Recursive Solution of Distributed Detection/Communication Problems

Each member of a team of decision agents receives a conditionally independent observation about some underlying discrete hypothesis. Subject to causality constraints, the agents seek to optimize a team performance index by making (discrete) decisions which are conveyed to other agents over capacity constrained communication links. This paper presents a general recursive methodology for computing the optimal decision rules for each node. This is achieved by recasting the stochastic optimal control problem as an equivalent deterministic one. Decomposability of the performance index permits the decomposition of the optimal control problem into a set of subproblems which can be recursively solved by employing a spatial dynamic programming algorithm.