LMS based algorithms for echo cancellation in telecmmunications
Date
2003
Authors
Sristi, Prasad
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The application of various popular least-mean-squares (LMS) algorithms to echo cancellation in telecommunication systems is studied. To start with, an analysis of the echo path impulse response in an HDSL2 environment is presented in order to provide a framework for studying the performance of the adaptation algorithms. Then three categories of LMS algorithms, namely, subband, variable step-size, and block LMS adaptation algorithms are researched. An extensive study of the effects of the filter banks used on the performance of the various configurations is carried out through mathematical analysis and simulations. A new method of designing optimum filter banks for usage in subband adaptive filters is described and the superior performance of the designed filter banks over other types of filter banks is illustrated through simulations.
Various existing variable step-size adaptation algorithms are described and simulated and their performance is studied in different echo cancellation scenarios. A new variable step-size LMS algorithm is proposed and simulations are included to show that superior performance of the proposed algorithm is possible through the proposed algorithm in conventional as well as subband echo cancellation configurations with respect to existing variable step-size algorithms. Block adaptive filtering algorithms in time and frequency domain are then discussed and simulated. The thesis concludes with comparisons between block LMS and conventional LMS adaptive filtering in terms of computational complexity and echo cancellation performance which are based on mathematical analysis and simulations.