Joint detection and estimation fusion in distributed multiple sensor systems

In this paper, we address the problem of the joint detection and estimation fusion in distributed multiple sensor systems. The objective is to decide between two hypotheses and make the corresponding estimates if the decided hypothesis has unknown parameters at the same time. The fusion system comprises N independent sensors, and each one sends a binary decision to the fusion center for integration. The traditional approaches treat the detection and estimation separately, and tend to place the primary emphasis on the detection part through maximizing detection probability and then treat the estimation part as suboptimal. The proposed approach treats these two parts simultaneously in a jointly optimal manner. The average estimated cost concerning about the random decision rule is minimized subject to the corresponding constraints with respect to the false alarm probability and the probability of missed detection. The resulting scheme shows that a trade-off between detection and estimation performance can be achieved. A hypothesis test example with unknown parameters based on multi-sensor quantized measurements is given, which demonstrates the flexibility and the superiority of the proposed approach compared with the classical Neyman-Pearson and the GLRT.