What the study found
White-box test case prioritization techniques were generally slightly better than black-box techniques for Simulink models. In white-box prioritization, the total greedy approach did better than the additional greedy strategy in models with higher block interactions and connections.
Why the authors say this matters
The authors say test optimization is important because Simulink-based cyber-physical system simulations are compute intensive and test execution is long, especially when tests are repeated at different in-the-loop levels. The study suggests that comparing and selecting prioritization techniques can help in this testing context.
What the researchers tested
The researchers carried out an empirical study comparing white-box and black-box test case prioritization techniques for MATLAB/Simulink models. They assessed 11 prioritization techniques, including traditional techniques and new approaches proposed for Simulink models, using six models of different sizes and complexities.
What worked and what didn't
In the comparison, white-box techniques were, on average, slightly better than black-box techniques. For white-box prioritization, total greedy outperformed additional greedy in models with more than 10 block interactions and more than 200 connections, which the abstract says better capture system coupling than raw block count. Black-box techniques were faster, although total greedy was still fast enough to be used in practice.
What to keep in mind
The abstract does not describe detailed limitations beyond the study using six Simulink models. The findings are limited to the techniques and model set examined in this empirical study.
Key points
- White-box test case prioritization was generally slightly better than black-box prioritization in the Simulink models studied.
- Among white-box methods, total greedy outperformed additional greedy in models with higher block interactions and connections.
- Black-box prioritization was faster than white-box prioritization.
- The study compared 11 prioritization techniques across six Simulink models of varying size and complexity.
- The abstract says block interactions and connections better capture system coupling than raw block count.
Disclosure
- Research title:
- White-box prioritization was slightly better in Simulink tests
- Image credit:
- Photo by Alexander Dummer on Pexels
Get the weekly research newsletter
Stay current with peer-reviewed research without reading academic papers — one filtered digest, every Friday.