An adaptive filtering subband structure with critical sampling

Abstract

This paper presents a new structure for adaptive filtering in subbands which employs critical subsampling of the analysis filter bank output signals. The proposed structure allows exact modeling of linear systems with finite impulse responses in system identification problems. An adaptation algorithm for the subfilters coefficients is derived which optimizes the mean-square value of the final error signal. The proposed adaptive subband structure is compared to previously proposed structures. Computer simulations are presented to illustrate the convergence behavior of the proposed adaptation algorithm.