SAT-based summarization for Boolean programs

  • Authors:
  • Gérard Basler;Daniel Kroening;Georg Weissenbacher

  • Affiliations:
  • Computer Systems Institute, ETH Zurich, Switzerland;Computer Systems Institute, ETH Zurich, Switzerland;Computer Systems Institute, ETH Zurich, Switzerland

  • Venue:
  • Proceedings of the 14th international SPIN conference on Model checking software
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

Boolean programs are frequently used to model abstractions of software programs. They have the advantage that reachability properties are decidable, despite the fact that their stack is not bounded. The enabling technique is summarization of procedure calls. Most model checking tools for Boolean programs use BDDs to represent these summaries, allowing for efficient fix-point detection. However, BDDs are highly sensitive to the number of state variables. We present an approach to over-approximate summaries using Bounded Model Checking. Our technique is based on a SAT solver and requires only few calls to a QBF solver for fix-point detection. Our benchmarks show that our implementation is able handle a larger number of variables than BDD-based algorithms on some examples.