Stochastic simulation results exhibiting non-monotonic behavior
Charles Sadowski, J. Eyink · 2002
A stochastic simulation analysis is presently being conducted which tests the performance characteristics of a generic, air-to-air combat identification (ID) system. This analysis is specifically addressing three performance factors involving the identification of aircraft by other aircraft: ID range, the ID declaration rate, and the confidence level that the (declared) ID is correct. Some ID performance factor combinations produced counterintuitive results. For example, there were cases where improved ID systems produced poorer mission effectiveness. This phenomena (non-monotonic behavior) has been seen in some deterministic modeling results, with experts postulating everything from 'this won't happen in stochastic models' to 'the phenomena will be worse in stochastic simulations.' This paper includes a background discussion of the problem, a description of the identification probability analysis and modeling/simulation techniques used, a description of the nonmonotonic data observations, efforts made to investigate the causes of these counterintuitive results, and conclusions based on experience to date.>