Book Guide-WDM en
Book Guide-WDM en
Book Guide-WDM en
K1
Y
M
IV
Andr Girard
Ghislain Lvesque
Peter Clark
Gregory W. Schinn
Olivier Plomteux
Don Galbraith
Michael Carlson
Stphane Chabot
Normand Cyr
Stphane Duquet
Daniel Gariepy
Nicholas Gagnon
Stphane Vigot
Andr Fougres
tienne Gagnon
Alain Chandonnet
Managing Editors:
Production Editors:
Copy Editors:
Design and Graphics:
Page Layout:
2008, 2000, 1998 EXFO Electro-Optical Engineering Inc., Quebec City, Canada
3rd edition
2nd edition
1st edition entitled Introduction to WDM Testing
No part of this book may be reproduced in any form or by any means without
permission in writing from EXFO.
VI
Acknowledgements
This book would not have been possible without the enthusiasm and teamwork of
the people who make up EXFO. In particular, the authors would like to thank
Sonia Blanger, Guylaine Bureau, Nathalie Duquette, Guy Fournier, Maryse Imbeault,
Dominique Landry, Claudia Lavoie, Guillaume Lgar, rick Pelletier, lie Prusse,
Jean-Pierre Simard, Pierre Talbot, and Elizabeth Tessier for putting their skills to
work for the Guide.
VII
VIII
TABLE OF CONTENTS
PREFACE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .XV
2 THE BASICS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7
2.1 OPTICAL TRANSMISSION TECHNOLOGIES . . . . . . . . . . . . . . . . . . . . . . . . .8
2.2 TIME-DIVISION MULTIPLEXING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8
2.3 WAVELENGTH-DIVISION MULTIPLEXING . . . . . . . . . . . . . . . . . . . . . . . . .11
2.4 DEFINITION OF A DWDM SYSTEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12
2.5 MAIN COMPONENTS DEFINED . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12
2.5.1 Transmitters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13
2.5.2 Receivers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16
2.5.3 Attenuators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16
2.5.4 Switches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .17
2.5.5 Optical Cross Connect (OXC) Units . . . . . . . . . . . . . . . . . . . . . . . . .17
2.5.6 Wavelength-Dependent Couplers . . . . . . . . . . . . . . . . . . . . . . . . . . .18
2.5.7 Dispersion Compensation Devices . . . . . . . . . . . . . . . . . . . . . . . . . .19
2.5.8 Multiplexers and Demultiplexers . . . . . . . . . . . . . . . . . . . . . . . . . . .19
2.5.9 Optical Add/Drop Multiplexers (OADMs) . . . . . . . . . . . . . . . . . . . .24
2.5.10 Optical Amplifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .24
2.5.11 Optical Fiber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .29
2.6 WDM AND TDM:A COMPARISON . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .31
2.6.1 Link Design Flexibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .31
2.6.2 Speed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .31
IX
TABLE OF CONTENTS
TABLE OF CONTENTS
4.4.1 Multiplexers/Demultiplexers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .68
4.4.2 Switches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .74
4.4.3 Optical Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .74
4.4.4 Receivers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .75
4.4.5 Optical Amplifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .75
4.4.6 Bragg Gratings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .83
4.4.7 Isolators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .83
4.4.8 Fiber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .84
4.5 AUTOMATED TEST SYSTEM FOR COMPONENT TESTING . . . . . . . . . . . . .90
4.6 ENVIRONMENTAL QUALIFICATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .91
4.7 FIELD TESTING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .92
4.7.1 Optical Loss Test Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .92
4.7.2 Optical Time Domain Reflectometer . . . . . . . . . . . . . . . . . . . . . . . .92
4.7.3 Backreflection Meter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .93
4.7.4 PMD Test Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .93
XI
TABLE OF CONTENTS
5.2.5 Architecture Interconnections . . . . . . . . . . . . . . . . . . . . . . . . . . . .107
5.3 COMMISSIONING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .107
5.3.1 Transmitters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .108
5.3.2 Receivers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .108
5.3.3 Optical Amplifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .109
5.3.4 Multiplexers and Demultiplexers . . . . . . . . . . . . . . . . . . . . . . . . . .109
5.3.5 System-Level Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .112
5.3.6 Alarm Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .114
5.4 MAINTENANCE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .115
5.4.1 Optical Signal-to-Noise Ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . .115
5.4.2 Losses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .115
5.4.3 Polarization Mode Dispersion . . . . . . . . . . . . . . . . . . . . . . . . . . . . .116
5.4.4 Chromatic Dispersion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .116
5.4.5 Optical Gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .116
5.4.6 Wavelength . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .116
5.4.7 Crosstalk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .116
5.5 MONITORING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .116
5.5.1 Optical Supervisory Channel . . . . . . . . . . . . . . . . . . . . . . . . . . . . .117
5.5.2 Remote Fiber Test System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .118
5.5.3 Monitoring and Troubleshooting:An Example . . . . . . . . . . . . . . . .120
TABLE OF CONTENTS
GLOSSARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .170
ACRONYM INDEX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .185
FURTHER READING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .189
XIII
TABLE OF CONTENTS
XIV
PREFACE
When a new technology comes along, the process of moving from theory and lab
experience to real-world deployment and practice takes time and hard work.The
people and companies that use new technology often need to go through a period of
learning and assessment to overcome both anticipated and unforeseen challenges.
WDM technology is no exception. Right now, the fiber-optic and telecommunications
industries are in the middle of that period of changing processes and reassessing
needs.And as WDM grows and becomes more widely implemented, more people and
companies than ever need access to practical, yet thorough, information about testing
WDM technology.
This is where EXFO's Guide to WDM Technology and Testing comes in.This book is
aimed at the engineers, technicians, and scientists throughout the telecommunications
industry who want to learn more about the practical aspects of WDM and DWDM
technology. In addition, system and component manufacturers, installers, service
providers, and private network operators will find useful information to inform
corporate decision-makers.Also, newcomers to the field, and people who know about
one aspect of WDM testing and would like to broaden their knowledge, are likely to
find what they need in this guide.
Chapter 1 begins with a basic overview of WDM technology: what it is, what it does,
and the challenges for the present and future. Chapters 2 through 5 cover the
requirements and the fundamentals of WDM testing, as follows:
Chapter 2 introduces definitions and concepts to people who are new to the
field.
Chapter 3 outlines critical WDM parameters according to component.
Chapter 4 addresses the specific testing and qualification needs of component
and fiber manufacturers.
Chapter 5 covers the installation, maintenance, and monitoring of WDM systems.
Chapter 6 talks about standards. It should be noted that standards for WDM
technology are still under development and, thus, subject to change. However,
advanced readers will be interested in the details about standards that this chapter
provides.
With WDM technology moving beyond long-haul links, it is important to outline the
next stage: metropolitan area networks (Chapter 7).The testing needs for this arena
can sometimes differ significantly from those of long-haul networks, so it was felt that
this topic deserved a chapter of its own.
With WDM technology continuing to evolve, Chapter 8 will interest readers who need
to keep an eye on the future: not just advances in technology, but also how testing
needs will change.And finally, Chapter 9 solidifies previous concepts for readers. Using
a case study of a telecommunications service provider deploying WDM on an existing
link, readers can see a real-life example of how testing works.
XV
CHAPTER 1
The TDM approach to increasing network capacity adds channels by subdividing those
that already exist, but at some cost in equipment and signal-protocol complexity.The
time slots of each channel must be carefully protected and identified, so that
individual channels can be recovered.The WDM approach, on the other hand, simply
transmits more signals along the fiber, with no timing or protocol implications.
Of course,TDM techniques are applied to individual WDM channels to provide greater
flexibility in the assignment of network bandwidth to particular users.
CHAPTER 1
1000
5120
2560
1280
640
640
320
160
160
40
320
160
80
100
1280
2004
2000
2002
1998
1996
1992
1994
1990
1988
1986
1982
BANDWIDTH (Gb/S)
10000
1984
1.3 APPLICATIONS
40
10
DWDM has not yet made a great impact on the incumbent local-exchange carrier
(ILEC) market, where laying new fiber is often the most economical way of increasing
bandwidth, although some use in point-to-point applications has been reported.
Products are now entering the market that specifically address the problems of
metropolitan area networks (MANs), products that provide ring and mesh architecture
protection at the optical layer. Products intended to serve the pure-data market
primarily Internet trafficare especially interesting because the reduced level of
protection that is needed for this form of data, compared to voice traffic, could
eliminate much of the equipment performing SONET functions. Direct optical
connections to end-users are also feasible, along with lease-a-wavelength marketing
strategies.
CHAPTER 1
The use of DWDM for data transmission over fiber links, using Internet Protocol (IP), is
another proven application of this technology. Potential reductions of data
transmission costs are estimated at 100% for end users (Internet gratuity).
DWDM technology developments give credence to the feasibility and practicality of
all-optical networks with no electronic signal regeneration at any point.Although
much work remains to be done to ensure that such networks can provide SONET-like
reliability, components are already available to split, combine, and multiplex optical
signals carried by optical fiber. Fixed-wavelength, optical add/drop multiplexers
(OADMs) have reached the market, and dynamically reconfigurable units are expected
in the near future. Optical cross connect (OXC) units that serve as network hubs are
also in development.A similar technology should eventually yield units to switch and
convert individual wavelength channels in a fiber.
1993
1995
1997
1999
2005?
2000
TDM
2002
OADM
OXC
CHAPTER 1
CHAPTER 1
CHAPTER 2
THE BASICS
1
2
1
Time-division
demultiplexer
4X1
Time frame
4
Time-division
demultiplexer
1X4
Figure 2.01 In TDM systems, channels are divided into time-slots propagating at the same
wavelength on the same fiber.
The duration of a time slot depends upon a number of different engineering design
factors and, in particular, upon the transmission speed needed for each link. Each
communication path is assigned a specific time slot, a TDM channel, during which it is
allowed to send data from the source to the user. No other source is permitted to
transmit at the same time.The multiplexer at the source end takes in data from the
sources connected to it, and inserts packets of data from each into the fiber during the
appropriate TDM time slot.The demultiplexer at the other end must recognize the
time slots, unload the data from each, and send it as a continuous stream to the
corresponding user.
The use of TDM techniques has increased the useful capacity of a single optical fiber
channel to more than 10 Gb/s, using technology and design methods that are familiar
to telecommunications system designers; systems at this rate are gradually replacing
early 2.5 Gb/s TDM implementations.The 10 Gb/s rate represents a turning point of
sorts. Below this speed, the governing characteristics of optical fiber have a reasonably
small impact on transmission quality; above it, their effects must be carefully taken
into account.Although TDM systems operating at up to 40 Gb/s may soon be widely
available, and research is underway into 100 Gb/s transmitters, further increases will
CHAPTER 2
THE BASICS
not be easy; existing electronic
techniques become prohibitively
complex and expensive at high
rates. New laser-modulation
methodswill be needed and effects
that can be neglected at low bit
rates will begin to degrade signal
transmission at high rates.
Optical
carrier (OC-x)
Synchronous transport
Line rate
module (STM-x)
(Gb/s)
STM-1
0.15552 (0.156)
OC-12
STM-4
0.62208 (0.625)
OC-48
STM-16
2.48832 (2.5)
OC-192
STM-64
9.9533 (10)
OC-3
STM-256
39.813 (40)
OC-768
Physical phenomena that can
safely be ignored at lower data
Table 2.01 Transmission data rates standardized
rates in short-distance systems
become limiting factors as data rates and transmission distances increase. Chromatic
dispersionarising from the variation of the effective index of refraction of the
optical fiber with wavelengthis one such parameter. Its effect on signal qualityit
spreads optical pulsesis 16 times greater at the current 10 Gb/s (OC-192) data rate
than it is at the 2.5 Gb/s OC-48 rate. Even though measures are available to control
chromatic dispersion, these measures lead to increased system loss, cost, and
complexity. For standard, step-index singlemode fiber (i.e., non-dispersion-shifted fiber
often referred to as G.652 fiber), 10 Gb/s transmission is limited to distances of 50 to
75 km without any compensation or correction.
CHAPTER 2
THE BASICS
Despite these difficulties, higher data rates are coming on the scene.The new 40 Gb/s
(OC-768/STM-256) rate was recently achieved and, even though commercial
implementations are unlikely before late 2001, major players in the industry have
announced successful transmission along links of more than 100 km, with hopes to
raise the limit to much farther distances. So far, trials have not included full SONET
support and these trials lack some of the features that will be needed for reliable
network operation. Results, however, are encouraging, and suggest that upgrading
10 Gb/s links will be possible much sooner than was originally thought possible.
Making appropriate and economic use of the full bandwidth that OC-768/STM-256
offers will be a challenge to network designers and operators, but help is on the way
in the form of an array of products to inject lower-rate tributaries (OC-48/STM-16 and
OC-192/STM-64) into high-capacity networks.The distance limitations expected of
early implementations will probably restrict their use to metropolitan areas, but wider
applications can be expected as distance limitations are eliminated and other limiting
optical phenomena are handled.
Regeneration can also be an
issue where the financial budget
allows it.
1
0
Non-Return to zero coding
1
0
Return to zero (50% cycle) coding
A soliton is a special kind of light pulse that maintains its shape (typically Gaussian) as
it travels over arbitrarily long distances, provided that the pulse is amplified at regular
spacing.The shape or waveform of a soliton allows for compensation of the dispersion
affecting traditional pulse waveforms; this compensation is based on the fact that the
fiber index of refraction has a non-linear term that is dependent upon the pulse peak
power.The soliton uses non-linear and chromatic dispersion effects to conserve its
shape.Another important property of solitons is that they have stable propagation
characteristics and are very robust to perturbations in the transmission path. Even if
this new propagation technique is limited by fiber attenuation, it may be of great help
for amplified, long-haul applications.
Whether TDM technology turns to a universal and independent protocol like IP or
adopts a largely used protocol like SONET or STM, it will remain exploited and
deployed by a significant number of operators for years to come.Advances in soliton
research have provided a second wind to a technology that we considered at its limits.
Therefore, whatever problems occur and whatever the solutions employed in TDM, no
other technology will ever replace it. In a best-case scenario, another technology will
be used side-by-side to increase efficiency:WDM.
10
THE BASICS
CHAPTER 2
11
CHAPTER 2
THE BASICS
1
2
OADM
2
2
The difference lies in the number
3
3
of simultaneous optical signal
3
1
channels carried by the fiber. In a
basic TDM system, there is only
1
one; in a DWDM system, there are
several channels.We can think of a
DWDM system as a number of
parallel TDM systems, sharing fiber Figure 2.03 Typical WDM system with add/drop and
and equipment.
cross connect capabilities
BANK
12
Although suppliers can be relied upon to provide material that has been tested at the
manufacturing plant to meet all specifications, the mere installation of many
components in the field can significantly degrade their performance. Furthermore, the
characteristics of individual components may interact in unexpected ways when they
are assembled in a system.As a result, reliable network services that satisfy given
requirements can be guaranteed only when the system as a whole is shown to meet
specifications, not merely when its components do so in isolation.
CHAPTER 2
THE BASICS
2.5.1 Transmitters
Hybrid modules containing integrated circuits and laser diodes, modulated by separate
integrated circuitry, have now largely replaced early fiber-optic transmitters made up
of discrete electrical and electro-optical devices.Very large scale integrated (VLSI)
circuit implementations have come along to meet the need for ever-higher modulation
rates and to improve reliability. Present-day transmitters, although usually hybrid in
construction, can safely be treated as black box modules at the system design level.
They are electro-optical transducers: sources of signal light whose intensity can be
modulated by digital electrical input signals, either internally (laser current) for low
bit-rate transmissions or externally (laser optical beam) for fast rates.
A transmitter for a single channel (wavelength) is typically made up of a distributed
feedback (DFB) laser (at least 0 dBm or 1mW) followed by a modulator, which is
usually external to the laser especially when modulation rates are high. Recent
developments in integrated optics have led to attractive and cost-effective transmitter
designs in which the laser chip, modulator, and a subsequent solid-state optical
amplifier, are combined in a package.Transmitter packages are now available that
incorporate multi-lasers, multiplexers, and a booster amplifier, which is usually a
semiconductor optical amplifier.
A variable attenuator covering the band of the modulator is sometimes mounted next
to a laser to tailor its power to a specific value.This value is chosen to suit the
characteristics of the first repeater down the transmission line, or when used with
other wavelength transmitters to make sure that all transmitters combine to produce
an equally flat spectral power distribution.
The aforementioned DFB laser is presently the source of choice for DWDM systems. In
this device, the usual two-mirror Fabry-Perot laser cavity is reduced and controlled,
and selecting the exact lasing wavelength through optical feedback is accomplished
by a longitudinal grating manufactured as part of the laser chip (see Figure 2.04).The
effect of the grating is to force the emission of a single, longitudinal lasing mode or
line with a very narrow line widthtypically less than 100 MHz full width half
maximum(FWHM)with sidebands or sidemodes suppressed by at least 40 dB
(see Figure 2.05). A sidemode suppression ratio is used to describe the relative
intensity of the highest sidemode
from the main peak.As with the
Fabry-Perot laser, the waveguide
geometry ensures a stable-oriented,
highly polarized output.
DFB laser
13
Waveguide
Active layer
CHAPTER 2
THE BASICS
14
Transmitter Wavelengths
The provision of interchangeable components for WDM use and the need for
interconnections among networks both lead to a requirement for the definition of a
set of standard frequencies to be used in such systems. Study Group 15 on Transport
Networks, Systems and Equipment in the ITU Telecommunication Standardization
Sector is the international standards body responsible for defining this set (the set is
defined in the G.692 Recommendation on Optical Interfaces for Multichannel Systems
with Optical Amplifiers).
CHAPTER 2
THE BASICS
(GHz)
(nm)
(GHz)
(nm)
(GHz)
(nm)
(GHz)
(nm)
197
197
196
196
196
196
196
196
196
196
196
196
195
195
195
195
195
195
195
195
195
100
000
900
800
700
600
500
400
300
200
100
000
900
800
700
600
500
400
300
200
100
1521.020
1521.792
1522.565
1523.338
1524.113
1524.888
1525.664
1526.441
1527.219
1527.997
1528.776
1529.556
1530.337
1531.118
1531.901
1532.684
1533.468
1534.253
1535.038
1535.825
1536.612
195
194
194
194
194
194
194
194
194
194
194
193
193
193
193
193
193
193
193
193
193
000
900
800
700
600
500
400
300
200
100
000
900
800
700
600
500
400
300
200
100
000
1537.400
1538.189
1538.978
1539.769
1540.560
1541.352
1542.145
1542.939
1543.733
1544.529
1545.325
1546.122
1546.920
1547.718
1548.518
1549.318
1550.119
1550.921
1551.724
1552.527
1553.332
192
192
192
192
192
192
192
192
192
192
191
191
191
191
191
191
191
191
191
191
191
900
800
700
600
500
400
300
200
100
000
900
800
700
600
500
400
300
200
100
000
900
1554.137
1554.943
1555.750
1556.558
1557.366
1558.176
1558.986
1559.797
1560.609
1561.422
1562.236
1563.050
1563.865
1564.682
1565.499
1566.317
1567.135
1567.955
1568.776
1569.597
1570.419
190
190
190
190
190
190
190
190
190
189
189
189
189
189
189
189
189
189
189
188
188
800
700
600
500
400
300
200
100
000
900
800
700
600
500
400
300
200
100
000
900
800
1571.242
1572.066
1572.891
1573.717
1574.543
1575.370
1576.199
1577.028
1577.858
1578.689
1579.521
1580.353
1581.187
1582.021
1582.856
1583.693
1584.530
1585.368
1586.206
1585.045
1587.885
(GHz) (nm)
188
188
188
188
188
188
188
188
187
187
187
187
187
187
187
187
187
187
186
186
186
700
600
500
400
300
200
100
000
900
800
700
600
500
400
300
200
100
000
900
800
700
1588.726
1589.568
1590.411
1591.255
1592.100
1592.945
1593.792
1594.639
1595.487
1596.337
1597.187
1598.038
1598.889
1599.742
1600.596
1601.451
1602.306
1603.163
1604.020
1604.878
1605.737
Cavity
Input
WAVELENGTH LOCKER
Output
15
CHAPTER 2
THE BASICS
2.5.2 Receivers
The receiver converts optical signals to an electronic form by detecting whatever
modulation has been applied to the light signal and demodulating it. Of course, the
receiver must be completely compatible with the transmitterboth its primary
wavelength and its modulation characteristicsand it must be designed to cope with
whatever signal degradations other network components may have introduced.
Traditional optical techniques are used to couple light energy from a fiber to a
detector, generally a photodiode.The resulting signal must be amplified
electronicallyadding as little noise as possiblewithin an electrical bandpass
appropriate to the expected signal. Electronic filtering may also be needed to flatten
the effective frequency response of the unit.All these operations are usually carried
out within a single hybrid module, including a receiver module whose input is light
from a fiber and whose output is a clean electrical signal that must be demodulated
appropriately.The complexity of the demodulation process depends on the
modulation used. For example, timing information must be extracted if the channel
uses TDM, and various error detection and recovery schemes may be required at this
level.
Two types of photodiodes are commonly used: positive-intrinsic-negative (PIN) diodes
and avalanche photodiodes (APD).The PIN device operates with standard low-voltage,
logic power supplies (5 V), but it is less sensitive and covers a narrower bandwidth
than the avalanche model. High-speed PINs were used in 10 Gb/s and 40 Gb/s
applications before the arrival of APDs.The latter is found mainly in long-haul
applications, where its higher cost and greater circuit complexity is justified. In many
cases, the use of an APD alone allows the user to forego the need for an optical
preamplifier with a PIN photodetector. It is, therefore, economically very justifiable.
Critical selection parameters for receivers include spectral response (A/W as a
function of wavelength, especially in relation to the detector used), sensitivity
(a measure of the level at which intrinsic detector noise masks the incoming signal),
both spectral and electronic bandwidth, dynamic range, and noise.Appropriate criteria
for each depend on the individual application. For example, noise characteristics are
more important than high output power in an optical preamplifier used just before a
channel receiver. Furthermore, the
need for optical filteringsuch as
2.5.3 Attenuators
16
dB
dB
THE BASICS
CHAPTER 2
dB
Attenuation
High-power lasers can be used in network transmitters to reduce the need for in-line
amplification.Attenuation may be needed in specific portions of the network to keep
optical components from being exposed to radiation strong enough that their
behavior becomes non-linear. Power tailoring is also often necessary to trim the gain
of EDFAs to provide an overall network response that is spectrally flat.This is
especially critical when a channel is added or dropped in EDFAs.
0 dB
Wavelength
2.5.4 Switches
Early switching in fiber-optic networks involved converting the optical signal to an
electronic one, performing the necessary switching operation, and reconverting the
signal to opticala cumbersome and expensive process that involves switching speed
limitations and reduces WDM operability.
Switches are used in WDM networks to divert signals to alternate routes in response
to network difficulties, or to reroute signals to other networks.As both network
complexity and reliability expectations have grown, the need for just a few switches
to offer simple rerouting possibilities has rapidly increased to a requirement for
complex N-by-N cross-connect units that permit extensive, all-optical reconfiguration.
The ability to switch signals easily is critical for the survival of the modern all-optical
network.As a result, optical cross-connect (OXC) routers are now essential to this
survival.
Standard optical switches used until now have tented to be 1xN switches, where an
electrically-controlled mechanism moves a flexible input fiber from-one output fiber
to another. In network applications, these switches have normally omly been used for
restoration purpose and not for dynamic bandwidth allotment or redistribution.
17
CHAPTER 2
THE BASICS
Pure optical solutions under study include waveguides, liquid crystals, or even bubble
technology, whose characteristics can be quickly modified by heat, an electrical signal,
or optical gates (individual optical amplifiers that can be turned on and off rapidly).
The ever-increasing need for fast, reliable, and inexpensive optical switching
equipment will drive research and development efforts for the foreseeable future.
Output
Output
OXC
Figure 2.09 Signal i coming to the input port of the OXC is rerouted to other ports.
In such a device, the wavelengths at each output depend on the wavelengths at the
input.A control port can be added to allow selective wavelength routing.
18
CHAPTER 2
THE BASICS
2.5.7 Dispersion Compensation Devices
Power
(dB)
Input pulse
DCD
Output pulse
0
Time
Figure 2.10 The broadening effect due to dispersion and its compensation
19
CHAPTER 2
THE BASICS
Multiplexing Technologies
Although a thorough discussion of
the mechanisms underlying
mux/demux operations is beyond
our scope, we will outline the
wavelength-dividing mechanisms
used in the more common types.
Thin-Film Filters
I1
Thin-film filters consist of a
number of layers of transparent
1
dielectric materials of differing
refractive indices deposited
sequentially on an optical
3
substrate. Since the index of
2
refraction changes at each
I3
I2
interface in the stack of layers, a
portion of the incident light is
2 = 1
reflected at each interface.This
reflected light interferes
constructively or destructively
with incoming light, depending
on wavelength.Through judicious Figure 2.12 Incident wavelength 1 is reflected 2 at the
same angle, and 3 is transmitted at a lower angle.
selection of the indices
of refraction and the thickness of
each layer, one can use this internal interference to produce a filter that will pass any
desired range of wavelengths and reflect the others (Figure 2.12).
20
The techniques for choosing design parameters and applying the coating have been
known in the optical industry for decades.The selection of coating materials is
limited. Many substances, whose optical properties are desirable, have physical
properties that are less than ideal. Generally, the more stringent the requirements, the
greater the number of optical layers needed. Despite these difficulties, filters with
many highly specialized spectral response functions can be produced inexpensively
without much variation during production.
CHAPTER 2
THE BASICS
2
used and, thus, alters, the bandpass
of the filter, so the tilt must be
Figure 2.13 Example of a cascaded thin-film filter for
factored into the design of the
multiwavelength demultiplexing
unit. In multi-stage units for
multichannel use, the beam
reflected from each filter is the
input for the next stage; thus, alignment is extremely critical.
Thin-film filters provide bandpasses narrow enough for use in mux/demux
applications in WDM systems of 16 or 32 channels. More densely spaced networks are
now turning to other technologies.
ADD
1, 2 n
Fiber Bragg grating
Circulator
21
CHAPTER 2
THE BASICS
Bragg gratings are seldom used alone as passive DWDM components. One common
application involves an add/drop multiplexer with one or two optical circulators, as
shown in Figure 2.14. On the drop side, the circulator recovers the reflected
wavelength and sends it to the drop port. On the add side, the circulator recombines a
new transmission signal at the same wavelength as that at which it was dropped.
Therefore, a specific wavelength channel is either added to or dropped from the
original transmitted signal.This technique is often used at the interface between a
long-haul and a metropolitan network, where the former typically uses as many
wavelengths as possible and the latter uses far fewer wavelengths.
Fiber Bragg grating devices are also beginning to appear in mux and demux devices in
Mach-Zehnder configurations and in hybrid combinations with other types of filters.
Aside from mux/demux applications, this narrowband filtering technology is also used
to perform EDFA gain flattening, wavelength stabilization, and wavelength locking.
Diffraction Gratings
A bulk-optic diffraction grating reflects light at an angle proportional to wavelength
and, once again, the underlying physical principle is the same: constructive and
destructive interference.
For each wavelength of incident
light, there is an angle for which
light waves reflecting from
individual grating lines will differ
in phase by exactly one
wavelength.At this angle, the
contribution from each line will
add constructively, so this will be
the angle of maximum
transmission for that specific
incident wavelength.
1
2
Incident light
Short wavelengths
Reflected light
Designing a mux or a demux
using a diffraction grating, is a
Long wavelengths
matter of positioning the input
and output optics to select the
Figure 2.16 The incident light is divided into all
desired wavelengths.Although
primary wavelengths 1, 2,n
they are difficult to manufacture
and expensive, devices based on diffraction gratings have an insertion loss that is
essentially independent of the number of channels, rendering this technology one of
the more promising for high channel count systems. However, polarization control
requires critical attention.
22
CHAPTER 2
THE BASICS
Input waveguides
11 12 13
21 22 23
31 32 33
Output waveguides
11 21 31
12 22 32
13 23 33
An AWG can be used either to reassign wavelengths on one set of fibers to a different
distribution on another set, or to demultiplex individual frequencies on a single fiber
(Figure 2.16).This is becoming a key technology for manufacturers of wavelength
multiplexers/demultiplexers. Since its architecture is easily scalable, it holds the
potential for applications in systems with hundreds of channels.
An AWG is also called a Dragone router, a phase array, or a phasar.
23
CHAPTER 2
THE BASICS
Arrays of these devices, sometimes in conjunction with Bragg gratings written in the
tapered section, can be used to extract individual frequencies from multichannel WDM
or DWDM systems, or to add new channels in the middle of networks. Since they are
completely passive and exhibit low losses, the use of large arrays is economically
feasible.
1 2 n
1
2
n
network, one may not want to
combine or disassemble the entire
channel structure, but merely add
a single channel to it or extract a
single channel from it with an
Inserted
Dropped
channel
channel
add/drop multiplexer.An optical
x
x
add/drop multiplexer performs
this operation without the need to Figure 2.19 Typical representation of on OADM device.
convert the signals in all the
channels to electrical form and
back again.
The technologies mentioned are all applicable to the construction of OADMs.Today,
devices are on the market to handle many commonly needed tasks including adding,
dropping, or replacing OC-48 channels at a point in the network without disturbing
the other channels carried in the same fiber.These devices provide major benefits for
both network reliability and security. Future developments will certainly concentrate
on making OADMs reconfigurable in the field, perhaps eventually remotely.
24
A traditional electronic repeater restores the level of signals over a long link by
extracting signal information from the fiber, converting it to electrical pulses,
amplifying them, and reinserting them into the following fiber section. In contrast, an
EDFA is transparent; that is, it is oblivious to signal protocol, format, bit rate, and
(within its limitations) to the wavelength of the optical signal.
CHAPTER 2
THE BASICS
Since the EDFA is insensitive toor unaware ofnetwork protocols, it can be directly
connected to varied equipment such as asynchronous transfer mode (ATM) switches
and Internet Protocol (IP) components without danger of interfering with their
functions.This flexibility is one of the major advantages of DWDM usage.The use of
EDFAs in a network, however, involves several new concerns, especially the
appropriate treatment of their non-uniform spectral gain and of the amplified
spontaneous emission (ASE) noise that they add.
EDFA-equipped networks offer great advantages, such as the ability to make low-cost
incremental upgrades, thus increasing capacity one channel at a time to meet demand.
All-optical networks are feasible, since electronic signal-processing components are
needed only at the points where the information is first inserted into the network and
finally retrieved. Start-up costs for DWDM systems can be quite low, since most
existing OC-48 network facilities can readily be incorporated into DWDM systems by
treating each OC-48 path as an individual wavelength channel.
Another example of the contribution of optical amplification is in CATV applications
where the user wishes to deliver (broadcast) a common signal to as many
subscribers as possible. Since direct fiber to the home (FTTH) is still rarely employed
today, the CATV signal is generally delivered to a local distribution point close to a
group of houses, from where the final connections are made via coaxial cable. Most
CATV signals are analog and they require higher signal-to-noise ratios at the receiver
end than digital signals; optical amplification can accomplish that task and is essential
if onethe user is trying to maximize the number of subscribers sharing the signal.
FTTH will become economically feasible only if subscribers can be equipped with
low-cost, limited-sensitivity receivers, so the highest power possible must be delivered
to network end points.With hundreds or thousands of subscribers competing for this
signal power, the need for transparent, easily distributed optical amplification is
obvious.
New research into high-power EDFA pumping schemes has led to an extended
rangeL-band or extended-bandunit operating from 1570 nm to 1605 nm. It is
often referred to as a long-wavelength EDFA (LWEDFA).
Pump Lasers
A critical component of an EDFA is the pump laser, the source of energy that the
amplifier adds to the signal. Pump energy is distributed among all optical channels in
the EDFA.Therefore, as channel counts increase, so does the need for higher pumping
power. EDFAs that can handle many channels frequently are designed to use more
than one pump laser.
Input
s
s + p
p
WDM
Output
s
Figure 2.20 Simplified view of an EDFA device with isolators, pump laser (p), and WDM
coupler
25
CHAPTER 2
THE BASICS
Both 980 nm and 1480 nm lasers are suitable for EDFA pumping because both
wavelengths correspond to the energy levels of excited ions and, therefore, are well
absorbed by the erbium-doped fiber. Naturally, trade-offs must be considered. High
channel count systems and pre-amplifiers can benefit from pumping at 980 nm
because these lasers provide the EDFA with lower noise figures than the 1480 nm
models. However, higher power at a lower cost is available at 1480 nm.The choice is
complicated by the fact that the pump laser must be selected at the beginning of a
network's life cycle, before the ultimate channel count can be known, so an
appropriate power-noise trade-off can be difficult to weigh. Some EDFAs are pumped
at both wavelengths to take advantage of both pump wavelengths.
If a laser transmitter transmits a signal of +16 dBm along a fiber having a typical
0.2 dB/km attenuation in the 1550 nm region, after 80 km the signal will have been
attenuated to 0 dBm (or 1 mW), ignoring any other causes of loss such as splices.
On the other hand, if the same laser emits 0 dBm, the signal will be down to 16 dBm
after 80 km along the same fiber. In the early stage of fiber-optic telecommunications,
lasers were able to emit relatively low power, and the signal had to be electronically
regenerated after distances much shorter than the 80 km range.These repeaters
received the optical signal, converted it to an electronic one, amplified it, and
reconverted it into an optical form.Although this technique was not spectrally
limitedit could easily regenerate signals at both 1310 nm and 1550 nmit was quite
complex and did not allow for
future upgrades of the system bit
rate without replacing the
repeaters.
In the late 1980s, Payne and
Laming of the University of
Southampton in the UK proposed
the use of erbium-doped fiber to
amplify signals optically.The age of
all-optical repeaters had arrived.
However, the technology came
with one small restriction: it
worked only in the 1550 nm
spectral region. Hence, erbium
only provides gain over a narrow
Figure 2.21 EDFA optical spectrum
wavelength band centered around
this region.
a) s
c) s
26
s + p
WDM
s
p
WDM
d) s
b)
s + p
WDM
s + p
s
WDM
s + p s + p
WDM
DCD
WDM
CHAPTER 2
A: Pre-amplifier region
B: In-line amplifier region
C: Booster amplifier region
Gain
P output
THE BASICS
P input
27
CHAPTER 2
THE BASICS
Various techniques have been proposed to flatten the gain and expand the DWDM
amplification window to 40 nm or more. Gain flattening is also necessary to eliminate
the distorted amplification of signals through cascaded, in-line EDFAs. Devices, such as
the LWEDFA described earlier, stretch the gain window to about 1610 nm, providing
the capability for bidirectional DWDM transmission in different non-interfering
windows, and also potentially relaxing the requirements for more densely packed
DWDM transmissions in a narrower window.
dBm
10
5
0
-5
-10
-15
-20
-25
-30
-35
a)
-40
1530
1535 1540
10
dBm
5
0
-5
-10
-15
-20
-25
-30
b)
-35
-40
1530
1535 1540
28
THE BASICS
CHAPTER 2
29
CHAPTER 2
THE BASICS
Although this standard fiber (G.652) exhibits zero chromatic dispersion at 1310 nm, it
shows high dispersion at 1550 nm (18 ps/nm.km). Despite this apparent
incompatibility with the EDFA window at 1550 nm, recent tests indicate that this fiber
can carry moderate-rate DWDM transmissions without loss of signal quality over
significant distances.This is largely due to the fact that, although dispersion is high at
1550 nm, it can be compensated with dispersion compensating fiber or other
compensation devices.
Dispersion (ps/nm.km)
30
THE BASICS
CHAPTER 2
2.6.2 Speed
The capacity increase that TDM offers results from pumping more information bits
along a link. How fast this can be done dependswithin some fundamental fiber-cable
limitson the electronic components used. Digital circuitry is required to accept data
from each source, to store it, to load it into appropriate time slots, to unload it, and to
deliver it to the correct user.All these digital components must operate at, or close to,
the speed of the composite link of the multiplexer; that is, each channel, regardless of
its net bandwidth, must be equipped with at least electronic equipment capable of
handling the overall bandwidth of the link.A fiber-optic cable transmission medium can
handle rates of several Gb/s, while the logic speeds of commercially available digital
electronics are of the order of 1 billion operations per second.Although electronic
speeds are bound to rise,TDM will forever be limited by the economics of needing
leading-edge electronics in each signal path, so the technique will unlikely provide a
composite link speed commensurate with the tremendous bandwidth presented by
fiber-optic cable.This limitation applies to the WAN as well as the premise
environment.
Although individual WDM channels are subject to the same end-of-link requirements for
supporting electronics as TDM channels, other equipment in the channel only need to
deal with its own low data rate, not the rate of the composite signal.The overall
channel capacity is not constrained by the speed of the supporting electronics.
Capacity can be increased at any time simply by adding more channels on demand.The
fastest TDM link design, using the most advanced electronics, can always be transmitted
over a single WDM link as one of many channels.WDM can provide composite link
speeds that are in line with the enormous bandwidth offered by fiber-optic cable.
31
CHAPTER 2
THE BASICS
In this chapter, we have presented the main building blocks of a DWDM system and we
have pointed out how TDM and DWDM technologies are being combined to provide
dramatic bandwidth upgrades.All these technological advances in a DWDM system
generate new testing methods to ensure that each component and set of components
work together to deliver expected performance. In the next chapter, we will describe
these critical testing parameters, which will be followed by a detailed discussion of
testing methods in Chapter 4.
32
CHAPTER 3
Power
The arrival of DWDM, as that of any other technology, introduces new difficulties as
well as powerful advantages.The main concern for users of the new DWDM system is
its reliability and stability over time.Testing optical quality parameters and system
behavior from component manufacturing up to the system integrator is an important
means of ensuring that a DWDM system will be commissioned with the right
specifications and will deliver robust performance for many years.
Laser power
Fiber attenuation
Component losses
Laser modulation
Fiber non-linearity
RIN
BER
Fiber PMD
Modal dispersion
Jitter
Transmission rate
Time
34
EDFA ASE
EDFA gain
Crosstalk
Four-wave mixing
Raman
Power
Laser modulation
Fiber non-linearity
RIN
BER
PDL
Laser power
Fiber attenuation
Component losses
Brillouin
effect
Wavelength
Time
Fiber chirp
Jitter
Transmission rate
PMD
stability
EDFA range
& BW
Laser chirp
Chromatic dispersion
Stability of opt. freq.
Phase noise
-dependent PMD
CHAPTER 3
On the wavelength axis, we are now faced with such critical elements as spectral
stability, EDFA spectral range, central wavelength, and bandwidth.At the corner of
wavelength and time, we encounter laser chirp, chromatic dispersion, stability of the
optical frequency, and phase noise (self-phase modulation and cross-phase
modulation).At the corner of wavelength and power, we find EDFA amplified
spontaneous emission (ASE), EDFA gain, crosstalk, four-wave mixing, and stimulated
Raman forward scattering.Where all three axes meet, we encounter stimulated
Brillouin backscattering.Although WDM technology makes networks more efficient by
adding bandwidth and channels, its effective application requires care. In the planning,
designing, manufacturing, and implementation stages, these factors must be considered
and appropriately countered.
n
a)
b)
We will describe, here, the most important parameters and phenomena to be tested.
These parameters are related to the fiber, optical components, effects of nonlinearities, effects of fiber dispersion and active components on the system.
CHAPTER 3
Channel
-3 dB
Centre
Wavelength
Channel
-3 dB
Centre
Wavelength
36
CHAPTER 3
The channel transmitter will be operating near nominal wavelength, the wavelength at
which the device is designed to operate, usually one of the standard ITU wavelengths.
So the central wavelength will have to be as close as possible to this frequency such
as that of the channel ITU wavelength.Cutoff wavelengths, upper or lower, are the
wavelengths at which the insertion loss reaches a specified value, usually 3 dB.
2. Channel spacing (which is intended to match the network standard channel grid)
In existing network systems, both evenly spaced and unevenly spaced channels are
used. Evenly spaced channels are most commonly in accordance with the ITU grid,
spaced at 100 GHz intervals.
Uneven channel spacing is used to minimize and predict non-linear effects such as
four-wave mixing. Four-wave mixing arises when two or more wavelengths interact to
generate new wavelengths.With even channel spacing, these new wavelengths fall on
top of existing channels and create crosstalk.With uneven spacing, four-wave mixing
causes noise between channels.
3. Bandwidth at -3 dB and at some
other appropriate value
(typically -0.5 dB, 20 dB or lower)
Bandwidth is the spectral width
over which the transmission (or, in
the case of reflective devices such
as Bragg gratings, the reflectance)
exceeds some stated value (-3 dB
from maximum or full width half
maximumFWHMfor example).
It is meaningless to cite bandwidth
without specifying a threshold level.
Bandwidth defines the spectral
range over which the device can be
used effectively.
Figure 3.06 Bandwidth measurement of a wide signal
37
CHAPTER 3
Some manufacturers use a figure of merit (FOM) to describe the shape of a filter. It is
defined as the ratio of the bandpass bandwidth to the rejection-band bandwidth. For
an ideal square-spectral-response filter, the ratio would be one. In general, a high-order
filter will have sharper transitions and thus a higher FOM.
4. Isolation and crosstalk (the amount of energy from one channel that appears in an
adjacent channel)
Broadly speaking, channel isolation and crosstalk describe the rejection of signals from
an adjacent or other channel in a multichannel device.The measurement takes the
bandpass characteristics of each channel into account and is normally specified under
worst-case conditions. Isolation and crosstalk have slightly different interpretations.
Whereas channel isolation describes the rejection of signal power from or to another
channel, crosstalk describes the power leaking through a channel band from other
channels. Isolation is the minimum value in dBm at which the DUT eliminates an
outside signal, while crosstalk is the difference in dB between the maximum input
power value and the minimum leaking power (isolation).
Figure 3.08 shows the loss spectra
for three channels: A, B, and C.The
operating wavelength limits for
channel B are also shown; the
channel-B transmitter may operate at
any wavelength in the range
indicated. Measuring isolation
determines how much channel-B
power will leak into channel A.
Placing the loss spectra of the two
channels over each other, we can
identify this value and express it in
dB. Normally, we will do so for the
worst possible case, in this case at
the band edge (a typical situation for
thin-film devices).With the type of
Figure 3.08 Loss spectra of a 3- device
device shown above, the isolation
between any two non-adjacent
channels is very high, and non-adjacent-channel crosstalk can be neglected.This may
not be so for some technologies such as arrayed waveguide gratings (AWGs).
38
CHAPTER 3
39
CHAPTER 3
Loss, bandwidth, and central frequency are particularly sensitive to the state of
polarization. Each must be measured at different states of polarization, using a
polarization controller.The variation in results is the polarization dependence of the
parameter in question. Ideally, measurements should be performed for all polarization
states, but a random selection usually suffices.The following sections describe the
three principal measurements.
Pulse dispersion
Propagation axis
PMD = Temporal delay
Slow axis
PMD
40
Detector
power
Signal response
CHAPTER 3
The most useful single measure of IL is the value at the wavelength where the loss is
greatest. Using this value, the designer can safely calculate a loss budget that will apply
at any transmission wavelength. In fact, this is the way most component manufacturers
specify channel insertion loss. Note once again, however, that the ITU center is not
necessarily the center of the bandpass.
This method is applicable when the channel width is known, but some components
are intended for general applications where the exact operating conditions cannot be
determined in advance. In such situations, the best way to describe the IL of the
device is to calculate the central wavelength of the bandpass and to determine the IL
at that wavelength.This method is often used to characterize single-channel devices or
those with symmetrical bandpass.
There is not yet an accepted standard as to how insertion loss should be defined in
the presence of PDL. One very reasonable definition has been proposed by standards
committees. First, it is necessary to measure insertion loss with a depolarized source
according to the formula above; the PDL will be the (max-to-min) variation about this
value.Alternatively, the insertion loss may be defined as the best-case loss as the input
state of polarization into the device under test (DUT) of a fully polarized source
(e.g., laser) is adjusted.The worst-case loss would then be the sum of the IL and PDL.
41
CHAPTER 3
Important factors to consider when comparing the IL values for different devices
include the effect of connectors and the uniformity of the IL across channels in a
multichannel unit.
3.2.4 Directivity
Directivity, which is sometimes
referred to as near-end crosstalk, is
a measure of the isolation between
the inputs ports of a multi-input
device (Figure 3.15). It is
particularly important in a
multiplexer, where power returned
to the system transmitters must be
kept to a minimum.
The users must provide a nonreflective termination on the
output side of the device when
measuring directivity. If not,
reflections at the fiber end-face
may degrade the measurement.
2
Non-reflective
termination
3
Reflected light can be re-reflected in the forward direction. Such forwardpropagating reflections lag the original signal, causing problems at the demodulation
stage.This phenomenon is termed multipath interference (MPI).
Reflections occurring inside an EDFA optical amplifierespecially if the gain
medium lies between themcan lead to additional multipath interference and can
contribute significantly to amplifier noise .
All things being equal, a component with a 55 dB reflectance would be preferable to
one with a reflectance of 50 dB.
CHAPTER 3
43
CHAPTER 3
Dispersion (ps/nm.km)
The chromatic dispersion of a fiber link is cumulative with distance, and is stated as
the change in group delay per unit wavelength, in ps/nm.The delay coefficient is the
dispersion for a particular fiber type and is generally quoted in units of ps/(nm.km).
S0
slope at zero dispersion
0
0
dispersion zero wavelength
Wavelength (nm)
material
dispersion given
dispersion-unshifted
waveguide
dispersion-shifted
waveguide
-1
-2
1100
1200
1300
1400
1500
1600
Wavelength (nm)
dispersion compensation
Chromatic dispersion requires more attention in WDM systems using G.652 fibers,
since the dispersion is large in the 1550 nm region.
44
Dispersion (ps/nm.km)
CHAPTER 3
Figure 3.19 PMD is affected by physical stress, temperature, and fiber imperfections.
Polarization mode dispersion is the primary mechanism through which these flaws
affect performance.
At any point along a fiber, a polarized light pulse can be decomposed into
components aligned with two local, orthogonal axes of the fiber: a fast axis and a slow
axis. It should be noted that these axes do not necessarily correspond to a linear state
of polarization.In the real world of cabled fiber, the orientation of these axes, and the
relative difference in propagation speed corresponding to each axis (directly related
to the magnitude of the local birefringence) changes along the optical path. In a
somewhat idealized scenario, different sections of the fiber will have different
orientations of these local birefringence axes. (The change in orientation of the local
principal axes is known as a mode-coupling event.) In each segment, a time delay will
be introduced between that portion of the light aligned with the local fast versus the
local slow birefringence axes. Since the relative orientation of these axes in adjoining
segments is different, the pulse will experience a statistical spreading over time.
For a particular wavelength, the input state of polarization of a light pulse launched
into the fiber can be adjusted so that the pulse undergoes no spreading (at least for a
measurement time sufficiently short that environmental perturbations can be
neglected). In fact, two such mutually orthogonal input states of polarization exist.
These are known as the input principal states of polarization, one of which
corresponds to the fastest and the other to the slowest pulse propagation time
through the fiber.The difference in these two propagation times is known as the
differential group delay (DGD) corresponding to that wavelength, and the PMD is
defined as the wavelength-averaged value of the DGD.
45
CHAPTER 3
Because the individual factors that cause PMD cannot be measured or even observed
in isolation, the phenomenon must be viewed as a constantly changing, unstable
stochastic process.This process results in a broadening of the pulses carrying
information, which can impair the ability of a receiver to decode them correctly. PMD
is thus a critical phenomenon that limits the transmission rate.
BIT
1
BIT
0
3>2
1
2>1
fast axis
2
3
slow axis
Figure 3.20 Transmission bits (0,1) get wider along the fiber so that 0 and 1 are undetectable.
PMDTOT = (N (PMDN)2)1/2
46
CHAPTER 3
ITU has proposed the criteria for PMD outlined in Figure 3.24 as guidelines to
maintain acceptable receiver error rates.
Proposed PMD coefficient for a 99.994%
probability that the power penalty will be less
than 1 dB for 0.1 of the bit period
Bit rate
(Gb/s)
Maximum PMD
(ps)
PMD coefficient
400 km fiber (ps/km
2.5
10
20
40
40
10
5
2.5
2.0
0.5
0.25
0.125
1/2
47
CHAPTER 3
5
4
10
3
5
2
2.5 Gb/s
1
0
As noted earlier, PMD affects
0
10
20
30
40
50
60
70
80
individual components, not just
Differential group delay (ps)
fiber. However, in the case of
component, PMD is generally
Figure 3.25 PMD-induced sensitivity penalty
constant, not quasi-random, and
can be reduced through quality
control during the manufacturing stage.
3.3.4 Non-Linearity
Non-linearity in fiber optics has similar effects
as non-linearity in other physical systems, be
they mechanical or electronic. It causes the
generation of spurious harmonic and sum and
difference frequencies.These added signals
cause unexpected loss effects in optical
communications networks.
48
Figure 3.26 Second-order PMD is the
variation of the polarization dispersion
vector orientation and its magnitude
(i.e., DGD) as a function of wavelength.
CHAPTER 3
Fiber non-linearity becomes noticeable when the laser signal intensity (the power per
unit area) reaches a threshold value.Also, non-linear effects generally become evident
after signals have passed through a length of fiber, depending on the characteristics of
its construction and the operating conditions under which it is placed.
Non-linear region
Linear region
where MFD is the mode field diameter. For G.653 dispersion-shifted fibers and G.655
non-zero dispersion-shifted fibers,Aeff is approximately equal to 50 to 60 m2, while
G.652 dispersion-unshifted fibers have Aeff approximately equal to 80 m2.The
effective fiber length Leff has the same effect as Aeff. For typical singlemode fibers, Leff is
typically equal to 20 km.
Non-linear effects fall into two categories, depending on the behavior of the non-linear
coefficient .These are scattering phenomena (when g is a real number and gives rise
to gain or loss) and refractive index phenomena (when g is an imaginary number and
gives rise to phase modulation).
1. With scattering phenomena, the laser signal is scattered by sound waves
(acoustic phonons) or fiber molecular vibrations (optical phonons) and is shifted
to longer wavelengths.The following are two different types of scattering
phenomena:
a)
b)
49
CHAPTER 3
2. With refractive index phenomena, the signal power is high enough that the
refractive index may no longer be assumed to be a constant, but instead is
approximated by:
n = n0 + n2I
where n0 is the fiber linear refractive index, I is the signal intensity, and n2 the nonlinear index (about 2 to 3 x 10-16 cm2/W for silica fiber).The following are the different
refractive index phenomena:
self-phase modulation, or the effect a signal has on its own phase
b)
c)
Power
a)
Wavelength
Power
For G.653 fibers at 1525 nm, for instance, the backscattered signal is downshifted by
about 10.7 GHz (+0.085 nm) with a bandwidth of about 60 MHz. For G.652 fibers in
the same window, the backscattered signal is downshifted by about 11 GHz (+0.088
nm) with a bandwidth of about 30 MHz.As a rule of thumb, SBS should be considered
as a potential problem if
monochromatic light of more than
about 6 dBm is launched into the fiber.
A number of techniques have been
developed to suppress SBS in real
systems.The most popular involves a
rapid (~50 kHz) dithering of the carrier
wavelength over a range of about 1
GHz, much greater than the 30 to 60
MHz SBS bandwidth.
50
CHAPTER 3
Self-Phase Modulation
When the intensity of the laser signal becomes too high, the signal can modulate its
own phase.This modulation broadens the signal spectrum and temporally broadens or
compresses the signal, depending on the sign (positive or negative) of the chromatic
dispersion it sees.A shift to short wavelengths occurs at the trailing edge of the signal
and a shift to long wavelengths at the leading edge.
In WDM systems, the spectral broadening created by self-phase modulation in a signal
channel can interfere with adjacent signals.
Self-phase modulation increases with
an increase in injected channel power in a fixed fiber with fixed effective area
an increase in the channel bit rate (higher bit rates have faster rise-and-fall bit
slopes)
negative chromatic dispersion
In WDM systems, self-phase modulation is more of an issue for G.652 fibers (zero
dispersion at 1310 nm) than for G.653 dispersion-shifted fibers at 1550 nm and G.655
non-zero dispersion-shifted fibers. It is not significantly influenced by a decrease in
channel spacing or by an increase in the number of channels.
Self-phase modulation decreases with
a zero or small positive value of chromatic dispersion
an increase in fiber effective area
dispersion compensation
Cross-Phase Modulation
In this case, the signal in one channel modulates the phase of an adjacent channel.
Cross-phase modulation is sensitive to the same factors as is self-phase modulation, as
well as to an increase in the number of channels. It is not significantly influenced by a
decrease in channel spacing, as is the case for self-phase modulation, but will decrease
with
an increase of the fiber effective area
dispersion compensation
Cross-phase modulation is less important in WDM systems using fibers with large
effective areas.
51
CHAPTER 3
Four-Wave Mixing
Four-wave mixing is one of the most disruptive non-linear effects in WDM systems.
When the intensity of the laser signal reaches a critical level, ghost signals appear,
some of which may fall within the true channels.The number of these ghost channels
is given by:
N2(N 1) / 2
Power
Power
w1
2w1-w2
w2
Wavelength
w1 w2 w3
Wavelength
2w2-w1
Figure 3.30 Four-wave mixing creates unwanted signals within the transmission spectral
range.
CHAPTER 3
Gain (dB)
40
35
P small-signal gain
30
25
20
15
10
5
0
-4
-2
10
12
14 16
Output (dBm)
P compression
P saturation
53
CHAPTER 3
input signals (typically >3 dBm in the case of an EDFA), the amplifier is in deep gain
saturation, and the output power corresponding to an input power level where the
gain 1 is termed the saturated output power.
In general, the values of these three key amplifier gain parameters, that is, the smallsignal gain, 3 dB gain compression, and saturated output power, vary with input signal
wavelength. Consequently, users wishing to fully characterize amplifier performance
will require a test system that performs measurements at different optical input
powers for a number of wavelengths of interest.
To measure saturated gain for DWDM applications, the amplifier is saturated at one
particular channel wavelength and the gain measurements are taken at other channel
wavelengths, using an input signal. Probing at all wavelengths over the operating
spectrum reveals the spectral sensitivity of this parameter.The measurement can also
be bidirectional, to provide so-called reverse saturated small-signal gain.
Gain Cross Saturation
Gain cross saturation is the change in the gain of one specific channel when the input
power in another channel (or several channels) is changed by a specific amount.
Gain Flatness
Gain flatness is the maximum difference among individual channel gains when their
input powers are equal.
Polarization Dependent Gain
Polarization dependent gain (PDG) characterizes the difference in gain for different
input polarization states. It is usually determined by scanning or sampling all possible
input polarizations and noting the lowest and highest gain values.
Signal Gain
The signal gain of an EDFA is the gain that applies to a large signal such as the traffic
signal, which, for a power amplifier application, normally drives the amplifier into
saturation.This is the main factor determining the operating point of the amplifier. In
contrast, noise gain is the gain that applies to a small signal that has one with little or
no impact on the amplifiers operating point, while another large signal is driving the
amplifier into saturation (Figure 3.32).
Profile
The term profile is often used to refer to the wavelength-dependence of a particular
property. For instance, the noise gain can be a simple value (expressed in dB) for one
wavelength, while the noise gain profile describes how the parameter varies with
wavelength.
Channel Gain
Channel gain is the gain of one particular signal in a DWDM system. In general, it is
different for different wavelengths or channels. Spectral flatness (or gain flatness)
describes the change in gain with wavelength, i.e., the slope of the gain profile, usually
in dB/nm. Gain tilt describes how this property changes as individual channel
wavelengths are added or removed.
54
CHAPTER 3
N3
Pump
(980 nm or
1480 nm)
N2
Stimulated
emission
(1550 nm)
R21
R12
N1
Figure 3.33
3-Level
population level system of a
stimulated
emission
typical EDFA
and absorption
Power
-20
Interpolated
output
noise Nout ()
Z Z+Z
Forward
ASE
Backward ASE
Optical
pump
Spontaneous emission
-10
As an optically-pumped amplifying
medium, erbium-doped silica has
what is known as a three-level
structure (Figure 3.33), where the
upper level in the 1550 nm band
transition is metastable and can give
rise to spontaneous and stimulated
emission.
Measured
noise values
Out
-30
ASE
-40
-50
-60
In
-70
o
1526
1528
1530
1532
1534
1536
1538
Wavelength
55
CHAPTER 3
56
Signal-spontaneous beat noise, generally termed ASE beat noise, arises from signal
heterodyne mixing with essentially white ASE noise. Since the beat frequency of the
signal with the surrounding ASE noise is within the frequency bandwidth of the
modulated information carrier, it cannot be removed by optically or electrically
filtering the signal. Measuring it is therefore important.The noise figure (NF)
specification of most EDFAs is a measurement of this effect.
CHAPTER 3
all expressed in dB, where SNRin is the signal-tonoise ratio at the amplifier input and SNRout is
the signal-to-noise ratio at its output. It is
assumed that narrowband filtering is used at
the receiver input.
With a 980 nm pump and a complete erbiumion population inversion, the ideal NFthe
smallest achievable noise figure for high-gain
amplifieris about 3 dB.
DFB
LED
FP
ASE
3.5 TRANSMITTERS
Figure 3.38 An LED, FP laser, and a DFB
Dense WDM systems owe their bandwidth
laser have very different bandwidths and
advantage to the use of multiple channels at
spectral shapes.
different wavelengths.All of these wavelengths
must fit into the operating band of the EDFAs. If there are many channels, each must be very
tightly controlled.The characteristics of the light source for each channel determine the
success of this control. Even in low channel count WDM systems, compatibility with
adjoining networks, potential for upgrading, and adherence to international standards all
imply tight control of transmitter characteristics.The following paragraphs discuss the major
requirements.
57
CHAPTER 3
The linewidth of the laser should be narrow to allow close DWDM channel spacing.
Narrow linewidth minimizes pulse distortion caused by fiber dispersion and allows
narrowband filters to be used in front of the receiver to improve the signal-to-noise
ratio.Also, very low residual sidebands are required to ensure that no source output
interferes with any other channel.
Operation in a single longitudinal mode is necessary to minimize the generation of
beat noise.The line broadening (chirp) tendency of lasers directly modulated by the
drive current makes them unsuitable for dense WDM systems. External modulators
that are physically integrated with the source are preferred.
Peak power, peak wavelength, spectral width, and susceptibility to chirp must stay
within acceptable bounds both in the short- and long-term, as the device ages. Shortterm drifts due to temperature and other ambient influences must also be sufficiently
low. If not, means must be provided to detect and counteract them. Laser sources must
be suitably protected if they are subject to backreflection, which can cause them to
become unstable.
Laser modules themselves are expensive, and their replacement can involve very
costly and complex mechanical disassembly and re-alignment of components that are
critical to the effective operation of a link.A long, stable life is evidently a prime
requirement.
Energy not converted to light results in heat that affects the characteristics
(wavelength and power) of the laser and creates instability.Thermoelectric coolers are
generally used to maintain the laser sources at the desired temperature.
The transmitter must remain within the system channel bandwidth and should not
drift in wavelength over time.The typical spectral drift in DFB lasers is less than
0.1 nm/oC and 0.01 nm/mA; the sensitivity to case temperature is typically
0.002 nm/oC and the aging drift should typically be no greater than 0.001 nm/year.
The output power should be stable over time and side modes should be suppressed to
better than 40 dB below the peak output.The laser is optically isolated and should not
be affected by spurious reflections from the transmission medium, especially those
coming from the first in-line EDFA.
The modulator should provide minimal laser chirp, ideally less than its modulation
bandwidth (in other words, the line broadening should only be due to the modulation
of the signal itself). Its insertion loss should be as low as possible.
We will take an individual look at the main transmitter components: the optical source
used and the modulator.
58
The source must be able to generate enough optical power to meet the bit error rate
(BER) requirement in the particular application, and its output must be easily
modulated at the required rate. Linearity can be important if output power is to be
varied or when analog modulation is performed.
Although both the light emitting diode (LED) and the laser diode (LD) meet these
requirements to a useful extent, the power and bandwidth advantages of the LD make
it the source of choice in demanding WDM systems.
CHAPTER 3
Sources to be used in a 2.5 Gb/s system can, in general, be directly modulated by their
current. Sources to be used at higher transmission rates require separate modulators to
minimize chirp. In either case, appropriate means must be provided to efficiently
couple the output energy to the next component, the modulator, and to the
transmission line. Heat-dissipating measures, i.e., thermometers, are usually required.
3.6 RECEIVERS
BER
The function of the receiver is to provide the demodulator with the cleanest electrical
signal it can extract from the
optical signal it receives.
Receiver performance is
measured by the bit error rate
(BER) it delivers.The result, for a
given received signal, depends, in
10-9
1e-09
turn, on the receivers sensitivity,
its bandwidth, and any noise it
adds to the signal before
demodulation.
The overall performance of a
receiver is described by its
sensitivity curve, which plots the
14.00
16.00
BER as a function of optical
Attenuation (dB)
power received for a given data
Figure 3.39 Typical BER screen result with the
rate.
important 10-9 mark.
Such a curve applies to a particular set of operating conditions and, thus, incorporates
the effects of bandwidth, detector noise, and demodulation techniques.
Mechanical and environmental factors must also be considered, including size, weight,
power needs, and possible temperature sensitivity (especially for avalanche
photodiodes), as well as ease of servicing and replacement.
59
CHAPTER 3
Dispersion
Fiber
Compensator
3.8 SWITCHES
Switches are used in the network to rearrange the links.They allow signal re-routing
and are used for configuring a path or restoring a link.They are also used in
conjunction with optical add/drop muxs.The key parameters that determine the
performance of switches and, thus, their suitability for particular network applications,
are
insertion and coupling loss
return loss
PDL
crosstalk and isolation
reliability
switching time
stability
60
CHAPTER 3
3.9 ATTENUATORS
The main application of attenuators in WDM systems is to tailor the power in each
channel so that the power delivered to the first in-line EDFA is spectrally flat.
Consequently, flatness over a channel width is a key performance parameter. Other
parameters, whose relative importance depends upon the specific system
applications, include
stability
reliability
ORL
PDL
PMD
accuracy
repeatability
insertion loss
3.10 ISOLATORS
Isolators are used in situations where backscattered light or the light reflected from an
interface can degrade the performance of a sensitive component such as a DFB laser.
The following are the critical parameters that determine the performance of isolators:
Wavelength-dependence, especially for so-called narrowband isolators that are
designed to operate in a spectral range narrower than 20 nm. Isolators are
described both by a peak reverse-direction attenuation figure and by the bandwidth
for which the isolation is within 3 dB of the peak value.
Small insertion loss <1 dB in the forward direction, but typically in excess of 35 dB
(single stage) or 60 dB (double stage) in the reverse direction, and low polarization
dependence
Polarization mode dispersion (PMD). Isolators are constructed using highlybirefringent elements, and are very prone to PMD (typically 50 to 100 fs), especially
for single-stage designs. Double-stage isolators can be designed so that the PMD
induced by the first stage is largely canceled by the second stage.
Polarization dependent loss (PDL), which also degrades the performance of optical
isolators
N/=on-PMD multipath interference dispersion (an additional dispersion effect
caused by spurious reflections in the device or unbalanced optical paths)
People involved with DWDM technology have very direct questions regarding DWDM
testing.They need to know what to test, how to test, and where to test. In this chapter,
we have answered the first question.The second one will be addressed in the next
chapter, where test methods for component characterization are described.The last
question is dealt with later on as part of installation and commissioning.
61
CHAPTER 3
62
CHAPTER 4
4.1 OVERVIEW
A whole new family of sub-systems and components is needed to permit and facilitate
the development, manufacture, and deployment of systems using dense WDM
technology.As is the case for any evolving technology, new terminology, specifications,
and measurement techniques have been developed to describe each of these
elements.This section describes the key measurement and testing techniques used to
characterize the major network elements of a dense WDM system.
The components of a DWDM network include transmitters (including lasers and
modulators), receivers (including filters and detectors), transponders, optical
amplifiers (including boosters, in-lines, and pre-amplifiers), multiplexers,
demultiplexers, optical add/drop multiplexers, routers and switches, fibers, cables, and
compensators.As dense WDM technology moves to closer and closer wavelength
spacing, the requirements and performance specifications for wavelength-selective
components become
increasingly demanding and
Broadband source
test procedures become
Sources
P
correspondingly complex.
White source
P
DUT
Tunable laser
Detector
64
Broadband sources are required in a measurement, and these should be reasonably flat
over their spectral range to minimize corrections. Non-coherent broadband sources,
including near-black-body sources such as high-intensity incandescent lamps, LEDs,
and amplified spontaneous emission (ASE) sources, are available covering all the
spectral ranges of interest both for WDM and more traditional components. Because
their output is non-polarized, or nearly so, measurements made will minimize any
polarization dependencies.
CHAPTER 4
ASE sources provide high-intensity light over wide wavelength bands, but they must
be selected to display a spectral power distribution as flat and as uniform as possible
over the wavelength range of interest.
Requirements for narrowband sources are usually met by external cavity lasers (ECLs),
whose precise lasing wavelength can be adjusted by mechanical means with
accuracies of a few picometers over a spectral range exceeding 120 nm. Such sources
are strongly polarized and very nearly monochromatic, making them suitable for
demanding wavelength-sensitive measurement situations. Spontaneous emission can
be controlled with appropriate filtering, and their noise performance can be adequate
for low insertion loss measurements. New tunable laser designs are now available,
including the fiber laser, which combines the advantages of the EDFA and narrow
tunable filtering technologies. It offers very low baseline noise compared to the ECL
(lower than 65 dB).
Tunable laser
80
ASE source
60
Broadband (LED) source
40
White source
20
0
700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800
(nm)
Figure 4.02 Spectral emitting region and dynamic range of different type of sources
Figure 4.02 indicates the spectral regions in which each of the source types discussed
is particularly applicable.
65
CHAPTER 4
Fixed mirror
Movable
mirror
Input
Detector
Detector
Input
Rotating
grating
Fixed mirror
Input
Fixed
grating
Fixed
grating
Movable
detectors
66
Fixed detectors
Figure 4.03-c The fixed grating divides the input into primary
elements before reaching the detectors.
CHAPTER 4
67
CHAPTER 4
and its ability to scan entire wavelength bands simultaneously make it a very useful
complement to the OSA.The wavelength meter, designed with its own internal
reference, is more likely to maintain calibration with heavy use.
Source
4.4.1 Multiplexers/Demultiplexers
Multiplexers and demultiplexers are important components of DWDM systems, and
their bandwidth, crosstalk, insertion loss, return loss, isolation, and polarization
properties are critical to network performance.The desirable spectral attributes of
several other DWDM components are very similar to those needed in multiplexers and
demultiplexers (or, more particularly, in individual mux/demux channels), and test
procedures for these devices are included in this section.
Switch
Source
Nx1
OSA
68
Figure 4.05 The ASE source shoots through the mux
while the OSA gets the signal for each output port.
Insertion Loss
Figure 4.05 shows a typical setup
to determine the insertion loss of
a DWDM mux/demux.A flat,
broadband ASE source is used,
whose output covers the DWDM
band.The OSA output will,
therefore, directly indicate the
shape of the insertion loss.A
suitable reference measurement,
made by connecting the source
directly to the OSA, can be used to
convert this relative loss curve to
an absolute one. In this case, the
spectral resolution of the
measurement is limited by the
OSA RBW and its shape.
The 1xN switch is included to ease or even automate the testing of multichannel
devices.All measurement components must have low polarization state sensitivity.
CHAPTER 4
Polarized
LED
source
OSA
Polarization
controller
69
CHAPTER 4
/4
Waveplate
/2
Waveplate
Input light
Output light
= 0
Polarized
LED
source
= 0
Horizontally polarized
(0)
= 0
Power
meter
Polarizer
= 0
= 0
= 225
Diagonally polarized
(45)
= 0
= 45
= 0
Circularly polarized
Figure 4.08 Three different waveplates are used to create different polarization states.
Four different input polarization states are needed to determine the four parameters
of the Mueller matrix.The polarization controller (Figure 4.08) generates those four
polarization states.After a reference measurement between the polarizer and the
detector, the DUT is inserted after the polarizer.The power is measured by the
detector with respect to every polarization state, and then the Mueller element is
calculated with the transmission coefficient (T). Since the Mueller element is related
to the DUT maximum and minimum transmission coefficient, the PDL is obtained
with:
PDL (dB) = -10 log Tmin
Tmax
Polarized
LED
source
1xN
switch
OSA
Polarization
controller
70
CHAPTER 4
Crosstalk
Crosstalk between DWDM
channels is measured by replacing
the ASE source used in the
Tunable
1xN
OSA
a) source
previous setups with a narrowlinewidth, tunable laser source.
Repeated OSA scans are
performed on all of the
multiplexer or demultiplexer
channels while stepping the
tunable source through the
wavelength region of interest at
N
b) Tunable
Power
the desired resolution.The
source
meter
measurement resolution is
determined by the tunable-laser
steps, not by the OSA, but loss
curves with a resolution as high as
0.001 nm can be obtained with
Figure 4.12 Crosstalk can be measured using a) a
excellent dynamic range.The
tunable laser with on OSA or b) a tunable laser with
procedure
is time-consuming,
power meters.
however, especially for
multichannel DUTs, unless a multichannel power meter is also available.A wavelength
meter should be used for calibration.
Optical Return Loss
Optical return loss (ORL) is measured using the combination of a source, coupler, and
photodetector, often referred to as an optical continuous wave reflectometer (OCWR).
After a calibration step in which a component with a known reflectance is substituted
for the DUT, the DUT is inserted.The detector then measures the power it reflects, a
correction factor based on the calibrating step is applied, and the ORL is displayed.
With a high-power, non-coherent optical source and a high-sensitivity, high-resolution
detection system, return losses of 70 dB or lower can be detected and measured with
this configuration.
71
CHAPTER 4
ASE
Broadband
coupler
ASE
source
OSA
OSA
Bandwidth
Optical bandwidth is measured
using a procedure very similar to
that used to measure insertion
loss.The spectral characteristics of
the measurement system can be
removed with a reference
measurement.A typical setup is
illustrated in Figure 4.14.
72
CHAPTER 4
73
CHAPTER 4
4.4.2 Switches
Optical switches are tested using a source and a
multichannel power meter, along with a suitable
switch controller (see the configuration shown in
Figure 4.17).
Switching speed is often a characteristic of
considerable interest, so a high sampling rate and
rapid stabilization are essential qualities in the
power meter. If tests are to be automated to any
extent, the power meter must be able to
recognize, and be triggered by, the electrical
signals that drive the switch.Appropriate
read-out facilities are also needed for both
switching rates and interchannel crosstalk.
N
Power
meter
Source
TTL
controller
The loss, crosstalk, and transient behaviors of switching elements can be critical in
network applications, so the instrumentation used to characterize them must offer
very high dynamic ranges at very high measurement rates.
74
2. The same trigger starts sampling with the power meter at the input channels. If the
rate of change of the drive current is constant, output power can be accurately
correlated with input current throughout the scan.With high-speed power meters,
many thousands of data points can be acquired in a second, enough to provide
adequate resolution in the threshold region, even in a short overall acquisition
period.
3. After the output-versus-drive-current characteristics (also called L-I curve) have
been obtained, the controller can fix the current of each device at a specific value,
slightly above the threshold, that represents a typical operating power level.The
wavelength meter can then be used to accurately measure the wavelength.
CHAPTER 4
4.4.4 Receivers
Linearity testing is needed for semiconductor photodetectors, which saturate as the
input power increases. Biasing the p-i-n junction improves this behavior, but increases
dark noise, which thus limits the measurement of very low power levels.Various
measures taken to increase dynamic range (trans-impedance pre-amplifiers, automatic
gain control circuits, or automatically selected fixed amplification steps), all tend to
increase non-linear behavior, especially at very low signal levels or if discretely
stepped amplification schemes are used and are inadequately calibrated.
Source
dB
dB
dB
Shutter
Receiver
75
CHAPTER 4
76
CHAPTER 4
ASE
source
a)
Source
ORL
meter
1x2
coupler
b)
OSA
Source
WDM
attenuator
OSA
N
Figure 4.23 The OSA measures the spectral gain
redistribution of the EDFA as a function of attenuated
input signals.
77
CHAPTER 4
This type of measurement lends itself well to automation, and test-set modules are
available to perform the appropriate measurements and analyze their results.
Backreflections are yet another potential disturbing influence on EDFAs.
Every active device in a network is subject to backreflections: power reflected
unintentionally from components farther along the link.This reflected energy may
affect such parameters as continuous power stability, wavelength stability, spectral
broadening, and multi-path interference.Although backreflection affects EDFAs in
particular, tests similar to the following may occasionally be called for to characterize
other devices.
The setup in Figure 4.24 tests the sensitivity of the power and wavelength
characteristics of devices to backreflection.
High-speed
power
meter
Variable
back
reflector
MWM
Figure 4.24 The high-speed power meter detects light reflected back to the EDFA.
By setting the variable back reflector to the maximum level likely in real life (the
reflectance of a glass-air interface: 4% or 13.9 dB), one can check for fluctuations in
the power and wavelength characteristics of the DUT, and thus set confidence limits
on its isolation behavior in actual use.The optical signal-to-noise ratio of a typical
wavelength meter is adequate to detect the presence of secondary peaks or
wavelength shifts of 1 pm.
The noise figure of an EDFA determines such cost-critical network characteristics as
the input power requirements and allowable repeater spacing.The accurate
measurement of noise figure is complicated by the fact that the major contributor to
noise, amplified spontaneous emission (ASE), varies with the signal level, so simple
methods of measuring noise figure that are adequate in most electronic applications
measuring the output in the absence of an input signalcannot be used.There are
four common ways to overcome this difficulty: three optical and one electrical.
The optical methods are interpolation, polarization nulling, and time-domain
extinction.
Interpolation involves measuring the amplifier output over a broad spectral range,
with a signal present.The ASE within the signal bandwidth is estimated by
interpolating output levels just outside it (Figure 4.25).
78
Pamp
CHAPTER 4
^
^
Psignal + PASE (s)
Optical
power
BW
s
Figure 4.25 Interpolation method for EDFA noise
figure determination
(s) = Ptot
G x P
where Pamp is the true amplifier ASE at the signal wavelength s, Ptot
(measured) ASE, G is the amplifier gain, and P is the source ASE.
is the total
This simple and inexpensive method is fast and cost-effective, but it is sensitive to
spontaneous emission in the source used.
Signal
Optical
power
Optical
power
ASE + GSSE
2
ASE
2
dB
Attenuator
Source
Polarizer
OSA
Polarization
controller
Polarizer
Figure 4.26 Polarization nulling method for EDFA noise figure determination
79
CHAPTER 4
The polarization nulling method is generally not sensitive to SSE and is particularly
useful in detailed ASE studies. However, it is difficult to automate and it may be
affected by PMD.
The third optical method, timedomain extinction, is based on the
fact that the mechanisms
governing changes in carrier
population in an EDFA are
relatively slow, and there is a
period of a millisecond or so after
a signal disappears before the
corresponding ASE generation
ceases.Thus, the ASE in a channel
bandwidth can be observed by
looking at the power output in
that channel immediately after the
channel input signal is cut off
(see Figure 4.27).
)
)
80
Shotnoise
limited
source
Calibrated
high-speed
detector
ESA
RF
Low-noise RF
amplifier
Power reference
Power reference
RIN (dB/Hz)
CHAPTER 4
NF (f)
100 kHz
1 MHz
10 MHz
100 MHz
1 GHz
The RIN method is particularly suited to detect multipath interference (MPI) effects
produced by reflections due to discontinuities or misalignment in the device; effects
that are not detected by the optical methods discussed. But RIN subtraction is
expensive to perform, as it requires very low noise instrumentation and extensive
calibration.
81
CHAPTER 4
Polarization Effects
A PMD analyzer based on an
interferometric technique can
Polarized
PMD
measure the inherent PDM of a
LED
analyzer
source
non-active EDFA module, as long
as the device does not contain
narrowband filters.A polarized
LED source is used, at a
Figure 4.31 Setup used when measuring PMD with
wavelength well above the strong an interferometry-based PMD analyzer
1531 nm absorption band of the
erbium (Figure 4.31). Such interferometric methods will also reveal multipath noise
interference effects, including those caused by spurious Fabry-Perot etalons resulting
from faulty components or fiber splices.
Strong MPI noise effects can be detected with a simple combination of a high-speed
photodiode detector and an oscilloscope. MPI will appear as a sinusoidal signal (with
harmonics) superimposed on the background noise.
Results can be improved by replacing the oscilloscope with an electrical spectrum
analyzer, but even so, valid quantitative measurements require careful calibration of the
entire measurement system across a wide RF frequency range.
Although an EDFA does not exhibit appreciable MPI noise in normal use, it may do so
if it is exposed to high levels of reflected signal power at its output.A variable
backreflector is used to test for this effect, and a well-isolated EDFA should remain
stable at backreflection levels approaching 100% (0 dB).
Tunable
source
Polarization
WDM
PMD
PMD
attenuator
Analyzer
Analyzer controller
PDL
meter
Figure 4.32 Measurement of the wavelengthdependent polarization dependent gain of the EDFA
Wavelength-dependent
polarization dependent gain
(PDG) is measured with a PDL test
set, an appropriate sourcea
selection of fixed models or a
tunable laserand a polarization
controller (Figure 4.32).An
attenuator is often needed to
maintain power levels typical of
operational networks.
Bidirectional Testing
EDFAs are increasingly used bidirectionally: amplifying signals simultaneously in both
directions on a link.All of the tests described can be performed in both directions,
often in quick succession using a switching arrangement like that shown in
Figure 4.33.
Source
2x2
switch
82
OSA
CHAPTER 4
Laser
source
Coupler
OSA
Typical setup loss per channel would be around 7-8 dB.The insertion loss measured
would be twice the one-pass value.
4.4.7 Isolators
Optical isolators are needed throughout fiber networks to block backreflections or
back-generated ASE, either of which can disrupt signal integrity. Both insertion loss
and, especially, return loss must be determined throughout the wavelength band.
These measurements may be carried out using either a tunable laser and a power
meter, or an ASE broadband source and an OSA.
Source
Polarization
controller
PDL
meter
Polarized
LED
source
PMD
analyzer
83
CHAPTER 4
Tunable
source
POL
PMD
PMD
controller
Analyzer
Analyzer
PDL
meter
1x2
switch
4.4.8 Fiber
Chromatic Dispersion
Chromatic dispersion is one of the
most fundamental properties
affecting the performance of fiber
in a communications link.
Although it can be measured in a
variety of ways, a method
particularly suited to the
investigation of behavior in the
1550 nm band is illustrated in
Figure 4.38.
MWM
Tunable
source
Fiber
PMD
PMD
Modulator
Analyzer
Analyzer
Receiver
RF
analyzer
84
Tunable
source
Wavelength-Scanning Methods
Both of the wavelength-scanning
techniques extrema counting
and Fourier transformare
performed using either of the two
setups shown in Figure 4.39.
Broadband
source
CHAPTER 4
Detector
Fiber
P
Fiber
OSA
The two measurements differ in the treatment of the output information. In the first,
the extrema over a given wavelength range are counted and related to PMD
mathematically. In the second, a Fourier transform (the square of the second moment
of the Gaussian fit) is applied to the wavelength scan to extract the PMD parameter.
85
CHAPTER 4
Interferometric Methods
PMD can be measured using a broadband, polarized source and analyzing the
transmitted radiation with a Michelson interferometer, as shown in Figure 4.41.
Fixed mirror
Source
Fiber
Input
Movable
mirror
or
Polarized
LED
source
Detector
Figure 4.41 PMD measured using a polarized signal and the interferometer
86
Source polarization is assured over the full spectral range by a separate polarizer,
which may be included in the source module.The detector must have adequate signalto-noise ratio and dynamic range to cover the measurement conditions, possibly with
some manual intervention.The fiber under measurement should be treated
appropriately: supported in a strain-free manner and held steady.Temperature control
is not usually needed, as the measurement can be made quite quickly.
A test-sample measurement is usually preceded by a source measurementdirectly
coupling the polarized source to the interferometer to check that it provides a single,
smooth, nearly symmetric autocorrelation peak.A real measurement can then be
made on a PMD emulatoreither fixed or variable.This device is deliberately
configured to prove a fixed amount or range of PMD.
CHAPTER 4
The fiber itself can then be measured, generally for a number of different input
polarization states (some of which will couple better to the principal axes of the test
fiber). If results are needed over a range of temperatures, steps of at least 5C are
recommended to produce significant variation.
PMD can be calculated from the resulting interferogram in two ways.The one to use
will depend on the level of mode coupling in the test fiber. If this coupling is
negligible, as will usually be the
case for a short fiber or one
designed to maintain polarization,
the interferogram will show a
strong central peak that represents
the autocorrelation of the source,
and two satellite peaks
symmetrically displaced from it by
the group delay of the fiber
(Figure 4.42).Thus, the PMD value is
this displacement (or half the total
displacement between the two
satellites) converted to time.The
figure quoted is usually normalized
to a unit length of fiber by dividing
by the sample length, giving units of Figure 4.42 Weak mode coupling PMD trace
ps/m or ps/km.
Appreciable random mode coupling
in the test fiber will wipe out the
satellite peaks, and the PMD must be
determined by analysis.The second
moment (the rms half-width) of the
autocorrelation function is observed
(Figure 4.43):
<2> = (3/4)*
87
CHAPTER 4
Polarimetric Methods
The PMD measurement techniques outlined so far are sufficient for many practical
applications, in particular to characterize fibers exhibiting strong, random mode
coupling with at least moderate levels of PMD.The coverage of all ranges of PMD in
fibers exhibiting arbitrary levels of coupling requires a much more detailed look at the
underlying polarimetric parameters that describe the medium.Two methods are
currently known: the Jones Matrix Eigen analysis (JME) and the Poincar Sphere
Analysis (PSA). Both methods involve injecting a specified set of polarized signals into
the fiber or other DUT and fully characterizing the polarization state of the output
signals to obtain the corresponding (normalized) Stokes vectors as a function of the
optical frequency.
In the JME method, all the input signals are linearly polarized, whereas in the original
PSA technique, one of them must be circularly polarized. However, the PSA method,
recently revised by EXFO has been shown to require only the same three linearly
polarized input signals as the JME. In fact, it is now recognized that the two methods
are fundamentally equivalent and they are in the process of being consolidated as a
recommended test method at the international level. Indeed, both methods determine
the DGD as a function of wavelength from the same set of raw data (output Stokes
vectors).
88
The JME method is commonly implemented using the tunable laser source configuration,
whereas the PSA method (as implemented by EXFO) is based on a combination of a
broadband source and an interferometer.A linear polarization state adjuster, the fiber or
component under test, and a four-port polarimeter complete the setup.
This setup record the Stokes parameters as a function of wavelength. From the Stokes
parameters, the output state of polarization (i.e., the output Stokes vectors) as well as
the degree of polarization can be calculated as a function of wavelength.
CHAPTER 4
DGD (ps)
89
CHAPTER 4
Polarization state adjuster (PSA):The PSA is used to produce four known states-ofpolarization (SOPs).A measurement scan is performed at each state and based on
the attenuation values for each SOP, and the PDL vs. wavelength is calculated in
accordance with the Mueller Matrix algorithms.The time required for a PDL
measurement is the time necessary to perform four sweeps.
Depolarizer unit:This component is used to condition the tunable laser signal when
PDL measurements are being performed. It could also be used for attenuation
measurements where unpolarized light conditions are required.
CHAPTER 4
The system software stores data in a separate database for each device type using a
database with commercially available software.These databases can be stored on each
local machine or can be connected to a LAN for storage on any network drive.Test
data can also be exported in a delimited .TXT format, which is then compatible with
many different storage and analysis programs.This automated testing system uses a
must-have software that offers component and system manufacturers the chance to
add reliability, repeatability, and speed to their testing procedure whether the tests are
performed in a laboratory, a central office, or an environmental room.
91
CHAPTER 4
Modern optical time domain reflectometers (OTDRs) often offer capabilities in the
fourth window region, at 1625 nm. In addition to the ability to test and troubleshoot
the important 1625 nm optical supervisory channel, using this wavelength presents
other important advantages. In
particular, in many circumstances,
live fibers may be tested at the
14
1625 nm wavelength while
normal dense WDM transmission
12
continues uninterrupted in the
10
EDFA spectral region. In addition,
8
because optical losses due to fiber
6
bending are more pronounced at
4
1625 nm than at the shorter
2
DWDM operational wavelengths,
0
OTDR testing at the long
-2
wavelength can reveal critical
1300
1350
1400
1450
1500
1550
1600
1650
1700
Wavelength (nm)
points in the installed fiber, that is,
points where the performance of
the fiber is acceptable at the time
of installation but could degrade
Figure 4.45 Comparison of bending loss for different
over time (Figure 4.45).
fiber types as a function of wavelength
92
CHAPTER 4
Power
93
CHAPTER 4
where PMD1+2 is the total PMD of the optical link, which is composed of links 1 and 2.
If the maximum allowable PMD for a link length of say, 40 km, is 2.5 ps (0.4 ps/km1/2);
and if two 40 km links are tested together, one with 1.5 ps (0.24 ps/km1/2) PMD
and the other with 3 ps (0.47 ps/km1/2), the result for 80 km will be 3.35 ps
(0.37 ps/km1/2). Both links will be declared good, although one of them clearly is not.
Similarly, if one link has a PMD of 2 ps (0.32 ps/km1/2) and the other, 4 ps
(0.63ps/km1/2), an 80-km result of 4.47 ps (0.50 ps/km1/2) will be obtained, indicating
that both links are bad; another incorrect conclusion. Care must be taken to test the
links in the configuration in which they are used.
There is no doubt that PMD will become critical in high-speed communication
networks like SONET.Although manufactured cable may be measured for PMD in the
plant, cabling and installation will often introduce mechanical bends and pressure
points that will alter its value.With the increasing use of EDFAs to increase fiber span,
all sources of dispersion must be controlled and minimized, otherwise the gains these
devices offer will be compromised by PMD effects.
The interferometric technique is best suited to measurements of PMD in the field. It is
fast, it requires no hardware communication or loopback, its wide dynamic range
allows large PMD values to be measured over long distances, and it is insensitive to
fiber vibration.
Manufacturing and qualifying DWDM components require new test instruments and
techniques.The most important ones have just been described. Nevertheless, rapid
developments in technology will certainly increase the number of new measurements
and methods.Also, some of these measurements will become more relevant as
technology evolves (i.e., PDL).These specialized techniques and equipment are rapidly
evolving from the quality-control laboratories and manufacturing verification
environment to the field-testing arena.The next chapter will deal with key testing
issues for system verification before final exploitation of the system begins.
94
CHAPTER 5
The rapid evolution of WDM technology and the need to quickly apply it in the field
have led to a requirement for new and advanced test procedures.WDM technology
has the potential to dramatically increase the capacity of existing networks, reduce the
cost of network upgrades by increasing the bandwidth of existing links, and offer new
options for all-optical network design.These goals can be met, but only by confirming
the correct operation of all the network components through extensive testing during
installation, and by checking ongoing performance through appropriate maintenance
procedures.
In the previous chapter, we discussed different WDM technology components. In this
chapter, we will focus on what links all these components, i.e., the WDM system.
Measuring the optical and electronic parameters of the link is vital to ensure that each
channel is operating at the specified wavelength and that all network elements are
spectrally aligned according to the design specifications. Obviously, once all critical
components are installed, they must be tested again to ensure that they still conform
to the system providers requirements.They will have to be tested before the system is
put into operation (commissioning) and on a regular basis after that (maintenance).
After the network is in service, operation, administration, and maintenance tests are
still necessary to confirm the proper working of network operations and all
management functions. Critical parameters must be monitored to ensure that they do
not stray from design limits.These system parameters are particular to the system used
and depend on several factors, including the length of the optical link, transmission
rates used, number of wavelengths, and channel spacing.
This chapter, thus, is relevant for system operators, installers, maintenance teams, and
all other personnel who want to know more about test systems.
96
CHAPTER 5
5. EDFA gain (Both amplitude and spectral width; small signal gains of 30 to 40 dB are
typical, while spectral widths of 40 nm are typical in non-extended-range EDFA.)
6. Receiver gain (as high and as stable as possible)
7. Fiber type and specifications (critical for data rate, dispersion, and maximum
channel possibility)
Even though it is possible to characterize system performance with the previous
parameters, many more considerations have to be taken into account to carry out a
complete study of the system performance.
Source
Receiver
Transmitter
User
Fiber-optic cable
Stimulus
Receiver
Pattern
generator
Detector
Fiber-optic link under test
Pattern, clock
To determine whether the tentative link design can deliver this power, the installer
must analyze it, i.e., step through the components of the link to determine just how
much power will reach the receiver.This is done with a fiber-optic data-link power
budget.
97
CHAPTER 5
ELEMENT VALUE
COMMENT
3 dBm
-5 dB
-1 dB
Transmitter to fiber-optic
cable with ST connector.
Value accounts for
misalignment.
Splice loss
-0.25 dB
Mechanical splice
-20 dB
-1 dB
Assumes an ST connector
and includes misalignment
-27.25 dB
-24.25 dBm
Total loss
Optical power delivered to the
receiver
Receiver sensitivity
-40 dBm
Loss margin
15.75 dB
The entries in Table 5.01 are self-explanatory. Clearly, the optical power at the receiver
is greater than that required to achieve the necessary BER.The loss margin entry is
significant since it specifies the amount by which the received optical power exceeds
that required. In this example, it is 15.75 dB. Good design practice requires that it be
at least 10 dB.Why? Because no matter how carefully the power budget is estimated,
some entries are always forgotten and some are too optimistic. In other cases, vendor
specifications turn out to be inaccurate and an allowance must always be made for
future maintenance.
CHAPTER 5
The signal will lose power, the integrity of the data will be damaged by multipath
interference, and the stability of the source will tend to become lower.
The loss is a function of the component quality and accuracy. It fluctuates from one
channel to another and this is the main reason why tests must be performed over the
entire spectral range.The high-performance lasers used in DWDM systems are very
sensitive to reflected light, which can degrade the stability of the laser and the signalto-noise ratio, or even damage the source. Reflections can appear in the EDFAs and
cause major increases in the noise figure.All these problems must be measured before
and after the decisive signal is put to the network.
99
CHAPTER 5
Optical noise, which has taken on new importance since the introduction of optical
amplifiers in transmission systems, is due mainly to amplified spontaneous emission
(ASE) in the EDFAs.Although the manufacturer has almost certainly tested the EDFAs
individually, it is important to check their performance on-site, with all optical
channels in operation and all cascaded amplifiers present, to confirm that overall
performance expectations are being met. Gain variation merits special attention in
multi-amplifier systems, as it will directly affect system power flatness.ASE noise
figures can be particularly significant in some configurations, since this phenomenon
degrades the signal-to-noise ratio in all optical channels.
System gain will vary over time because of temperature changes, local stress,
component degradation, and network modifications.
The OSNR must be determined for each populated channel in a WDM system.
Referring to Figure 5.05, it is defined as:
OSNR = worst-case value of 10 Log P + 10 Log B dB
N
B
where P is the optical signal power in the ith channel. B is the reference optical
bandwidth. (B and B may be either frequency or wavelength); B is typically 0.1 nm.
N is the interpolated value of noise power measured in the noise equivalent
bandwidth, B, for the ith channel:
N = N( - ) + N( + )
2
where is the interpolation offset equal to or less than one-half the channel spacing.
100
CHAPTER 5
5.1.5 Gain
Optical signals are modified many
times in passing from a source to
an ultimate receiver in a network.
In an optical network using
EDFAs, the gain for individual
channels depends on the channel
population at that particular
point, so means must be available
to check that the signal-to-noise
ratio of every channel remains
adequate at the receiver;
aggregate measurements are not
sufficient.And, of course, means
must be available to determine
Figure 5.04 Gain is not flat over the EDFA
the cause of any discrepancies.
bandwidth.
Although the overall go/no-go
criterion may be based on total power and the power in each channel, tracing the
cause of weak channels may well require an ability to chek gain flatness at individual
amplifiers, as well as to investigate a variety of polarization and non-linear effects.All
require accurate spectral information about the signal.
101
CHAPTER 5
Since a 100 GHz channel separation implies very narrow channel bandwidths, spectral
drifts in the distributed feedback lasers used as transmission sources can have
devastating effects on signal levels at the receiver end.Therefore, source stability and
spectral purity are of paramount importance. Sidelobes are also of particular concern,
since they can add noise to adjacent dense WDM channels.Although older,
unstabilized lasers used in SONET networks cannot meet these requirements, existing
transmission equipment can be upgraded for dense WDM use by replacing these
sources with spectrally stabilized DFB models. In some cases, installed equipment can
be converted to dense WDM by directly inserting OC-48 laser signals into the
stabilized, modulated DFB.
5.1.7 Drift
Real optical sources are not
absolutely stable because both
their output power and central
wavelengths can be expected to
drift. Figure 5.06 shows a typical
result of monitoring a source over
a 12-hour period. Drift is caused
by such factors as temperature
changes, backreflection, and laser
chirp phenomena.
The primary concern with drift is
that the signal must remain within
the acceptable channel limits at all
times, under all operating
Figure 5.06 The wavelength and power vary over time.
conditions. Excessive drift may
cause the loss of the signal in the
affected channel.The drifting source may even spill over into an adjacent channel and
interfere disastrously with the information transfer there. Drift must be measured and
controlled to avoid loss of data.
5.1.8 Crosstalk
Crosstalkthe unwanted contribution of energy in one channel to othersis another
effect whose magnitude is difficult to accurately predict from pre-installation data; it
must be observed in real life, with real (or at least simulated) signals present.
Crosstalk calculations were
discussed in section 4.4.1.They
involve a detailed examination of
the shape of the passbands of two
adjacent channels to check that the
amount of signal leaking from one
to the other is negligible. Normally, a
minimum of 25 dB is required
between channels (Figure 5.07), but
13 dB might suffice for a submarine
link, and 17 dB values have been
tolerated in very long haul
applications.
102
Figure 5.07 The bold portion of peak 2 represents
crosstalk in adjacent channels.
CHAPTER 5
Four-Wave Mixing
Although four-wave mixing is also
a non-linear effect, it requires
special attention because of the
seriousness of its consequences on
a network.When two optical
signals interact in this way, energy
extracted from them can turn up
at wavelengths in use by other
channels, thus seriously disturbing
network operation.Although the
effect is similar to crosstalk, its
effects are not limited to the two
conflicting channels. Its magnitude
depends on the relative
polarizations of the two
interacting signals as well as their
magnitude and channel spacing.
103
CHAPTER 5
We noted earlier that the character of PMDand even the units most useful to
express itdepends upon the degree of mode coupling in the optical fiber. In
practical terms, the transition between the two states occurs at distances equaling the
coupling length, Ic, of the fiber, that is, the length of fiber at which energy transfer
between the polarization states becomes appreciable.Values of Ic seldom exceed a
kilometer, and values below and around 300 meters are common, so most real-life
networks operate under conditions of strong random mode coupling, in which PMD is
expressed as an expected time delay per square root of the unit length (e.g., ps/ km).
PMD is of particular significance in old cables laid in the late 80s when manufacturing
techniques and processes were not as advanced as they are now and the phenomenon
and its causes were not well recognized or understood. Now fiber manufacturers are
aware of the importance of maintaining near-ideal core geometry throughout the
length of fiber, and cable manufacturers take care to avoid placing undue stress on the
fiber when winding cable. Finally, carriers and installers now appreciate the effect that
the operating environment can have on PMD. But these kilometers of cable laid in the
late 80swith different types and ages intermixedcreate complexity and difficulty
for network operators who are planning capacity upgrades to OC-192/STM-64.
104
CHAPTER 5
Network compatibility
Commissioning
Maintenance and monitoring
Inter-vendor operability
Architecture interconnections
105
CHAPTER 5
5.2.2 Commissioning
Transmission performance tests measure transmitter and receiver performance, optical
amplifier performance, error behavior, and related effects that accumulate through
WDM network elements. Laser wavelengths may vary because of manufacturing
variations, thermal shifts, or frequency broadening.When a spectrally broadened signal
propagates through a dispersive medium, such as conventional fiber, error rates rise.
End-to-end optical testing should be performed at the system level, with all the
building block components working together in the optical link.The net dispersion
between the transmitter and receiver for all channel wavelengths, including those
added or dropped at intermediate points, should be checked to ensure that it is within
specified levels. Optical power and OSNR at the receiver should be measured.To
achieve balanced OSNR levels among all the channels in DWDM systems, an OSA can
be used to monitor the changes in OSNR as the transmitter levels are varied.
The number of wavelengths available to a WDM system is limited by amplifier,
passband, crosstalk, component drift, fiber non-linearity, and cascaded filter alignment
tolerances. Optical amplifiers are widely used to compensate for component and
transmission losses, and the impact of noise accumulation and cross-gain saturation
needs to be carefully considered to minimize signal power variations.Transmission
performance tests check end-to-end optical system performance. Channel wavelength,
channel drift, optical signal-to-noise ratio, channel spacing, channel isolation, insertion
loss, amplifier noise, gain flatness, bit error rate, jitter, and transmission delay are all
critical parameters to be measured. Each of these tests will be defined in greater detail
in section 5.3 where we will discuss commissioning tests.
106
CHAPTER 5
elements (NE/NE) and the operation systems and network elements (OS/NE) via the
embedded data communications channel (DCC), a local area network (LAN), or an
optical supervisory channel (OSC). Maintenance tests are discussed in detail in
section 5.4.
Alarm surveillance and performance monitoring are included in this category.Access
to key features and parametersincluding system administration and security
functionsmust be checked, both locally and remotely, according to the
specifications.These tests should be comprehensive enough to ensure that the
network operator is able to monitor and control the state of the network. It is possible
to use a channel dedicated to performing surveillance or transmitting information
about the systems status.This supervisory channel, as well as a complete supervisory
system (RFTS), will be looked at in more detail in section 5.5.
5.3 COMMISSIONING
A number of critical parameters should be measured and recorded when a network is
first commissioned to check start-up performance and to provide a base line for future
troubleshooting or upgrade planning.This testing period is so crucial that system
vendors supply their own test sequences.The tests will be carried out on individual
components and also on the system in operation.The objective of the commissioning
tests is to demonstrate the general performance and integrity of the system. Basically,
there are three series of tests.The first step consists of testing each system component
107
CHAPTER 5
to check their performance and optimize data.The second step focuses on the
complete system.We ensure that its performance conforms to requirements.The last
step is often forgotten, but is, however, a very important function. It involves
disruptive tests. Once all these tests have been carried out and when we can confirm
that the entire system and its components conform to standards, we are ready to start
operating the system and take advantage of WDM technology.
5.3.1 Transmitters
The transmitter output power should be
measured, using a power meter or an OSA and,
possibly, a calibrated variable attenuator.An
estimate should be made of the probable error
and the result quoted appropriately; i.e.,
XY dBm.
Source
Power meter
5.3.2 Receivers
The receiver sensitivity parameter
of most value is the input power
for a specified BER.
It is measured
P
using a BER test set, a variable
attenuator, and a power meter. It
should be measured for all the
ITU channels likely to be used in
the application.
3 dB
BERT
dB
Receiver
Splitter
OSA
MUX
108
DEMUX
CHAPTER 5
Tunable
laser
source
IN
OUT
Power
meter
The gain flatness for large signals can be measured with the same setup, varying the
input wavelength through the 1550 nm window in 1 nm steps and noting the output
power at each wavelength.
IN
OSA
OUT
Output
P
1540-1560
-25 dBm
980
1480
ASE
Pump
Remnant pump power is a measure of pump energy that escapes from the EDFA. It is
measured separately at the input and output of the unit, using an OSA.The input
measurement is made at 980 nm, the output at 1480 nm.
Amplified spontaneous emission (ASE) is measured similarly, at both the input and
output, using an OSA and scanning the complete gain band.The result is generally
given as a power spectral density, i.e., in dBm/nm (or, often, in dBm/0.1 nm).
109
CHAPTER 5
Transponder
MUX
dB
BER
tester
DEMUX
Power
meter
MUX
DEMUX
The corresponding parameter in a demultiplexer is the output power level for each
channel, measured using a calibrated power meter.The usual acceptance level is
20 dBm.
110
CHAPTER 5
OSA
OSA
MUX
MUX
Power
meter
DEMUX
b)
a)
Figure 5.18 Insertion loss is measured using a) an OSA at the mux level and b) a power
meter at the demux level.
The central wavelengths and the bandwidths of the individual channels of both
multiplexers and demultiplexers are important characteristics, and crosstalk can be
estimated from them.These spectral qualities are measured using an OSA, with a
wavelength meter for calibration. In most applications, the central wavelengths should
correspond with ITU grid values.The bandwidth of each channel (the width at 3 dB)
should be less than 0.2 nm.
OSA
OSA
MWM
MUX
DEMUX
a)
MUX
ASE
source
MWM
DEMUX
b)
Figure 5.19 Channel central wavelengths are measured using a) an ASE source and an OSA
for the demux and b) the OSA with the system sources for the mux.
111
CHAPTER 5
Directivity is of particular
importance in multiplexers, and it
is usually checked during
commissioning by connecting a
laser to an input of the multiplexer
and by connecting an OSA at
every other input port.
OSA
MUX
DEMUX
112
Source
N
OSA
Figure 5.21 The OSA is used at any point along the link to characterize the system.
OSA
MUX
DEMUX
CHAPTER 5
The uniformity of the transmitter power is also of interest. It is generally quoted as the
difference in power between the strongest and weakest channels, measured with an
OSA at the output of the first optical amplifier (fed by the multiplexer).The OSNR
should be high at this point (>28 dB for a 0.1 nm bandwidth) and the power variation
among channels should not exceed 2 dB.
Both the short- and long-term error performance should be checked.A standard link
configuration is used to measure the short-term rate: six channels spaced at 50 GHz,
each having a span loss of 25 dB, for example. Using a BER test set, the user can
confirm that the bit error rate in each channel is acceptable (under 1012 for a 25 dB
span loss).
Loopback
Protocol
analyzer
Optical link
MUX
MUX
DEMUX
Figure 5.23 A protocol analyzer displays the performance of every channel over a loopback
optical link.
The long-term error performance is also checked with a BER test set, at a realistic
signal level (15 dBm).The test should span a 24-hour period, so it is usually
impractical to check every channel individually.Two good candidates are the channel
that exhibits the lowest OSNR and the one with the highest power.
113
CHAPTER 5
An idea of the fragility of the link can be obtained by disturbance testing: removing
and reinserting modules from channels at random while observing the signals in the
remaining channels. Only the signal in the channel suffering the disturbance should be
affected.
The wavelength or frequency stability of the channels should be checked using a
wavelength meter.The frequency of each channel should stay within 5 GHz of the
nominal frequency for systems using a channel separation of 50 GHz.
A system-level cold-start test should also be performed.With BER monitoring
equipment on a suitable number of channels, all the power should be turned off.After
a suitable period (which will depend on the time constants of the power supplies and
the maximum possible delays in control paths, but which should be at least a minute)
the power should be restored. It is important to note the time taken for the signal
paths to be restored and for all appropriate management functions to return to
normal.There should be no spurious alarms.
114
5.4 MAINTENANCE
AFTER
Wavelength drift
P1
t
1
1 2 3 4
Gain tilt
EDFA noise
Crosstalk
or
CHAPTER 5
5.4.2 Losses
Losses in WDM systems must be examined channel by channel, and the best
instrument for the purpose is the OSA, possibly backed up by a wavelength meter for
spectral calibration and verification.
Standard methods of characterizing optical return loss have been discussed, using a
source, coupler, and photodetector to measure the power reflected by the link, which
would not be useable during the test.An OSA may be used as a detector to investigate
wavelength dependence. However, in the field, useful results can usually be obtained
more simply, using a small, portable backreflection meter or at the system level with
the remote fiber test system (RFTS) working with actual network signal traffic as
much as possible.
115
CHAPTER 5
5.4.6 Wavelength
Accurate measurement of central wavelengths calls for interferometric techniques, i.e.,
a wavelength meter. For field use in dense WDM systems, the absolute wavelength
accuracy should be better than about 0.005 nm. Signals may be weak, so a wide
dynamic range, 30 to 40 dB, is needed.An OSA may be needed, in addition to the
wavelength meter, to meet this requirement.
A wavelength meter can also be used to monitor channels over time to detect
wavelength and power drifts.
5.4.7 Crosstalk
Measurements of crosstalk in the field are facilitated by the data-storage capability of
modern OSAs. By using an optical switch, the operator can test each channel in
sequence without changing connections between measurements.The power within
each channel is stored and used to calculate crosstalk after all channels have been
probed.
5.5 MONITORING
116
Monitoring is not a test in itself, but rather the continual verification of live channels
through whatever individual tests are feasible and appropriate. It is used both to
detect variations in any optical characteristic of the network that might affect system
performance and to gather statistical information. Monitoring functions range from the
scheduling and management of tests to the analysis of the data acquired.
CHAPTER 5
Monitoring is particularly important in dense WDM systems, which are very sensitive
to optical phenomenal such as dispersion, crosstalk, central wavelength drift, and so
on.The ability to perform repeatable tests on critical parameters is essential if the user
is to be able to recognize and respond quickly to indications of instability.A good livechannel monitoring system provides constant information about system performance,
both specific and statistical.
CHAPTER 5
OSS
GIS map
RTU
Dial-up
PSTN
Remote optical
network controller
RTU
Eth
ern
et T
CP
/IP
Central station/server
(TSC)
RTU
Dial-up
PSTN
RTU
Portable test
system
RFTS units are seen as key elements in providing many of the supervisory and
maintenance functions needed in fiber networks.Their widespread use can be
attributed to very real cost/benefit advantages:
- A strong call for quality of service (QoS) continuous monitoring of a network is
an important strategic advantage for service providers.
- A significant increase in revenue per fiber the speed advantages of dense WDM
networks (both OC-48/STM-16 and OC-192/STM-64) can be realized in
information-carrying terms only if performance can be guaranteed at the channel
level.
- The high cost of maintenance a test system that provides automatic surveillance
of the physical network eases the workload of maintenance crews and directs
their attention where it is most needed.
Fast and easy fault localization is the key benefit to an effective monitoring system, but
the capability to check the suitability of installed fiber before performance
enhancements are undertaken can also often be significant.
118
However, few, if any, of the RFTSs now in use on non-DWM networks are actually used
to identify transmission faults. Instead, standard transmission protocols like SONET or
SDH perform that function, reacting quite rapidly to interruptions (i.e., in less than
120 ms). SONET switching and network topology protocols can then be used to heal
the network without interrupting service to its users.
CHAPTER 5
119
CHAPTER 5
Source
2
N
2
4
3
1. The first test should be performed at the demultiplexer at the end of the link, to see
if any further tests are needed.A wavelength meter can be used to check the OSNR
and the wavelength accuracy of each channel.The OSNR at the end of the link
should be greater than 18 dB for an acceptable BER (1013, for example); OSNRs of
22 dB or less are typical. Depending on the system and the detector type, the
channel power should be about 8 dBm.The channel wavelength should be within
a few tens of pm of the nominal value (a measurement of this accuracy calls for a
wavelength meter).
CHAPTER 5
2. If the first step indicates a problem in the link, the transmitter power level at the
output of the multiplexer is the next parameter to test: its specified value might be
from 5 to 2 dBm, depending on the system provider. If there is a discrepancy at
this level, the multiplexer should be checked for excessive loss.
3. The specific results of the previous tests will probably have identified the channel
or channels in difficulty.The power and central wavelength of the appropriate
source lasers should be checked.
121
CHAPTER 5
4. EDFAs are the next candidates: their gain should be linear and uniformly distributed
among the channels.A gain tilt of more than 3 dB may exceed specifications.
5. Digital test systems (SDH, Sonet, IP, or others) can help ensure transmission quality.
The minimum input power for a good BER should be over 20 dBm at the detector
level.The test is no longer of optical concern and electrical equipment is needed.
After checking all these previous parameters and procedures, the system integrator is
ready to "turn-key" the new DWDM system and fine-tune the last details before final
commissioning. Now that the system is working, it will surely be audited by the end
user via a quality control procedure.The idea is to check-list client requirements.
Those requirements are strongly influenced by both the recommendations of the
standardization bodies and industry trends.The next chapter will take a look at these
standardization issues and their impact on customer performance requirements.
122
CHAPTER 6
N E W S TA N DA R D S F O R D E N S E W D M S Y S T E M S
Product-Oriented
Level
ITU
IEC
International
ETSI
CENELEC
Europe
TTC
JISC
Japan
EIA/TIA
Table 6.01 Standards development organizations
124
North America
CHAPTER 6
N E W S TA N DA R D S F O R D E N S E W D M S Y S T E M S
125
CHAPTER 6
N E W S TA N DA R D S F O R D E N S E W D M S Y S T E M S
CHAPTER 6
N E W S TA N DA R D S F O R D E N S E W D M S Y S T E M S
Each of these study areas must be examined in considerable depth, with particular
attention paid to the way technology is evolving, so that the lag between the leading
edge and the commonplace is kept to a minimum. In the area of characterizing optical
fibers and cables, for example, the WP recently reviewed and altered recommended
test methods to include those pertinent to new fiber types, and added new methods
dealing with non-linearity and non-uniformity.
Detailed attention is also paid to the way complex phenomena, such as chromatic
dispersion and polarization-mode dispersion, are defined and measured, so that
component and system measurements can be exchanged among suppliers and users
with confidence.
An area of particular interestbecause of the phenomenal growth rate in Internet
accessis the use of the Internet Protocol (IP) directly in optical networks.The WP is
studying the question and has recommended a structured approach, one aimed to
ensure that routing equipment under development or being installed will now be
compatible with future steps to provide IP-over-WDM, which promises a dramatic
reduction in the bandwidth needed for Internet communication.
Another area under study by the WP is the development of standards for new, highcapacity WDM network applications.A wavelength grid with spacing of 50 GHz has
been proposed that is compatible with the present 100 GHz grid. Standards for
systems up to 128 channels are included, as are those for lower capacity inter-office
use and for unequally spaced channels (i.e., partially populated bands).
Work has also been undertaken towards developing specifications for TDM systems
operating at bit rates greater than 10 Gbps, with due consideration to the eventual
upgrade of these links to WDM. In particular, the status of key technologies to support
40 Gbps TDM transmission systems is under review.
Of particular interest to WDM testers is work currently underway on the uses of the
1625 nm region (1620 to 1660 nm) for remote fiber test systems (RFTSs), which are
discussed in section xx, and for other monitoring applications.
127
CHAPTER 6
N E W S TA N DA R D S F O R D E N S E W D M S Y S T E M S
Major decision-making at the IEC follows an established chain of command from the
governing committee, known as the IEC Council, through other administrative bodies,
down to technical committees (TCs), subcommittees (SCs), and WGs.There are
approximately 200 TCs and SCs, and about 700 WGs, representing the knowledge of
about 10 000 experts worldwide, who prepare the technical documents needed in
drafting international standards.Any IEC member country, as well as any organization
with official links to the IEC, may contribute to standards.
The TCs cover a multitude of topics, from household appliances to electro-medical
equipment to superconductivity.TC 86, Fiber Optics, is a large committee covering
fiber-optic communication systems and components including specifications and test
and measurement methods. Its structure is outlined in Table 6.02.
Working Groups
WG 1: terminology and symbology
WG 4: fiber-optic test equipment calibration
WG 7: discrete/integrated optoelectronic semiconductor devices for fiber-optic
communications, including hybrid modules
Subcommittees
SC 86 A: fibers and cables
WG 1: fibers
WG 3: cables
SC 86 B: fiber-optic interconnecting devices and passive components
WG 4: test methods
WG 6: interconnecting device specifications
WG 7: component specifications
SC 86 C: fiber-optic systems and active devices
WG 1: systems and active devices
WG 1: systems
WG 3: optical amplifiers
Table 6.02 International Electrotechnical Commission (IEC) structure
128
CHAPTER 6
N E W S TA N DA R D S F O R D E N S E W D M S Y S T E M S
As one would expect, the interests of those developing standards for products are
unlikely to be vastly different from the interests of those developing standards for
applications.Therefore, it is not surprising that many of the TC 86 activities are on
subjects already mentioned as current ITU areas. Such topics include PMD,WDM,
optical amplifiers, and fiber specifications.
The ITU work on choosing a reference test method for measuring PMD is reflected at
a more practical level by the development by IEC TC 86 of PMD-related standards. For
example, a technical report (IEC 61282-3) on calculating PMD in fiber-optic systems
has been prepared. Published standards include PMD Measurement Techniques for
Singlemode Optical Fibers (IEC 61941), which discusses all the commercially available
PMD measurement techniques for singlemode optical fibers. Proposals for statistical
PMD cable specifications, PMD test methods for optical amplifiers, and PMD
measurement methods for optical components are currently under review.
Other recent work includes dispersion compensation, PMD in digital and analog
systems, and the development of PMD power-penalty relationships.
A recently formed sector board, SB 4, addresses several topics related to the
infrastructure of telecommunications networks that are relevant to WDM systems,
including
standardizing installation conditions and instructions
DWDM techniques for optical amplifiers and fibers
xDSL techniques for installed and new passive infrastructures
major changes in fiber types and their influence on standards
twisted pairs for outdoor applications, optimized for ADSL
new developments in multimedia home cabling
Another concern is the attenuation (insertion loss) of DWDM devices. Document
IEC 61074 from the IEC SC 86B defines terminology in this area.The same committee
is working with the ITU-T on a rigorous definition of central frequency.
Generic specifications for optical amplifiers are provided in IEC Document 61291-1.
A performance specification template for multichannel optical amplifiers,
IEC 61291-4, is under discussion.A black box" approach is taken to this and other
product specifications: specifications that are as free as possible of application detail.
IEC 61291-4 discusses automatic gain control and the relevant operating input power
range, gain equalization, clamping criteria and spectral hole criteria. It provides generic
specifications andthrough related standardsperformance specification templates
for digital, analog, and multichannel optical amplifiers, specific to various application
areas (power amplifier, pre-amplifier, line amplifier, etc.).
The most important performance characteristics of optical amplifiers are covered:
1. Gain (IEC 61290-1)
a) Small-signal gain in the forward and reverse directions
b) Maximum value of the small-signal gain (absolute value, spectral profile
and variation as a function of temperature)
c) Overall spectral variation and wavelength band variation of the small-signal gain
d) Small-signal gain stability
e) Polarization-dependent gain variation
129
CHAPTER 6
N E W S TA N DA R D S F O R D E N S E W D M S Y S T E M S
130
The North American standards and conformity assessment body for all areas of
telecommunications, the TIA, has two committees dedicated to fiber-optic
telecommunications issues: FO-2.0 (Optical Communications Systems), responsible for
system and applications level standards and FO-6.0 (Fiber Optics), responsible for
product standards. Each has a number of active subcommittees and working groups,
among which FO-2.1.1 (Optically Amplified Devices, Subsystems, and Systems),
FO-2.1.2 (Working Group on Single-Mode Transmission Design), FO-6.3.5 (Passive
Fiber-Optic Branching Devices), and FO-6.6.5 (Single Mode Fibers) are the most
involved with WDM-related issues.TIA is an independent organization under the
American National Standards Institute (ANSI), with both private and public
membership.
Many TIA members are also involved in ITU and IEC, and the Association maintains
direct liaison with both.
CHAPTER 6
N E W S TA N DA R D S F O R D E N S E W D M S Y S T E M S
FO-2.1.1 has recently addressed issues relating to optical amplifiers, including the
definition and measurement of noise figure, round-robin test results, and relative
intensity noise (RIN) artifacts.
A draft entitled Optical Fiber Standard Test Procedure, OFSTP-6Measurement of
Optical Signal-to-Noise Ratioproduced by FO-2.1.2 proposes a scheme using an OSA
based on a Michelson interferometer, a diffraction grating, and/or a Fabry-Perot
interferometer. It provides values of the resolution bandwidth required to receive less
than 0.1 dB error in the signal-power measurement. For example, for a 10 Gbps
channel, a bandwidth of 0.2 nm is needed; and for 2.5 Gbps or lower, 0.09 nm is
required.
Guidelines and methods for the measurement of PMD and PDL have received recent
attention from FO-6.3.5, especially for components to be used in WDM systems.
FO-6.6.5 has addressed such issues as non-linear effects in fibers, new low-OH fibers
and PMD measurements based on the Jones Matrix Eigenanalysis (JME), and Poincar
Sphere Analysis (PSA) methods.
STRUCTURE OF TIA FO-2.0 AND FO-6.0 COMMITTEES
Liaisons
US Department of Defense (DOD)
Telecommunication Standardization Sector of the ITU (ITU-T)
IEC TC-86, SC86A, SC86B, SC86C, through technical advisory groups (TAGs)
FO-2.1
Subcommittee on Singlemode
Systems
FO-6.3.6
FO-6.3.8
FO-2.1.1
FO-6.3.10
Connector Specifications
FO-6.6
FO-2.1.2
FO-6.6.1
FO 2.2
FO-6.6.2
FO-2.2.1
FO-6.6.3
Sensor Fibers
FO-2.3
FO-6.6.5
FO-6.6.6
Step-Index Fibers
FO-2.4
FO-2.5
FO-2.6
FO-6.1
FO-6.6.7
Fiber Coating
FO-6.6.8
FO-6.6.9
Subcommittee on Reliability of
Fiber-Optic and Active Optical
Components
Fiber-Optic Field Tooling & Test
Instrumentation
FO-6.1.1
FO-6.3
Interconnecting Devices
FO-6.3.1
FO-6.3.3
Ferrule/Fiber Geometrical
Measurements
FO-6.3.4
Intermateability Standards
FO-6.3.5
FO-6.7
Fiber-Optic Cables
FO-6.7.1
FO-6.7.10
FO-6.7.14
Hydrogen Susceptibility
FO-6.7.15
IEC Liaison/Ribbons
FO-6.7.16
FO-6.7.17
FO-6.8
FO-6.9
Fiber-Optic Sensors
131
CHAPTER 6
N E W S TA N DA R D S F O R D E N S E W D M S Y S T E M S
132
CHAPTER 7
Long-haul >200km
Interaction 50 to 200 km
With or without amplifier
Access <60 km
Without amplifier
BANK
BANK
Access <60 km
Without amplifier
In principle, DWDM techniques can provide the needed increase in capacity for metro
areas.The problem has been to do so at a reasonable cost because of the high quantity
of equipment to be installed in the metro networks in comparison with the long-haul
scenario. However, DWDM equipment is now available for this new market, whose
importance will grow significantly: estimated sales of about US$200 million in 1998
are expected to reach close to $2 billion by 2003.
Although only about 15% of clients in the residential market are expected to require
1.5 Mbps service between now and 2005, with only 20% of that group needing
service at 2 Mbps or faster, demand in the small business sector will be more
spectacular. By the end of 2005, up to 85% of small-business users will require data
rates better than 2 Mbps, with 20% going as high as 45 Mbps.
Copper-based access networks cannot easily accommodate a capacity growth of this
magnitude, so, to a significant extent, they are capping the growth in demand and
customers are slow to ask for services that are not available for demonstration. Fiber
needs to get into the local access network before metro WDM will take off.
134
CHAPTER 7
That is not to say that metro DWDMs will necessarily be simpler than their long-haul
counterparts. In fact, flexibility will be much more significant for local carriers, and
particular attention will be needed to design-in as many reconfiguration possibilities as
can be imagined, with as few assumptions about transmission protocols as can be
managed.A great deal of emphasis must be given to providing sophisticated
management tools that will allow this flexibility to be used easily, quickly, andabove
allcost-effectively.
Installing metro DWDM does not necessarily imply replacing existing SONET
networks: the two can live side by side, with parts of the metro DWDM carrying
SONET traffic. Instead, good use must be made of the raw bandwidth offered by
DWDM technology. For example, by using the many available channels to deliver
servicesSONET, IP, or ATMeach in its native format; metro DWDM systems can not
only can accept traffic from devices supporting SONET interfaces, such as ATM
switches and routers, but can also accommodate other protocols such as Gigabit
Ethernet (GigE), FDDI, and ESCON.Although it may not be cost-effective to transmit
time inefficient signals, such as Gigabit Ethernet, in their raw form on a dedicated
long-haul DWDM channel, it may make sense to do so over the short distances typical
of a metro system.
As the SONET portion of such a network reaches its limits, it can by converted, link by
link, for DWDM use, which can grow as needed, at least while channel space is
available.The old traffic can still be handledSONET channels can be provided
while entirely new business is added at very low incremental costs.
Although metro DWDM link spans are characteristically short, they are not necessarily
short enough to avoid the need for in-line optical amplification (by EDFAs, usually).
Because needing or not needing EDFAs in the link represents a major cost break-point,
metro DWDM networks are usually classified in one of two ways: interoffice or access.
Link distances in the former generally exceed 50 km, and they are subject to many of
the optical and equipment constraints typical of long-haul systems. However, the
requirement for flexibility remains high. For example, they may provide the top-level
service of a CLEC.Although these link distances may require in-line optical amplifiers,
they can often operate cost-effectively at low data rates, thus reducing system costs.
The short links of an access metro system provide compound benefits. First, EDFAs are
not needed, saving considerable cost. Next, because EDFAs are not present, signal
characteristics, such as wavelength tilt, are much less critical, so component selection,
qualification, and maintenance are all simpler (although with no in-line amplification,
insertion and other losses may actually be more critical in these components).
Furthermore, non-linear effects in components or fiber are unlikely to present
problems. Finally, the removal of EDFAs frees the system designer from using
wavelengths outside the EDFA window, either for additional channels or to provide a
wide channel spacing that will further relax the component quality needed.Taking full
advantage of the potential simplifications offered in an access metro environment
requires the investigation of a set of equipment and configuration options that is quite
different from the choices that are relevant to the design of long-haul systems.
DWDM equipment for metro use is often configured similarly to the existing network,
whose capacity it is supplementing in rings (either two counter-rotating rings or a
single bidirectional cable) that are compatible with the SDH ring network
architecture. Monitoring and protection schemes are available to provide rapid path
switching (tens of milliseconds) should a component or the fiber itself fail.
135
CHAPTER 7
1... 4
A
.9
5x
SONET
..
12
OADM
s
2
/3
16
-3
OC
TDM
OC-48
C12
-3
OC
a)
TDM
OC-48
O
C3
TDM
OC-48
Optical
ADMs
C
.5
B
..
OADM
16
/3
2
s
C3
-3
OC
16
/3
2
s
OADM
s
2
/3
16
TDM
OC-48
b)
OADM
1... 12
Figure 7.02 a) The TDM metropolitan ring is overloaded between A an D and b) WDM
topology is the logical solution.
DWDM MANs (Figure 7.02-b) also offer an economic advantage over traditional links
(Figure 7.02-a) because the complexity of equipment needed to add or drop a
particular signal depends only on the characteristics of that signal.The add/drop
equipment does not have to deal with the full bandwidth of all the information on the
link. Costs are, therefore, closely related to actual benefits.And, of course, DWDM
services can be added to a network link without disturbing the existing traffica
consideration that is increasingly important as the need grows to provide day-to-day
alterations in service.
The information on each channel and the integrity of transmission can be protected
individually, to suit specific needs. In particular, present-day protection protocols such
as SONET can be handled without special provisions. Routing can be flexible, to
accommodate applications from point-to-point communications to data collection and
broadcasting.
CHAPTER 7
DWDM MANs offer all these advantages, plus the ability, in many cases, to interface
signals easily with compatible long-haul networks, without concern for bit-rates and
protocols. Components and systems are now becoming available for these
applications: two-fiber, interoffice spans cost approximately US$30 000 to $50 000
wavelength, and access metro systems are somewhat less.
Extensive planning is needed to ensure that DWDM upgrades can provide all the
SONET-like protection mechanisms that current customers require, while still offering
the service-expanding new features inherent to DWDM technology. Price, immediate
functionality, and future expansibility are all important considerations.
137
CHAPTER 7
138
CHAPTER 8
THE FUTURE
PRESENT
FUTURE
Fiber
- NZDSF
- Dispersion
compensation
- Dispersionmanaged fiber
- Better
compensation
- Lower loss
- Multicolor
- Ultra-low loss
- Solitons
Transmission
- 0.5 Tb/s
- 2.5 Gb/s per
- DWDM 32
-1.2 Tb/s
-10 Gb/s per
- DWDM 128
- 2 to 5 Tb/s
- 40 Gb/s per
- Soliton
- UDWDM 200+ s
Bandwidth
Management
- Electrical
- Async, SONET
- Interfaces
<10 Gb/s
- STS granularity
- Electrical, optical
- Multiprotocol, ITU
grid
- Optical
- Multiprotocol, ITU
grid
- Interfaces up to
10 G b/s
STS granularity
- DCS
- Interfaces up to 40 Gb/s
- STS-3c granularity
- Fast switching
THE FUTURE
CHAPTER 8
8.2 FIBER
The theoretical capacity of fiber is enormousthat of a singlemode fiber is perhaps
50 THz, under realistic assumptions. However, these astronomical rates will require
considerable technological developments. Deployment of a 400 channel system seems
feasible in the near future, and with each operating at a conservative 2.5 Gbps
(allowing for an optical signal-to-noise ratio of 20 dB) a total bandwidth of 1 Tbps does
seem attainable. Further foreseen developments will soon raise that figure to
2.5 Tbps.
DWDM will continue to be the technology of choice, and the fixed add/drop
multiplexers that are now being introduced will give way, first to field-reconfigurable
units and later to remotely controllable ones. Network complexity will grow, especially
as specialized services requiring specific logical configurations (CATV, for example)
are overlaid on traditional services. Channel counts will rise, as will the requirements
for laser stability and filter accuracy, and for sophisticated management tools for
power and noise. DWDM-based architectures based on packet and cell switching and
offering SONET,ATM, SDH, Gigabit Ethernet and IP interfaces will allow large
businesses to own individual wavelengths on a net, thereby going a long way toward
satisfying the growing demand for secure yet flexible communications among
company sites.
The transmission of Internet Protocol (IP) via optical fiber is on the immediate
horizon, and the dramatic savings in possible bandwidth will spur its eventual
adoption at all levels of Internet service.
Over the past few years, the increase in installed fiber has doubled bandwidth capacity
about every six months.To maintain this rate, cables containing many hundreds of
fibers will be common in new installations, and fully deployed 128-channel DWDM
systems will be widespread.
8.3 COMPONENTS
More and more network overhead functions will be built into transmission
components: transponders, in particular. Such features as SONET processing, forward
error correction, automatic protection switching, bit error rate checking, and signal
tracing will be included, all monitored by automated functions provided in an optical
supervisory channel.
Networks will become increasingly all-optical as research leads to new devices able
to provide functionalities now available only in electronic equipment. Removing the
need to recover and regenerate user traffic to pass through such a device will reduce
the hardware complexity of networks, but it will increase the importance of the
effects of various optical phenomena on signal integrity because, over increasingly
long distances, signals will exist only as optical pulses.
But a possible damper on the widespread use of all-optical networks exists: it is still
too early to say if transparent pipelines can ever satisfy the needs of service suppliers
and regulatory agencies to monitor, test, and manage information transfer right down
to the bit level.
DFB lasers are expected to be the transmitting sources of choice for dense WDM links
for some time, especially as on-chip features are added to improve their performance
and to ease their integration into networks. Nevertheless, other technologies are on
the horizon.
141
CHAPTER 8
THE FUTURE
8.4 TRENDS
Even if soliton-based communication systems are not yet commercial realities, many
indicators point to a rapid move from the laboratory to the service providers
networks.The strongest force for implementation of this technology comes from the
ultra-long-haul submarine cable applications.As this market is increasing, the need for
a low-cost ultra-distance solution like soliton technology is expected to strongly
increase as well.
These types of systems will lead to a decreased cost per bit because of the elimination
of repeaters or optical amplifiers.The soliton-based system will have to wait sometime
before becoming a commercial reality because the optical transmission systems are
still in evolution, but the volume of publications and research that is being conducted
by both universities and industry is a clear indication that this technology is here to
stay.
Fiber to the home (FTTH) is a technology in the wings. Its major impediment is
transceiver cost: it needs a laser in each home.Although this requirement is currently
cost-prohibitive, it is possible that wireless technology (fiber almost to the home)
will overcome at least some of the difficulties in the short term, providing adequate
data rates without the need for extensive deployment of a completely new fiber
infrastructure. In 10 years, 50% of homes will have FTTH using some form of wireless
or fiber final link. Inexpensive vertical cavity surface emitting lasers (VCSELs) are
already available for short-wavelength use, and developments are likely for the
1310 nm and 1550 nm spectral regions. Multimode (MM) WDM is another technology
with promise for LAN and other short-haul applications, eventually including FTTH.
Potentially it offers both lower cost and better reliability than its current competitors.
One thing is certain: the increasing need for bandwidth is not going to abate.The
phenomenal growth of Internet traffic has given service providers a taste of what can
happen when a totally new service is offered. HDTV is just beginning to demonstrate
its appeal to consumers, and millions of home viewers may ultimately decide that an
unobtrusive cable is easier to deal with than a rooftop satellite dish. Past attempts to
popularize such bandwidth-intensive services as video telephones have failed at least
partly because of compromises needed to meet severe bandwidth constraints.When
these constraints finally crumble, there is no telling what new conveniences
consumers will demand.
142
THE FUTURE
CHAPTER 8
Even without considering growth at the consumer level, business use of data networks
for everything from inter-office communication to plant monitoring shows no sign of
flagging. Indeed, one has only to recall the recent fears of a Y2K meltdown to
underline the dependence of modern industry on reliable communications.And the
ever-increasing cost of travel will make video conferencing a more and more attractive
option for businesses large and small, especially if the one-view-at-a-time nature of
present techniques can be overcome.Although the ultimate capacity of fiber-optic
systems may seem able to meet any conceivable demand safely, we must not forget the
predictions made early in the digital computer revolution that, perhaps, half a dozen of
these strange devices would serve all the needs of North America.
143
CHAPTER 8
144
THE FUTURE
CHAPTER 9
146
Qualifying the optical link is absolutely essential before buying and installing a new
system.The physical link plays an important role in system limitations and upgrading
capabilities. Using an analogy, even the most performing cars could not reach their full
capacity on roads that are not made for them.When testing the link, not only current
needs but also future improvements must be taken into account. In this case, testing
will be carried out in anticipation of the 128 channels at a transmission rate of
40 Gb/s.
CHAPTER 9
In the majority of cases, qualification of the link will be performed inside a central
office (CO) or a secondary building, where the space available for test equipment can
vary from comfortable to very limited. In this case, the company only has a small
space inside a secondary building that is not close by.This is why portable instruments
are often the units of choice for this first series of tests. Before deploying the system,
the customer will have no instruments in place.Therefore, a company such as EXFO
must supply all the necessary testing equipment, i.e., sources, couplers, switches, and a
variety of detectors.
To properly qualify the optical link, we suggest the following tests:
loss
ORL
PMD
chromatic dispersion
non-linear effects
Since a system is already working on the link, it is not necessary to perform continuity
tests. However, the link must be inspected. It is a good idea to take advantage of the
fact that the other system is shutting down to perform a full series of tests.
Each test will require connections. It should be noted that the instrument connectors
as well as the patchcords connecting the system or the optical link must always be
clean.The use of a talk set will facilitate the coordination of testing procedures at both
ends of the fiber.All results obtained should be saved for future use or simply for
internal use.This is what we propose:
Talk set
Talk set
Figure 9.01 The talk set allows two or more operators to communicate over the cable under
test.
147
CHAPTER 9
1. Foot print:The OTDR test is required to establish the links foot print, ideally at
1310 nm, 1550 nm, and 1625 nm.The OTDR is connected to the patchcord going
outside. Users choose parameters in Expert mode or using the one-button testing
feature in Novice mode. Once the measurement has been taken in one direction, the
operator at the other end can do the same.ToolBox Process software can then be used
to perform a bidirectional analysis of the link.The OTDRs bidirectional analysis is
useful for qualifying the loss of certain events according to the signals direction.
Mini-OTDR
OTDR
Figure 9.02 OTDR establishes the links foot print at given wavelength.
2. Loss:The OLTS test is necessary because this technique gives a much more accurate
loss measurement than that with the OTDR. It can also take advantage of the
bidirectional measurements. Using the MultiTest or MaxTester, 1550 nm and 1625 nm
wavelengths are chosen.The source is connected to the detector for auto-calibration;
FasTesT should analyze the loss in both directions.
OLTS
AS
ES
OLTS
148
Figure 9.03 Losses are accurately measured using the OLTS and a complete report is
given for bidirectional testing.
CHAPTER 9
3. ORL: By using the ORL test set, the user can measure the ORL level.This
measurement is very important because when it is too high, ORL can cause problems
such as source instability or high BER at the receiver level.The instrument must
therefore be connected to the patchcord going outside; the wavelengths employed in
the system must be used. For ITU wavelengths, a 1550 nm wavelength should be used.
149
CHAPTER 9
4. PMD:Although PMD is a complex phenomenon, the PMD test is made easy with the
PMD analyzer.The PMD value is critical to determine the maximum transmission rate
that can be used on the link.The polarized source (1550 nm) is connected to the other
end of the link and the analyzer at this end. No communication is necessary between
the two, but the source must be turned on before taking an acquisition.When the
source is turned on, the user must check that there is enough power on the PMD
power bar.The total length of the link obtained with the OTDR is entered to obtain the
PMD coefficient.The user can personalize the configuration or use Automatic mode,
but more accurate values will be obtained by using the exact measurement range.
PMD analyzer
PMD emulator
Figure 9.05 The emulator qualifies the PMD analyzer used to characterize optical link
performance.
150
CHAPTER 9
OSA
IQ-2400
IQ-2600
WDM
coupler
Figure 9.06 Bring signals from the tunable laser and DFB laser close together and check if
side peaks appear.
The link seems to be of sufficient quality for the 128 WDM channels, but the
transmission rate must be limited to 10 Gb/s, since the average PMD of 8.3 ps is
higher than the ITU standard of 2.5 ps for 40 Gb/s rates.An order for the system can
now be placed.
151
CHAPTER 9
152
For all possible component tests, it is preferable to use a laboratory instrument with
the highest technical specifications.The following is the preferred assembly for these
types of tests:
CHAPTER 9
1- Central wavelength and channel spacing: Since the primary function of the
multiplexer is to combine all wavelengths (separate in the case of the demux), it is
essential to qualify the component as a function of the wavelength.These two
important parameters will be measured by using a broadband ASE source and an OSA.
The source is connected to one of the input ports of the mux and the OSA to the
output port.The central wavelength is then deduced by taking the halfway distance
between the wavelengths where there will be 3 dB less than the maximum power.The
spacing between each central wavelength will provide the channel spacing.The
measured wavelengths should correspond as much as possible to those that will be
used in the system, i.e., those in the ITU grid.
OSA
IQ-5240
ASE
source
IQ-2300
Switch
IQ-9100
2- Ripple:The ripple will be measured using the same setup.This time, the analysis will
concentrate on the shape of the spectral response. If the mux were perfect, the filter
would have a perfectly rectangular
shape at the output port, but in
reality, there is a deformation at the
channel peak that must be
examined. Ripples are typically due
to parasitical Fabry-Perot etalons
that produce different spectral
modulation noises. Ripples are
usually sensitive to the input light
state of polarization.A nonpolarized ASE source provides an
average measurement of the ripple
amplitude. If this variation in
power is too great, there will be
significant variations in signal
Figure 9.08 Ripple causes variation in signal power.
power for a small variation in
wavelength.
153
CHAPTER 9
where IL is insertion loss in decibels, Pin (dBm) is input power in decibels referenced
to a milliwatt and Pout (dBm) is output power referenced to a milliwatt.This equation
quantifies the power loss in the device at a particular wavelength or in a given
spectral region. Normally, insertion loss should be as small as possible. For
components, it is not uncommon to measure insertion loss of 0.10 dB or even less.
The critical instruments for accurately measuring insertion loss are a stable light
source and a linear power meter. Other important points to consider include the type
of fiber, type of source, launching conditions, mode filters, and polarization
dependence.
154
CHAPTER 9
Tunable
source
Power
meter
IQ-203
IQ-2600
IQ-1500
MUX
155
CHAPTER 9
Source
Power
meter
IQ-2600
MUX
156
IQ-1200
CHAPTER 9
PDL controller
IQ-5100
IQ-3400
Tunable
source
IQ-2600
MUX
157
CHAPTER 9
8- ORL: Optical return loss (ORL) is a measure of the amount of optical power
reflected by a component. It is expressed in dB.
ORL = 10 log (incident power/reflected power)
A device that has an ORL of 0 dB reflects 100% of the incident power, while a device
with 20 dB ORL reflects 1%. Most passive components are designed to minimize
reflected power, and it is not uncommon to have components with >60 dB ORL.
At 60 dB ORL only 0.0001% of the optical power is reflected. Because of the low
reflected power levels, accurately measuring high ORL poses some interesting
challenges for instrument manufacturers. For accurate and stable return loss
measurements, it is important to have a high-power, stable, and (ideally) non-coherent
light source. High and stable power is important to achieve the required sensitivity
(80 dB); the coherence properties of the source are necessary to avoid interference
problems. If the coherence properties of the source are inadequate, large fluctuations
will be observed on the measured ORL.The return loss meter, when used with the
light source (optimized for ORL measurements), can perform stable ORL
measurements to 80 dB.When used with the medium-coherence tunable laser source,
ORL can be measured across the DWDM wavelength range.
Most of these tests can be carried out automatically and quickly by using a test system
dedicated to passive component testing.This assembly includes broadband sources
(DFB and tunable) so that all types of tests can be performed, as well as a polarization
controller, couplers, switches, a PDL tester, and an enormous capacity to save data for
complete offline analysis.
The multiplexer is of very good quality and will be just as efficient in the system as it
is alone. Once all the other tests have been performed and the system has been tested
in the factory, it will be delivered to the customer, who will have to perform his own
series of tests.
9.3 COMMISSIONING
When the company has received the system and installed it, it is important to perform
a series of tests to check that the connected components provide the required
performance and meet the installation requirements. In certain cases, the sum of all
the small losses may introduce significant problems in the system.Therefore, we
suggest critical testing procedures that will allow users to determine if they are ready
or not to start up the system while maximizing the important parameters.These are
the last tests that will be performed before the WDM system is turned on for
operation.
When the system is installed, the equipment in place (sources and multiplexer) can be
used for testing.The environment in which the tests will be performed may be clean
or dusty, very large or very restricted.Therefore, a good battery-operated portable
instrument is definitely a must.The following will be tested:
transmitters
detectors
amplifiers (EDFA)
multiplexers/demultiplexers
158
CHAPTER 9
We suggest that users follow the test procedure proposed by the system vendor once
the system has been delivered.To give an idea of the types of operations that the user
will have to perform, here is an EXFO procedure to test the systems important
components:
1- Transmitters:The power and wavelength of each transmitter must correspond to the
data supplied by the system vendor and be in accordance with the ITU standards.A
power meter is used to accurately measure the power of each laser and a
multiwavelength meter for high wavelength accuracy.
OR
Power meter
MUX
DEMUX
OSA or MWM
159
Figure 9.14 Characterize the system source with a power meter, OSA, or MWM.
CHAPTER 9
2- Detectors: Since the detectors must be able to detect the channel power to which
they are each dedicated without experiencing crosstalk from other channels, an
instrument that can make this type of calculation automatically must be used.To do
this, we will use the system sources and an OSA.All the sources are connected and a
measurement is taken with the OSA.Then, we terminate a source and measure the
power coming from other sources inside this channel.
MUX
DEMUX
FLS-2600
OSA
ASE source
160
CHAPTER 9
FLS-2600
Tunable source
Attenuator
EDFA
PUMP
GAIN MAIN
IN
GAN
OUT
OR
OSA
Power meter
161
CHAPTER 9
MUX
DEMUX
OSA
or
MWM
Figure 9.17 The multiwavelength meter (MWM) provides very accurate wavelength measurements.
162
5- General system:The system tests will be performed with all the new systems
components working together.To do this, an OSA is connected to each of the systems
physical ports to measure the OSNR of each channel. Depending on the system
vendors specifications and where users perform tests, values will vary between 18 dB
at the very end of the system and 40 dB at the beginning.At the EDFA level, a certain
gain should be recorded depending on the type of amplifier. Finally, a long-term test is
preferable to check channel stability. By using the multiwavelength meter in drift
mode, users will be able to follow the system wavelength drift with extreme
wavelength accuracy.
CHAPTER 9
6- Alarms:To check that all the systems alarms work properly, simply disconnect each
system component one after the other (sources, mux, demux, amplifier, laser pump,
and detector) to check that an alarm will be displayed.
All tests have been successful and the results are within the limits suggested by the
system vendor.We can therefore start operating the system and take advantage of the
significant increase in bandwidth (i.e., 32 wavelengths).
9.4 MONITORING
1- Generic Overview
The basic idea is very simple.The intelligence of RTUs, which used to be the dummy
units, is maximized thanks to their PC-based architecture. RTUs are located at strategic
points throughout the fiber-optic network to cover the maximum of the users fiberoptic network.
The RTUs include advanced OTDRs as well as optical switches to maximize the use of
OTDRs. Each fiber connected to a port of the optical switch is then monitored 24/7 or
according to an established schedule, by comparing OTDR acquisitions to standard
OTDR responses, also called reference traces. For each of those fibers, upper and
lower limits, or thresholds, are entered, and if one of the OTDR acquisitions exceeds
the threshold, an alarm is automatically generated.This alarm usually includes different
generic fields such as date and time, optical distance to the fault (OTDR
measurement), loss, CO, affected cable and fiber, and much more.
As shown in Figure 9.18, the carefully positioned RTUs report to a central
station/server, also called test system controller (TSC).All information sent by RTUs is
stored in a relational database for treatment and further SQL queries of all types.This
central station also executes all communication functions with RTUs, workstations,
and supervisory stations for remote users located anywhere in the field.
SOL database
OSS
GIS map
ONC workstation
RTU
RTU
RTU
RTU
RTU
163
CHAPTER 9
Locating Faults
This is where the RFTS really shows its strengths. Indeed, through the powerful
graphical information system (GIS) embedded in the central station RFTS software,
such as GeoMedia or Small World, users have the capability to map their fiber-optic
networks and link them to specific data such as
type of fiber
manufacturer
type of connectors
installation date
to and from locations
fiber count
spare fibers
clients affected, etc.
Users also have the possibility of entering landmark data, such as manholes, COs,
towers, etc., with the exact locations in GPS or other types of coordinates.
The optical distance to the fault allows the system to actually extrapolate the physical
location of the fault by calculating the optical-to-physical ratio between every
landmark of the network.The optical distance can be extremely different from the
physical one (on the map) for many possible reasons:
fiber slacks left in manholes or poles
access fiber meandering along streets
helix factor, etc.
2- Connecting and Operating the System
Physical Monitoring
Install the RTU (RFTS rack), and connect all fibers to be monitored to the optical
switch.These fibers can be active or dark.
164
IQ-7000
OTDR
IQ-9000
Switch
CHAPTER 9
Take reference traces for all the fibers to be monitored, with the OTDR connected
to the optical switch at the RTU level.The reference trace is used to set upper and
lower limits, or thresholds, and determine when alarms of different severity (minor,
major, critical, etc.) will be generated.These thresholds can be set for
total link power loss
reflective event degradation
non-reflective event degradation
section attenuation
new event detection
launch level degradation, etc.
165
Figure 9.21 The reference trace is the basis for
comparative analysis.
CHAPTER 9
Spectral monitoring
Install the RTU (RFTS rack), and
connect all fibers to be monitored to the
optical switch.These fibers have to be
active.
IQ-5220
IQ-9100
166
OSA
Switch
CHAPTER 9
Take reference traces for all the fibers to be monitored, with the OSA connected to
the optical switch at the RTU level.The reference trace is used to set upper and
lower limits and determine when alarms of different severity (minor, major, critical,
etc.) will be generated.These thresholds can be set for
central wavelength drift
SNR
signal power
167
CHAPTER 9
Hopefully troubleshooting will represent only a fraction of the time that you will
spend working with this system!
GLOSSARY
Absorption
Loss of light in fiber caused by impurities, resulting from
conversion of optical power into heat.
Accuracy
Defines how close a measurement is to its true value.
Add/drop multiplexer (ADM)
A component used to multiplex lower-speed electrical and/or optical signals
into a high-speed optical channel and vice versa.An add/drop multiplexer
can support either time-division multiplexing (TDM) or dense wavelengthdivision multiplexing (DWDM). It also links individual lines to backbone
trunks.
Amplified spontaneous emission (ASE)
The light emitted from the decay of the upper level of a lasing transition
without stimulated emission. In general, this emission is spectrally broad and
unpolarized.
Amplitude modulation (AM)
Transmission technique where the information is encoded in the amplitude
of the carrier.
ANSI
American National Standards Institute.
Arrayed waveguide grating (AWG)
Also known as a phasar.An integrated optical component which serves as an
optical multiplexer/demultiplexer. It is based on the phase differences
experienced by different input wavelengths to separate the channels, much
like the classical diffraction grating.
ASCII
American Standard Code for Information Interchange.
Asynchronous transfer mode (ATM)
A high-speed transmission scheme providing bandwidth on demand for
multimedia (voice, video, or data).
Asynchronous transmission
Occurs when the beginning and end of the data unit being transmitted is
individually signaled by the transmitter with start and stop bits.
Attenuation
The diminution of average optical power.Attenuation results from
absorption, scattering, and other radiation losses.Attenuation is generally
expressed in dB without a negative sign.
170
Attenuator
An optical device that reduces the intensity of a light beam passing
through it.
Avalanche photodiode (APD)
A photodiode that produces current through internal amplification.This is
referred to as avalanche multiplication.
GLOSSARY
Backbone
A network designed to interconnect lower-speed channels.
Backscattering
The portion of scattered light that returns in a generally opposite direction
to that of propagation. See Rayleigh scattering.
Bandpass
Defines the range of frequencies that pass through a filter or other devices.
Bandwidth
Measure of information-carrying capacity; the greater the bandwidth, the
greater the information-carrying capacity.
Bend loss
The result of macrobends (curvature of fiber) or microbends (small
distortions in the fiber) producing increased attenuation by coupling light
energy from the fiber core to the cladding.
Birefringence
The property whereby the effective propagation speed of a light wave in a
medium depends upon the orientation of the electric field
(state of polarization) of the light.
Bit error rate (BER)
The number of digital highs on a transmission link that are interpreted as
lows, and vice versa, divided by the total number of bits received. In modern
networks, BERs much better than 10-9 are expected.
Brillouin scattering
In stimulated Brillouin backscattering (SBS), the laser signal creates periodic
regions of altered refractive index, that is, a periodic grating that travels as an
acoustic wave away from the signal.The SBS effect can result in a very noisy
and unstable forward-propagating signal, since much of the optical energy is
backscattered.
Broadband
Supporting a wide range of carrier frequencies (e.g., voice, data, or video).
Buffer coating
Protective material that covers and protects a fiber.The buffer has no
optical function.
CATV
Acronym for cable television.
C-band (conventional band)
The spectral window from about 1525 nm to 1565 nm corresponding to the
strong amplifying range of the silica-based erbium fiber.
171
GLOSSARY
Central wavelength
Wavelength of a sources peak power.
Channel isolation
The ratio of the light in an optical channel that corresponds to the desired
spectral range against the contribution due to light at other wavelengths.
Chirp
A change in the characteristic optical frequency of a device as a function of
time (e.g., modulated diode laser) or position
(e.g., chirped fiber Bragg grating).
Chromatic dispersion
A phenomenon caused by the wavelength dependence of group velocity in
an optical fiber. Since any practical light source has a certain spectral width,
chromatic dispersion results in pulse broadening.The coefficient describing
chromatic dispersion per unit length is generally given in units of
ps/(nm.km).
Cladding
Low refractive index material surrounding the core of a fiber.
Coherence
A phenomenon whereby the phases of the photons (or constituent wave
trains) of a light beam maintain a definite relationship with each other.A
narrow linewidth laser is said to exhibit a high degree of coherence.
Connector
Hardware installed on cable ends to connect cables to a transmitter, a
receiver, or another cable.
Connector adapter
A device that allows a connectorized fiber to interface with a power meter.
Core
Center section of the optical fiber carrying and guiding light.
Coupler
Optical device containing several input and output ports to distribute
optical power.
CPU
Central processing unit.
Crosstalk
Undesirable signals in a communication channel due to leakage or coupling
from another channel.
172
Cutoff wavelength
The shortest wavelength for which an optical fiber can only support the
propagation of a single transverse mode.
Dark current
Thermally induced current of a detector in the absence of incident light.
GLOSSARY
dBDecibel
Standard logarithmic unit for the ratio of two quantities. In fiber optics, the
ratio is optical power and represents loss or gain.
dBc
Decibel referenced to the carrier optical power.
dBm
Decibel referenced to a milliwatt.
Demultiplexer (Demux)
Separates X different channels (wavelengths) of a signal into X
different signals.
Dense wavelength-division multiplexing (DWDM)
WDM technology where channels are closely spaced and typically
concentrated within the 1550 nm wavelength region (C-band).
Detector
Transducer that converts incident optical energy to an electrical signal at a
receiver device.
Differential group delay (DGD)
There are two mutually orthogonal input states of polarization.These are
known as the input principal states of polarization, one of which
corresponds to the fastest and the other to the slowest pulse propagation
time through the fiber.The difference in these two propagation times is
known as the differential group delay (DGD).
Diffraction grating
A bulk optical element that causes a reflection often at one or more
wavelength-dependent angles.This element is based upon coherent
scattering from an array of fine, parallel, equally spaced reflecting or
transmitting lines.
Directivity
The ratio of light that passes through the desired port against the light that
leaks through to an undesired output port in a multiport device
(e.g., optical circulator).
Dispersion
Signal distortion caused by differing path lengths of the modes in the fiber,
which results in the broadening of an input pulse along the distance of the
fiber and limits the bandwidth.
Distributed feedback laser (DFB laser)
An injection laser diode that has a Bragg reflection grating in the active
region to suppress multiple longitudinal modes and enhance a single
longitudinal mode.
DSn
173
The hierarchy of channel capacity for the transmission of digital signals used
in North America.A DS1 channel supports 1.544 Mb/s and holds 24 DS0
channels at 64 Kb/s.A DS3 is made up of seven DS2 channels
(four DS1 channels) that are multiplexed and transmitted at 44.736 Mb/s.
GLOSSARY
Dynamic range
For an optical instrument, generally defined as the ratio (in dB) of the
smallest signal that can be observed (at a specified wavelength separation) in
the presence of a strong, nearly saturating signal.
EDFA
Erbium-doped fiber amplifier.
Edge-emitting LED (EELED)
A type of LED that emits via a rectangular-shaped edge facet. Designed for
WDM component manufacturing and testing.
EIA
Electronic Industries Association.
Electromagnetic interference (EMI)
Any electrical or electromagnetic interference that causes degradation, failure
in electronic equipment, or undesirable response. Optical fibers neither emit
nor are affected by EMI.
Erbium-doped fiber amplifier (EDFA)
An optical amplifier that uses active erbium-doped fiber and a pump source
(laser) to boost or amplify the optical signal.
ESA
Electrical spectrum analyzer.
Fabry-Perot (FP) laser
Laser with multi-longitudinal modes.
Fault
Break in the continuity of the optical fibers normal performance.
Fiber Bragg grating
A spectral filter based on a periodic variation in the refractive index in an
optical fibers core. It is a key component in devices such as optical
multiplexers, dispersion compensators, and gain-flattened EDFAs.
Fiber coating
Material immediately around the cladding covering the optical fiber to
preserve the integrity of the fiber.
Fiber distributed data interface (FDDI)
ANSI architecture for a metropolitan area network (MAN); a network based
on the use of optical fiber to transmit data at 100 Mb/s.
Fiber in the loop (FITL)
Deployment of fiber cable in the local loop, that is, the area between the
telephone companys central office and the subscriber.
174
Flatness
Spectral uniformity of the WDM optical signal.
FOA
Fiber-optic adapter.
GLOSSARY
Frequency modulation (FM)
Transmission technique where information is encoded in the frequency of
the carrier.
Fresnel reflection
Reflection of a portion of the light incident on a planar interface between
two homogeneous media having different refractive indices. For a
perpendicularly-cleaved fiber terminated in air, the Fresnel reflection is
-14.6 dB if the fibers index of refraction is 1.46.
Fused biconic tapered (FBT) device
A type of optical coupler based on the melting together of two optical fibers
along a certain predetermined length (typically 2 to 5 mm).This fusing of
fibers results in a certain transfer of energy from one fiber to the other.
FWHM
Full width half maximum.
Gain
Gain is the ratio between the average output and input powers, omitting the
contribution of the amplified spontaneous emission (ASE) of the
amplifier itself.
GPIB
General purpose interface bus.
Graded index fiber
An optical fiber whose core refractive index uniformly decreases from its
center out to its edge. Normally used to refer to multimode fibers.
Group index
Ratio of speed of light in a vacuum against the speed of propagation of a
light signal in a fiber.
HeNe
Helium-neon laser.
HiBi
High birefringence fiber.
IEEE
Institute of Electrical and Electronics Engineering. It is a professional body
that is very active, among other fields, in fiber optics and opto-electronics.
Index-matching gel
Material with an index of refraction that is almost equal to that of the fiber
core; used to reduce Fresnel reflections.
Index of refraction
Ratio of the speed of light in a vacuum to the speed of light in a
given material.
Insertion loss (IL)
Loss of optical energy resulting from the insertion of a component or device
into the optical path.
175
GLOSSARY
International Telecommunication Union (ITU)
International standards development body for telecommunications, and the
source of many network standards.
L-band (long band)
The spectral window from about 1568 nm to 1610 nm.
Local area network (LAN)
A communications network that links data processing and
telecommunications equipment within a confined geographical area.
Different elements include servers, workstations, a network operating system,
and a communications link.
Laser
Acronym for light amplification by stimulated emission of radiation. Source of
highly coherent light via stimulated emission. Semiconductor lasers find
widespread use in the fiber-optic industry.
Light-emitting diode (LED)
A semiconductor device that emits incoherent light.
Linearity
For optical measurement instruments, generally used to refer to the deviation
of a measured change of a performance parameter (e.g., power, wavelength)
from the expected change.
Loss
See attenuation.
Mach-Zehnder interferometer
A device that divides an optical signal into two optical paths having
different, generally variable, path lengths.When these two beams recombine,
they interfere.These devices are often used as external intensity modulators.
Macrobending
Curvature of a fiber that causes loss of light.
Mandreling
Wrapping an optical fiber or cable around a cylinder to induce changes in
optical propagation (e.g., loss in singlemode fiber, mode mixing in
multimode fiber).
Material dispersion
Dispersion caused by the wavelength dependence of the optical fibers index
of refraction.
176
Measurement range
Difference between the launch signal level at the interface to the fiber under
test and the minimum level at which an event can be accurately identified
and measured within defined limits.
Metropolitan area network (MAN)
A stretched LAN providing data communication over a distance of about
50 km, generally associated with the IEEE 802.6 MAN standard.
GLOSSARY
Microbending
Microscopic bends or bumps in fiber; these cause loss of light by transferring
light energy from the guiding core to the cladding.
Mode
Distribution of electromagnetic energy in an optical fiber, which satisfies
Maxwells equations; path followed by light in an optical fiber.
Mode coupling
Energy exchange between modes of light propagating in an optical fiber.
Mode field diameter
Characteristic diameter of optical energy in an optical fiber.
Mode stripper
Device used to fully attenuate modes of light propagating in the cladding of
the optical fiber.
Multimode fiber
Optical fiber supporting more than one spatially propagating mode.
Multipath interference (MPI)
Interference that arises from multiple reflections in an optical path.These
reflections cause part of a detected signal to be dephased, which leads to
pulse spreading and degraded system performance.
Multiplexer (Mux)
A device that combines several different signals (typically at different
wavelengths) on a single fiber.
Narrowband wavelength-division multiplexing (NBWDM)
Type of WDM technology where channels are concentrated around the
1550 nm wavelength. See DWDM.
Network architecture
Description of how communication is established by its terminals, protocols,
and software between data-processing equipment at remote sites.
National Institute of Standards and Technology (NIST)
American national standards laboratory that maintains standards for a variety
of industries including fiber optics. One such example is standard setting for
optical power.
Node
Termination point for two or more communication links.
Noise floor
Optical power level where the signal cannot be distinguished from noise.
SNR=1.
177
Nominal wavelength
An approximate or target wavelength that is characteristic of an optical
device (e.g., filter, laser). It is not guaranteed to be its true wavelength.
GLOSSARY
Numerical aperture (NA)
Light-gathering capabilities of a fiber described by the relative maximum half
angle (in units of radius) at which light is accepted and propagated through
the fiber. It is a function of the refractive indices of the fiber core and
cladding. Standard singlemode fiber has an NA of about 0.12.
Optical carrier (OC)
The main unit used in SONET (synchronous optical NETwork). OC denotes
an optical signal and the number following OC represents increments of
51.84 Mbps (capable of holding DS3 signals), which is the minimum
transmission rate.The standard SONET format for this transmission rate is
called OC-1. Higher transmission rates are exact multiples of OC-1
(e.g.,OC-12, OC-48, and OC-192) Transmission is typically carried out using
the protocol at the following rates: OC-3, OC-12, OC-48, and OC-192.
OH-peak
A wavelength range of around 1390 nmcorresponding to an absorption
peak of the hydroxyl ion (OH-)where silica-based optical fibers tend to
exhibit enhanced attenuation.This is caused by water contamination in the
manufacturing process. Some manufacturers have managed to reduce this
peak, permitting newer fibers to be used across a wide spectral range.
Optical cleave
The breaking of an optical fiber to predictably produce flat end surfaces that
are perpendicular to the longitudinal axis of the fiber. Sometimes referred to
as a mirror-like surface across the entire end surface.
Optical continuous wave reflectometer (OCWR)
An instrument for measuring the total backreflection returning to the source
along an optical fiber. For optical components (e.g., isolators, filters, etc.), the
OCWR is used to measure return loss (the fraction of light reflected versus
the fraction of light transmitted). It can also be used to measure Fresnel
reflections of mated connectors.
Optical cross connect (OXC)
Generally refers to a non-blocking NxN, reconfigurable optical switch where
the optical signal entering any input port can be directed to any desired
output port.
Optical fiber amplifier (OFA)
A term used to include all optical amplifier technologies, including EDFA.
Optical fiber
Fiber made of dielectric material and consisting of the core (light-carrying
medium), and the cladding (protective layer) allowing total internal
reflection of the light for propagation.
178
GLOSSARY
Optical time domain reflectometer (OTDR)
An instrument used to characterize an optical fiber wherein an optical pulse
is launched into the fiber and the resulting backscatter and reflections are
measured as a function of time. It is used to identify faults and other
localized losses and to estimate loss.
Photodiode
Device that absorbs light energy and produces a photocurrent.
Photon
The packet or element of light exhibiting features of both particles and
waves, also referred to as quantum of electromagnetic energy.
Peak wavelength
The wavelength corresponding to the maximum value of a spectral
performance characteristic (e.g., the peak wavelength of an optical source,
bandpass filter transmission, etc.).
Pigtail
Short length of fiber attached to a connector, source, detector, or coupler.
p-i-n photodiode
A type of photodiode (positive/insulating/negative) with a large intrinsic
region in between p- and n-doped semi-conducting regions.
Poincar sphere (PS)
A three-dimensional graphical representation of the state of polarization of a
light beam.
Point-to-point
A two-station communications system that directly links two terminals.
Polarization
A term used to describe the orientation of the electric and magnetic field
vectors of a propagating electromagnetic wave.
Polarization controller
A device that transforms the input state of polarization (SOP) of a light beam
into a different, adjustable output SOP.
Polarization dependent bandwidth
The dependence of the spectral width on the incident state of polarization.
Polarization dependent central wavelength
The dependence of the peak transmission or reflection wavelength on the
state of polarization of the incident light.
Polarization dependent loss (PDL)
The difference in dB between the maximum and minimum values of loss
(attenuation) due to the variation of the polarization states of light
propagating through a device.
179
GLOSSARY
Polarization mode dispersion (PMD)
Dispersion of light causing a delay between the two principal states of
polarization propagating along a fiber or through a device due to the
birefringence property of the material.
Polarizer
Device that transmits light having one electric field orientation and blocking
all others.
Port
Opening for light traveling to and from a component.
Principal states of polarization (PSPs)
The two generally orthogonal states of polarization of a monochromatic light
beam launched into a fiber (input PSP) that will propagate through the fiber
without spreading or distortion.The SOP of this light beam as it exits the
fiber will be in one of two, generally orthogonal, output PSPs. In general, the
output PSPs are not the same as the input PSPs and the orientation of these
PSPs changes with wavelength. Not to be confused with birefringence. Only
in the spectral case of a single HiBi fiber are the PSPs and the axes of
birefringence the same.
Protocol
A set of procedures required to establish, maintain, and control
communications.
Pulse width
Length of the optical probe pulse that is directly related to the
pulse duration.
Rayleigh scattering
A fundamental, wavelength-dependent scattering process that depends
strongly on inhomogeneities in material density smaller than a
wavelength in size.
Receiver
Terminal equipment of the fiber network that converts optical signals to
electrical signals.
Reference power
Power level set as a datum power level; in loss testing it is the power level of
the test light source.
Reflectance
Ratio of reflected power against incident power at a single reflection point,
or from a device; measured in negative dB units.
180
Reflection
A change in the direction of light at an interface between two dissimilar
materials so that light returns into the material from which it originated.
Reflection generator
Device that reflects a fraction of the optical power incident on the device
back into the fiber.
GLOSSARY
Refraction
A bending of the light propagation direction at an interface between two
materials having different indices of refraction.
Refractive index
See index of refraction.
Relative intensity noise (RIN)
A measure of noise in the baseband frequencies of an optical carrier.
Repeatability
Variation of measured quantity when measurement conditions are changed
and restored.
Repeater
Device that regenerates the bits into a digital signal, normally by converting
the optical signal to an electronic signal, retimes the bit transitions, and
retransmits a new optical signal.
Resolution
Minimum value of the quantity measured by a test instrument.
Responsivity
Ratio of photodetectors electrical output against optical input in
amperes/watts.
Retardation
In a birefringent material such as a waveplate, the difference in phase shift
experienced by light propagating along the slow and fast axes.
Retardation plate
An optical component with two main axes.They separate an incident
polarized beam into two perpendicular polarized beams that recombine to
form a particular single polarized beam.A retardation plate generates full-,
half-, and quarter-wave retardations.
Ripple
An approximately concentric wave that appears on a surface that has not
been polished with an oscillating polishing lap.
Sensitivity
Minimum optical power required for a specified level of performance.
Signal-to-noise ratio (SNR)
Ratio of signal power against noise power for the receiver.
Singlemode fiber
Optical fiber supporting only one spatial mode of light propagation.
SOA
181
GLOSSARY
SONET
Synchronous optical network.A network that describes a set of common
characteristics for optical transmission of digital signals; a Bellcore-proposed
protocol for fiber networks.
Source
Light-emitting diode or laser diode that emits light for launch into an optical
fiber; characterized by its central wavelength and spectral width.
Spectral width
Measure of the wavelength extent for an optical device. In a source, it is the
width of the light at half-peak power (full width at half maximum), and in
power meters, it is the detector spectral sensitivity range.
Splice
Method to join two ends of optical fiber in a permanent or temporary
manner.Types include fusion splices and mechanical splices.
State of polarization (SOP)
The orientation of the electric field vector of a propagating optical wave. In
general, this vector will trace an ellipse as it propagates. In special cases, it
will remain oriented in one direction (linear polarization) or will trace out a
circle (left or right circular polarization).
Step-index fiber
Fiber having a fixed, flat index of refraction within the fiber core and a lower
(step) index at the cladding.
STM
Synchronous transfer mode.
Switch
Device that transfers light from one or multiple input ports to one or
multiple output ports.
Synchronous
Digital communication having a constant time interval between successive
bits or characters and using no redundant information, such as start and stop
bits, to identify the beginning and end of the data unit.
Synchronous digital hierarchy (SDH)
A worldwide, high-speed synchronous protocol standard transmitting up to
40 GB/s, known as SONET in North America. See SONET.
Talk set
Test set similar to a telephone that uses an optical fiber to establish voice
communication.
182
GLOSSARY
Topology
Physical arrangement of stations on a network.
Total internal reflection
The physical process responsible for keeping a light beam guided within the
core of an optical fiber. It arises when light in a higher refractive index
medium is incident upon the surface of a lower refractive-index medium at
an oblique angle greater than a characteristic critical angle.
Transistor-transistor logic (TTL)
Type of signaling in which a nominal +5 V is equated with logic 1 and a
nominal 0 V is equated with logic 0.A common semiconductor technology
for building digital logic integrated circuits.
Transmitter
Terminal equipment of the fiber network that emits optical signals in
response to electrical signal input.
Trunk
A single circuit between two points, both of which are switching centers
and/or individual distribution points.
Tunable laser
A laser capable of having its central wavelength varied to optimize it for a
given application.
Wide area network (WAN)
A network that links data processing and telecommunications equipment
over a larger area than a single work site or metropolitan area. It usually links
cities and is based on X.25 packet switching. It may be implemented by a
private corporation or a public telecommunications operator.
Waveguide dispersion
Pulse spreading caused by the dependence of the phase and group velocities
on a wavelength due to geometric properties of the waveguide.
Wavelength-division multiplexing (WDM)
Optical transmission technique that uses different light wavelengths to send
data. Combination of two or more optical signals for transmission over a
common optical path.
Wavelength-independent coupler (WIC)
An optical splitter connecting inputs to outputs having a suppressed
insertion loss dependence on a wavelength.
183
GLOSSARY
184
ACRONYM INDEX
ADM:
Add/Drop Multiplexer
ADSL:
APS:
ATM:
BER:
BDLA:
BML:
CLEC:
CD:
Chromatic Dispersion
CO:
Central Office
CORBA:
DBFA:
DCD:
DCF:
DCOM:
DSF:
Dispersion-Shifted Fiber
DWDM:
EDFA:
EML:
FDF:
FEC:
FTTC:
FWM:
Four-Wave Mixing
GIS:
HDSL:
HFC:
IP:
Internet Protocol
IXC:
Inter-eXchange Carriers
LAN:
MAC:
MIB:
NDSF:
Non-Dispersion-Shifted Fiber
NE:
Network Element
185
ACRONYM INDEX
186
NEL:
NML:
NOC:
NRZ:
Non-Return to Zero
NZDSF:
OADM:
OC:
Optical Carrier
OC-3:
OC-12:
OC-48:
OC-192:
OC-768:
OCH:
Optical Channel
OLTE:
ONC:
OO:
Object-Oriented
OPS:
OSC:
OSNR:
OSS:
OTAU:
OTDR:
OXC:
PDH:
PMD:
POP:
Point of Presence
POP3:
PSTN:
QoS:
Quality of Service
RFTS:
RTU:
SBS:
SDH:
SMTP:
ACRONYM INDEX
SNMP:
SONET:
SPM:
Self-Phase Modulation
SQL:
SRS:
STM-1:
STM-4:
STM-16:
STM-64:
STM-256:
TCP/IP:
TDM:
Time-Division Multiplexing
TMN:
TSC:
VDSL:
VPN:
WAN:
WDM:
Wavelength-Division Multiplexing
XPM:
Cross-Phase Modulation
187
ACRONYM INDEX
188
FURTHER READING
Agrawal, G.P. 1992. Fiber-Optic Communication Systems. New York,Toronto:Wiley &
Sons.
_________. 1989. Nonlinear Fiber Optics. New York:Academic Press.
Anderson, G.E.,Telcordia Technologies Inc. (formerly Bellcore). September 26 to 30,
1999. SONET & WDM:A Shotgun Marriage? Technical Proceedings. Chicago:
National Fiber Optic Engineers Conference. 1:121-32.
Becker, P.C., N.A. Olsson, and J. R. Simpson. 1999. Erbium-Doped Fiber Amplifiers,
Fundamentals and Technology, New York:Academic Press.
Bellcore. 1996. GR-2919-CORE. Generic Criteria for SONET Point to Point Wavelength
Division Multiplexed Systems in the 1550 nm Region. Piscataway, NJ.
_________. December 1994. GR-253-CORE. Synchronous Optical Network (SONET)
Transport Systems: Common Generic Criteria. Piscataway, NJ.
Born, M., and E.Wolf. 1964. Principles of Optics. Oxford: Pergamon Press.
Boyd, R.W. 1992. Nonlinear Optics. New York:Academic Press.
Bruyere, F., and O.Audoin. 1994.Assessment of System Penalties Induced by
Polarization Mode Dispersion in a 5 Gb/s Optically Amplified Transoceanic
Link. IEEE Photonics Tech. Lett. 6:443.
Buck, J.A. 1995. Fundamentals of Optical Fiber. New York: John Wiley & Sons.
Cao, X.D., and Meyerhofer. 1994. Frequency-Domain Interferometer for Measurement
of the Polarization in Mode Dispersion in Single-Mode Optical Fibers.
Opt. Lett. 19:1837.
CCITT Comit Consultatif International Tlgraphique et Tlphonique.
Recommendation G.652. Characteristics of a Single-Mode Optical Fibre
Cable.
Chabot, S., EXFO. Network Reliability: Damocles Sword? Technical Proceedings.
Chicago: National Fiber Optic Engineers Conference. 2:433-39.
Chon, J., H. Luo, C. Huang, R. Huang, J. Chen, and J. Bautista,Wavesplitter Technologies,
Inc. September 26 to 30, 1999. Expandable 50-GHz and 100-GHz Dense
Wavelength Division Multiplexers Based on Unbalanced and Cascaded-Fiber
Mach Zehnder Architectures. Technical Proceedings. Chicago: National Fiber
Optic Engineers Conference. 1:89-96.
Clark, R.C. 1947.A New Calculus for the Treatment of Optical Systems. J. Opt. Soc.Am.
37:110.
Cooperson, D.A., and M. Steinberg, Ryan Hankin Kent, Inc.The Evolution of DWDM
into Optical Networking Will Develop the Metro Market in 2000. Technical
Proceedings. Chicago: National Fiber Optic Engineers Conference. 2:500-08.
Curti, F. et al. 1990. Statistical Treatment of the Evolution of the Principal States of
Polarization in Single-Mode Fibers. J. Lightwave Tech. 8:1162.
189
FURTHER READING
Curtis, M.,Ameritech. September 26 to 30, 1999. Enabling Metropolitan Area DWDM
Networks. Technical Proceedings. Chicago: National Fiber Optic Engineers
Conference. 2:522-30.
Cyr, N., G.W. Schinn, and A. Girard, EXFO. September 26 to 30, 1999. Stokes Parameter
Analysis Method:The Consolidated Test Method for PMD Measurements.
Technical Proceedings. Chicago: National Fiber Optic Engineers
Conference. 2:280-89.
Damle, R., Sprint Advanced Technology Labs; R. Freund, D. Breuer, Broadband Network
Design GmbH (Germany). September 26 to 30, 1999. Outside-In Evaluation
of Commercial WDM Systems. Technical Proceedings. Chicago: National
Fiber Optic Engineers Conference. 1:450-54.
Desurvire, E. 1994. Erbium-Doped Fiber Amplifiers: Principles and Applications. New
York: John Wiley & Sons.
DiMichele,A.,Abou-Arrage, and B. Kent, Bell Canada; J. Schallenberg, Stentor Canadian
Network. September 26 to 30, 1999. Building High Capacity Broadband
Metro Networks Using Optical Add-Drop Multiplex Systems. Technical
Proceedings. Chicago: National Fiber Optic Engineers Conference. 2:639-48.
DiPasquale, R.J.,A.B. Sripad, K.C. Budka, and B.W.A. Rijsman, Lucent Technologies, Inc.
September 26 to 30, 1999. Breakthroughs in Robustness & Performance of
Network Elements Allow Centrally-Managed Operations of Large SONET
Networks. Technical Proceedings. Chicago: National Fiber Optic Engineers
Conference. 1:506-15.
Electronic Industries Association (EIA). FOTP 50 Standard. Light Launch Conditions
for Long Length Graded-Index Optical Fiber Spectral Attenuation
Measurements.
Escobar, H.E., P. Morkel, and L. Steinhorst, Fujitsu Network Communications; D.
Maruhashi, K.Yamane,T.Tsuda, and K.Takatsu, Fujitsu Limited (Japan).
September 26 to 30, 1999. High-Capacity DWDM System Performance
Results of 32 Channels OC-192 Over NZ-DSF. Technical Proceedings.
Chicago: National Fiber Optic Engineers Conference. 2:53-60.
European Telecommunications Standards Institute. May 1996. Swiss Telecom PMD
Specification and Information on PMD Statistics.
EXFO Electro-Optical Engineering Inc. 1997. Introduction to WDM Testing. Qubec:
Imprimeries Transcontinental Inc.
Foschini, G.J., and C.D. Poole. 1991. Statistical Theory of Polarization Dispersion in
Single Mode Fibers. J. Lightwave Tech. 9:1439.
Galtarossa,A. et al. 1996. In-Field Comparison among Polarization Mode Dispersion
Measurement Techniques. J. Lightwave Tech. 14:42.
190
Gisin, N. 1991. Solutions of the Dynamical Equation for Polarization Dispersion. Opt.
Comm. 86:371.
Gisin, N., et al. 1996. How Accurately Can One Measure a Statistical Quantity Like
Polarization Mode Dispersion? Photon. Tech. Lett. 8: 1671-1673.
FURTHER READING
Gisin, N., and J.P. Pellaux. 1992. Polarization Mode Dispersion:Time Versus Frequency
Domains. Opt. Comm. 89:316.
Gisin, N. et al. 1994. Definitions and Measurements of Polarization Mode Dispersion:
Interferometric Versus Fixed Analyzer Methods. IEEE Photonics Tech. Lett.
6:730.
Gisin, N. et al. 1995. Definition of Polarization Mode Dispersion and First Results of the
COST 241 Round-Robin Measurements. Pure App. Opt. 4:511.
Gisin, N. et al. 1991. Experimental Comparison between Two Different Methods for
Measuring Polarization Mode Dispersion in Singlemode Fibers. Electron.
Lett. 27:2292.
Gisin, N. et al. 1993. Experimental Investigations of the Statistical Properties of
Polarization Mode Dispersion in Single Mode Fibers. IEEE Photonics Tech.
Lett. 5:819.
Gisin, N. et al. Oct. 17-18, 1991. Polarization Mode Dispersion Measurements in Optical
Fibers, Cables, Installed Terrestrial Cables and Fiber Ribbon. Proceedings of
Optical Fiber Measurement Conference (OFMC).York, England. 85-88.
Gisin, N. et al. 1991. Polarization Mode Dispersion of Short and Long Single-Mode
Fibers. J. Lightwave Tech. 9:821.
Gisin, N. et al. 1993. Polarization Mode Dispersion for Single-Mode Fibers with
Polarization-Dependent Losses. Opt. Comm. 101:333.
Heffner, B.L. 1993.Accurate,Automated Measurement of Differential Group Delay
Dispersion and Principal State Variation User Jones Matrix Eignenanalysis.
IEEE Photonics Tech. Lett. 5:814.
_________. 1992.Automated Measurement of Polarization Mode Dispersion Varies
Jones Matrix Eigenanalysis. IEEE Photonics Tech. Lett. 4:1066.
_________. 1992. Deterministic,Analytically Complete Measurement of PolarizationDependent Transmission through Optical Devices. IEEE Photonics
Technology Letters. 4 (5): 451-54.
Henmi, N., S. Nakamura, and S. Hasegawa, NEC Corporation (Japan); S. Cortez, NEC
America. Multiple WDM Ring-Based Transport-Network Architecture for
Future Broadband-Data Services. Technical Proceedings. Chicago: National
Fiber Optic Engineers Conference. 2:213-23.
Hewlett-Packard Company. 1998. Fiber Optic Test and Measurement. Edited by D.
Dickerson. Upper Saddle River, NJ 07458: Prentice Hall PTR.
Huard, S. 1997. Polarization of Light. New York:Wiley & Sons.
IEC Technical Committee No. 86: Fiber Optics,Working Group 4, Subgroup 5.
Calibration of Optical Spectrum Analyzers. Geneva, Switzerland.
International Telecommunications Union (ITU): ITU Study Group 15. March 1996.
Contribution 301. First Report of ITU-T International PMD Round Robin
Measurements in Asia (Japan), Europe, and USA.
191
FURTHER READING
_________. Nov. 15, 1995. Contribution D, 523 (4/15). PMD: Comparison of Fiber and
Cable Results.
_________. May 1996. Polarization Mode Dispersion Measurement for Single Mode
Optical Fibers by Interferometric Method.
Kashyap, R. 1999. Fiber Bragg Gratings. New York:Academic Press.
Laude, J.P., ISA; Jobin,Y., Horiba Group (France); Lange, K., Instruments Sa, Inc.,
Horiba Group. September 26 to 30, 1999. Dense Wavelength Division
Multiplexers and Routers Using Diffraction Gratings. Technical Proceedings.
Chicago: National Fiber Optic Engineers Conference. 1:83-88.
Leung, J. and J. Nikolopoulos, Nortel Networks (Canada);T. Ray, MCI WorldCom.
September 26 to 30, 1999. Implementation of Level 2 Routing in DWDM
Networks for Enhanced OAM&P Performance. Technical Proceedings.
Chicago: National Fiber Optic Engineers Conference. 1:548-55.
Marsh, J.A., GN Nettest. DWDM System Testing: Deployment and Maintenance Issues.
Technical Proceedings. Chicago: National Fiber Optic Engineers
Conference. 2:199-205.
McGuire,A., BT Laboratories (UK); P. Bonenfant, and J. Manchester, Lucent
Technologies. September 26 to 30, 1999. IP Over WDM:The Missing Link.
Technical Proceedings. Chicago: National Fiber Optic Engineers
Conference. 1:97-108.
Mitchell,A.F., Z. Zhu, and X. Chang, MFB Communications Inc. September 26 to 30,
1999.Advances in Repeaterless Systems. Technical Proceedings. Chicago:
National Fiber Optic Engineers Conference. 1:430-41.
Mochizuki, K. et al. 1981. Polarization Mode Fibres. Electron. Lett. 17:153.
Noutsios, P., M. O'Sullivan, D. Beckett, and S. Corbeil, Nortel Networks (Canada).
Experimental and Theoretical Investigations of High PMD Impairments on
an OC-192 Field System. Technical Proceedings. Chicago: National Fiber
Optic Engineers Conference. 2:290-303.
O'Connor, D., L. Mo, and E. Catovic, Fujitsu Network Communications Inc. September
26 to 30, 1999. Packet Over SONET and DWDMSurvey and Comparison.
Technical Proceedings. Chicago: National Fiber Optic Engineers
Conference. 1:109-20.
Ocakoglu, G., N.Wauters, and P. Falcao, Hermes Europe Railtel (Belgium). September 26
to 30, 1999. Evolution of the Largest Pan-European DWDM Network:
Stepping into the 21st Century. Technical Proceedings. Chicago: National
Fiber Optic Engineers Conference. 1:455-64.
192
FURTHER READING
Poole, C.D., and D.L. Favin. 1994. Polarization Mode Dispersion Measurements Based on
Transmission Spectra through a Polarizer. J. Lightwave Tech. 12:917.
Poole, C.D. et al. 1991. Dynamical Equation for Polarization Dispersion. Opt. Lett.
16:372.
Redifer, G.B., Ericsson Inc. DWDM in the Metro MarketplaceDoes It Really Cost In?
Technical Proceedings. Chicago: National Fiber Optic Engineers
Conference. 2:239-43.
Rodio, R.J., CIENA Communications, Inc. September 26 to 30, 1999.There's More to
Optical Networking than Buying "ITU-Compliant" Lasers. Technical
Proceedings. Chicago: National Fiber Optic Engineers Conference. 1:442-49.
Ross, J. and R. Bryttegard, Fujitsu Network Communications, Inc.; J. Schmit, and L.
Chow,Tellabs Operations, Inc.; S. Kalambar, Lucent Technologies, Inc.; G.
Buehler, Siemens AG; N. Davis, Nortel Networks. September 26 to 30, 1999.
Achieving Integrated Management for Multi-Vendor, Multi-Technology
Transport Networks. Technical Proceedings. Chicago: National Fiber Optic
Engineers Conference. 1:183-88.
Saito,T.,T. Ota, H. Ogoshi, and T.Tsuda,The Furukawa Electric Co., Ltd. (Japan).
September 26 to 30, 1999. 50-GHz Channel Spacing
Multiplexer/Demultiplexer Combined by Two 100-GHz Channel Spacing
AWGs and FBGs. Technical Proceedings. Chicago: National Fiber Optic
Engineers Conference. 1:73-82.
Sexton, M., and A. Reid. 1992. Transmission Networking: SONET and the Synchronous
Digital Hierarchy. MA, USA and London, England:Artech House.
Smith, K., and L. McInerney,TTC. September 26 to 30, 1999.A Comprehensive Strategy
for Testing DWDM Systems. Technical Proceedings. Chicago: National Fiber
Optic Engineers Conference. 2:187-98.
Sudo, S. 1997. Optical Fiber Amplifiers. MA, USA and London, England:Artech House.
Telecommunications Industry Association (TIA). FOTP 113. Polarization Mode
Dispersion Measurement for Single-Mode Optical Fibers by Fixed Analyzer.
_________. FOTP 124. Polarization Mode Dispersion Measurement for Single-Mode
Optical Fibers by Interferometric Method.
_________. FOTP 122. Polarization Mode Dispersion Measurement for Single-Mode
Optical Fibers by Jones Matrix Eigenanalysis.
Tetu, M. 1997. Absolute Wavelength Stability Ions. Optical Fiber Communications
Conference. Dallas,TX: Optical Society of America.Tutorial FEl: 167-220.
Tyrone, Jr., B.H., Southwestern Bell Technology Resources; X. Lee, Nortel Networks; S.
Vigot, EXFO. September 26 to 30, 1999.Testing Requirements for Dense
WDM Optically Amplified Links. Technical Proceedings. Chicago: National
Fiber Optic Engineers Conference. 2:181-86.
193
FURTHER READING
Walsh,A.,Telcordia Technologies, Inc. (formerly Bellcore); S. Gloeckle,Ameritech.
September 26 to 30, 1999. Management of Bandwidth on SONET Ring
Architectures in the Network View. Technical Proceedings. Chicago:
National Fiber Optic Engineers Conference. 1:357-67.
Wilder, M.R.,Anritsu Company.Testing Fibers Carrying DWDM Traffic. Technical
Proceedings. Chicago: National Fiber Optic Engineers Conference. 2:422-26.
Wright,A., R.Yarlagadda,Williams Communications. September 26 to 30, 1999. Field
Trail of OC-48 ATM Directly Over DWDM. Technical Proceedings. Chicago:
National Fiber Optic Engineers Conference. 1:133-37.
Wu, L.-R., Hitachi Telecom (USA), Inc. September 26 to 30, 1999. Increasing WDM
Throughput Efficiency with Generic TDM. Technical Proceedings. Chicago:
National Fiber Optic Engineers Conference. 1:306-15.
194