OS characterization for local CFAR detection

An order-statistic (OS) characterization for modeling clutter statistics at local detectors is presented. Important features of the OS characterization include its structure, which allows for parallel computations, and its degrees of freedom, which enable it to track various changes in the clutter statistics. Constant-false-alarm rate (CFAR) performance for local detectors designed from the OS characterization is compared to that of conventional CFAR detectors that operate locally at each receiver. The results demonstrate the ability of the local OS detectors to adapt to changes in the skewness of the clutter distribution. For the cases tested, the local OS detectors maintain the expected value for the false alarm probability better than the local conventional CFAR detectors by factors ranging from 2 to 100. OS detectors also provide information to the data fusion center that can be useful in discriminating different types of interference and detecting small targets in clutter. Examples of utilizing OS detections at the data fusion center are discussed