The normalized least mean square (NLMS) algorithm is an important variant o
f the classical LMS algorithm for adaptive linear filtering. It possesses m
any advantages over the LR IS algorithm, including having a faster converge
nce and providing for an automatic time-arising choice of the LMS step-size
parameter that affects the stability, steady-state mean square error (MSE)
, and convergence speed of the algorithm. An auxiliary fixed step-size that
is often introduced in the NLMS algorithm has the advantage that its stabi
lity region (step-size range for algorithm stability) is independent of the
signal statistics.
In this paper, we generalize the NLMS algorithm by deriving a class of nonl
inear normalized LMS-type (NLMS-type) algorithms that are applicable to a n
ide variety of nonlinear filter structures. We obtain a general nonlinear N
LMS-type algorithm by choosing an optimal time-varying step-size that minim
izes the nest-step MSE at each iteration of the general nonlinear LMS-type
algorithm, As in the linear case, we introduce a dimensionless auxiliary st
ep-size whose stability range is independent of the signal statistics. The
stability region could therefore be determined empirically for any given no
nlinear filter type, We present computer simulations of these algorithms fo
r two specific nonlinear filter structures: Volterra filters and the recent
ly proposed class of Myriad filters. These simulations indicate that the NL
MS-type algorithms, in general, converge faster than their LMS-type counter
parts.