Requirements oriented testing with Markov chain usage models in the automotive domain


As a result of the integration of increasingly elaborate and distributed functionality in modern automobiles the amount of installed electronic and software continuously grows. The associated growth in system complexity makes it inevitable that the test methods used for verification and validation keep pace with this development. Nowadays the test routine in industry usually requires each test case to be crafted manually by a test designer. The test case execution itself and test result evaluation usually are performed in an automated manner. This procedure has many drawbacks, as the crafting of single test cases is apparently awkward and error-prone and it is impractical to calculate test management criteria such as test coverage.
Within this project a method is developed that overcomes these drawbacks. Markov chain usage models (MCUM) constitute the central role within this project. MCUMs are employed to describe the possible usage of the System-under-test and to derive test cases from them. On the one hand the integration of MCUMs makes it possible to develop methods to integrate test requirements formally, as to improve traceability. On the other hand they provide the basis to incorporate algorithms or strategies that allow the generation of test cases fitting to various test requirements in the automotive domain. These comprise e.g. different coverage criteria under usage or system oriented aspects. Moreover established methods exist that allow the calculation of dependability measures based on results obtained from test cases automatically generated from MCUMs. Also the test planning can be supported by indicators that are derived during the test process.
The project aims for developing a method to describe test requirements formally by building a model. This model allows the derivation of test suites considering various testing aims and constraints. The tools themselves should form a part of the ITF (Integrated Testing Framework) and the process extend the current one described by EXAM and employed within the Volkswagen AG.

  • Schlüsselwörter: Automotive Technology; Dependable Systems; Model Driven Test Generation; Requirements; Usage Model; Markov Chain; Statistical Testing; Embedded Systems; Test Automation; Model Based Testing
  • Projektdauer: 2008-11-01 - 2011-10-31

Projektmitglieder

Förderer

  • AUDI AG (Ingolstadt)
  1. Sebastian Siegl, Winfried Dulz, Reinhard German und Gerhard Kiffe, "Model-Driven Testing based on Markov Chain Usage Models in the Automotive Domain," Proc. of the 12th European Workshop on Dependable Computing, Toulouse, France, Mai 2009