AuthorsS. Ali
TitleScalable Model-Based Robustness Testing: Novel Methodologies and Industrial Application
Afilliation, , Software Engineering
Project(s)The Certus Centre (SFI)
StatusPublished
Publication TypePhD Thesis
Year of Publication2012
PublisherUniversity of Oslo
Place PublishedOslo, Norway
Thesis Typephd
ISBN NumberISSN 1501-7710
Abstract

Embedded systems, as for example communication and control systems, are being increasingly used in our daily lives and hence require thorough and systematic testing before their actual use. Many of these systems interact with their environment and, therefore, their functionality is largely dependent on this environment whose behavior can be unpredictable. Robustness testing aims at testing the behavior of a system in the presence of faulty situations in its operating environment (e.g., sensors and actuators). In such situations, the system should gracefully degrade its performance instead of abruptly stopping execution. To systematically perform robustness testing, one option is to resort to Model-Based Robustness Testing (MBRT), which is a systematic, rigorous, and automated way of conducting robustness testing. However, to successfully apply MBRT in industrial contexts, new technologies need to be developed to scale to the complexity of real industrial systems. This thesis presents a solution for MBRT on industrial systems, including scalable robustness modeling and executable test case generation. One important contribution of this thesis is a scalable RobUstness Modeling Methodology (RUMM), which is achieved using Aspect-Oriented Modeling (AOM). It is a complete, automated, and practical methodology that covers all features of state machines and aspect concepts necessary for MBRT. Such methodology, relying on a standard (Unified Modeling Language or UML) and using the target notation as the basis to model the aspects themselves, is expected to make the practical adoption of robustness modeling easier in industrial contexts. The applicability of the methodology is demonstrated using an industrial case study. Results showed that the approach significantly reduced modeling effort (98% on average), improved separation of concerns, and eased model evolution. The approach is further empirically evaluated using two controlled experiments involving human subjects and results showed that the proposed methodology significantly improves the readability of models as compared to modeling using standard UML notations. Another important contribution of this thesis is an efficient approach for solving constraints (written in Objects Constraint Language (OCL)) on the operating environment of a system, which is mandatory for emulating faulty situation in the environment for the purpose of MBRT. A set of novel heuristics is devised for various OCL constructs, which are required for the application of search algorithms. The heuristics have been empirically evaluated on an industrial case study for robustness testing and the results showed to be very promising and significantly better than the existing works in the literature on OCL constraint solvers. A final contribution of the thesis is robustness test case generation from the models developed using RUMM. Test case generation also includes scripts generation for environment emulation, which is mandatory for automated robustness testing again using an industrial case study. In preliminary experiments, the execution of test cases found one critical, robustness fault in a deployed industrial system.