Large Scale Simulation of Manufacturing and Communications Systems

Regents Of The University Of Michigan - Ann Arbor, Ann Arbor MI

Investigators

Abstract

This grant provides funding for research on methods for performance evaluation of large network models of complex systems that arise in manufacturing and communication systems. The systems in mind are so complex that simulation is typically the only viable tool for benchmarking the system. Unfortunately, simulation is not a particularly scalable technology. The research focuses on developing fast and efficient simulation methods for complex systems, by exploiting prior knowledge of the system's dynamics. Such prior knowledge may be obtained from examination of a "fluid model", for example, which is an idealized model of complex systems where all entities are modeled as fluids flowing through pipes. A great deal is known about fluid models, and the hope is that this information can lead to very efficient simulation methods. If successful, the results of this research will ensure that simulation becomes a more scalable tool than before for a large class of systems. The central idea is that more accurate results may be obtained in (far) less time than would otherwise be the case. One can then implement more efficient algorithms for searching over different operating policies to determine the best operating policy, or at the very least one that is better than the existing policy.