A framework for adaptive synchronization of distributed simulations

Increased complexity of simulation models and the related modeling needs for global supply chains have necessitated the execution of simulations on multiple processors. While distributed simulation promises reduced complexity (as the result of decomposition), increased parallelism and convenient analysis of geographically distributed systems, it poses a challenging problem: synchronizing the distributed simulation federates. This paper discusses a new discrete event distributed simulation framework, which is designed with two goals in mind: (1) easy and fast development of distributed simulations and (2) efficient adaptive synchronization of simulation processes. This research uses state machine models for the automated synthesis of so called ´local synchronization agents´ and an adaptive synchronization algorithm has been developed based on pacing of simulation processes using real-time. Upon completion, this scalable framework is expected to shorten the lead-time to develop distributed simulation systems with reasonable performance characteristics.