Study Of Adaptive Channel Equalizers Based On State-space Realizations

With the development of digital communications technology, the inter-symbol interfer-ence (ISI) and inter-channel interference (ICI) caused by the limited transmission bandwidthand distortion a?ect communication quality a lot. In a communication system, the magnituderesponse is not constant and the phase response is not linear, both varying with the environ-ment. All of these can lead to distortion of the signals transmitted. Research of suitableequalizers which can eliminate these in?uence has theoretic and practical importance. Nowa-days, adaptive channel equalizers are one important part of signal processing, they providea superior scheme for processing distorted signals, yielding a performance better than otherfixed filters.An adaptive equalizer actually is an adaptive filter, as it is used in channel balance, peoplecall it channel equalizer. This thesis firstly describes some basic knowledge on adaptive filters,analyzes a class of filter's parametrizations and structures, and investigates some adaptivealgorithms and criterions. Secondly, the structure and performance of adaptive lattice filters areintroduced, in which a normalized lattice infinite impulse response (IIR) equalize is studied.Then we analyze an internally stable lattice filter which is based on a normal lattice IIR filter.Its application on adaptive channel equalization is also investigated, where both Bussgang-type adaptation and Riemannian-gradient-type adaptation are exploited. The correspondingscheme is implemented on MATLAB to simulate the performance of the entire system. Then,based on the internal balance realizations (IBR) we propose a new adaptive filter scheme.Gradient-based algorithm and recursive Gauss-Newton prediction error (RPE) algorithm areused to update filter's parameter vectors. Once again, the whole scheme is implemented onthe MATLAB platform to simulate the performance of the proposed scheme. The study ismainly based on second order filter sections and for high order adaptive equalizers a cascadeof 2nd order sub-filters is adopted. Finally, we compare the performance of the internallystable lattice adaptive equalizer and the proposed IBR adaptive equalizer. Simulations showthat our proposed IBR adaptive equalizer yields a better performance.