Query-driven petri net reduction for analysis in Ada tasking

We have illustrated methods to address three types of problems in static analysis for Ada tasking: quantitative questions, safety problems and MAY-happen event problems. We have applied a two-phase methodology to automate analysis: first deriving a semantically rich model independent of any specific analysis issue, that is the original Ada nets, and then manipulating this model with algorithms that are designed for the specific analysis issue of concern. We call such a methodology the query-driven net reduction. The philosophy behind this methodology is that different analyses demand different aspects of information from the system. An optimized analysis model should only contain the necessary information. In addition to reachability graph generation, the linear algebraic method is also investigated as a follow-up analysis technique. Experiments show that the net reduction technique substantially enhances the analysis ability of both state space generation approaches and linear algebraic methods.<
>