A Novel Technique for Well-Optimized Content Replication in Fog Computing

A fog computing architecture processes huge amount of data by using edge devices. The cloud computing paradigm's fog layer helps to improve quality of service (QoS) in time-sensitive applications. Because devices at the edge of the cloud are very diverse and dynamic, content replication is vital. Replication is essential for consistent service delivery in fog computing. Time-sensitive requests received scant consideration in many of the early approaches. To improve fog computing replication, this research suggests a novel content replication mechanism using KNN and Naïve Bayes classification models. The recommended approach consists of two steps: The first assigns a failure risk classification to the fog servers that are available for replication. In the second phase, we suggest size-based content replication, in which the file or content is replicated on the less likely-to-fail fog server. We have evaluated the performance using metrics like utilization, throughput, and latency. The usefulness of the proposed strategy is validated by simulation results comparing it with the Replica-3 method. Comparing this to the Replica-3 technique, better resource usage, higher throughput, and lower latency are therefore attained. By allowing the system to handle more queries in parallel across multiple computers or processors, KNN's parallel processing capabilities of CPUs and GPUs ensures better resource utilization and greatly increasing throughput as compared to Naïve Bayes model. Additionally, KNN classification offers a useful data structure that significantly reduces latency by minimizing the time required to find the nearest neighbours.