Pmd ile sınırlı fiber optik haberleşme sistemlerinin hız ve performanslarının artırılması için düşük karmaşıklıklı uyarlamalı bir sistem

In this thesis, an adaptive signal processing based framework for theimprovement of fiber optical communication systems is presented. We usePolarization Division Multiplexing (PDM) technique in conjunction with theproposed blind source separation algorithms to increase the bandwidth efficiencyof the system where two independent signals are transmitted over two orthogonalpolarizations. We classify the PDM adaptive receiver design problem into twoseparate cases corresponding to low and high symbol rates. In the low symbolrate case, the main problem is to separate the transmitted sequences from theirinstantaneous mixtures. In the high symbol rate case, the goal is not only theseparation of mixing in space but also the compensation of the mixing in time,i.e., the Inter Symbol Interference (ISI), caused mainly by the Polarization ModeDispersion (PMD). Major constraints in the development of adaptive structuresfor these goals are the limitations imposed by the existing technology in terms ofprocessing very high rate fiber communication signals. In fact, the approachproposed in this thesis is centered around the development of low complexitymethods to address this issue. The proposed structures and algorithms arebased on the special class of Blind Source Separation (BSS) algorithmsexploiting magnitude boundedness of the digital communication signals. Asshown in the thesis, the use of the Magnitude Bounded BSS (MB-BSS) approachenables the development of low complexity algorithms and the correspondinghardware structures to counteract the mixing of digital communication signals inboth time and space caused by the communication medium. As the mostchallenging task is the design of adaptive structures for compensatingconvolutive mixtures corresponding to the high symbol rate, a special attention isgiven to this problem. For this purpose, the algorithmic extensions of the basicMB-BSS algorithm for instantaneous mixtures that make use of the novelparaunitary mapping approaches have been developed. Furthermore, theparametric versions of adaptive paraunitary MB-BSS algorithm are introduced forthe purpose of reducing complexity and enabling all-analog compensators.Simulations are used to illustrate the convergence speeds and robustness of theproposed algorithms whose results indicate that MB-BSS algorithm for both ofthe cases show nice convergence behaviors. We also present simulation resultsto demonstrate the successful tracking capability of the algorithm against timevariations of the fiber channel. All simulation results indicate that MB-BSS basedapproach provides a promising framework for the rate and performanceenhancement of optical communication systems which suffer from PMD.