Counter example-based error localization of behavior models

  • Authors:

  • Tsutomu Kumazawa;Tetsuo Tamai

  • Affiliations:

  • Graduate School of Arts and Sciences, The University of Tokyo, Tokyo, Japan;Graduate School of Arts and Sciences, The University of Tokyo, Tokyo, Japan

  • Venue:
  • NFM'11 Proceedings of the Third international conference on NASA Formal methods
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

Behavior models are often used to describe behaviors of the system-to-be during requirements analysis or design phases. The correctness of the specified model can be formally verified by model checking techniques. Model checkers provide counterexamples if the model does not satisfy the given property. However, the tasks to analyze counterexamples and identify the model errors require manual labor because counterexamples do not directly indicate where and why the errors exist, and when liveness properties are checked, counterexamples have infinite trace length, which makes it harder to automate the analysis. In this paper, we propose a novel automated approach to find errors in a behavior model using an infinite counterexample. We find similar witnesses to the counterexample then compare them to elicit errors. Our approach reduces the problem to a single-source shortest path search problem on directed graphs and is applicable to liveness properties.