Making abstraction-refinement efficient in model checking
COCOON'11 Proceedings of the 17th annual international conference on Computing and combinatorics
An efficient approach for abstraction-refinement in model checking
Theoretical Computer Science
Detecting spurious counterexamples efficiently in abstract model checking
Proceedings of the 2013 International Conference on Software Engineering
| Hi-index | 14.98 |
Model checking for large-scale systems is extremely difficult due to the state explosion problem. Creating useful abstractions for model checking task is a challenging problem, often involving many iterations of refinement. In this paper we consider techniques for model checking in the counterexample-guided abstraction refinement. The state separation problem is one popular approach in counterexample-guided abstraction refinement, and it poses the main hurdle during the refinement process. To achieve effective minimization of the separation set, we present a novel probabilistic learning approach based on the sample learning technique, evolutionary algorithm, and effective heuristics. We integrate it with the abstraction refinement framework in the VIS [1] model checker. We include experimental results on model checking to compare our new approach to recently published techniques. The benchmark results show that our approach has overall speedup of more than 56 percent against previous techniques. Our work is the first successful integration of evolutionary algorithm and abstraction refinement for model checking.