Adaptive Target Detection in Sensor Networks

In a sensor detection network, each sensor makes a binary decision based on its own observation, and then communicates its binary decision to a fusion center, where the final decision is made. To implement an optimal fusion center, the performance of each sensor 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, I will introduce a recursive algorithm that approximates these values on-line and adapts the fusion center. This approach is based on time-averaging of the local decisions and using them to estimate the error probabilities and a priori probabilities of the hypotheses. This method is efficient and its asymptotic convergence is guaranteed.