Analysis of an LMS-based algorithm using reduced number of computations

Many algorithms have been used to improve the performance of the LMS algorithm. One such algorithm is the one given by Mayyas and Khasawnch (1993). In this paper, we give a mathematical analysis for this algorithm, derive a bound on its convergence parameter, /spl delta/, and modify it to reduce the number of computations from (6N-3) to (4N+1). Simulation results presented in this paper serve to justify the improved performance of this algorithm over its predecessor: the LMS algorithm.