Bounding CPU utilization as a part of the model design and the scenario design of a large-scale military training simulation

Goals of large scale training simulations include scalability, composability, and extensibility in affordable open systems architectures. System designs that employ Optimistic Time Warp, predictive contracts, interest expressions, and distributed systems methods offer great potential to accomplish this goal. However, the combination of these concepts produces an extremely complex and dynamic solution space. The paper demonstrates a method that can aid in both the design decisions for the level of detail of models, and the composition of scenario models. A basic simulation model that approximates one node of a highly interactive distributed military training simulation with platform level entities is developed. The entities will be effected by events that represent fire, sensing, and command organization. The simulation outputs will then be used to develop regression equations. The regression equation predicts performance of scenarios within the experimental region.