Performance Improvements of Adaptive Fir Equalizer Using Modified Version of VSSLMS Algorithm

Abstract

In this paper possible improvements in the performance of adaptive Linear Equalizer (LE) and Decision Feed Back Equalizer (DFE) are reported. A modified Least Mean Square (LMS) algorithm incorporating a recursively adjusted adaptation step size based on rough estimate of the performance surface gradient square is proposed. The first proposed algorithm was called Adjusted Step Size LMS (ASSLMS) which used single adjusted step size for all weight coefficients. The second proposed algorithm was called Distributed Step Size LMS algorithm (DSSLMS). This algorithm (i.e. DSSLMS) will distribute the resultant variable step size in an exponential form among all weights of the adaptive filter such that each weight coefficient has its own step size. These proposed algorithms through computer simulation results shows favorable performance than traditional LMS algorithm and another Variable Step Size LMS (called VSSLMS) algorithm in terms of fast convergence time, less miss-adjustment in steady state, and good tracking ability.