摘要
随着人工智能技术的突破,语义通信成为6G通信网络关键研究方向,但无线信道的广播特性带来隐私风险,亟需强安全的解决方案。现有方案多依赖网络层加密,存在操作复杂、密钥泄露等隐患,而物理层安全可利用信道时变性与随机性规避复杂密...展开更多
随着人工智能技术的突破,语义通信成为6G通信网络关键研究方向,但无线信道的广播特性带来隐私风险,亟需强安全的解决方案。现有方案多依赖网络层加密,存在操作复杂、密钥泄露等隐患,而物理层安全可利用信道时变性与随机性规避复杂密码机制。基于此,提出一种信道容量约束的物理层安全语义通信框架。其核心设计目标是在将遍历安全容量限制于预设安全阈值的同时,最大化合法信道的信道容量。通过引入可微分的互信息估计器,建立了包含合法用户和窃听者的双信道容量估计模型,该模型可与语义通信系统进行联合优化。此外,还创新性地提出了安全句子相似度指标,以量化评估语义通信的安全性能。仿真结果表明,所提方案以传输精度的轻微下降为代价,换来了通信安全性的大幅增强。收起
With advancements in artificial intelligence,semantic communication has emerged as a key research direction for 6G networks.However,the broadcast nature of wireless channels introduces privacy risks,necessitating secure solutions.Existing methods often rely on network-l...MORE
With advancements in artificial intelligence,semantic communication has emerged as a key research direction for 6G networks.However,the broadcast nature of wireless channels introduces privacy risks,necessitating secure solutions.Existing methods often rely on network-layer encryption,which suffers from operational complexity and potential key leakage.Physical layer security,leveraging the channel's time-variation and randomness,can circumvent complex cryptographic mechanisms.In response,this paper proposes a physical layer secure semantic communication framework based on channel capacity constraints.The core design goal is to maximize the legitimate channel's capacity while restricting the overall security capacity to a predefined threshold.A differentiable mutual information estimator is introduced to construct a dual-channel capacity estimation model that includes both legitimate users and eavesdroppers,which can be jointly optimized with the semantic communication system.Additionally,an innovative security sentence similarity metric is proposed to quantify the security performance of semantic communication.Simulation resullts demonstrate that the proposed scheme significantly enhances communication security at the cost of a slight decrease in transmission accuracy.FEWER
作者
刘宇翔
王旭
施政
王金涛
杨光华
马少丹
LIU Yuxiang;WANG Xu;SHI Zheng;WANG Jintao;YANG Guanghua;MA Shaodan(The School of Intelligent Systems Science and Engineering,Jinan University,Zhuhai 519070,China;State Key Laboratory of Internet of Things for Smart City,University of Macao,Macao 999078,China)
出处
《移动通信》
2026年第1期35-43,共9页
Mobile Communications
基金
国家自然科学基金面上项目“面向超可靠低延时通信的叉包HARQ传输理论与方法”“智能反射面辅助的安全通信理论与关键技术研究”(62171200,62171201)
国家自然科学基金国际(地区)合作与交流项目“AI驱动的6G无线智能空口传输理论与关键技术”(62261160650)
广东省自然科学基金面上项目“叉包HARQ辅助太赫兹通信的传输理论和方法研究”(2023A1515010900)
中央高校基本科研业务费项目“基于深度学习的端到端语义通信可靠性研究”(21625361)
高校青年教师科研创新能力支持项目(SRICSPYF-BS2025134)
暨南大学研究生拔尖创新人才项目资助(2025CXY358)。
关键词
语义通信
物理层安全
信道容量
互信息估计
semantic communication
physical layer security
channel capacity
mutual information estimation