가변 parameter reweighting 기반 p-norm constraint 가변 스텝사이즈 LMS 알고리즘

Abstract

In this paper, a variable parameter reweighting for p-norm constraint variable step-size least mean square (p-NCVSSLMS) algorithm is proposed to improve the convergence performance. More specifically, the proposed variable reweighted parameter is determined according to a power of a ratio of absolute value of the individual coefficient to expected absolute value of the coefficients. As a result, it is demonstrated from the simulation results that, compared with the conventional sparsity-aware algorithm, the proposed algorithm yields better convergence performance in various sparse system environments.