Markov Decision Process for sonobuoy transmission scheduling

Assumed here is a multi-static sonar system consisting of a distributed field of emitter and receiver sonobuoys within a defined search area. The field is deployed to search for and estimate the location, heading, and speed of underwater targets within the search area. We describe a method for optimizing, over a long time horizon, choice of transmission locations, and of waveforms from a small waveform library to detect and estimate the states of multiple targets. The method used is based on Markov Decision Processes as a structure to describe the overall scheduling problem. The key innovations are in simplifications of the state space used in seeking the optimal schedule, which enable a computationally feasible algorithm.