Performance analysts of dynamic synthetic environment data caching in real-time HLA simulations

In the past, most Distributed Interactive Simulations (DIS) restricted environmental data to being static, i.e. once a simulation started, an entity's environment was unalterable. Although this restriction has initially been carried over to the next generation of DIS - based on the High Level Architecture (HLA) - more recent developments include support for dynamic effects, such as cratering, smoke, etc. To prevent unacceptably high additional network traffic due to communicating dynamic environment changes to all federates in an HLA federation, a caching mechanism based on the notion of an entity's area of interest has been proposed. Using this approach, each HLA federate only maintains an up-to-date copy of the environment data for that part of the battlefield which represents the area around the entities simulated by the federate. In this paper, we describe results of a performance study which compares area of interest based caching with caching all battlefield environment data on all federates. Based on an actual prototype implementation of an HLA simulation at the Defence Evaluation gr Research Agency (DERA), Malvern, a simple model of an HLA simulation has been developed and analysed using discrete event simulation. A number of experiments with varying model parameters were carried out and the results are encouraging, i.e. our findings suggest that the proposed area of interest based caching strategy helps significantly to reduce network and federate overheads.