BUCS: Patterns and Robustness: Experimentation with Safety Patterns in Safety-Critical Software Systems

Abstract

In modern society, we rely on safely working software systems. This is the final report in a masters degree project to reveal key issues in the science field of computer software architecture and design of safety-critical software systems. A pre-study of a navigation system implied that functionality related problems and safety-critical problems do not stack one to one, but rather is a case of solving these aspects in different layers. This means that changes in software systems functionality do not necessary mean that change in safety-critical modules has to be done as well, and visa versa. To further support the findings in the pre-study, an experiment was created to investigate these matters. A group of twenty-three computer science students from the Norwegian University of Science and Technology (NTNU) participated as subjects in the experiment. They were asked to make two functional additions and two safety-critical additions to a software robot emulator. A dynamic web tool was created to present information to the subjects, and they could here answer surveys and upload their task solutions. The results of the experiment shows that there were not found any evidence that the quality attributes got affected by the design approaches. This means that the findings of this study suggest that there is difficult to create safety-critical versions of software architectural design patterns, because all design patterns have a set of additions and concequences to a system, and all sides of the implications of the design pattern should be discussed by the system architects before used in a safety-critical system.