基于深度强化学习的移动边缘计算安全传输策略研究  

Research on secure transport strategy of mobile edge computing based on deep reinforcement learning

作　　者：王义君[1] 李嘉欣 闫志颖 吕婧莹 钱志鸿[2] WANG Yijun;LI Jiaxin;YAN Zhiying;LYU Jingying;QIAN Zhihong(College of Electronic and Information Engineering,Changchun University of Science and Technology,Changchun 130022,China;College of Communication Engineering,Jilin University,Changchun 130012,China)

机构地区：[1]长春理工大学电子信息工程学院,吉林长春130022 [2]吉林大学通信工程学院,吉林长春130012

出　　处：《通信学报》2025年第4期272-281,共10页Journal on Communications

基　　金：国家自然科学基金资助项目(No.61540022);吉林省科技厅重点研发基金资助项目(No.20230203091SF)。

摘　　要：在移动边缘计算中,任务卸载过程中会面临信息泄露和被窃听等安全问题。为了提高移动边缘计算系统的安全传输效率,提出了无人机辅助物理层安全传输策略。首先,构建了无人机(UAV)搭载的移动边缘计算系统,由I个用户设备、M架合法无人机(L-UAV)和N架窃听无人机(E-UAV)构成;其次,保证L-UAV在规定周期内完成卸载任务的同时,以通信系统安全传输效率最大化为目标,采用引入注意力机制的多智能体深度确定性策略梯度(A-MADDPG)算法进行问题求解与优化;最后,在保证卸载前提下实现用户的机密信息不被窃听者窃听和安全计算效率最大化,保障系统整体安全性。仿真结果表明,所提算法相较于其他基准算法展现了更佳性能,在安全传输效率方面表现优越。In mobile edge computing,the process of task unloading will face security problems such as information leakage and eavesdropping.To improve the unloading efficiency of mobile edge computing system,the physical layer security transmission strategy of mobile edge computing was proposed.Firstly,the mobile edge computing system based on unmanned aerial vehicle(UAV)was studied,which was composed of I user devices,M legal UAV(L-UAV)and N eavesdropping UAV(E-UAV).Secondly,while ensuring the unloading of L-UAV within a specified period,the multi-agent depth deterministic policy gradient(Attention-MADDPG)algorithm with the addition of attention mechanism was adopted to solve and optimize the problem with the aim of maximizing the safety unloading efficiency of the communication system.Finally,under the premise of ensuring uninstallation,the user’s confidential information was not eavesdropped by the eavesdropper,and the secure computing efficiency was maximized to ensure the overall security of the system.Simulation results show that compared with other benchmark algorithms,the proposed algorithm has better performance in terms of secure transmission efficiency.

关 键 词：移动边缘计算 物理层安全 深度强化学习 无人机辅助卸载 

分 类 号：TN918.91[电子电信—通信与信息系统]