A new proportionate normalized least mean square algorithm for high measurement noise

摘要

In this paper, we derive a new improved proportionate normalized least mean square (IPNLMS) algorithm with unconventional minimization criterion that minimizes the summation of each squared Euclidean norm of difference between the currently updated coefficient vector and past coefficient vectors, which is called the improved IPNLMS (I-IPNLMS) algorithm. Simulation results demonstrate that the proposed I-IPNLMS algorithm has the superiority of the lower misalignment than the conventional IPNLMS algorithm in the context of sparse system identification with a low signal-noise-ratio (SNR).