Model-based Testing
This page contains papers on Model-based testing and the Model-based testing tool MTest. To master the growing complexity of embedded software in control systems, the latter is increasingly developed in a model-based way. This means that typically modeling and simulation tools such as Simulink/Stateflow® are used to create an executable model of the system under development at an early stage of the development process. This model then serves as a basis for the future control unit software.
The test process accompanying the model-based development process (Model-based testing) can profit from the existence of such an executable model in various ways - some of these are discussed in the following papers.
Robinson-Mallett, C.; Hierons, R.M.; Poore, J.; Liggesmeyer, P.: Using communication coverage criteria and partial model
generation to assist software integration testing. Software Quality Journal, Springer, April 2008-
Robinson-Mallett, C.; Liggesmeyer, P.; Mücke, T.; Goltz, U.: Extended state identification and verification using a model checker.
University of Kaiserslautern, Fraunhofer IESE, Germany, 2006.
Robinson-Mallett, C.: Modellbasierter Modultest technischer Softwaresysteme, Phd's Thesis, Hasso-Plattner-Institute at University of Potsdam,
Oktober 2005. (in German).-
Conrad, M.; Dörr, H.; Fey, I.; Yap, A.: Model-based Generation
and Structured Representation of Test Scenarios. Workshop on Software-Embedded
Systems Testing (WSEST), Gaithersburg (MD), USA, November 1999.
-
Conrad, M.; Sadeghipour, S.: Einsatz von Überdeckungskriterien auf Modellebene
- Erfahrungsbericht und experimentelle Ergebnisse. Softwaretechnik-Trends,
Band 22, Heft 2, (ISSN 0720-8928), Mai 2002. (in German)
The development of software models is done successively in multiple levels, where each lower level describes the model in more detail. This approach demands a proof of correctness of the refined model stages compared with the original model's functions. The evaluation of such a test, where a huge amount of test data may be generated, can be automated by the tool MEval which is based on a new developed signal comparison method.
MEval is commercially available at IT Power Consultants.
The new signal comparison method is discussed in the following paper.