Two parameter CFAR detector for weibull clutter based on a combination of tail extrapolation and importance sampling techniques

It is well known that the use of optimal CFAR detectors in two parameters clutter is problematic due to the high computational load they demand. Normally, some kind of suboptimal approach will he used. In [l] an approach based on the use of tail extrapolation technique (a variance reduction technique used in rare event simulation) is demonstrate to work well from a practical point of view: presents a good trade off among detection losses, spatial resolution of clutter patches and computational burden. In this paper we will present a new detector based in the same philosophy but using Importance Sampling (other variance reduction technique used in PFA evaluation) for the estimation of detection threshold. This detector has a higher computational burden but it reduces the detection losses with respect to the detector presented in [l] for CFAR references of moderate size in spiky clutter conditions.