Text

  • Study location Sal Alfa, campus Västerås
Date
  • 2022-10-03 13:15–14:15

Muhammad Nouman Zafar’s licentiate thesis in Computer Science

The public defense of Muhammad Nouman Zafar’s licentiate thesis in Computer Science will take place at Mälardalen University, room Alfa (Västerås Campus) at 13:15 on 3rd October 2022.

The title of the thesis is “Model-based System Testing of Safety-Critical Embedded Software”.

The faculty examiner is Assoc. Professor Cyrille Artho, KTH, Sweden.
The examining committee consists of Assoc. Professor Cyrille Artho, KTH, Sweden; Professor Serge Demeyer, University of Antwerp, Belgium and Dr. Markus Borg, RISE, Sweden.

Reserve is Dr. Emil Alégroth, BTH, Sweden.

The licentiate thesis has serial number 328.


Summary

Testing of safety-critical embedded systems, such as those found in cars and trains, is a complex and expensive activity. Test automation can help find more faults and can reduce testing costs. Model-based Testing (MBT) is one of the automated testing techniques whose strengths and weaknesses in industrial settings need to be thoroughly investigated. The main objective of this thesis is to propose and evaluate an automated, cost-effective, and resource-efficient model-based test generation approach to test a safety-critical embedded system. As a start, we have explored multiple state-of-the-art MBT tools and evaluated the selected tool (i.e. GraphWalker (GW)) in terms of modeling notations, generation algorithm, stopping conditions, and model completeness to generate test cases. Then, we proposed and implemented a Model-Based Test scrIpt GenEration fRamework (TIGER), based on GW, to generate system-level test scripts. We evaluated the generated test scripts by executing them at Software-in-theLoop (SiL) level and by inducing model-level mutations. Finally, we performed a comparative analysis between test suites developed using TIGER, Combinatorial testing (CT), and manual industrial practices in terms of (i) Modified Decision and Condition Coverage (MC/DC) and requirement coverage, (ii) efficiency (i.e. test generation and execution time) and (iii) fault detection effectiveness through mutation analysis.

The results showed that the MBT-generated test suite using edge coverage criterion tends to cover each requirement multiple times while achieving the same level of requirement coverage achieved by the manually written test suite. The test suite generated using TIGER also provides higher MC/DC test coverage than the test suites generated using CT and written manually by industrial practitioners. However we also found some merits of higher strength CT. Higher strength CT was the most efficient technique in terms of the time required to generate and execute tests as well as the achieved fault detection rate when compared with manual and MBT-generated test suites.

Contact Information