Optimization of a parallel distributed detection system using distance measures

In this paper, we use both JJ-divergence and Bhattacharyya coefficient as distance measures to optimize a parallel distributed detection system consisting of NN local detectors and a global decision maker. The optimization is obtained by allowing the thresholds of the local detectors to be functions of the global probability of false alarm. Numerical results for the case of four and six local detectors are obtained. These results indicate the superiority of our approach over those reported in the literature in which the local thresholds were specified without taking the level of global probability of false alarm into account.