Adaptive Slepian-Wolf Decoding using Particle Filtering based Belief Propagation

A major difficulty that plagues the practical use of Slepian-Wolf coding (and distributed source coding in general) is that the precise correlation among sources need to be known a priori. To resolve this problem, we propose an adaptive Slepian-Wolf decoder using particle filtering based belief propagation. We show through experiments that the proposed algorithm can simultaneously reconstruct a compressed source and estimate the joint correlation between the sources. Further, comparing to the conventional Slepian-Wolf coder based on standard belief propagation, the proposed approach can achieve higher compression under varying correlation statistics.