窃听攻击下基于传感器传输策略的信息物理系统隐私保护  

Privacy protection of cyber-physical system under eavesdropping attack based on sensor-based transmission strategy

作　　者：金增旺 刘奕彤 刁靖东 王震[1] 孙长银 刘志强[1] Zengwang JIN;Yitong LIU;Jingdong DIAO;Zhen WANG;Changyin SUN;Zhiqiang LIU(School of Cybersecurity,Northwestern Polytechnical University,Xi’an 710072,China;Yangtze River Delta Research Institute of NPU,Northwestern Polytechnical University,Taicang 215421,China;Ningbo Institute of NPU,Northwestern Polytechnical University,Ningbo 315103,China;Qian Xuesen Laboratory of Space Technology,China Academy of Space Technology,Beijing 100094,China;School of Artifcial Intelligence,Anhui University,Hefei 230031,China)

机构地区：[1]西北工业大学网络空间安全学院,西安710072 [2]西北工业大学长三角研究院,太仓215421 [3]西北工业大学宁波研究院,宁波315103 [4]中国空间技术研究院钱学森空间技术实验室,北京100094 [5]安徽大学人工智能学院,合肥230031

出　　处：《中国科学:信息科学》2025年第3期619-638,共20页Scientia Sinica(Informationis)

基　　金：国家重点研发计划(批准号:2022YFB3104005);国家自然科学基金(批准号:U21B2008,U23B2039);2022年度太仓市基础研究计划(批准号:TC2022JC17);宁波市自然科学基金(批准号:2021J046)资助项目。

摘　　要：信息物理系统(cyber-physical system, CPS)的开放性和无线传感器的普及,使得CPS受到窃听攻击的风险加剧,因此CPS隐私保护成为亟待解决的重要问题.本文针对窃听攻击下CPS中的隐私保护问题利用远程状态估计的方法,设计了一种将传输调度与噪声注入相结合的传感器传输策略.该策略允许传感器在发送数据包和保持静默间选择,并在发送数据时以一定概率传输噪声以混淆数据.本文引入开环预测的概念作为系统基准,构建合法用户与窃听者的马尔可夫链模型,推导出期望稳态估计误差协方差,确定了噪声传输概率的范围.本文优化合法用户和窃听者估计误差协方差的负线性组合,达到最小化合法用户的误差并使得窃听者的误差保持在开环估计之上的效果,以此实现隐私保护和能量管理的平衡.此外,本文证明了所提方案具有阈值结构,并通过基于动态规划的算法在仿真环境下对算法进行验证与分析.The openness of cyber-physical systems(CPS)and the widespread use of wireless sensors elevate the vulnerability to eavesdropping attacks on CPS.Consequently,the privacy preservation of CPS becomes an urgent and critical concern.In this paper,a sensor transmission strategy combining transmission scheduling and noise injection is designed using the remote state estimation method to protect the privacy of CPS under eavesdropping attacks.This strategy enables the sensor to transmit data or maintain silence,utilizing noise transmission to confuse the data when sending it.This paper introduces the concept of open-loop prediction as the system reference.It constructs Markov chain models for the legitimate user and eavesdropper,deriving the expected steady-state estimation error covariance.Consequently,the range of noise transmission probability is determined.The objective is to minimize the error of the legitimate user while maintaining the error of the eavesdropper above the open-loop estimation,thereby striking a balance between privacy protection and energy management.In addition,the proposed scheme is demonstrated to possess a threshold structure,and its effectiveness is validated and analyzed within a simulation environment by a dynamic programming algorithm.

关 键 词：信息物理系统 隐私保护 远程状态估计 传感器传输策略 噪声注入 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置] TP309[自动化与计算机技术—控制科学与工程]