Joint Source Channel Decoding of Turbo Codes

We present techniques for modifying a turbo decoder to incorporate a hidden Markov source model, a Markov channel model, or both. This allows the receiver to utilize the statistical characteristics of the source and/or channel during the decoding process, and leads to significantly improved performance relative to systems in which the Markov properties are not exploited. When applied for binary-input, binaryoutput Markov channels, the methods presented here outperform techniques based on conventional channel interleaving, as they can allow decoding at rates which are above the capacity of a memoryless channel with the same stationary bit error probability as the Markov channel.