Time-Varying Proportionate Least Mean Square Algorithm for an Adaptive Noise Canceller

Abstract

A novel time-varying proportionate least mean square algorithm (TVPLMS) is proposed in this paper which depends on a time-varying step size during adaptation. The proposed algorithm addresses the issues arising due to the presence of sparsity and the effect of Gaussian noise in physical systems. The proposed TVPLMS algorithm is a modified version of the proportionate least mean square algorithm (PNLMS) which uses a gain matrix proportional to the weight coefficients. Moreover, the PNLMS algorithm is a prominent algorithm developed for systems with sparse impulse response. In this paper, the proposed TVPLMS algorithm is used for an adaptive noise canceller (ANC). Simulation experiments to analyze the behavior of the proposed TVPLMS algorithm for the sparse system show its enhanced behavior.