Cell Averaging CFAR Detector with Scale Factor Correction through the Method of Moments for the Log-Normal Distribution

Abstract

The new LN-MoM-CA-CFAR detector is introduced, exhibiting a reduced deviation of the operational false alarm probability from the value conceived in the design. The solution solves a fundamental problem of CFAR processors that has been ignored in most proposals. Indeed, most of the previously proposed schemes deal with sudden changes in the clutter level, whereas the new solution has an improved performance against slow statistical changes that occur in the background signal. It has been proven that these slow changes have a remarkable influence on the selection of the CFAR adjustment factor, and consequently in maintaining the false alarm probability. The authors took advantage of the high precision achieved by the MoM (Method of Moments) in the estimation of the Log-Normal (LN) shape parameter, and the wide application of this distribution to radar clutter modeling, to create an architecture that offers precise results and it’s computationally inexpensive at the same time. After an intensive processing, involving 100 million Log-Normal samples, a scheme, which operates with excellent stability reaching a deviation of only 0,2884 % for the probability of false alarm of 0,01, was created, improving the classical CA-CFAR detector through the continuous correction of its scale factor.