Unified Bayesian Multisensor-Multitarget Sensor Management for BMD

"Multisensor-multitarget sensor management presents amajor theoretical and practical challenge for ballistic missile detection,tracking, and discrimination. A common approach is to assemble a patchworkof heuristic "bottom up'' techniques---e.g., loosely integrating manydistinct algorithms, each of which addresses some specific part of theproblem (detection, tracking, sensor cueing, allocating and schedulingsensor dwells, allocating and scheduling platform flight paths, etc.).Scientific Systems Co., Inc. (SSCI) and its subcontractor Lockheed MartinTactical Systems (LMTS) believe that a theoretically rigorous approach tomultisensor-multitarget sensor management for ballistic missile defense isnow feasible. Sensor management is inherently a problem innonlinear adaptive control theory in which the observations, the datasources being controlled, and the targets being tracked by the controlprocess are all randomly-varying multi-object systems. We use aBayesian approach, meaning in particular that any control-theoreticobjective function must be some statistical moment of the time-evolvingmultitarget posterior distribution. We propose the use of these objectivefunctions in conjunction with Multi-Hypothesis Correlator (MHC) algorithms,in which case they become potentially computationally tractable. Detection and tracking are someof the key technologies for global surveillance, precision strike, airsuperiority and defense, which are three of