Analysis of non-coherent CFAR detectors in sea-clutter: A comparison

In radar systems, detection performance is always related to target and clutter models. The probability of detection is shown to be sensitive to the degree of estimation accuracy of clutter levels. In this work, the performances of logt-CFAR, zlog(z)-CFAR and Bayesian-CFAR detectors are investigated using both simulated and real data. The clutter is assumed to be log-normal, Weibull or Pareto type II distributed. The dependence of the false alarm probability is presented. From simulated data, CFAR detectors provide fully CFAR decision rules. From IPIX real data with different range resolutions, it is shown that the Bayesian-CFAR algorithm exhibits a small deviation of the false alarm probability.