Abstraction refinement by controllability and cooperativeness analysis

  • Authors:
  • Freddy Y.C. Mang;Pei-Hsin Ho

  • Affiliations:
  • Advanced Technology Group, Synopsys, Inc.;Advanced Technology Group, Synopsys, Inc.

  • Venue:
  • Proceedings of the 41st annual Design Automation Conference
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

We present a new abstraction refinement algorithm to better refine the abstract model for formal property verification. In previous work, refinements are selected either based on a set of counter examples of the current abstract model, as in [5][6][7][8][9][19][20], or independent of any counter examples, as in [17]. We (1) introduce a new "controllability" analysis that is independent of any particular counter examples, (2) apply a new "cooperativeness" analysis that extracts information from a particular set of counter examples and (3) combine both to better refine the abstract model. We implemented the algorithm and applied it to verify several real-world designs and properties. We compared the algorithm against the abstraction refinement algorithms in [19] and [20] and the interpolation-based reachability analysis in [14]. The experimental results indicate that the new algorithm outperforms the other three algorithms in terms of runtime, abstraction efficiency (as defined in [19]) and the number of proven properties.