A new convergence factor for adaptive filters

Abstract

It is proposed that the convergence factor mu in the adaptive filter can be made time-varying according to the gradient of the performance surface. Computer simulations show that when a suitable damping parameter, alpha , is used, the algorithm possesses a fast convergence rate and yields very small misadjustment. The algorithm works well under nonstationary environments, in sharp contrast with the two-stage method.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>