Performance study of three different sparse adaptive filtering algorithms for echo cancellation in long acoustic impulse responses

Abstract

In this paper, the problem of echo cancellation in long acoustic impulse responses (AIRs) is highlighted. Three of the mostly-used recent NLMS-based sparse adaptive filtering algorithms are presented; and their performances in the context of acoustic echo cancellation (AEC) are studied and compared. The algorithms of interest include the improved proportionate normalized least mean square (IPNLMS), its sparseness-controlled (SC) upgrade (SC-IPNLMS) as well as the so-called variable-step-size reweighted zero-attractor NLMS (VSS-RZA-NLMS) which is based on the compressive sensing (CS) framework. Series of simulations were carried out both in synthetic and real different-sparseness long acoustic impulse responses with stationary and non-stationary inputs in order to effectively analyze, evaluate and compare the strengths and the weaknesses of these algorithms in terms of convergence speed, steady-state performance and computational complexity.