A Monitor Method based on Adaptive Frequency for Self-Adaptive Software

Abstract

Self-Adaptive Systems (SASs) are required to adapt to the complex changes with different characteristics by frequent monitor. The problem with gathering and updating changes information frequently at runtime is that it may cause computing and communication overhead, which affects the real-time of SASs. However, if the frequencies are reduced, it is difficult to ensure the accuracy of changes identifying. So it is necessary to solve the trade-off between the accuracy and expensive overhead. Existing methods based on data processing improve accuracy by reducing potential uncertainties. Other methods based on adaptive frequency lack quantification and are difficult to ensure accuracy. We expect to combine them and break through the latter. To this purpose, this paper proposes a method based on adaptive frequency for SASs. The method can ensure the accuracy of adaptive adjustment by comprehensively analyzing the influencing factors of monitoring frequency, and can quantify the monitoring frequency in real time by calculating these factors according to the monitoring data. Finally, we exemplify these methods with an e-commerce System.