Handling Uncertainty Online for Self-Adaptive Systems

Self-Adaptive Systems (SASs) are required to adapt to the frequent changes from external environments, user requirements and their own. However, the potential uncertainties challenge the SASs. In which, the uncertainties in Monitor is mainly that the monitoring data is inaccurate, and the uncertainties in Analyze is mainly that the abnormal state is difficult to match accurately. These uncertainties lead to identify changes inaccurately and directly affect the reliability of SASs. So they are particularly critical. To deal with them, current adaptive methods, which either consider only uncertainties at design or rely on specific scenarios without versatility. This paper deal with both two uncertainties of Monitor and Analyze at runtime, and introduces Fuzzy Theory and Semantic Distance Technology to handle them online to ensure the reliability of SASs, and our methods are common to all SASs. Finally, we exemplify these methods with the Bookstore System, which proves the effectiveness of the methods.