Multiterminal Gaussian source coding using pairwise orthogonal transforms

Abstract

The multiterminal source coding problem has gained prominence recently due to its relevance to sensor networks. In general, rate-distortion in multiterminal source coding is an open problem. However, variations of multiterminal source coding, such as the chief executive officer problem, are addressed in the literature. In addition, full and partial cooperation among the sensors, and reliability and robustness against sensor failures are also being investigated in the literature. In this thesis, multiterminal Gaussian source coding was investigated in the context of sensor networks. The rate-distortion region was parameterized for a pair of correlated Gaussian sources using an implicit transform parameter. The approach made use of established results in multiple descriptor coding using pairwise orthogonal transforms.