A Robust, Iteration Dependent Variable Step-Size (RID-VSS) Least-Mean Square (LMS) Adaptive Algorithm

Abstract

A robust variable step-size LMS algorithm is proposed. The variable step-size is a weighted running average of the squared error signal which varies as the square of the estimated error changes. The weighting factor ensures stability and convergence. Robustness of the algorithm is achieved through the step-size inherent bounded nature and independence from the initial condition. The algorithm is compared with two benchmark VSS algorithms. Convergence and steady-state behavior of the proposed adaptive filter are analyzed. Simulation for the system identification scenario is carried out and the performance of the proposed algorithm is compared.