Multiterminal source coding: Theory, code design and applications

Summary form only given. Distributed source coding (DSC) considers the problem of separate compression and joint decompression of multiple correlated sources. Berger and Tung first introduced the multiterminal (MT) source coding problem and studied the region of all possible encoder rate tuples such that the sources are reconstructed under distortion constraints. One can loosely view MT source coding as the lossy version of Slepian-Wolf coding. Applications of MT source coding include distributed video sensor networks and MT/multiview video coding for camera arrays. Despite recent progresses, finding the sum-rate bound of quadratic Gaussian MT source coding with more than two terminals is still an open problem. In this talk, I will present our recent theoretical results on quadratic Gaussian MT source coding, limit-approaching code designs, and MT video coding (including the case with depth camera assistance).