A runtime approach for software fault analysis based on interpolation

In the application system, obtaining the information of the system at runtime and analyzing them are important for system adjustment. Many runtime metrics can be collected from software systems, and some statistical relationships exist among these metrics. Extracting the information of these metrics from the monitoring data and then analyzing the relationships between these metrics is an effective way to detect failure and diagnose fault. This paper proposes a fault analysis approach for the system at runtime which gets the information of the system by monitoring. We demonstrate this approach in a case study which shows that our approach is effective and is beneficial to find the relationship between the fault and the component.