A symbolic simulation mechanism for well-formed coloured Petri nets

Abstract

Coloured Petri nets are well suited to the modelling of realistic distributed systems. Such models are however usually too large and complex to analyze and simulate using standard techniques. Originally the model symmetries inherent to the colour definitions of stochastic well-formed nets were exploited for Markovian analysis. The authors propose a similar technique to substantially reduce the average length of the event list in case of large and symmetric models. Examples of real simulation models of multiprocessor system architectures show that, after a given threshold, the symbolic simulation time may be virtually independent of the actual size of the system to be modelled, thus making the simulation of systems with thousand processors as inexpensive as the simulation of systems with only few processors.<>