Closed-loop control for joint air operations

This paper focuses on the problem of providing real-time, closed-loop feedback control of joint air operations (JAO) via near-optimal mission assignments. For this application, a rollout algorithm is employed which is based on the theory of stochastic dynamic programming. The primary benefits of this technology are agile and stable control of distributed stochastic systems. The rollout algorithm is applied to a small JAO, scenario that includes limited assets' risk/reward that is dependent on mission composition, basic threat avoidance routing, and multiple targets, some of which are fleeting and emerging. Simulation results illustrate the benefits of the closed-loop feedback control. It is shown that the rollout strategy provides statistically significant performance improvements over an open-loop feedback strategy that uses the same baseline heuristic. The performance improvements are attributed to the fact that the rollout algorithm was able to learn near-optimal behaviors that were not modeled in the baseline heuristic.