分布式自动删除平均恒虚警率检测技术

Distributed Automatic Censored Cell-averaging CFAR Detector

根据自动删除平均算法提出了一种新的分布式多传感器的目标检测算法.在该方法中,首先根据自动删除平均算法(Censored cell-averaging,CCA)得到各传感器的杂波/噪声电平估计,然后将检测单元电平与得到的杂波/噪声电平估计值相比较,得到有无目标的局部判决,并将其传送到融合中心.融合中心采用"k/N"融合准则得到有无目标的全局判决.其中,自动删除平均算法的优势明显,它不需要干扰的先验信息,可以容纳的干扰目标数不会像顺序统计量OS(k)(Order statistics)方法那样受指定k值的限制,更接近实际.自动删除平均算法还可以检测本身可能是目标的干扰.在假定目标服从Swerling 2型起伏的情况下,导出了相应的检测概率与虚警概率解析表达式.多种检测器数值和图表分析的比较结果表明了该方法的有效性和优越性.

Based on automatic censored cell-averaging constant false alarm rate （CCA-CFAR） technique, a new distributed CCA-CFAR detector is presented. In the scheme, every local detector employs automatic censored cell-averaging CFAR algorithm to form estimate of clutter power level, and compares it with the test sample. Then each local detector makes local binary decision, and transmits it to the fusion center. Finally, the fusion center makes the overall decision based on the total local decisions. The overall decision, which is zero or one, is obtained at the dada fusion center based on the ＂k/N＂ fusion rule. We analyze the performance of the detector in various environments including homogeneous, multiple target, and clutter edge cases. The results show that for the multiple interfering target situation, it exhibits good robustness. The attractive feature of the proposed detector is that they do not require a priori knowledge about the interference in order to perform well. Under Swerling 2 assumption, the analytic expressions of detection probability and false alarm probability are derived.