Approximate Distributed Discrete Event Simulation using Semi-Conservative Look-Ahead Estimation

Abstract

A novel way of distributed discrete event simulation, called approximate distributed discrete event simulation, is presented in this paper. Compared with the classic simulation, the models for approximate simulation give some kind of free margin to the simulator during the execution. This can be used in some cases to reduce the overhead of the simulation, especially the execution time. Since the margin can be adjusted arbitrarily in the range, a trade-off between the simulation precision and the execution time can be achieved this way. It's well known that the execution time of distributed discrete event simulation can't be reduced significantly compared with a sequential simulation when the logical processes are tightly coupled and the look-ahead is very short. In this study, a framework of approximate distributed discrete event simulation with some novel algorithms is developed, which is aimed to provide a longer look-ahead and further the trade-off between the simulation precision and the execution time using the free margin provided by the model.