Radar Resource Management for Target Tracking-A Stochastic Control Approach

This chapter develops a stochastic control formalism for radar resource management. It formulates the sensor management problem as a two‐timescale stochastic control problem. Two alternative formulations are given for the micromanagement problem, one formulation deals with maximizing the mutual information of multiple targets, the other formulation deals with a partially observed Markov decision processes (POMDPs) setup involving a finite state Markov chain. The chapter illustrates how to parameterize such a monotone policy, so that the optimal parameterized policy can be computed via simulation‐based stochastic optimization. The chapter gives sufficient conditions under which the optimal micromanagement policy is monotone with respect to the state. It presents an alternative formulation of sensor management for a maneuvering target modeled as a jump Markov linear system (JMLS) and describes several myopic sensor management algorithms. The micromanagement problem deals with scheduling the optimal Bayesian filter while the macromanagement problem deals with allocating target priority.