Permutation test algorithms for nonparametric radar detection

In this paper, we present two new permutation test algorithms for applications to radar detection. The first algorithm is a direct realization of the permutation test definition with very high computational complexity in the most cases, the second algorithm is based on histogram computations and, consequently, computationally very efficient. As is well known, detectors based on permutation tests have constant false alarm rate (CFAR) characteristic under clutter with the exchangeability property. The exchangeability property means that the multidimensional distribution of a block of clutter (noise) samples is invariant under permutations of its arguments.