Sensor Calculus: Considerations in the Design of Distributed Systems for Detection, Discrimination and Decision

This research applies optimal control and experimental design theories to management of a multidetector system including the cost of false alarms, misses, communications and improvements in operating characteristics. The successful interaction of multiple detectors is applicable to battle management and control. A powerful new framework is presented for the analysis of distributed detection networks, supported by a compact notation for the description of complicated networks. The new methods are applicable to problems with any number of threats, any number of messages and any number of available actions. As illustrations the Sensor Calculus is applied to reveal interesting features of the case of two-fold threats, messages and actions, including symmetry breaking and suboptimality of maximum entropy principle, multidetector systems, optimal control theory probability, programming, signaling and action set, theory of experimental design.