Autonomous Reconfigurations in Ad-Hoc Cloud Computing Continuums

Network metrics collected from a Kubernetes-based application are used to model and continuously monitor the system for anomalies. Application instances are automatically retrieved, and for each instance, the most relevant network metrics are gathered within a defined time window (files labeled “model”). These metrics are then combined (file labeled “model_.csv”) to train a behavioral model. Subsequently, a new time window containing the latest metric data is analyzed to infer and assess whether the observed anomaly rate exceeds a predefined threshold, thereby indicating a potential issue.