Maximum likelihood CFAR for lognormal clutter with censored samples

In this paper the structure of Maximum Likelihood (ML) CFAR for lognormal clutter is derived assuming that some reference samples are censored to avoid destructive effect of interfering targets. It is shown that under some acceptable approximations we will achieve in a simple structure for this detector which is suitable for real time processing. The proposed detector is compared with non censored ML CFAR (i.e. logt CFAR) via simulation. It is shown that while the proposed detector is a little weaker than logt CFAR in homogeneous clutter, in non homogeneous clutter it outperforms the logt CFAR.