RELIABILITY-AWARE MONITORING FOR CLOUD–FOG ARCHITECTURES USING LIGHTWEIGHT MACHINE LEARNING
Saravanan Raj
Abstract
Cloud-fog architectures allow high-scale and latency-aware services through the distribution of the computation to nearer data sources, still, the provision of reliable and efficient monitoring is a challenging issue because of resource limitation, heterogeneity, and workload dynamism. Redundancy A paper highlighting a dependable platform of monitoring in cloud-fog ecosystems merging optimistic machine learning with agile sampling and selective reporting is introduced in this paper. The estimation of local reliability at the fog nodes is useful in dynamic setting the intensity of monitoring depending on the predicted operational stability to minimize the overhead unnecessarily and maintain observability. The framework lays more emphasis on critical conditions by employing the reliability-based adaptation and only transmits small-sized summaries to the cloud in cases of degradation. Large-scale testing based on heterogeneous monitoring data shows that it is characterized by substantial benefits in average monitoring latency, stability, scalability, coverage, and reliability detection accuracy in comparison with well-established fog computing, edge computing, adaptive sampling, and lightweight anomaly detection strategies. The findings uphold that reliability awareness and lightweight learning are effective measures of scalability, efficiency, and resilience of monitoring in cloud-fog architectures.