Dynamic Relevance Filtering in Asynchronous Transfer Mode-Based DistributedInteractive Simulation Exercises

摘要

As Distributed Interactive Simulation (DIS) exercises continue to grow in scale,the need to support a large number of players has become apparent. The demands on the network and the simulation hosts in large exercises, though, have proved to be prohibitive, requiring significant computational overhead to filter through the information and extract what is relevant to a particular simulation. Some mechanism is needed to reduce irrelevant network traffic received by a system, while increasing the bandwidth available for the DIS exercise. Previous research efforts in this area have centered primarily on fixed geographic partitions of the battlespace to reduce the traffic at a given host. This geographic partitioning cannot adapt to the changing battlespace, and requires relatively significant pre-exercise setup and coordination. Our research has been to implement a DIS exercise system using native ATM interfaces, and to determine if a dynamic partitioning system is feasible and will provide a sufficient reduction in network traffic to allow DIS exercises to scale to the target 100,000 entities. A support infrastructure for DIS over ATM was developed and tested with current AFIT DIS applications, and a prototype dynamic partitioning system using geographic criteria was implemented.