Optimising Test Execution Times in Test Suite Generation

Abstract

In the last decade, enormous progress has been made in generating test cases for coverage criteria automatically via model checking. However, many approaches suffer from generating large test suites with long test execution times. In recent work, we developed a method to generate test suites from state-based design models with the shortest test execution time satisfying a given set of coverage criteria. We were able to show that valuable improvements can be achieved. But space complexity of generating the fastest test suite turned out to be too high; thus our approach was limited to small models. Alternatively, a model checker can be called several times, each run generating a single test case. Thereby, the space complexity problem is solved. To build a test suite with minimised execution time, optimisations have to be applied afterwards. Here we present and compare several such algorithms to optimise test suites with respect to execution time.