Non-Functional Requirements Discovery and Quality Assurance Using Goal Model for Earthquake Warning System in Operation

Many industrial systems that are developed without proper engineering guidance due to a lack of expertise or resources suffer from failures and maintenance problems in their evolving lifecycle. For mission-critical systems, in particular, ensuring high quality of non-functional requirements in a rapidly changing domain environment is of the utmost importance. In this paper, we report our case study with an industry system, a sensor-based earthquake warning system that was developed without a rigorous engineering process. Therefore, no requirements documents are available for future maintenance and verification. In this study, we used various types of software analysis methods such as stakeholder interviews, document reviews, source code analysis, model checking, and software testing for discovering requirements and also for verification purposes. We used a goal modeling approach to gather a set of initial requirements as though they had been elicited using an appropriate requirements engineering method in the early stage of the development process. Furthermore, software testing and model checking were iteratively used to verify and clarify unknown-source, uncertain, and unconfirmed requirements during the revision of the goal model. This study also provides an architectural improvement of the system through the discovery of requirements conflicts and violations. The experience and findings of this study, which demonstrate the effectiveness of applying diverse software engineering techniques in maintenance, can contribute to the analysis and evolution of systems developed without a proper engineering process, by discovering and verifying some critical requirements specifications.