IDENTIFICATION OF NONMINIMUM-PHASE FIR SYSTEMS VIA 4TH-ORDER CUMULANTS AND GENETIC ALGORITHM

In this paper, the problem of estimating the parameters of an FIR syst em from only the fourth-order cumulants of the noisy system output is considered. The FIR system is driven by a symmetric, independent, and identically distributed (i.i.d) non-Gaussian sequence. We propose a ne w formula called Weighted Overdetermined C(q,k) (WOC(q,k)) by extendin g the conventional C(q,k) formula. The optimal selection of the weight s in WOC(q,k) is performed by using the Genetic Algorithm (GA) optimiz ation method which minimizes a nonlinear error function based on the f ourth-order cumulants alone. Simulations are provided to reveal the ef fectiveness and the superiority of this novel technique over the C(q,k ) and other existing techniques. (C) 1997 Elsevier Science B.V.