毫米波车联网多基站多用户下的安全传输方案
Secure transmission scheme for millimeter-wave Internet of vehicles with multiple base stations and users
俱莹 1陈宇超 2田素恒 1刘雷 1李赞 1裴庆祺 1王明阳3
作者信息
- 1. 西安电子科技大学通信工程学院,陕西 西安 710077
- 2. 深圳市国电科技通信有限公司,广东 深圳 518000
- 3. 北京跟踪与通信技术研究所,北京 100085
- 折叠
摘要
针对移动场景下毫米波安全波束形成的时效性和毫米波动态窃听场景下安全连接的鲁棒性等问题,提出了一种基于决斗双重深度Q网络(D3QN)-深度确定性策略梯度(DDPG)算法的多智能体安全协作通信方案.该方案利用路侧单元(RSU)辅助的协作干扰技术降低窃听者对合法信号的接收质量,并通过联合优化车辆用户(VU)的基站与波束连接控制、阻塞RSU的选择,以及RSU的协作干扰方向和发射功率控制,使得所有合法车辆的总保密传输速率最大化.在此基础上,针对车联网的高动态性,通过构建基于D3QN的VU智能体和基于D3QN-DDPG的RSU智能体,实现了实时的离散-连续混合决策.最后,通过多维度的性能分析和方案对比实验,验证了所提方案的有效性.
Abstract
For the timeliness of millimeter-wave secure beamforming in mobile scenarios and the robustness of secure connections in millimeter-wave dynamic eavesdropping scenarios,a multi-agent secure cooperative communication scheme based on a dueling double deep Q network(D3QN)-deep deterministic policy gradient(DDPG)algorithm was proposed to address communication security issues.The scheme utilized road side unit(RSU)-assisted cooperative jam-ming technology to reduce the eavesdropper's reception quality of confidential signals.The optimization problem was formulated to maximize the total secrecy rate of all legitimate vehicles by optimizing the joint base station and beam connection control of the VUs,the selection of the jamming RSUs,and the cooperative jamming direction and transmit power of RSUs.Furthermore,for the challenges posed by the high dynamics of vehicular networks,the scheme achieved a fusion of real-time discrete and continuous decision-making by creating a VU agent grounded in D3QN and an RSU agent harnessing D3QN-DDPG capabilities.Finally,through multi-dimensional performance analysis and scheme comparison experiments,simulation results demonstrate the effectiveness of the proposed scheme.
关键词
毫米波车联网/多智能体/物理层安全/深度强化学习Key words
millimeter-wave Internet of vehicles/multi-agent/physical layer security/deep reinforcement learning引用本文复制引用
基金项目
国家自然科学基金(62102301)
国家自然科学基金(62132013)
国家杰出青年科学基金(延续资助)(62425103)
科学探索奖基金()
出版年
2024
国家科技期刊平台
NSTL
万方数据