Parallel and distributed simulation of parallel DEVS models

Abstract

Distributed simulation can speed up the execution of models significantly. We introduce a new simulation algorithm and present partitioning and load balancing techniques that are tailored to the efficient distributed execution of PDEVS. We base our elaborations on the idea of minimizing interprocessor communication, since this is a major bottleneck in distributed PDEVS simulation. Additionally, experimental results are provided which compare the performance of this new approach to alternative algorithms.