Exploring Static Analysis for Understanding Self-Adaptive Systems

Self-adaptive systems (SAS) detect environmental changes, prevent or react to them, and adjust their own configuration and behavior to ensure their quality. Understanding how SAS are designed and work/function, as well as their complexity and quality features, is necessary for their utilization, reuse, maintenance, and further evolution/development. A key role in understanding SAS is played by the analysis tools. Currently, there are no available tools developed specifically for SAS analysis, i.e., tools which focus on the adaptivity aspects including design and implementation features, or performance achievements. Therefore, the research question we address in this paper is: How to analyze/understand SAS and their adaptive features with available general-purpose tools?. Adaptivity can be analyzed (1) statically, starting from the source code, and generating higher abstraction related models concerning design and implementation solutions, and (2) dynamically, starting from software execution and focusing on the performances obtained due to the SAS adaptivity. In this paper, we perform static analysis of two SAS examples with the general purpose Understand tool. We use the Understand features to generate various graphs of the software architecture/structure, to analyze its complexity, and to compute various metrics. Furthermore, we propose metrics specific to self-adaptation. We compute these metrics using data gathered from Understand. The defined metrics are implemented as a plugin for Understand. We conclude that the Understand tool is suitable to perform static analysis on SAS. It may be also a candidate tool to be further extended and personalized to become also a SAS specific analysis tool.