Interval Abstraction Refinement for Model Checking of Timed-Arc Petri Nets

Abstract

Abstract. State-space explosion is a major obstacle in verification of time-critical distributed systems. An important factor with a negative influence on the tractability of the analysis is the size of constants that clocks are compared to. This problem is particularly accented in explicit state-space exploration techniques. We suggest an approximation method for reducing the size of constants present in the model. The proposed method is developed for Timed-Arc Petri Nets and creates an under-approximation or an over-approximation of the model behaviour. The verification of approximated Petri net models can be considerably faster but it does not in general guarantee conclusive answers. We implement the algorithms within the open-source model checker TAPAAL and demonstrate on a number of experiments that our approximation techniques often result in a significant speed-up of the verification.