Algorithm for the Iterative Design of Observer Field Tests
ARMY TANK-AUTOMOTIVE AND ARMAMENTS COMMAND WARREN MI
Pagination or Media Count:
In a previous paper, the author described a technique for designing an observer test in an iterative manner. In field tests to compare the observability of combat vehicles, the test designer must select the optimum number of observation opportunities to balance collecting enough data to draw valid conclusions against the high cost of supporting vehicles and personnel at a test site. The test designer can select the number of observations, N, so that a given experimental difference in detectability will be statistically significant at a given confidence level. Alternatively, the test designer can select N so that the probability of rejecting a given underlying difference in detectability is less than a given amount. The test designer, however, generally lacks key parameters for the efficient design of the test. Namely, the designer lacks the detection probabilities of the vehicles at each range. The standard deviation of the difference in detection probability depends upon the detection probability itself. Therefore, the test designer must select the number of observations for each range based upon the conservative assumption that the probabilities are near 50, the probability for the maximum standard deviation. In the previous paper, an iterative technique of test design was described. In this technique, the test designer modifies the test matrix as the test progresses. Early test results yield estimates of the probability of detection for each vehicle at each range. Based on these estimates, the test designer reallocates the number of observations among the ranges, improving the efficiency of the test. In this paper, the author presents an algorithm to implement this iterative technique.
- Test Facilities, Equipment and Methods
- Miscellaneous Detection and Detectors
- Combat Vehicles