Symbolic Test Case Generation for Time-Discrete Hybrid Systems

In this article we present a model-based test case specification and associated test data generation methods for embedded systems processing Boolean, integral and real-valued variables. Testing experts are relieved from the task of constructing input data to the system under test in an explicit way and manually calculating the expected reactions. Instead, test cases are specified by means of temporal logic formulae using the Duration Calculus, allowing to describe classes of concrete runs which are considered equivalent for the test objectives to be investigated. The concrete input data to the system under test and its expected reactions are automatically generated using an approach based on interval analysis. Following this approach, test data generation is handled as an interval constraint solving problem. The basic solution technique based on pavings of the solution set and bi-partitioning algorithms are accelerated by using a tightly integrated combination of interval constraint propagation and a variety of novel techniques originating from Boolean SAT solving methods, which have been adapted for the mixed Boolean, integer and real-valued variable setting.