Polynomial-based acoustic echo cancellation
Acoustic echo is an annoying, nevertheless common, phenomenon in the modern communication systems such as cellular telephones or video conference systems.This thesis introduces methods for polynomial-based acoustic echo cancellation which are used in telecommunication systems, how acoustic echo is formed and how to model and reduce it with the means of signal proeessing. The author describes the principles of adaptive filtering, presents the best known adaptive filtering algorithms and introduces some more sophisticated algorithms based on these. The aim is to develop non-linear models for modelling echo path alongside traditional linear models. These non-linear models try to model the non-linear distortion, which is inflicted into the sound by the small and inexpensive audio equipment. The aim is also to consider side-effects of these non-linear models and how these side-effeets eould be avoided.Since these non-linear models are quite sensitive to the input signals, the mechanisms for controlling adaptive filtering algorithms are studied.