基于强化学习的移动边缘计算任务卸载方法
2023,31(10):306-311
摘要：最佳卸载策略直接影响移动计算任务卸载的时延与能耗，因此提出基于强化学习方法的移动边缘计算任务卸载方法。首先对移动设备的计算任务卸载形式展开具体分析，并基于分析结果获取计算任务卸载能量消耗、发射功率、传输速率等相关参数值，以此建立移动边缘计算任务卸载模型。最后基于建立的卸载模型结合Q-Learning算法对计算任务实施强化学习，找出计算任务的最佳卸载策略，从而实现移动边缘计算任务的实时卸载。实验结果表明，使用强化学习方法开展移动边缘计算任务卸载时，卸载能耗低、时延小。
关键词：强化学习方法;Q-Learning算法;移动边缘;计算任务卸载;卸载模型
Offload method of mobile edge computing task based on reinforcement learning
Abstract：The optimal unloading strategy directly affects the time delay and energy consumption of computing task unloading, so a mobile edge computing task unloading method based on reinforcement learning method is proposed. Firstly, the form of computing task offloading of mobile devices is analyzed in detail, and the energy consumption, transmission power, transmission rate and other related parameter values of computing task offloading are obtained based on the analysis results, so as to establish a mobile edge computing task offloading model. At last, based on the established unloading model and Q-Learning algorithm, we implement reinforcement learning on computing tasks to find the best unloading strategy of computing tasks, so as to realize real-time unloading of mobile edge computing tasks. The experimental results show that when using this method to unload mobile edge computing tasks, the unloading energy consumption is low and the delay is small.
Key words：Strengthening learning methods; Q-Learning algorithm; Move the edge; Calculation task unloading; Unloading model
收稿日期：2023-05-29
基金项目：国家自然科学基金项目（面上项目，重点项目，重大项目）
