Net level aggregation using nonlinear optimization for the solution of hierarchical GSPN in performance evaluation

Abstract

An approach for the hierarchical solution of large generalized stochastic Petri net models is presented. The method is based on the aggregation of submodels to substitute networks. The stochastic equivalence between these models is achieved by matching the flow time distributions of tokens in the submodel and in the aggregate net. This leads to a nonlinear optimization problem for finding the best aggregate net. As the main result, some insight is provided into the crucial point of estimating the parameters of a suitable aggregate net from a flow time distribution of the original net. The approach is demonstrated by means of an example.<>