Joint source-channel decoding of correlated sources over ISI channels

In this paper we propose to combine Turbo equalization and source-channel coding for the transmission of correlated sources over sensor networks. Concretely, we study an asymmetrical scheme where two correlated sources, S ₁ and S₂, have to be transmitted to a central node. Source S₁ is compressed and transmitted over a Gaussian ISI channel, whereas the other source S₂ is assumed to be available as side information at the receiver. The correlation between sources has memory and is modelled by a Hidden Markov Model (HMM). For Maximum a Posteriori (MAP) and Minimum Mean Squared Error (MMSE) equalization techniques, we propose a joint equalization and decoding scheme that achieves reliable communication of source S₁ at signal to noise ratios close to the theoretical bounds set by the combination of the Slepian-Wolf and Shannon theorems.