The detection performance of a cell-averaging CFAR detector is analysed for spatially correlated K-distributed clutter. In thermal noise or uniform clutter, the performance of the cell-averaging CFAR can be compared with that of the ideal fixed threshold, but with an associated 'CFAR loss', quantified by the increase in average signal-to-clutter ratio required to achieve a given probability of detection and probability of false alarm. In some types of clutter, knowledge of the overall amplitude distribution alone is not sufficient to assess the performance of a cell-averaging CFAR. In particular, sea clutter may exhibit significant spatial correlation, often associated with the sea swell. In such spatially varying clutter, the cell-averaging CFAR may be able to follow the local fluctuations, giving a considerable improvement in performance compared to the 'ideal' fixed threshold. This is quantified as a 'CFAR gain' with the limit of such improvement given by the performance of the 'ideal CFAR' detector. The CFAR gain or loss is illustrated in different clutter conditions for a range of cell-averager configurations. It is shown that significant CFAR gain, compared with the commonly accepted CFAR loss, can be achieved in some circumstances.
展开▼
