Title

ON THE ANALYSIS OF MISMATCHED DPCM FOR GAUSS MARKOV SOURCES.

Document Type

Conference Proceeding

Publication Date

12-1-1986

Abstract

We study a DPCM system with a first order predictor and a binary symmetric quantizer. The input process is assumed to be first order autoregressive source. In contrast with the previous work we have not assumed that the predictor coefficient is matched to the source. The stationary distribution of the joint input-reproduction process is studied. It can be shown that for small prediction coefficients this stationary distribution is singularly continuous with respect to the Lebesgue measure. For large prediction coefficients we believe that the stationary distribution is absolutely continuous with respect to the Lebesgue measure. For the later case and with the assumption of absolute continuity we use the projection method to solve the Chapman Kolmogorov equation for the stationary probability density function of the input-reproduction process. We then evaluate the distortion incurred between the input process and the reproduction process. Next we use a recursive algorithm to optimize the quantizer.

First Page

409

Last Page

414

This document is currently not available here.

COinS