Distributed Radar Target Detection with Ordered Local Statistics*

In distributed radar with local radar sites using cell-average constant false alarm rate (CA-CFAR) algorithm, the signal fusion based detection algorithm has been studied. However, the order statistic (OS) CFAR algorithm, as another important detection algorithm, has not been considered in distributed detection with signal fusion. This paper presents two algorithms for distributed detection with OS-CFAR test statistics. The first algorithm is that it randomly selects multiple values of order statistics to estimate the noise power through the maximum likelihood estimation (MLE) criterion. The second algorithm selects one of order statistics to estimate the noise power, and the optimum detection statistic for the fusion center is developed based on the Neyman-Pearson criterion. In numerical results, an iterative method is used to estimate the decision thresholds and the location of the selected value in order statistic has been verified to have a significant impact on detection performance. In particular, performance of the optimum detection statistic in distributed OS-CFAR is superior to suboptimum statistic and nonparametric statistic when choosing value at the front of the order statistics, while the conclusion is opposite when choosing value at the end of the order statistics.