Federated Learning-Based Computation Offloading for Low-Bandwidth Edge Internet of Things

Abstract

Internet of Things (IoT) devices generates massive amounts of data continuously, making it difficult to combine this data at the edge node or central edge server for AI techniques to be applied, especially on low-bandwidth networks. Traditional computation offloading involves extensive resource management, queuing, and high bandwidth usage to send data from an edge device to a server. To mitigate this challenge, federated learning is applied in this paper, to offload the computation of the edge server. Simulation results show the efficiency of the proposed method over the centralized method.