Key points are not available for this paper at this time.
ABSTRACT A statistical summary of the standard clutter models is given - Gamma, Inverse Gaussian, Log-normal, Weibull - and it is shown how mixtures of these distributions can be used to represent measured clutter data. This is then related to the problem of detecting a target embedded in clutter and it is shown how this type of clutter characterization can be used to construct a general CFAR detection method. This involves an adaptive adjustment of the detection threshold based on measured characteristics of the clutter. The procedure will maintain a constant false alarm rate as the received clutter data varies over different mixtures of distribution types.
Jack M. Miller (Sun,) studied this question.