Orthogonalization of Gaussian signals for Volterra system identification

Abstract

This paper presents a very simple method for orthogonalizing Gaussian input signals for identification of truncated Volterra systems. The orthogonalization requires a linear lattice predictor and some nonlinear processing of the backward prediction error signals of the various orders. However, the nonlinear processors do not depend on the statistics of the input signals, and consequently, are easy to design and implement.<>