Adaptive volterra filtering using M-band wavelet transform

Abstract

A new LMS adaptive Volterra filtering in the M-band wavelet transform domain is presented, where the input is pre-processed with MDWT (M-band discrete wavelet transform) being followed by power normalization. In particular, the pre-processing procedure leads to effective reduction of the eigenvalue spread of a Volterra input auto-correlation matrix, and thus improves the convergence rate of the adaptation process even in case of wide classes of input processes. To demonstrate the performance of the proposed approach, some simulation results are provided.