Adaptive Decision Fusion in Detection Networks

In a detection network, the final decision is made by fusing the decisions from local detectors. The objective of that decision is to minimize the final error probability. To implement and optimal fusion rule, the performance of each detector, i.e. its probability of false alarm and its probability of missed detection as well as the a priori probabilities of the hypotheses, must be known. However, these statistics are usually unknown or may vary with time. In this paper, we develop a recursive algorithm that adapts the fusion center. This approach is based on the time-averaging of local decisions and on using the analytic solutions that guarantee the asymptotic convergence. Also a simple method is proposed that enables the algorithm to track changes faster. Simulation results are presented to demonstrate the efficiency and convergence properties of the algorithm.