Ieee Transactions on Systems Man and Cybernetics - Part A

| The purpose of this paper is to propose a solution methodology for a missile defense problem involving the sequential allocation of defensive resources over a series of engagements. The problem is cast as a dynamic program-ming/Markovian decision problem, which is computation-ally intractable by exact methods because of its large number of states and its complex modeling issues. We have employed a Neuro-Dynamic Programming (NDP) framework, whereby the cost-to-go function is approximated using neu-ral network architectures that are trained on simulated data. We report on the performance obtained using several different training methods, and we compare this performance with the optimal.