Runtime Models for Analysing and Evaluating Quality Attributes of Self-Adaptive Software: A Survey

Abstract

Self-adaptive software has the capability to sense a change in its environment and its own behaviour, and then adjust itself accordingly during runtime to meet the desired requirements. Analysing and evaluating quality attributes is popular in self-adaptive software research. In recent years, several studies have proposed runtime models which analyse and evaluate quality attributes of self-adaptive software. This paper focuses on runtime models used for analysis and evaluation of quality attributes of self-adaptive software. Firstly, self-adaptive software, runtime models, analysis and evaluation of quality attributes based on runtime models, and related concepts are introduced. Studies describing runtime models are investigated and the meaning of runtime models used in analysis and evaluation of quality attributes of self-adaptive software is clarified. Two types of typical construction methods and their general construction processes are described. Runtime models were analysed and categorized considering multiple aspects, including type, modelling language, application scenarios, and relationship between runtime models and quality attributes. The extension mechanisms for runtime models were also analysed and extracted into two types. The weaknesses of current research are listed and analysed, with future research directions suggested.