Performance assessment of CFAR processors in pearson-distributed clutter

An analytical study on the performance of three existing classes of constan t false alarm rate (CFAR) detectors operating in heavy tailed distributed d ata is presented. In particular, we study the performance of the cell avera ging (CA), order statistics (OS), and p-percent truncated mean (PTM) CFAR p rocessors when the output measurements of the square-law detector can be mo deled as positive alpha-stable (P alphaS) random variables with a shape par ameter (characteristic exponent) equal to 0.5. We derive the exact expressi ons for the detection and false alarm probabilities of the detectors, and c ompare their performance by means of their corresponding receiver operating characteristics (ROC).