l/sub 1/-optimal deconvolution with stable predictor polynomials

Abstract

The problem of l/sub 1/-optimal deconvolution with stable predictor polynomials is addressed. It is shown that the problem of l/sub 1/ deconvolution can be formulated as a standard model-matching problem that can be solved using some results from the literature. The authors consider how to stabilize the resulting predictor polynomial in the event that it is unstable. Two equivalent existence conditions which indicate when it is possible to stabilize the predictor are presented. The Nehari theorem is then used to give a procedure to find a filter whose first M coefficients match those of the l/sub 1/-optimal filter, but which will exhibit a stable predictor polynomial.<>