Convergence Evaluation of a Variable Step-Size LMSE Adaptive Switching Algorithm

A simple and robust variable step-size normalized switching adaptive algorithm is proposed here. The use of variable step-size in the adaptation process of least mean switched error (LMSE) algorithm (VSS-LMSE) is investigated. The switching algorithm consists of applying the least mean fourth (LMF) algorithm and switching to the least mean square (LMS) algorithm when the absolute value of error is greater than 1. The LMSE algorithm with a fixed step-size usually results in a trade-off between the residual error and the convergence speed of the algorithm. The VSS-LMSE algorithm presented here will eliminate much of this trade-off. In this paper the MSE of using the VSS-LMSF algorithm in the adaptation process of system identification over a dispersive channel is investigated. The step-size variation makes it possible for the VSS-LMSE algorithm to converge faster and to a lower steady state error than in the fixed step-size case. Moreover the proposed VSS-LMSE algorithm has a much lower steady state error than that in the case of VSS-NLMS algorithm.