A new structured HMM model applied to burst errors in Viterbi decoding

A novel structured Markovian model for burst-error channel has been proposed, which mimics the generation of typical patterns of burst-error sequences and has only 7 parameters to be adjusted. Simple expressions for ML estimation of those parameters have been derived and applied to model the burst error processes at the output of a Viterbi decoder. Numerical results of a comparison with an usual Fritchman model showed that the proposed model is superior in terms of reproducing the statistical properties of this processes. A significant reduction in the computational effort for parameter estimation was also obtained.