Vector Wyner-Ziv coding for vector Gaussian CEO problem

Wyner-Ziv coding (WZC) is a compression technique which uses decoder side-information to help reconstruction. The vector WZC is a key element to solve the vector Gaussian CEO problem for sensor networks. Compared with the scalar CEO problem, in the vector CEO problem sensor node array is used by each agent and a vector source is to be estimated. In this paper, we present a nested-lattice based vector WZC structure, which is a generalization of the scalar WZC proposed by Zamir et al. The key is to explore the duality between the WZC problem and the dirty paper coding (DPC) problem. This coding structure is shown to be able to achieve the Berger-Tung sum rate, which is the best known compression rate for the CEO problem.