Task Offloading in Edge Computing Considering the Dynamics of Tasks and Networks Based on Grey Numbers

Task offloading from edge devices to the cloud is typically modeled using directed acyclic graphs (DAGs), which assume predefined task details and stable network conditions. However, obtaining detailed task information is often impractical, particularly in dynamic network environments. This article addresses these challenges by utilizing grey number theory to effectively capture the uncertainties inherent in both task characteristics and network conditions. Through a comprehensive comparative analysis of grey numbers, a deep Q-network (DQN) offloading method based on grey numbers is proposed, significantly improving the scheduling process. This study is the first to explore task offloading for edge devices while considering task uncertainty and network dynamics using grey number theory. Simulation results confirm the efficiency of the proposed method in selecting solutions that meet predefined criteria, with the DQN approach demonstrating significant improvements in scheduling performance.