Monopulse detection analysis of the trimmed mean CFAR processor in nonhomogeneous situations

The false alarm regulation capabilities and the detection performance of a CFAR processor depend on the robustness of the noise level estimation. The well known cell-averaging (CA) CFAR detector exhibits severe performance degradation in the presence of an interfering target return in the reference window or in regions of abrupt change in background clutter power. The ordered statistics (OS) CFAR processor designed to alleviate these problems resolves multiple targets quite well, but it lacks effectiveness in preventing excessive false alarms during clutter power transitions. A modified OS-CFAR detector, known as the trimmed mean (TM) CFAR processor, has been shown to give a marginal improvement in the false alarm rate performance in the presence of clutter edges. The CA- and OS-CFAR schemes are special cases of the TM-CFAR scheme. The detection performance of this processor, in nonhomogeneous environments, was previously studied by means of computer simulation. The author provides a complete detection analysis, for a Swerling II target fluctuation model, for the TM-CFAR processor when more than one target return is present within the reference window and in clutter power transitions