Model validation using automatically generated requirements-based tests

In current model-based development practice, validation that we are building a correct model is achieved by manually deriving requirements-based test cases for model testing. Model validation performed this way is time consuming and expensive, particularly in the safety critical systems domain where high confidence in the model correctness is required. In an effort to reduce the validation effort, we propose an approach that automates the generation of requirements-based tests for model validation purposes. Our approach uses requirements formalized as LTL properties as a basis for test generation. Test cases are generated to provide rigorous coverage over these formal properties. We use an abstract model in this paper-called the Requirements Model - generated from requirements and environmental constraints for automated test case generation. We illustrate and evaluate our approach using three realistic or production examples from the avionics domain. The proposed approach was effective on two of the three examples used, owing to their extensive and well defined set of requirements.