Reconfigurable stochastic Petri nets: A new formalism for reconfigurable discrete event systems

Multi-agent systems (MASs) represent a suitable model for open, dynamic and flexible systems (as the case of Reconfigurable Wireless Sensor networks, Reconfigurable Manufacturing Systems, etc). Indeed, the nature of multi-agent systems allows them to model systems where the number of components and their interconnection change over time. However, MAS models are often informal without mathematical background. On the other hand, Stochastic Petri Nets (SPNs) are among the popular formalism in modelling and verification of a various types of systems. Nevertheless, their rigid structure restricts their use in the analysis of reconfigurable systems. Despite the fact that many works introduce dynamic changes in PN, there exist few preliminary researchers that have tackled the reconfiguration in SPN. The aim of this work is to propose an extension to reconfigurable Petri net, called reconfigurable SPN (R-SPN) in order to provide a suitable tool for the formal modelling and verification of reconfigurable systems. The R-SPN consists of a SPN model, a set of reconfiguration rules and an algorithm used to transform R-SPN into an equivalent SPN, such that one can reuse and apply classical verification methods proposed for SPN on R-SPN. Thus, both structural and behavioural analysis can be carried out. A demonstration is given to illustrate the feasibility of the proposed extension on a reconfigurable manufacturing systems.