Performance Analysis of Permutation and Rank Tests under K-Distributed Clutter in Radar Detection

Permutation tests for nonparametric detection

“…From results not shown here (see [32,34,42,43]), for spiky clutter (v=0.5), the degradation in SCR of each detector when P fa changes from 10~3 to 10~6 is about 5dB, and from P /a =10~6 to 10~8, the degradation is about 2 dB. On the other hand, for Rayleigh clutter, there are SCR-degradations of 2 dB approximately from P/ a =10~3 to 10~6, and 1 dB approximately from P/ a =10~6 to 10~8.…”

“…Similar tendencies are observed for the SWII case when it is compared with the NF case, although the SCR required in SWII case is greater than the SCR required in the NF case for the same P d . Other results have been published in [42,43].…”

“…Additional results can be found in [42,43]. The clutter shape parameter (v) has been v=0.5 as a typical value for spiky clutter, and v = co for Rayleigh clutter.…”

“…Now, from (40) and (37) (42) Then, r(X!,x 2 ,.. .,x w ) = YA = I xf* an d it should be compared with (24). Also, considering any block permutation in (42), we have r k (X!,x 2 ,.. .,x N )= YA = I *&., where k=(fe 1 , k 2 k N ), fc,-=l, 2 M+l, and i=\,'l N; now, it should be compared with (25).…”