Effect of frequency-averaging on estimation of clutter statistics used in setting CFAR detection thresholds

Abstract : Incoherent averaging of radar returns measured at different frequencies has been shown to eb an effective method of improving fixed target detection performance. This improvement results from the fact that frequency- averaging narrows the amplitude probability distributions of both the target and clutter returns(after averaging), thereby allowing returns from targets to the more easily distinguished from those of clutter. This report examines the effect of frequency averaging on decreasing the variance of the estimated clutter reflectivity. The attainable reduction in clutter sample variance that results from frequency-averaging is shown to be strongly dependent upon the spatial non- uniformity of the clutter. The more non-uniform the clutter, the less effective frequency-averaging is in reducing the clutter sample variance and corresponding CFAR (Constant False Alarm Rate) threshold setting. An analytic expression is derived for the amount of sample variance reduction provided by frequency averaging, as a function of spatial standard deviation of mean clutter reflectivity and the number of frequencies averaged. The amount of variance reduction predicted by this expression is confirmed by experimental measurements (at Ku band) of clutter returns, at 16 and 32 different frequencies, for a variety of different clutter types.