The Improved μ-law PNLMS Algorithm with A Variable Parameter

摘要

Inspired by adaptive MPNLMS (AMPNLMS) algorithm, we propose an algorithm to improve the convergence performance of the improved u-law PNLMS algorithm (IMPNLMS) for non-sparse impulse responses. In this paper, a variable parameter u of the u-law compression in AMPNLMS algorithm for sparse impulse response is incorporated into the IMPNLMS algorithm. In proposed algorithm, the parameter u is not constant, but time varying and adaptive. Compared with IMPNLMS algorithm, simulation results demonstrate the proposed IMPNLMS algorithm has better convergence performance. Line segment approximation is discussed to bring down the computational complexity.