Cognitive Approaches to Detection of Small Targets

This paper shows the early results from an investigation into the potential gain in detection performance achievable by optimizing the parameterization of a CA-CFAR detector to local clutter behaviour. The effect of varying CFAR window length is studied in particular for range correlated time independent samples, and the CFAR Loss (measured by increasing the RCS of a non-fluctuating target simulated at each range cell) of an optimal "fixed" CFAR is compared to a varying window length that minimizes CFAR Loss for each range cell, showing potential performance increase.