Load balancing in fog networks using digital twins
Chi-Yu Wu, Ting-Ting Yang · IET conference proceedings. · 2025
The emergence of fog computing bridges the gap between cloud functionalities and end devices in the traditional cloud computing architecture, enabling services to be deployed closer to the users. Unfortunately, the inherent environmental differences, the heterogeneity of fog nodes, and the spatiotemporal variation in service demands from devices create a load imbalance among fog nodes. This study proposes a fog computing architecture that integrates digital twin and reinforcement learning technologies, aiming to implement a task offloading strategy to promote load balancing within fog networks. Through digital twin technology, the load of fog nodes can be monitored and predicted in real-time, while reinforcement learning is tasked with making appropriate task offloading decisions. The results confirm that this method effectively facilitates load balancing in fog networks.