A new variable step size least mean squares (LMS) algorithm using fuzz
y logic (FVSS-LMS) is presented. The change of the step size at each i
teration, which increases or decreases according to the degree of misa
daption, is computed by a proportional fuzzy logic controller. As a re
sult the algorithm has very good convergence speed while preserving lo
w misadjustment. The norm of the cross correlation between the estimat
ion error and input data is used as a measure of misadaption. Simulati
on results are presented to verify the performance of the proposed alg
orithm. (C) 1996 John Wiley and Sons, Inc.