Use of hidden Markov chain models for evaluating and selecting codes for fading channels and erasure prediction decoding co-operated with interleaver and HMC

The authors developed a decoder cooperating with an interleaver and a hidden Markov chain modeler (DIHMC) for the decoding of linear block codes on frequency flat correlated Rayleigh fading channels. DIHMC acquires the structure of symbol errors at the demodulator output and exploits it in the decoding process. DIHMC has two decoders. The first decoder corrects only errors. The second decoder re-decodes only the error words out of the first decoder. The hidden Markov process is described as a way of approximating the statistical structure of errors. The DIHMC decoding system, which can acquire the characteristics of the error process on the training model, generate reliability values of symbols, and exploit them in the decoding process, is developed. DIHMC was simulated on a frequency flat correlated Rayleigh and Rican fading channel. DIHMC computes the reliabilities of symbols using the available error record and error structure, and symbols having small reliability are erased.< >