Optimization of IIR polynomial predictive filter magnitude response

Polynomial Predictive Filters (PPFs) are capable of predicting future values of a signal that can be modeled as a polynomial of a low degree. FIR PPFs succeed at predicting the signal, but suffer from poor stopband attenuation and exhibit a disturbing passband gain peak, The introduction of feedback terms to the basic FIR structure effectively remedies these problems, but no well-developed method has been presented for finding the optimal feedback coefficients. In this paper, we present a practical method for optimizing IIR PPF feedback coefficients based on an arbitrary, lowpass-type magnitude response function supplied by the user. A general p-norm error criterion is used, which enables targeting both L-2 and near-L-infinity solutions. The proposed method is illustrated by the means of a practical example. (C) 1997 Elsevier Science B.V.