Analysis of the censored-mean level CFAR processor in multiple target and nonuniform clutter

Abstract

The performance of the cell-averaging (CA) CFAR processor degrades rapidly in nonideal conditions caused by extraneous targets and nonuniform clutter. The ordered-statistics (OS) CFAR algorithm, on the other hand, trades a small loss in detection performance in homogeneous environments for much less performance degradation in nonhomogeneous backgrounds. As a hybrid between CA and OS schemes, the censored mean-level (CML) CFAR technique has been introduced. This processor is designed to accommodate interfering targets in the reference window as well as control false alarms in the presence of clutter boundaries. The author provides a complete radar detection analysis for this scheme in nonhomogeneous environments and for a Swerling II target fluctuation model. Analytical results of performance are presented in both multiple target environments with one or more interfering targets and in regions of clutter power transitions. All the results for a single pulse detection appear as explicit analytic expressions. Integration of M pulses is also studied in nonideal conditions. The processor performance for that case is characterised by an integral equation which is evaluated numerically. As expected, lower threshold values and better detection performances, under both homogeneous and nonhomogeneous models, are obtained by increasing the number of noncoherently integrated sweeps.