Use of common random numbers in comparing alternatives

A fundamental use for system simulation is to compare a number of different approaches to achieve a stated objective. The author presents methods for making comparisons by either ranking alternatives or selecting the best alternative with respect to a single output performance measure. To increase the effectiveness of the comparisons, these methods are able to use the variance reduction technique of common random numbers. It is assumed that the simulation output is stochastic and has a distribution that is unknown to the analyst. The finite horizon simulation case is considered, and examples are given to illustrate the application and utility of the methods