Improving Invariant Mining via Static Analysis
Christoph Schulze, Rance Cleaveland · ACM Transactions on Embedded Computing Systems · 2017
This paper proposes the use of static analysis to improve the generation of invariants from test data extracted from Simulink models. Previous work has shown the utility of such automatically generated invariants as a means for updating and completing system specifications; they also are useful as a means of understanding model behavior. This work shows how the scalability and accuracy of the data mining process can be dramatically improved by using information from data/control flow analysis to reduce the search space of the invariant mining and to eliminate false positives. Comparative evaluations of the process show that the improvements significantly reduce execution time and memory consumption, thereby supporting the analysis of more complex models, while also improving the accuracy of the generated invariants.