Distributed expert systems for network performance optimization

A distributed expert system that uses a simulation-based optimization methodology for queuing networks and whose architecture permits parallel simulation of multiple configurations is presented. A knowledge-based search that is a randomized combination of steepest descent and branch and bound algorithms in which the generating function of new states uses qualitative reasoning and the gradient of the objective function is estimated using a heuristic score function method is shown to drive the performance of the network. A random search based on the relative order of the performance gradient components is shown to be a powerful qualitative reasoning technique. The system is implemented as a loosely coupled expert system with components written in Prolog, Simscript, and C. The efficacy of the method is demonstrated through an example from the domain of Jackson (1963) queuing networks.<
>