Analysis and application performance simulation of subband adaptive filtering structures

Abstract

Subband adaptive filtering reduces computational complexity and improves convergence performance. Here three kinds of subband adaptive filters (SAF) are investigated. The basic structure is simply composed of subbands decomposition and adaptive filters with low computation power. Based on the basic one, the second structure added by cross-filters is to eliminate subbands aliasing. The last one is an efficient method using polyphase decomposition and identities equations. Moreover, we analyse their advantages and defects on the computation complexity on echo cancelling and convergence performance on system identifying in virtue of MATLAB. The computer simulation results illustrate that compared to the fullband filter, SAF requires less computation power and possesses better convergence performance