Optimization of distributed detection systems under the minimum average misclassification risk criterion

A common model for distributed detection systems is that of several separated sensors each of which measures some observable, quantizes it, and communicates to a fusion center the quantized observation. The fusion center collects the quantized observations and takes the decision. The article deals with the design of the quantizers and of the fusion center under a rate constraint. The system of interest allows soft nonbreakpoint quantizers and nonindependent observations. Our finding is that locally optimal design of the distributed detection system is feasible via alternate minimization of the average misclassification risk.