A Price-based Energy-aware Offloading Mechanism Using Q-learning in Fog-Cloud Environments

In this research, we present an approach based on reinforcement learning to encourage users to offload their tasks instead of running locally. After formulating the problem with queuing theory, an algorithm using Q-learning is proposed. Evaluating the performance of the proposed method against traditional and state-of-the-art methods shows that it has a significant advantage. The proposed method, on average, consumes less energy compared to other methods. Also, it reduces the execution time of tasks and leads to less consumption of network resources.本研究提出一种基于强化学习（Reinforcement Learning）的方法，旨在鼓励用户将任务卸载至远端而非本地运行。在运用排队论（Queuing Theory）对该问题进行形式化建模后，本研究提出了一种采用Q学习（Q-learning）的算法。将所提方法与传统方法及当前最优方法开展性能对比，结果显示该方法具备显著优势：相较于其他方法，其平均能耗更低，同时能够缩短任务执行时长、降低网络资源消耗。