Coverage Criteria for Model-Based Testing using Property Patterns
This work addresses model-based testing for software systems, particularly in industrial contexts, but appears incremental as it builds on existing property pattern formalisms.
The paper tackles the problem of generating test cases from UML/OCL models using property patterns, proposing new automata coverage criteria and mutation techniques to ensure system robustness, with implementation and experimentation on realistic industrial case studies.
We present in this paper a model-based testing approach aiming at generating test cases from a UML/OCL model and a given test property. The property is expressed using a dedicated formalism based on patterns, and automatically translated into an automaton. We propose new automata coverage criteria that are tailored to the property automata we consider. These criteria are based on the coverage of a relevant subset of the transitions related to the original property, aiming at producing test cases that illustrate the dynamics of the system described in the property. In addition, we propose to produce test cases that can ensure the robustness of the system w.r.t. the property, by mutating the property automaton, in order to try to provoke events that would be forbidden by the property. This approach has been implemented into a large tool set and it has been experimented on realistic case studies, in the context of industrial research projects.