Resource Utilization Prediction and Planning for Complex Simulations

WarpIV Technologies has developed a federation performance resource utilization tool to help scenario planners prepare for missile defense simulation events. The algorithms designed and/or prototyped in Phase I consists of nine new capabilities: (1) highly configurable mock federation for representing realistic MDA federations during algorithm development and testing, (2) data logging interfaces for capturing relevant information about each federate during standalone operation, (3) integration of the data logger with the HPC-RTI to demonstrate future integration with OSF, (4) a variety of algorithms for analyzing logged federate behavior data and establishing accurate computational and message-sending models, (5) integration of federate behavior models with the Missile Defense Engagement Model (MDEM) to log actual federate behaviors that are tied to a specific scenario (scripted behavior), (6) Forecaster that combines hardware configuration with scripted behavior to performance bottleneck analysis for making logical-time and real-time performance predictions, (7) Predictor, similar to the Forecaster except that it uses a high-fidelity parallel discrete-event simulation to predict federation performance, (8) Radial Basis Function Network (RBFN) to provide performance predictions based on machine learning using Forecaster, Predictor, or actual federation performance data, and (9) federation performance optimization using grid-computing and a genetic algorithm to determine optimal federate-to-machine mappings. Approved for Public Release | 19-MDA-9932 (21 Feb 19)