Clutter-map CFAR detection for range-spread targets in non-Gaussian clutter. II. Performance assessment

The performance assessment of the clutter-map constant false-alarm rate (CFAR) procedure presented in Part I [1] is introduced. At first, the performance in the presence of point targets is investigated, subject to both Weibull and lognormal clutter, so as to elicit the influence of the system parameters. The study essentially relies on Monte-Carlo simulations, but approximations for the detection rate are also developed. Admittedly, a global figure of performance, the Average Threshold Deviation (ATD), is introduced and its worth as a design tool is validated. Next, the effect of extended as well as multiple targets is considered, showing that the system can be designed so as to avoid self-masking from either range-spread targets persisting on the same map cell, or spurious targets in close spatial proximity to the cell under test. Finally, the system robustness is demonstrated through a sensitivity study, aimed at investigating the effect of a possible mismatch between the design and the actual clutter distribution, that of the clutter correlation and of a thermal noise component in the overall disturbance.