Model-Based Testing of Industrial Transformational Systems

Abstract

We present an approach for modeling and testing transformational systems in an industrial context. The systems are modeled as a set of boolean formulas. Each formula is called a clause and is an expression for an expected output value. To manage complexities of the models, we employ a modeling trick for handling dependencies, by using some output values from the system under test to verify other output values. To avoid circular dependencies, the clauses are arranged in a hierarchy, where each clause depends on the outputs of its children. This modeling trick enables us to model and test complex systems, using relatively simple models. Pairwise testing is used for test case generation. This manages the number of test cases for complex systems. The approach is developed based on a case study for testing printer controllers in professional printers at Océ. The model-based testing approach results in increased maintainability and gives better understanding of test cases and their produced output. Using pairwise testing resulted in measurable coverage, with a test set smaller than the manually created test set. To illustrate the applicability of the approach, we show how the approach can be used to model and test parts of a controller for ventilation in livestock stables.