Grouped Content Compression Coding for Wireless Communications Networks

We propose content compression coding for massive users, where correlated contents of grouped users can be utilized to achieve high compression efficiency. We divide the users into several groups. For each group, we select the group head, encode the information of group head independently, and encode other users referring to the group head. We formulate the grouping optimization problem, transform it to a zero-one discrete optimization problem as well as a subset selection problem, and provide upper and lower bounds on the compression ratio. We investigate the sufficient condition and necessary condition on the optimality of one user group. We also investigate two specific type of sources, with star-type and chain-type user statistics, and provide the optimal user grouping. The compression ratio is also studied from both theoretical and experimental perspectives.