An adaptive estimation method using instrumental variables and deconvolution

Abstract

AbstractThe performance of the methods of recursive instrumental variables (RIV) and minimum variance deconvolution (MVD) is examined when the methods are interwoven so as to provide adaptive estimation of the ARMA model parameters of a system driven by white unmeasurable noise. It is shown that the methods have some similar properties which permit their successful combination in most cases. The asymptotic properties of the resulting estimation approach are evaluated in terms of the ARMA model parameters or the pole‐zero locations, the signal‐to‐noise ratio and the frequency bandwidth of the system impulse response. An analysis of the second‐order ARMA model and various simulation examples are presented which illustrate the derived results. The method has been successfully applied to the problem of adaptive seismic signal deconvolution.