New Performance Enhancement of Adaptive IIR Filtering Applications

Abstract

In this paper, a modified version adaptive Infinite Impulse Response Least Mean Square (IIR-LMS) is presented. This new proposed algorithm tries to enhance the performance of Previously Proposed LMS (PPLMS) algorithm by overcoming and avoid some of its drawbacks which are a higher level of miss-adjustment at steady state, and the need to know a statistical feature of the input signal in order to calculate the diagonal convergence factor matrix (MMAX). The new proposed algorithm is called Fast Adaptive LMS (FALMS), which uses an appropriate time-varying value of the M(k)MAX instead of fixed value(i.e., MMAX). M(k)MAX will be defined by the energy of the input signal. The FALMS express performance improvement such as fast convergence speed and minimize the level of miss-adjustment compared to IIR-LMS, IIR-NLMS(Normalized LMS) and PPLMS for system identification IIR application.