Errors-and-erasures decoding of block codes in hidden Markov channels

Abstract

Most investigation on the performance of block codes assume perfect interleaving and memoryless channels that introduce only random errors. Some authors investigate the effect of finite interleaving size on the decoder performance, using a two-state Gilbert-Elliott (GE) model to characterize the channel effect. However, there are circumstances that the burst-error channel of concern can not be properly described by a two-state GE model. This paper extends earlier analysis to investigate the performance of block codes in an environment that can be modeled as an arbitrary finite-state Markov chain.