In the last decade, discrete-event distributed simulation has been a research topic of several researchers. However, there has been no commonly accepted solution, neither a commonly useful implementation. The main objectives of this project are: (1) Increasing the concurrency of distributed simulation by employing efficient graph algorithms, (2) Finding an efficient distributed-deadlock detection algorithm for a conservative distributed simulation, (3) Developing a suboptimal method to eliminate cycles from a process graph by replicating some nodes and edges of this graph, (4) Developing an efficient clustering algorithm for mapping of processes to processors, (5) Investigate the problems of flow control, load sharing, and memory management, (6) The implementation of (1) and (4) on the network of SUNs and/or on the Encore Multimax multicomputer, (7) The evaluation of the proposed methods by simulation (changing different parameters in the implementation, using different input models, comparing the performance to a sequential simulation executed on a uniprocessor).
