Simulation of correlated non-Gaussian interference for radar signal detection

The authors present canonical computer simulation procedures for generating correlated non-Gaussian radar clutter which can be characterized by spherically invariant random processes (SIRPs). The probability density function (PDF) of every random vector obtained by sampling a SIRP is a function of a nonnegative quadratic form. The PDF of this quadratic form is used for assessing the performance of the simulation procedures. It is pointed out that use of this technique reduced the goodness of fit test from a multivariate testing procedure to a univariate testing procedure resulting in tremendous processing simplicity