The Binary Regenerative Channel

The Gilbert burst-noise channel introduced the idea of error states. 1 The error states represent different error processes, each of which generates independent errors. Gilbert's model yields a "renewal error process," that is, an error process for which the gaps between successive errors are independent random variables with the same probability distribution. Elliott 2 introduced a generalization which yields what we shall call a "Markov error process," that is, an error process for which the gaps between errors are dependent random variables with probability distributions which depend only on the last gap between errors. More recently, Elliott used a renewal error process, with component error processes which clo not generate independent errors.3 In order to match experimental data for block * This paper is based on material taken from a dissertation submitted to the Polytechnic Institute of Brooklyn in partial fulfillment of the requirements for the degree of Doctor of Philosophy in 1967. 1713