Extended Look-Ahead Sensor Management for Missile Defense with a Bayes Net Battle Model

"Effective missile defense relies on an array of sensors performing diverse collection tasks to successfully engage a missile attack. Sensors cannot cover the entire battle volume continuously, and Sensor Management is needed to allocate sensors to tasksproviding the greatest mission utility. Prior research has shown two ways to achieve effective Sensor Management: (1) integrate sensor allocation with weapon-target pairing and (2) use extended look-ahead horizons to make optimal use of defense-in-depthtactics. Tight integration of sensor and weapon tasking yields superior task plans, but has computational requirements that grow exponentially with the length of the look-ahead horizon, and is impractical for real-time execution.Our innovation for Sensor Management is to use extended look-ahead horizons, and partially de-couple the sensor and weapon tasking problems. Partially de-coupling the sensor and weapon task planning into separate, coordinating sub-problems reducescomputational requirements by several orders of magnitude while preserving much of the mission effectiveness of tight integration. Partially de-coupling the problems makes it possible to use extended look-ahead horizons, and thereby capture the benefitsof defense-in-depth tactics. Sensor Management is an essential activity in complex military and commercial systems, and is most effective when sensor tasking is aligned with the mission needs for information. Our innovat