Distributed Simulation is becoming an increasingly important aspect of DoD training, analysis, and acquisition processes. However, certain technical hurdles have yet to be overcome. This project addresses specific problems in meeting message passing bandwidth and latency requirements. New time management paradigms have yet to be exploited by existing HLA RTIs. These include approximate time protocols, repeatable real-time techniques, and synchronized data distribution management. Embedding these technologies in an RTI gives a federation new control of its execution environment. By understanding underlying network infrastructure and specific simulation requirements, we can make use of these new controls, optimizing the performance of the system, and ensuring correct system behavior.
Benefits: The synchronization primitives explored in this project are generally applicable to many distributed computational systems. The problems in distributed simulation are, of course, closely related to on-line gaming and entertainment. The research is also applicable to collaborative design efforts, where distributed simulation is used as an engineering tool. The research outlined here is especially relevant now, as we anticipate the advent of the Next Generation Internet. Active management approaches outlined in this project can exploit new quality of service guarantees, available with the NGI. The techniques we explore can be modified, extended, or refined to be appropriate to many types of distributed computing. The completion of this project will allow both the construction of new, high performance RTIs for DoD and commercial use, and will permit the formalization of the management algorithms to be included in existing products.
Keywords: High Level Architecture; Federated Simulation Models; Federation Object Model; Simulation; Modeling; DIS
