A Better and Faster Design and Evaluation Tool for Large Systems

Reference#: P01980

Complex systems have a large number of design parameters that control or affect the performance of the overall system. As systems become larger, the possible number of combinations of these design parameters explodes combinatorially. Evaluating potential system performance with current techniques such as Modeling and Simulation (M&S) requires exploring all possible combinations, which becomes cost and time prohibitive to run. In addition, these large numbers of compute runs and voluminous output becomes difficult or impossible for analysts to examine and thoroughly understand. Queuing models scale linearly with system size and provide more direct insight into how the system is actually operating, but provide only estimates of potential system performance.

The Johns Hopkins Applied Physics Lab has invented technology that combines speed, linear scalability and insight of queuing models with the greater accuracy and generality of M&S tools to rapidly and cost-effectively explore the design parameter decision space, leading to a better and more robust system design. Basically, the queuing models are used at the beginning to examine the entire design space and focus the M&S tools on the critical regions and, subsequently, to help interpret the M&S results.

Ms. H. L. Curran
Phone: (443) 778-7262