Remote source coding with two-sided information

Basak Guler, Ebrahim MolavianJazi, Aylin Yener

Abstract: This paper studies the impact of side information on the lossy compression of a remote source, one which is indirectly accessed by the encoder. In particular, we identify the conditions under which sharing side information between the encoder and the decoder may be superior or inferior to having two-sided, i.e., correlated but not identical, side information. As a special case, we characterize the optimal rate-distortion function for a direct binary source with two-sided information by proposing an achievable scheme and proving a converse. This example suggests a hierarchy on the impact of side information, in that the performance is mainly determined by how well the decoder learns about the source and then by how well the encoder learns about the decoder's observation.