Iterative joint channel decoding of correlated sources

In this article we exploit the potential correlation existing between multiple information sources to achieve additional coding gains from the channel codes used for data protection. We do not assume the existence of, nor do we use channel sideinformation at the receiver. Instead, empirical estimates of the cross-correlation are used in partial decoding steps in an iterative joint soft decoding paradigm. Experimental results suggest that relatively few iterations (2 to 4) are sufficient to reap significant gains using this approach specially when the sources are highly correlated. Finally, we provide analytical performance bounds of the proposed technique showing a close match with the simulation results at sufficiently high SNR.