Distributed Signal Detection with Information Sharing Among Local Sensors

The shared nature of wireless channel motivates the study of distributed detection scheme with information sharing among local sensors. The optimal form of local decision rule of binary distributed detection of a known signal is considered, where some sensors broadcast their local decisions to the other sensors, which combine their observations with the information broadcasted by the other sensors to make their local decisions. We show that, in this case, the optimal form of local processing under the Bayesian criterion is a modification to the canonical likelihood ratio test (LRT) and the optimality of LRT quantization do not hold true for sensors listening to another sensor's transmission. Also, a design procedure using the obtained results is developed. Numerical results illustrated the effectiveness of the result in obtaining optimal decision parameters for local sensors.