Efficient Verification of Process Models Supporting Modifications of Data Values

Verification techniques detect undesirable behaviour of process models before their execution. In many use cases, data-value functions are essential. A data-value function modifies the values of data objects in a process model, e.g., increases the price of a product. Supporting such functions when verifying process models is challenging. This is because data objects with large domains often lead to state-space explosion. In this paper, to address this issue, we propose a novel approach using a binary encoding technique. We make use of Binary Decision Diagrams (BDD) to map the semantics of data-value functions into a Petri Net. This allows using the existing BDD reduction techniques to reduce the number of edges and nodes in BDDs and, ultimately, of places and transitions in Petri Nets. One can now map process models with data-value functions into much smaller Petri Nets, whose verification is feasible. We show that this is indeed the case, by verifying properties of an important real-world application, the German 4G spectrum auction.