QOS Technologies and Services in Edge/Fog Computing

Abstract

The thesis addresses the critical challenge of ensuring Quality of Service (QoS) in resource-constrained edge and fog computing environments. Through empirical experimentation with Raspberry Pi 4 devices, it collects and analyzes 275 TCP traffic samples to extract key statistical features. The core research applies unsupervised machine learning clustering algorithms to identify distinct traffic patterns, such as stable and high-throughput flows. The results demonstrate that TCP behavior is largely stable, whereas UDP exhibits high burstiness and self-similarity. These findings provide a practical, data-driven framework for optimizing resource allocation and enhancing network reliability in distributed computing systems.