Proceedings 16th Workshop on Parallel and Distributed Simulation (2002)
May 12, 2002 to May 15, 2002
Dimosthenis Anagnostopoulos , University of Athens
Faster-than-real-time simulation (FRTS) can be used for the performance evaluation of systems behavior in real time, providing significant capabilities for studying systems with a time-varying behavior. FRTS enables model validation through comparing simulation results with the corresponding system observations. However, experimentation proves to be rather demanding, as both delivering output results and ensuring their reliability must be accomplished within a predetermined time frame. Output analysis of system observations and model results and relevant timing issues are discussed. A method is introduced that determines whether it is possible to execute the ?optimal? faster-than-real-time experiment, in which case multiple replications are scheduled for execution, or a compromise has to be made between the ability to predict for the long future and the degree of reliability achieved for predictions. FRTS experimental results are also presented to support the effectiveness of the proposed method.
D. Anagnostopoulos, "Experiment Scheduling in Faster-than-Real-Time Simulation," Proceedings 16th Workshop on Parallel and Distributed Simulation(PADS), Washington, D.C., 2002, pp. 163.