虚假数据注入攻击信号的融合估计  

Fusion Estimate of FDI Attack Signals

作　　者：翁品迪 陈博[1,2] 俞立[1,2] WENG Pin-Di;CHEN Bo;YU Li(Institute of Cyberspace Security,Zhejiang University of Tech-nology,Hangzhou 310023;College of Information Engineering,Zhejiang University of Technology,Hangzhou 310023)

机构地区：[1]浙江工业大学网络空间安全研究院,杭州310023 [2]浙江工业大学信息工程学院,杭州310023

出　　处：《自动化学报》2021年第9期2292-2300,共9页Acta Automatica Sinica

基　　金：国家自然科学基金(61673351,61973277)资助。

摘　　要：研究了信息物理系统中假数据注入(False data injection,FDI)攻击信号的检测问题.在分布式融合框架下,首先将FDI攻击信号建模为信息物理系统模型中的未知输入,从而使得攻击信号的检测问题转化为对FDI攻击信号的实时估计问题.其次,在每个传感器端设计基于自适应卡尔曼滤波的FDI攻击信号的局部估计器;在融合中心端引入补偿因子,设计分布式信息融合准则以导出攻击信号的融合估计器.特别地,当FDI攻击信号是时变情况时,融合过程中补偿因子的引入可以大大提高对攻击信号的估计精度.最后,通过两个仿真算例验证所提算法的有效性.This paper is concerned with the estimation problem of false data injection(FDI)attacks in cyber-physical systems,where the false data is injected into the actuator.Under the distributed fusion framework,the attack signal is modeled as an unknown parameter in the addressed system models,and each local estimate of the attack signal at the sensor is obtained based on adaptive Kalman filtering method.Subsequently,by introducing the compensation factor,an optimal weighing fusion criterion is designed to improve estimation precision of attack signals.Particularly,when the FDI attack signal is time-varying,the introduction of compensation factor during fusing process can indeed improve the estimation precision largely.Finally,two illustrative examples are used to show the effectiveness of the proposed methods.

关 键 词：自适应Kalman滤波 假数据注入攻击 攻击信号估计 信息融合 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]