CFAR detection for multiple target situations

In the paper, three detectors are proposed and studied in the presence of interfering targets in the reference cells. In the weighted cell-averaging constant false-alarm rate (WCA-CFAR) detector the means generated by the leading and the lagging range cells are weighted in accordance with the level of the interference. Optimum weights that maintain CFAR are obtained while the probability of detection is maximised. The censored mean level detector (CMLD) is considered next, this requires exact knowledge of the number of interfering targets. The detection performance of the CMLD is studied for Swerling IV targets and compared to that with Swerling II targets. Then, the generalised censored mean-level detector (GCMLD) is proposed. The GCMLD does not require a priori knowledge of the number of interfering targets and achieves robust performance. The number of interfering targets is determined and their corresponding samples are then censored. Exact expressions for the probability of censoring the interfering targets and the overall probability of detection are derived. A comparison of the performance of the GCMLD with that of a recently proposed CFAR detector is also presented to demonstrate the robustness of the GCMLD.