Preface
The first generation of fiber-optic communication systems debuting in 1980 operated at a meager bit rate of 45 Mb/s and required signal regeneration every 10 km or so. However, by 1990 further advances in lightwave technology not only increased the bit rate to 10 Gb/s (by a factor of 200) but also allowed signal regeneration after 80 km or more. The pace of innovation in all fields of lightwave technology only quickened during the 1990s, as evident from the development and commercialization of erbium-doped fiber amplifiers, fiber Bragg gratings, and wavelength-division-multiplexed lightwave systems. By 2001, the capacity of commercial terrestrial systems exceeded 1.6 Tb/s. At the same time, the capacity of transoceanic lightwave systems installed worldwide exploded. A single transpacific system could transmit information at a bit rate of more than 1 Tb/s over a distance of 10,000 km without any signal regeneration. Such a tremendous improvement was possible only because of multiple advances in all areas of lightwave technology. Although commercial development slowed down during the economic downturn that began in 2001, it was showing some signs of recovery by the end of 2004, and lightwave technology itself has continued to grow. The primary objective of this two-volume book is to provide a comprehensive and up-to-date account of all major aspects of lightwave technology. The first volume, subtitled Components and Devices, is devoted to a multitude of silica- and semiconductor-based optical devices. The second volume, subtitled Telecommunication Systems, deals with the design of modern lightwave systems; the acronym LT1 is used to refer to the material in the first volume. The first two introductory chapters cover topics such as modulation formats and multiplexing techniques employed to form an optical bit stream. Chapters 3 through 5 consider the degradation of such an optical signal through loss, dispersion, and nonlinear effects during its transmission through optical fibers and how they affect the system performance. Chapters 6 through 8 focus on the management of the degradation caused by noise, dispersion, and fiber nonlinearity. Chapters 9 and 10 cover the engineering issues related to the design of WDM systems and optical networks. This text is intended to serve both as a textbook and a reference monograph. For this reason, the emphasis is on physical understanding, but engineering aspects are also discussed throughout the text. Each chapter also includes selected problems that can be assigned to students. The book's primary readership is likely to be graduate students, research scientists, and professional engineers working in fields related to lightwave technology. An attempt is made to include as much recent material as possible so that students are exposed to the recent advances in this exciting field. The reference section at the end of each chapter is more extensive than what is common for a typical textbook. The listing of recent research papers should be helpful to researchers using this book as a reference. At the same time, students can benefit from this feature if they are assigned problems requiring reading of the original research papers. This book may be useful in an upper-level graduate course devoted to optical communications. It can also be used in a two-semester course on optoelectronics or lightwave technology. A large number of persons have contributed to this book either directly or indirectly. It is impossible to mention all of them by name. I thank my graduate students and the students who took my course on optical communication systems and helped improve my class notes through their questions and comments. I am grateful to my colleagues at the Institute of Optics for numerous discussions and for providing a cordial and productive atmosphere. I thank, in particular, Renè Essiambre and Qiang Lin for reading several chapters and providing constructive feedback. Last, but not least, I thank my wife Anne and my daughters, Sipra, Caroline, and Claire, for their patience and encouragement. Govind P. Agrawal Rochester, NY |
TABLE OF CONTENTS Chapter 9.6.3 - PMD and Polarization-Dependent Losses
9.6.3 PMD and Polarization-Dependent LossesAs discussed in Section 3.4, fluctuations in the residual birefringence of optical fibers change the state of polarization (SOP) of all channels in a random fashion and also distort optical pulses because of random changes in the speeds of the orthogonally polarized components of the same pulse (PMD). In a realistic WDM system, one should also consider the effects of polarization-dependent loss (PDL) associated with various optical components (see Section 3.5). Moreover, when optical amplifiers are used periodically for loss compensation, the polarization-dependent gain (PDG) of such amplifiers can also degrade a WDM system. For this reason, considerable attention has been paid to understanding the impact of PMD, PDL, and PDG on the performance of WDM systems [157]-[163]. The main problem from the standpoint of system performance is that fiber birefringence can change with time in a random fashion owing to variations in temperature and stress along the fiber link. As a result, both PMD and PDL can lead to channel outage – a phenomenon in which the BER of a channel increases so much that it is effectively out of use. Such an outage can occur at random times for random durations. The only solution to this problem is to design the WDM system such that the outage probability remains below a certain value. Acceptable values of the outage probability are below 10-5, a value that corresponds to an outage of less than 5 min/year. As seen in Section 3.4, the impact of PMD depends on the differential group delay (DGD) that leads to distortion and broadening of optical pulses. The outage probability depends on the average value of DGD and can be reduced by controlling it below a certain value. The origin of PDL-induced outage is quite different and is related to how PDL affects the optical SNR as the signal traverses the fiber link. To understand why PDL changes the SNR, one should consider what happens to the signal and noise as they pass through an optical element with different losses along its principle axes. Since noise is unpolarized, on average half of the noise is polarized along the signal and the other half is orthogonal to it. Because of this feature, a part of the orthogonal noise component is transferred to the signal, and it affects the SNR by an amount that depends on the SOP of the signal before it enters the lossy element. Since signal SOP is random for different PDL elements, the SNR at the end of the fiber link fluctuates in a random fashion. When the number of PDL elements is relatively large, optical SNR as well as the Q factor follow a Gaussian distribution [162]. Because of PDL, the Q factor may increase or decrease, and channel outage occurs when the reduction in Q exceeds a certain value. One can introduce the concept of the PDL-induced penalty as the maximum change in the Q factor for a given amount of PDL. However, since a Gaussian distribution has long tails, the penalty can be arbitrarily large albeit with a lower and lower probability. In one set of numerical simulations, the penalty was defined with 99% confidence, that is, the probability was at most 1% that the actual penalty may exceed the predicted value [162]. Figure 9.33 shows how the Q-factor penalty increases with the average value of accumulated PDL for several values of root-mean-square DGD. The simulations were performed for a 10-Gb/s channel carrying an RZ bit stream with 33% duty-cycle Gaussian pulses and operating over 4,000 km with 40 fiber spans. Each span had three PDL elements whose principle axes were oriented randomly. The dotted lines show the improvement that can be realized by employing a PMD compensation scheme. Such a scheme is beneficial but it does not affect the PDL-induced penalty. As seen in Figure 9.33, PDL can degrade the system even in the absence of PMD, and PMD compensation cannot reduce this penalty. Moreover, PDG can lead to additional penalties [158].
Figure 9.33: Q-factor penalty as a function of the average value of PDL along the fiber link for several values of DGD. Dotted lines show the improvement realized with PMD compensation. (After Ref. [162]; ©2003 IEEE.)
|
