Distributed CFAR Detector Based on Local Test Statistics

This paper studies the design of distributed constant false alarm rate (CFAR) detection systems where the fusing center receives all of the local test statistics. On the assumption that the clutter power level of the background in each local detector is identical and the returns of the test cells of different sensors are all independent and identically distributed, we first work out the closedform expressions of the probabilities of detection (PD) and probabilities of false alarm (PFA) for the fusing center when the local processors use cell average (CA) detectors. By using the differential characteristics of frequencydomain in Laplace transform the universal fusing method is extended to other local CFAR detectors such as maximum likely (ML) and ordered statistic (OS) detectors of any amount or form. This method successfully avoids the assumption of equivalent signaltoclutter ratio (SCR) in local detectors for traditional fusing center which uses MOS method. The experimental results indicate the validity of the new algorithm proposed in the paper.