0.1 



i/iS/ 



0.01 



0.001 





^ D 





A 



X 



X 



X 



I 



X 

 I o ^ 



1^ 



D D 



a 



D 

 D 

 X n 



^ A A 



A A 



D 

 D 



D 



a 



"a^b 



RICE 



RAYLEIGH 







10 

 10 

 10 



-6 

 -8 

 -12 



A 

 X 



O 

 D 





0.2 



0.4 



0.6 



0.8 



1.0 



Figure 14. Ratio of per-stage miss probabilities vs single-scan detection 

 probability for the optimized tests of figures 12 and 13. 



and small-sample nature of the tests) even as the SNR (and therefore Pq), the 

 FAR, the underlying distribution, and N vary. (Fig. 15 is for fixed-sample stages, 

 ^A ~ 1' '^A " -*■' "a^B ~ '^ ^ ^^'°' optimum /Vg and K'g per point. Successive 

 segments correspond to increasing values of Ng; the ratio increases fairly 

 smoothly with N for constant Ng.) 



30 



