Bipartite HMM Model for Burst Errors Focused on the Generation of Gaps and Clusters

A new Hidden Markov Model (HMM) for burst errors is proposed. This model is based on a compact representation of the error sequence in terms of succeeding pairs of clusters and gaps lengths. Its Markov chain has two classes of states associated to the generation of gaps and clusters lengths, respectively. The proposed model may be characterized by few parameters. An algorithm for ML (“Maximum Likelihood”) estimation of these parameters on the grounds of the EM (“ExpectationMaximization”) approach is derived. Some preliminary results of performance evaluation show that the model and the estimation algorithm here presented provide a flexible and efficient tool for capturing and reproducing statistics of interest in the context of burst errors modelling.