Iterative decomposition and aggregation of labeled GSPNs

The use of Stochastic Petri Nets for performance analysis is limited by the state explosion of the underlying Continuous Time Markov Chain. A class of analysis methods to overcome this limitation are based on repeated decomposition and aggregation. In this paper, we propose a general framework for these kinds of solution methods and extend known techniques by introducing new classes of aggregates to reduce the approximation error. Aggregation relies on a formal definition of equivalence of Stochastic Petri Nets, which allows us to build aggregates at several levels of detail. The approach has been completely automated and allows the analysis of large and complex models with a low effort.