I=
Pakistan Research Repository Home
 

Title of Thesis

Mitigation of SPM and GVD Effects in Fiber Optic Communications by Dispersion- and Power-Map Co-Optimization Using Genetic Algorithm

Author(s)

Muhammad Yousaf Hamza

Institute/University/Department Details
Department of Computer Science / Lahore University of Management Sciences, Lahore
Session
2009
Subject
Computer Science
Number of Pages
163
Keywords (Extracted from title, table of contents and abstract of thesis)
Nonlinearity, Compensation, Intuitive, Optimum, Optimization, Justification, Coaxial, GVD, SPM,Power, Map, Genetic, Algorithm, Established, DCF

Abstract
Optical fibers have several advantages over conventional transmission systems based on coaxial cables, radio and microwave links etc. These benefits mainly include extremely large bandwidth, high data transmission rates, electrical isolation, immunity to cross talk, signal security, low loss, system reliability etc. In spite of these benefits, fiber optic transmission systems are not perfect and have many limitations and challenges to overcome in long-haul communications. Optical pulse degradation is
one such limitation. An optical pulse propagating through fiber experiences degradation that may cause communication errors limiting the overall system performance. Fiber loss, dispersion and nonlinearity are the main factors that lead to such degradation even in single mode fibers (SMFs). For single channel communications, group velocity dispersion (GVD) and self-phase modulation (SPM) are the most significant dispersion and nonlinear effects respectively. GVD arises due to wavelength dependence of refractive index; it causes pulse broadening and thus limits the reach because of inter-symbol interference (ISI). SPM arises due to intensity dependence of refractive index; it introduces chirp that causes pulse spectrum broadening. These phenomena impose limits on the transmission rate and overall performance of the optical communications; therefore it is highly desired to mitigate these effects in fiber optic communications.
Dispersion compensating fibers (DCFs) are used to compensate the GVD effect in long-haul communication systems. Dispersion compensation (DC) can completely eliminate GVD effects if SPM can be ignored. However, this is not possible for cases when SPM is significant as the pulse shape also depends upon SPM through pulse
power. Hence, the pulse degradation is sensitive to the net amount of dispersion compensation, dispersion-map and the values of launch power. As a result, exact pulse recovery may not be possible by dispersion compensation only. In view of these facts, a system designer faces issues about (i) the net amount of dispersion compensation (ii) the distribution of the compensating elements and (iii) the value of launch power, for the best system performance. As will be discussed in this thesis, these are tough questions that have not been addressed yet because the answers are not intuitive; analytical treatment is not feasible; and the optimization using exhaustive search is prohibitive due to the extremely large number of possibilities. Minimization of pulse degradation requires multidimensional simultaneous optimization of net residual dispersion, dispersion-map and launch power.
This thesis provides a thorough background about the importance of optimization of dispersion compensation and values of launch power to minimize pulse degradation. As a next step, it reports an approach that can be used to co-optimize the net amount of residual dispersion, dispersion-map and launch power, in a reasonable time, to mitigate the effects of both GVD and SPM.
For investigations in these directions, first of all, a single fiber span was considered to explore the effects of pre- and post-compensations along with the launch power on pulse degradation. These investigations clearly showed that optimum dispersion-map and optimum launch power play a key role in minimizing optical pulse degradation due to GVD and SPM in fiber optic communications. As a next step, a 2-spans fiber link was considered as the simplest case that employed pre-, in-line, and post-DCFs. For launch power optimization, two approaches were considered. In the first approach, launch power was optimized under the conventional assumption that same value of power is launched into each fiber span. This approach is referred as ‘launchpower optimization’. In addition, the effects of launching different values of power in each span were proposed and investigated. We named this as ‘power-map
optimization’. This link was optimized for two cases: (i) under the constraint of 100% dispersion compensation using brute force, and (ii) without any apriori assumption of
the net compensating value. For case (ii), genetic algorithm was used due to its various advantages over conventional optimization techniques. The results obtained for 2-spans system clearly indicate that (i) power-map optimization yields better results than launch-power optimization, while (ii) dispersion-map optimization without any apriori assumption of the net amount of DC produces superior results than pre-fixed 100% compensated system. It is also pointed out that SPM and GVD effects are rather small in the 2-spans system and therefore the optimization has limited scope; however exhaustive investigations of the 2-spans system provided a basis to verify the GA-based optimum outcomes for the same system.
The above mentioned results established the need and justification of such an optimization approach for a larger system, for which as an example, 1600 km long fiber link consisting of 20-spans along with 21 DCFs was considered; and optimal solutions for the net amount of dispersion compensation, dispersion-map, lunch power and power-map were explored. It is important to note that exhaustive search for this intractable problem was impossible and therefore GA was used to obtain the near optimal solution in a reasonable time. A comparison of the results obtained through the method proposed in this thesis versus the results obtained through already proposed methods in the literature is also provided. The method proposed in this thesis showed a significant improvement over the traditional approaches as it recovered the input Gaussian pulse almost in its original form at the output of a 20- spans×80km/span link. Finally a discussion of these results in terms of improvement in Q-factor and the resulting bit error rate (BER) is also provided.

Download Full Thesis
1,253 KB
S. No. Chapter Title of the Chapters Page Size (KB)
1 0 CONTENTS

 

viii
98 KB
2

1

INTRODUCTION

1.1 Importance of Fiber Optics
1.2 Brief Historical Perspective
1.3 Main Advantages of Optical Fiber
1.4 Problem Background
1.5 Execution Procedure

1
187 KB
3 2 SOME DEGRADATION EFFECTS IN SINGLE MODE OPTICAL FIBERS

2.1 Fiber Loss

2.2 Dispersion
2.3 Nonlinear Phenomena
2.4 Summary

18
257 KB
4 3 PULSE PROPAGATION EQUATION AND SPLIT-STEP FOURIER METHOD

3.1 Pulse Propagation Equation
3.2 Split-Step Fourier Method
3.3 Summary

32
294 KB
5 4 GVD AND SPM-INDUCED DEGRADATION EFFECTS ON A GAUSSIAN PULSE

4.1 Dispersion-Induced Pulse Broadening
4.2 SPM-Induced Spectral Broadening
4.3 Interaction of GVD and SPM
4.4 Interplay of Loss, GVD and SPM
4.5 Summary

53
381 KB
6 5 IMPORTANCE OF DISPERSION AND LAUNCH POWER OPTIMIZATION

5.1 Importance of Optimization of Dispersion-Map and Launch Power
5.2 DCF Length Required for Perfect Compensation
5.3 Quantitative Description of Pulse Degradation
5.4 Optimization of Dispersion Compensation and Launch Power in one Span Link
5.5 Summary

72
276 KB
7 6 EXHAUSTIVE SEARCH BASED OPTIMIZATION OF A FIBER LINK

6.1 System Description
6.2 Optimization Procedure
6.3 Results
6.4 Summary

87
437 KB
8 7 GENETIC ALGORITHM BASED OPTIMIZATION

7.1 Genetic Algorithm: An Overview
7.2 Brief Historical Perspective of GA
7.3 Working Principle of GA
7.4 Applications
7.5 Dispersion and Launch Power Optimization of 2-Spans System Using GA
7.6 Summary

105
226 KB
9 8 OPTIMIZATION OF DISPERSION-MAP, LAUNCH POWER AND POWER-MAP FOR 20- SPANS SYSTEM USING GENETIC ALGORITHM

8.1 System Description
8.2 Dispersion-Map and Launch Power Optimization Using GA
8.3 Dispersion-Map and Power-Map Optimization
8.4 Q-Parameter Improvement
8.5 Summary

121
210 KB
10 9 DISCUSSIONS, CONCLUSIONS AND FUTURE RECOMMENDATIONS

9.1 Discussions and Conclusions
9.2 Future Recommendations

133
139 KB
11 10 APPENDIX

139


96 KB