AN EQUIVALENT MARKOV MODEL FOR BURST ERRORS IN DIGITAL CHANNELS

A Hidden Marker Model for burst errors is specified by a probability t ransition matrix Pt an initial probability vector p, and the state dep endent probability of error matrix B. Several procedures are available for estimating P, p and B from a given error (observation) sequence. However, even with some restrictions on the structure of the underlyin g Marker models, the estimation procedures are computationally intensi ve particularly when the observation sequence contains long strings of identical symbols. In this paper we show that, under some mild assump tions, a Markov model with an arbitrary transition matrix P is equival ent to a Markov model with a unique ''block diagonal'' transition matr ix Lambda. We also present a computationally very efficient algorithm for estimating Lambda from a set of observations using a modified Baum -Welch algorithm.