PERFORMANCE EXTRAPOLATION IN DISCRETE-EVENT SYSTEMS SIMULATION

The use of simulation as an engineering tool to design complex compute r stochastic systems is often inhibited by cost. Extensive computer pr ocessing is needed to recompute performance functions for changes in i nput parameters. Moreover, simulation models are often subject to erro rs caused by the input data in estimating the parameters of the input distributions. The 'what if' analysis is needed to establish confidenc e in a model's validity with respect to small changes in the system's parameters. To solve the 'what if' problem for several scenarios requi res a separate simulation run for each scenario. A method to estimate the performance for several scenarios using a single simulation run ba sed on the efficient score function is developed. The approximate func tion is an exponential random function which is tangential to the perf ormance function in expectation. We study the algorithm's properties a nd examine the validity of the estimates based on this single run proc edure by performing some experimental studies on some simple queueing and pert models. Implementation 'details' necessary to package this me thod with existing simulation software are provided. Finally, there is a set of recommended directions for future research.