Information integrity between correlated sources through Wyner-Ziv coding

In this paper, the problem of achieving information integrity in the classical Wyner-Ziv lossy source coding model is considered. A discrete memoryless source encodes its source sequence using the standard Wyner-Ziv coding technique and forwards the bin index to another source that employs the bin index and its own correlated sequence to decode the first source's sequence. There is an adversary who may arbitrarily modify the bin index sent to the decoder. It is assumed the adversary may base the attack on the true bin index, its own sequence (which may be correlated to both the encoder and decoder sequences) as well as the known codebook used by the sources. It is shown that a non-simulatability condition on the joint source distribution is necessary and sufficient to achieve information integrity. However, the additional requirement of information integrity may incur a penalty on the rate-distortion performance.