Turbo compression/joint source–channel coding of correlated binary sources with hidden Markov correlation

We propose a system to perform compression and joint source–channel coding of correlated binary sources when the correlation between the sources is defined by a hidden Markov model. In the case of source coding, punctured turbo codes are used as source encoders, with the objective of compressing the information at rates close to the Slepian–Wolf theoretical limit. The same system structure can be utilized for the transmission of the correlated sources through noisy channels (joint source–channel coding), so that the desired information rates are achieved by puncturing. In both cases, no a priori knowledge about the correlation statistics is required in the encoding process. The source decoder utilizes iterative schemes and does not present significant performance degradation when the correlation parameters are not known in the decoder, since they can be estimated jointly with the iterative decoding process.