A Fast and Low Complexity Adaptive Filters Implementation in Wavelet Packet domain with Partial updating and Selective Regressors

Abstract

Selecive regressors and selective coefficient update
(partial updating) of affine projection algorithm (APA) is two
effective method for reducing computational load and power
consumption in adaptive filter implementations. In addition to
selective regressors and selective coefficient update,
computational load of APA can be reduced by making use of
sub-band processing techniques and applying set-membership
concept for sub-bands. Considering these matters, first, the
new implementation of adaptive algorithms technique is
supposed in wavelet packet domain using multiple independent
adaptive filters instead of one. Afterwards, we apply the APA
to the proposed structure and incorporate a data selection
strategy based on the set-membership concept along with the
reduced computational complexity from selective coefficient
update and selective regressors in each pair of packets,
simultaneously. Also selective coefficient update NLMS
Algorithm is used in this prposed structure for further
confirmation of proposed algorithm. Simulation experiments
confirm the effectiveness of the proposed algorithm in terms of
reduced computational complexity and rate of convergence.