Detection improvements with CFARs having expanded amplitude inputs

This paper addresses some differences in performance of cell averaging, constant false alarm rate (CA-CFAR) processors as a function of whether the input signal has linear, square-law, or a higher order response relative to the output of a linear receiver. For brevity, these CFAR inputs are herein referred to as A, A/sup 2/, A/sup 4/, etc., signals. Specifically, the CFAR input voltage is assumed to be of amplitude A/sup N/, where A is the amplitude of a linearly received signal. When using conventional in-phase (I) and quadrature (Q) channels, A = (I/sup 2/ + Q/sup 2/)/sup 1/2/. Emphasis is on CA-CFAR performance when the threshold is set for rejecting very spiky clutter which, for this study, is defined as having Weibull statistics with the parameter a = 5. Under these conditions, analyses indicate that for detecting a target in noise, a square-law (N=2)CA-CFAR input signal may provide 9.3 dB better detection sensitivity than does a linear (N=1) CFAR input. Also for targets in noise and a Weibull a = 5 clutter threshold setting, it appears that fourth order (N = 4) CFAR input may yield nearly 20 dB improvement over linear input.