Constraint-Based Approach for Analysis of Hybrid Systems

  • Authors:
  • Sumit Gulwani;Ashish Tiwari

  • Affiliations:
  • Microsoft Research, Redmond WA 98052;SRI International, Menlo Park CA 94025

  • Venue:
  • CAV '08 Proceedings of the 20th international conference on Computer Aided Verification
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper presents a constraint-based technique for discovering a rich class of inductive invariants (boolean combinations of polynomial inequalities of bounded degree) for verification of hybrid systems. The key idea is to introduce a template for the unknown invariants and then translate the verification condition into an 茂戮驴 茂戮驴 constraint, where the template unknowns are existentially quantified and state variables are universally quantified. The verification condition for continuous dynamics encodes that the system does not exit the invariant set from any point on the boundary of the invariant set. The 茂戮驴 茂戮驴 constraint is transformed into 茂戮驴 constraint using Farkas lemma. The 茂戮驴 constraint is solved using a bit-vector decision procedure. We present preliminary experimental results that demonstrate the feasibility of our approach of solving the 茂戮驴 茂戮驴 constraints generated from models of real-world hybrid systems.