Authors: Sundari R, Assistant Professor Vijaya E
Abstract: The growing number of IoT devices has resulted in increasing demands on low latency and real-time processing requirements. Conventional cloud-based infrastructure does not meet the demand because of natural limitations with respect to bandwidth and time delay associated with communication processes. This research proposes a holistic edge computing framework that enables intelligence of IoT applications by utilizing federated learning techniques to perform privacy-preserving AI inference, load-balancing algorithms, and fault tolerance structures. The new architecture deals effectively with such important issues as scalability, resource management, energy efficiency, and reliability. It offers impressive performance gains compared to other architectures, with improvements in latency, throughput, and energy use by 45%, 38%, and 29%, correspondingly. System availability is ensured up to 99.97%.
