Robust Constrained Recursive Least P-Power Algorithm for Adaptive Filtering

Abstract

In this paper, we develop a novel constrained adaptive filtering algorithm called constrained recursive least p-power (CRLP) algorithm, which incorporates a set of linear constraints into the least mean p-power error (LMP) criterion to solve a constrained optimization problem directly. Compared with the conventional constrained adaptive filtering algorithms including constrained least mean square (CLMS), constrained recursive least square (CRLS) and constrained least mean p-power (CLMP), CRLP can achieve better performance under non- Gaussian noises. Simulation results are presented to confirm the superior performance of the new algorithm.