International Conference on Computer and Knowledge Engineering

Home / 15th International Conference on Computer and Knowledge Engineering

Online Task Offloading and Scheduling in Fog-Cloud ‎Environment based on Reinforcement Learning

Authors :

Ali Sheidaee¹ Leili Farzinvash² Alireza Sokhandan³

1- University of Tabriz 2- University of Tabriz 3- university of tabriz

Keywords :

Internet of Things (IoT)‎،Cloud computing‎،Fog computing‎،Task scheduling،Task Offloading،Reinforcement Learning

Abstract :

the rapid expansion of IoT devices has reshaped everyday ‎environments into data-intensive ecosystems, but this ‎transformation has also amplified the demand for real-time ‎processing and energy-efficient resource management. Meeting ‎these demands requires more than static scheduling policies; it ‎calls for intelligent, adaptive strategies capable of keeping pace ‎with dynamic and unpredictable workloads. In this study, we ‎introduce a reinforcement learning–based framework for task ‎offloading and scheduling within an integrated IoT–fog–cloud ‎architecture. Our approach builds upon Double Q-Learning, ‎enhanced with parallelized decision-making that enables fog and ‎cloud servers to jointly evaluate CPU demand, energy ‎consumption and response time. To further improve adaptability, ‎we incorporate an adaptive learning mechanism that ‎continuously balances exploration and exploitation, allowing the ‎system to respond effectively to fluctuating conditions and ‎heterogeneous resources. Through iterative refinement, the ‎framework learns execution policies that strike a balance between ‎responsiveness and sustainability. Experimental evaluations ‎demonstrate that our method reduces average task response time ‎by 5% and energy consumption by 11% compared with leading ‎alternatives. These results highlight the potential of parallelized ‎and adaptive reinforcement learning as a powerful solution for ‎managing large-scale IoT ecosystems, where efficiency and agility ‎are equally critical.‎