms Protection
Real-time video 10 Mb/s 200 ms Content distribution
Video on demand 10 Mb/s/ch 10 s Low packet loss
HDTV 10 Mb/s/ch 10 s Multicasting
Network-hosted software 25 Mb/s 200 ms Security

e-learning, interactive games, VoIP, and others. Table 1.1 lists common end-user ap­plications and their bandwidth requirements. As a result of the constantly increasing
bandwidth demand,users may require more than 50 Mb/s in the near future. However,
the current copper wire technologies bridging users and core networks have reached
their fundamental bandwidth limits and become the first-last-mile bottleneck. Delays
in Web page browsing, data access, and audio/video clip downloading have earned
the Internet the nickname “World Wide Wait.” How to alleviate this bottleneck has
been a very challenging task for service providers.

1.2.2 Access Technologies Compared

For broadband access services, there is strong competition among several technolo­gies: digital subscriber line, hybrid fiber coax, wireless, and FTTx (fiber to the x,
x standing for home, curb, neighborhood, office, business, premise, user, etc.). For
comparison, Table 1.2 lists the bandwidths (per user) and reaches of these com­peting technologies. Currently, dominant broadband access technologies are digital

TABLE 1.2 Comparison of Bandwidth and Reach for Popular Access Technologies

Service Medium Downstream (Mb/s) Upstream (Mb/s) Max Reach (km)

ADSL Twisted pair 8 0.896 5.5
ADSL2 Twisted pair 15 3.8 5.5
VDSL1 Twisted pair 50 30 1.5
VDSL2 Twisted pair 100 30 0.5
HFC Coax cable 40 9 25

BPON Fiber 622 155 20
GPON Fiber 2488 1244 20
EPON Fiber 1000 1000 20

Wi-Fi Free space 54 54 0.1
WiMAX Free space 134 134 5

6 BROADBAND ACCESS TECHNOLOGIES: AN OVERVIEW

subscriber loop and coaxial cable. For conventional ADSL (asymmetric DSL) tech­nology, the bandwidth available is a few Mb/s within the 5.5-km range. Newer VDSL
(very high-speed DSL) can provide 50 Mb/s, but the maximum reach is limited to

1.5 km. On the other hand, coaxial cable has a much larger bandwidth than twist
pairs, which can be as high as 1 Gb/s. However, due to the broadcast nature of CATV
system, current cable modems can provide each user with an average bandwidth of a
few Mb/s. While DSL and cable provide wired solutions for broadband access, Wi-Fi
(wireless fidelity), and WiMAX (worldwide interoperability for microwave access)
provide mobile access in a LAN or MAN network. Even though a nominal band­width of Wi-Fi andWiMAX can be relatively higher(54 Mb/s in 100 mfor Wi-Fi and
28 Mb/s in 15 km for WiMAX), the reach of such wireless access is very limited and
the actual bandwidth provided to users can be much lower, due to the interference in
wireless channels. As a LAN technology, the primary use of Wi-Fi is in home and of­fice networking. To reach the central office or service provider, multiple-hop wireless
links with WiMAX have to be adopted. An alternative technology that is also under
development is MBWA (mobile broadband wireless access, IEEE 802.20), which is
very similar to WiMAX (IEEE 802.16e). Compared to the fixed access solutions,
the advantages of the wireless technologies are easy deployment and ubiquitous or
mobile access, and the disadvantages are unreliable bandwidth provisioning and/or
limited access range.

The bandwidth and/or reach of the copper wire and wireless access technology is
very limited due to the physical media constraints. To satisfy the future use demand
(>30 Mb/s), there is a strategic urgency for service providers to deploy FTTx net­works. Currently, for cost and deployment reasons, FTTx is competing with other
access technologies. Long term, however, only optical fiber can provide the unlim­ited capacity and performance that will be required by future broadband services.
FTTx has long been dubbed as a future-proof technology for the access networks.
A number of optical access network architectures have been standardized (APON,
BPON, EPON, and GPON), and cost-effective components and devices for FTTx
have matured. We are currently witnessing a worldwide deployment of optical access
networks and a steady increase in FTTx users.

1.3 DIGITAL SUBSCRIBER LINE

Digital subscriber line (also called digital subscriber loop) is a family of access
technologies that utilize the telephone line (twisted pair) to provide broadband access
service. While the audio signal (voice) carried by a telephony system is limited from
300 to 3400 Hz, the twisted pair connecting the users to the central office is capable
of carrying frequencies well beyond the 3.4-kHz upper limit of the telephony system.
Depending on the length and the quality of the twisted pair, the upper limit can
extend to tens of megahertz. DSL takes advantage of this unused bandwidth and
transmits data using multiple-frequency channels. Thus, some types of DSL allow
simultaneous use of the telephone and broadband access on the same twisted pair.

DIGITAL SUBSCRIBER LINE 7

Modem

Central

Office

FIGURE 1.2 DSL access networks.

Figure 1.2 shows the typical setup of a DSL configuration. At the central office, a
DSLAM (DSL access multiplexer) sends the data to users via downstream channels.
At the user side, a DSL modem functions as a modulator/demodulator (i.e., receives
data from DSLAM and modulates user data for upstream transmission).

1.3.1 DSL Standards

DSL comes indifferent flavors, supporting various downstream/upstream bit rates and
access distances. DSLstandards are defined inANSI T1, andITU-T Recommendation
G.992/993. Table 1.2 lists various DSL standards and their performance. Collectively,
these DSL technologies are referred to as xDSL. Two commonly deployed DSL
standards are ADSL and VDSL.

As its namesuggests, ADSLsupports asymmetrical transmission.Since the typical
ratio of traffic asymmetry is about 2 :1 to 3 : 1, ADSL becomes a popular choice for
broadband access. In addition, there is more crosstalk from other circuits at the
DSLAM end. As the upload signal is weak at the noisy DSLAM end, it makes sense
technically to have upstream transmission at a lower bit rate. Depending on thelength
and quality (such as the signal-to-noise ratio) of the twisted pair, the downstream bit
rate can be as high as 10 times the upstream transmission. The maximum reach of
ADSL is 5500 m. While ADSL1 can support a downstream bit rate up to 8 Mb/s and
an upstream data rate up to 896 kb/s, ADSL2 supports up to 15 Mb/s downstream
and 3.8 Mb/s upstream.

To support higherbit rates, theVDSL standard wasdevelopedafter ADSL. Trading
transmission distance fordata rate, VDSLcan support amuch higher datarate but with
very limited reach. VDSL1 standards specify data rates of 50 Mb/s for downstream
and 30 Mb/s for upstream transmission. The maximum reach of VDSL1 is limited
to 1500 m. The newer version of VDSL standards, VDSL2, is an enhancement of

8 BROADBAND ACCESS TECHNOLOGIES: AN OVERVIEW

VDSL1, supporting a data rate up to 100 Mb/s (with a transmission distance of
500 m). At 1 km, the bit rate will drop to 50 Mb/s. For reaches longer than 1.6 km,
the VDSL2 performance is close to ADSL. Because of its higher data rates and
ADSL-like long reach performance, VDSL2 is considered to be a very promising
solution for upgrading existing ADSL infrastructure.

ADSL and VDSL are designed for residential subscribers with asymmetric band­width demands. For business users, symmetrical connections are generally required.
Two symmetrical DSL standards, HDSL and SHDSL, aredeveloped for businesscus­tomers. While HDSL supports a T1 line data rate at 1.552 Mb/s (including 8 kb/s of
overhead) with a reach of about 4000 m, SHDSL can provide a 6.696-Mb/s data rate
with a maximum reach of 5500 m. However, HDSL and SHDSL do not support si­multaneous telephone service, as most business customers do not have a requirement
for a simultaneous voice circuit.

1.3.2 Modulation Methods

DSL uses a DMT (discrete multitone) modulation method. In DMT modulation,
complex-to-real inverse discrete Fourier transform is used to partition the available
bandwidth of the twisted pair into 256 orthogonal subchannels. DMT is adaptive to
the quality of the twisted pair, so all the available bandwidth is fully utilized. The
signal-to-noise ratio of each subchannel is monitored continuously. Based on the
noise margin and bit error rate, a set of subchannels are selected, and a block of data
bits are mapped into subchannels. In each subchannel, QAM (quadrature amplitude
modulation) with a 4-kHz symbol rate is used to modulate the bit stream onto a
subcarrier, leading to 60 kb/s per channel. Typically, the frequency range between 25
and 160 kHz is used for upstream transmission, and 140 kHz to 1.1 MHz is used for
downstream transmission.

1.3.3 Voice over DSL

DSL was designed originally to carry data over phone lines, and DSL signal is
separated from voice signal. Recently, new protocols have been proposed to merge
voice and data at the circuit level. With advanced coding technologies, a 64-kb/s
digitized voice signal can be compressed to 8 kb/s or less, thus allowing more voice
channels to be carried over the same phone line. A voice over a DSL (VoDSL)
gateway converts and compresses the analog voice signal to digital bit streams, so
that calls made over VoDSL are indistinguishable from conventional calls. Usually,
12 to 20 voice channels can be carried over a single DSL line, depending on the
transmission distance and the signal quality. A VoDSL system can be integrated
into higher-layer protocols such as IP and ATM. Early DSL networks used ATM
to ensure QoS, where ATM virtual circuits were used for the voice traffic. ADSL
and VDSL networks migrate to packet-based transport, and they use packet-switched
based virtual circuits instead of ATM ones.

HYBRID FIBER COAX 9

1.4 HYBRID FIBER COAX

Cable networks were originally developed for a very simple reason: TV signal dis­tribution. Therefore, cable networks are optimized for one-way, point-to-multipoint
broadcasting of analog TV signals. As optical communication systems were devel­oped, most cable TV systems have gradually been upgraded to hybrid fiber coax
(HFC) networks, eliminating numerous electronic amplifiers along the trunk line.
However, before cable access technology can be deployed, a return pass must be
implemented for upstream traffic. To support two-way communication, bidirectional
amplifiers have to be used in HFC systems, where filters are deployed to split the
upstream (forward) and downstream (reverse) signals for separate amplification.

Figure 1.3 presents the network architecture of a typical HFC network. In HFC
networks, analog TV signals are carried from the cable headend to distribution nodes
using optical fibers, and from the distribution node, coaxial cable drops are deployed
to serve 500 to 2000 subscribers. As shown in the figure, an HFC network is a shared
medium system with a tree topology.In sucha topology, multiple users share the same
HFC infrastructure, so medium access control is required in upstream transmission
while downstream transmission uses a broadcast scheme. A cable modem deployed
at the subscriber end provides data connection to the cable network, while at the
headend, the cable modem termination system connects to a variety of data servers
and provides service to subscribers.

Compared withthe twisted pairs in a telephone system, coaxial cables have amuch
higher bandwidth (1000 MHz), thus can support a much higher data rate. Depending
on the signal-to-noise ratio on the coaxial cable, 40 Mb/s can be delivered to the
end users with QAM modulation. For upstream transmission, QPSK can deliver up
to a 10-Mb/s data rate. However, as cable systems are shared-medium networks, the
bandwidth is thus shared by all the cable modems connected to the network. By
contrast, DSL uses dedicated twist pairs for each user, thus no bandwidth sharing
for different users. Furthermore, as the transmission bandwidth must be shared by
multiple users, medium access control protocol must be deployed to govern upstream
transmission. If congestion occurs in a specific channel, the headend must be able to
instruct cable modems to tune its receiver to a different channel.

Primary

hub

Master

headend

FIGURE 1.3 HFC access networks.

Secondary

hub

Fiber
node

RF amplifier

10 BROADBAND ACCESS TECHNOLOGIES: AN OVERVIEW

1.4.1 Cable Modem

Cable modems were developed to transport high-speed data to and from end users
in an HFC network. Traditional TV broadcasting occupies frequencies up to 1 GHz,
with each TV channel occupying 6 MHz of bandwidth (Part 76 in the FCC rules). A
cable modem uses two of those 6-MHz channels for data transmission. For upstream
transmission, a cable modem sends user data to the headend using a 6-MHz band
between 5 and 42 MHz. At the same time, the cable modem must tune its receiver to
a 6-MHz band within a 450- to 750-MHz band to receive downstream data. While a
QAM modulation scheme is used for downstream data, a QPSK modulation scheme
is usuallyselected for upstream transmission, as it is more immune to the interference
resulting from radio broadcasting.

1.4.2 DOCSIS

DOCSIS (Data Over Cable Service Interface Specifications), developed by Cable­Labs, a consortiumofequipment manufactuers, isthecurrent standard forcableaccess
technology. DOCSIS defines the functionalities and properties of cable modems at a
subscriber’s premises and cable modem termination systems at the headend. As its
name suggests, DOCSIS specifies the physical layer characteristics, such as trans­mission frequency, bit rate, modulation format, and power levels, of cable modem
and cable modem termination systems, but also the data link layer protocol, such as
frame structure, medium access control, and link security. Three different versions of
DOCSIS (1.0/2.0/3.0) was developed during the past decade and were later ratified
as ITU-T Recommendation J.112, J.122, and J.222. Although some compromise is
needed as cable networks are a shared medium, DOCSIS offers various classes of
service with medium access control. Such QoS features in DOCSIS can support
applications (such as VoIP) that have stringent delay or bandwidth requirements.

Physical Layer

The upstreamPMD layersupports twomodulation formats:QPSK
and 16-QAM, and the downstream PMD layers uses 64-QAM and 256-QAM. The
nominal symbol rate is 0.16, 0.32, 0.64, 1.28, 2.56, or 5.12 Mbaud. Therefore, the
maximum downstream data rate is about 40 Mb/s and the upstream data rate is
about 20 Mb/s. To mitigate the effect of noise and other detrimental channel effects,
Reed–Solomon encoding, transmitter equalizer, and variable interleaving schemes
are commonly used.

Data Link Layer

The DOCSIS data link layer specifies frame structure, MAC, and
link security. The frame structure used in HFC networks isvery similar to the Ethernet
in both the upstream and downstream directions. For the downstream direction, data
frames are embedded in 188-byte MPEG-2 (ITU-T H.222.0) packets with a 4-byte
header followed by 184 bytes of payload. Downstream uses TDM transmission
schemes, synchronous to all modems. In the upstream direction, TDMA or S-CDMA
are defined for medium access control. An upstream packet includes physical layer
overhead, a unique word, MAC overhead, packet payload, and FEC bytes. MAC

OPTICAL ACCESS NETWORKS 11

layer specifications also include modem registration, ranging, bandwidth allocation,
collision detection and contention resolution, error detection, and data recovery. An
access security mechanism in DOCSIS defines a baseline privacy interface, security
system interface, and removable security module interface, to ensure information
security in HFC networks.

1.5 OPTICAL ACCESS NETWORKS

Due to their ultrahigh bandwidth and low attenuation, optical fibers have been widely
deployed for wide area networks and metro area networks. To some extent, mul­timode fibers were also deployed in office buildings for local area networks. Even
though optical fibers are ideal media for high-speed communication systems and net­works, the deployment cost was considered prohibitive in the access area, and copper
wires still dominate in the current marketplace. However, as discussed in Section 1.2,
emerging multimedia applications have created such large bandwidth demands that
copper wire technologies have reached their bandwidth limits. Meanwhile, low-cost
photonic components and passive optical network architecture have made fiber a
very attractive solution. In the past few years, various PON architecture and tech­nologies have been studied by the telecom industry, and a few PON standards have
been approved by ITU-T and IEEE. FTTx becomes a mature technology in direct
competition with copper wires. In fact, large-scale deployment has started in Asia,
North America, and Europe, and millions of subscribers are enjoying the benefit of
PON technologies.

1.5.1 Passive Optical Networks

Figure 1.4 illustrates the architecture of a passive optical network. As the name
implies, there is no active componentbetween the central office andthe user premises.
Active devices exist only in the central office and at user premises. From the central
office, a standardsingle-mode opticalfiber (feeder fiber) runs toa 1 : N passiveoptical
power splitter near the user premises. Theoutput ports of the passive splitterconnects
to the subscribers through individual single-mode fibers (distribution fibers). The
transmission distance in a passive optical networks is limited to 20 km, as specified
in current standards. The fibersand passive components between the centraloffice and
users premises are commonly called an optical distribution network. The number of
users supported by a PON can be anywhere from 2 to 128, depending on thethe power
budget, but typically, 16, 32,or 64.At the central office, anoptical lineterminal (OLT)
transmits downstream data using 1490-nm wavelength, and the broadcasting video is
sent through 1550-nm wavelength. Downstream uses a broadcast and select scheme;
that is, thedownstream data andvideo are broadcast to eachuser with MAC addresses,
and the user selects the data packet–based MAC addresses. At the user end, an
optical network unit(ONU), also called an optical network terminal (ONT), transmits
upstream data at 1310-nm wavelength. To avoidcollision, upstreamtransmission uses
a multiple access protocol (i.e., time-division multiple access) to assign time slots to

12 BROADBAND ACCESS TECHNOLOGIES: AN OVERVIEW

TDM PON Standards
BPON: ITU G.983
GPON: ITU G.984
EPON: IEEE 802.3ah

10 ~ 20 km

ONU

Fiber to the office/business

OLT

Central Office

OLT: Optical Line Terminal
ONU: Optical Network Unit

Coupler

FIGURE 1.4 Passive optical networks.

ONU

Fiber to the node/curb/neighborhood

VDSL, WiFi, etc.

ONU

Fiber to the home/user

each user. This type of passive optical network is called TDM PON. The ONU could
be located in a home, office, a curbside cabinet, or elsewhere. Thus comes the so­called fiber-to-the-home/office/business/neighborhood/curb/user/premises/node, all
of which are commonly referred to as fiber to the x. In the case of fiber-to-the-
neighborhood/curb/node, twisted pairs are typically deployed to connect end users to
the ONUs, thus providing a hybrid fiber/DSL access solution.

1.5.2 PON Standard Development

Early work of passive optical networks started in 1990s, when telecom service
providers and system equipment vendors formed the FSAN (full service access
networks) working group. The common goal of the FSAN group is to develop truly
broadband fiber access networks. Because of the traffic management capabilities
and robust QoS support of ATM (asynchronous transfer mode), the first PON stan­dard, APON, is based on ATM and hence referred to as ATM PON. APON supports

622.08 Mb/s for downstream transmission and 155.52 Mb/s for upstream traffic.
Downstream voice and data traffic is transmitted using 1490-nm wavelength, and
downstream video is transmitted with 1550-nm wavelength. For upstream, user data
are transmitted with 1310-nm wavelength. All the user traffic is encapsulated in
standard ATM cells, which consists of 5-byte control header and 48-byte user data.
APON standard was ratified by ITU-T in 1998 in Recommendation G.983.1. In the
early days, APON was most deployed for business applications (e.g., fiber-to-the­office). However,APON networksare largely substitutedwith higher-bit-rate BPONs
and GPONs.