Adaptive system identification using fuzzy inference based LMS algorithm

Abstract

A fuzzy LMS algorithm is presented. The algorithm uses fuzzy inference to obtain the adaptive gain at every iteration of the LMS algorithm. The adaptive gain is obtained by a fuzzy system which has two inputs, an error signal and a one-step previous adaptive gain. This algorithm is applied to adaptive system identification of a hybrid system in a telecommunication network. In the experimental results, the proposed fuzzy LMS adaptive algorithm shows a better performance than than the LMS algorithm in terms of the convergence speed.