If A Fails, Can B Still Succeed? Inferring Dependencies between Test Results in Automotive System Testing

Abstract

In this paper we propose an approach that, given a structured requirements specification, allows the automatic online detection of a redundant test case. This means that, at each time point during a testing phase, one automatically infers the failure of a test case from the current status of successful tests and failed tests. By a structured requirements specification we mean that one uses a hierarchical structure and types to document the (natural language) formulation of requirements. We have implemented the approach. The evaluation of our implementation in a case study in the context of the development process for Mercedes-Benz vehicles at Daimler AG indicates the practical potential of our approach.