基于信道特征的物联网设备物理层认证  

Physical layer authentication of Internet of Things devices based on channel features

作　　者：江凌云[1] 史秀秀 JIANG Lingyun;SHI Xiuxiu(School of Communications and Information Engineering,Nanjing University of Posts and Telecommunications,Nanjing 210003,China)

机构地区：[1]南京邮电大学通信与信息工程学院,江苏南京210003

出　　处：《南京邮电大学学报(自然科学版)》2025年第1期21-28,共8页Journal of Nanjing University of Posts and Telecommunications:Natural Science Edition

基　　金：江苏省重点研发计划(BE2020084-4);国家自然科学基金(92067201)资助项目。

摘　　要：目前的物联网设备处在复杂的环境中且资源有限,基于信道特征的被动型物理层认证(Physical Layer Authentication,PLA)方式非常适合应用于目前的物联网设备。而传统基于信道特征的PLA采集到的是静态特征,导致现实中的时变信道认证概率较低。针对这一问题,使用支持向量机(Support Vector Machine,SVM)对时变信道下提取的信道特征进行分类认证,并使用在线学习随机梯度下降(Stochastic Gradient Descent,SGD)来更新SVM模型,实现了分类模型随着信道的变化而更新。此外,使用了鲁棒主成分分析(Robust Principal Component Analysis,RPCA)对提取的信道特征进行降维处理,降低获取SVM模型的复杂度并抑制了信道噪声的干扰。仿真结果表明,方案改善了时变信道下的认证概率,提高了鲁棒性。Current Internet of Things devices are arranged in complex environments with limited re⁃sources.The channel-based passive physical layer authentication(PLA)consumes less resources,and is more suitable for current Internet of Things devices.The traditional channel-based PLA collects static fea⁃tures,resulting in low time-varying channel authentication probability in practice.To solve this problem,the support vector machine(SVM)is used to classify the extracted channel features under time-varying channels.The online learning stochastic gradient descent(SGD)is used to update the SVM model,thus the SVM model can be updated with the channel change.Moreover,the robust principal component analysis(RPCA)is used to reduce the dimensionality of the extracted channel features,and thus the complexity of the SVM scheme is decreased and the interference of channel noise is suppressed.Simula⁃tion results show that the scheme improves the authentication probability and the robustness under timevarying channels.

关 键 词：物理层认证 支持向量机 随机梯度下降 鲁棒主成分分析 

分 类 号：TN918[电子电信—通信与信息系统] TP181[电子电信—信息与通信工程] TP212[自动化与计算机技术—控制理论与控制工程]