Abstract
This paper describes on-going research, where we compare, via simulation experiments, a stochastic system to a standard. We are particularly interested in a subset of standards we call limit standards. A limit standard is a maximum or minimum benchmark derived from requirements, another model, or the actual system. The problem is to determine if a system meets the limit standard at customer-defined probabilities. Then, for those systems that meet the limit standard, identify which system is the "best," or results in the lowest probability of reaching the standard. Current comparison methods are based on expected value and cannot solve this type of problem. We outline a two-step approach, using methods from acceptance sampling and ordered statistics, to solve this problem.