Comparison of Systems 



In figures 16 through 19, single-scan detection probabilities are plotted 

 for a single-stage system, a two-stage system with fixed-sample stages, and two 

 cases of two-stage systems with sequential first stages. In each figure, the two 

 sequential cases (C= 4, Ng = 10) and {C= 6, Ng = 8), require approximately 

 the same average dwell-time in the signal case, and for large signals the values 

 of N(S) for the sequential cases and for the fixed-sample case are all approxi- 

 mately 13. A comparison of figure 16 with figure 17 will show that the power 

 savings of the two-stage systems over the single-stage system increase if the 

 FAR is reduced by decreasing a^ og and a. Figures 16 through 18 do not allow 

 a completely fair comparison of the single-stage case with the two-stage cases 

 because of the larger number of opportunities to detect the target in an N = 1 

 single-stage case than in /V = 1.2 cases, so a comparison for N = iV = 2 was 

 included as figure 19. 



As was seen in figure 7, the sequential test with C = 6 is capable of 

 providing a smaller a^ than the C = 4 test with approximately the same E„(n). 

 However, since a very small a^ is not efficiently utilized in a two-stage system, 

 the detection probability of a system employing the C = 6 test is only slightly 

 better than that of a system employing the C = 4 test and the same Ng. The 

 C = 4 test is generally a good choice since tests with larger C result in larger 

 sample sizes in clutter or signal situations. The tendency of the sequential 

 system toward a long first stage when clutter is present or the signal is small is 

 evident in figure 20. 



32 



