Distortion sum-rate performance of successive coding strategy in quadratic gaussian CEO problem

摘要

We consider a distributed sensor network, modeled by the CEO problem, in which each sensor communicates its observation to the fusion center (FC) using limited transmission rate. Based on the successive coding strategy, we obtain the optimal rate allocation strategy for the Gaussian CEO problem which minimizes the average distortion in the source estimate produced by the FC. This strategy can be simplified in a general parallel sensor network with L sensors by assigning equal rates to sensors if the sum-rate R macr is very large given a fixed L or if L is very large given a fixed R macr.