Desenvolvimento e analise de novos algoritmos preditivos para equalização cega

AUTOR(ES)
DATA DE PUBLICAÇÃO

1997

RESUMO

In this work, the blind equalization of a communication channel will be considered. The work will focus on the discussion of blind equalization techniques that are based on linear prediction theory. These are called predictive techniques for blind equalization. Firstly, the need for an equalizer in a communication system will be discussed, as well as the advantages and disadvantages of blind equalizers as opposed to supervised ones. Some traditional blind equalization techniques will be presented. Then, the principIes of predctive blind equalization techniques will be studied, and some adaptive algorithms for predictive equalizers will be presented. These algorithms are based on the steepest descent method (LMS). After this introductory presentation, this work will focus on a predictive technique based on magnitude and phase equalization. In this technique, a linear IIR preditor is used to compensate for the magnitude distortion caused by the channel, and a non-linear all-pass filter is used to compensate for the phase distortion introduced by the cascade of the channel and the predictor. Two new adaptive algorithms for this structure will be presented, one based on the least-squares method (RLS, Recursive Least Squares) and another one based on the Gauss-Newton method (GN). Some theoretical analysis of both the phase equalizer and the algorithms for this structure will then be carried on. It will be shown that the minimization criterion associated with the phase equalizer is unimodal. Furthermore, it will be shown that the proposed algorithms do converge to a point where they equalize the channel. Finnaly, some simulation results will confirm the effectiveness of these algorithms

ASSUNTO(S)

filtros adaptativos algoritmos comunicações digitais

ACESSO AO ARTIGO

Documentos Relacionados