Simple and fast approximations for generalized stochastic Petri nets

A primary problem with generalized stochastic Petri nets (GSPNs) is the exponential explosion in the number of reachable states. This limits the GSPN modeling capability. We present an algorithm that circumvents the problem by not enumerating the entire state space to find a solution. Instead, it considers token flow balance and the preferred cycle heuristic to reduce the number of reachable states by an order of magnitude. This provides fast approximations of performance measures of systems modeled as GSPNs. Comparisons show how accurate the approximations are. We also give criteria that help system modelers ensure high approximation accuracy.