On distributed arithmetic codes and syndrome based turbo codes for Slepian-Wolf coding of non uniform sources

We consider the use of source codes and channel codes for asymmetric distributed source coding of non uniform correlated sources. In particular, we use distributed arithmetic codes as source codes and syndrome based turbo codes as channel codes. We compare the advantages and drawbacks of the two systems for different source probabilities and different compression ratio. We show that prior knowledge of the source distribution improves the performance of both approaches in terms of their distances to the Slepian-Wolf bound. Turbo codes are better when the puncturing is low, while distributed arithmetic codes are less impacted by the change of compression rate.