Wild and Pignatiello describe a reverse simulation technique to find feasible values for system design targets in a simulation that lead to a stable system design, a system that conforms to known performance constraints. The technique is a heuristic procedure, based on an expert system, which adjusts the system design to conform to user performance targets. The user can then perform further experiments with the resulting stable system.
The reverse simulation technique is described through two simple queueing examples, and a prototype implementation and its performance metrics are described. The presentation and the content of the paper suffer as a consequence of this: the work is intended as a research overview. It covers the simulation and artificial intelligence areas, yet provides scant details of either area. A number of general issues are not tackled, such as how the rule base of an expert system is derived and how reverse simulation can be applied to less regular simulations than queueing problems. What the authors have provided, however, is a good introduction to the reverse simulation area and a commendable supporting reference list.