Methodology for the Design and Optimal Placement of Point Detectors in aDistributed Detection System for Remote Defence Against Biological Warfare Agent Releases

摘要

This report deals with the design and performance evaluation of a distributeddetection system for a dispersing biological warfare (BW) cloud embedded in the natural aerosol component of the background air. The distributed system employs a number of physically separated BW agent point detectors (sentries) located within some target region and a data fusion center that provides the final decision as to the presence or absence of the bio-target by combining the individual localized decisions from the various point detectors using a prespecified combining strategy. In this system, each detector implements a generalized likelihood ratio test on its own localized observations to test for the presence or absence of a bio-target. These localized detection decisions are then transmitted to a data fusion center where they are logically combined to yield a global detection decision for the distributed system. The optimization of the global detection performance of the distributed system is derived by application of the Lagrange multipliers method, whereby the global probability of detection for the system is maximized subject to the constraint that the global probability of false alarm is maintained at a prespecified level (viz., at a constant and tolerable false alarm rate). A number of combining strategies are investigated in order to determine some overall system optimality for detection. It is found that the optimal detection strategy for each individual BW agent detector in the distributed system depends on the strategy of all the other detectors, as well as on the structure of the data fusion center.