欺骗攻击下具备隐私保护的多智能体系统均值趋同控制  

Privacy-preserving Average Consensus Control for Multi-agent Systems Under Deception Attacks

作　　者：应晨铎 伍益明 徐明[1] 郑宁[1] 何熊熊[2] YING Chen-Duo;WU Yi-Ming;XU Ming;ZHENG Ning;HE Xiong-Xiong(School of Cyberspace,Hangzhou Dianzi University,Hangzhou 310018;College of Information Engineering,Zhejiang University of Technology,Hangzhou 310023)

机构地区：[1]杭州电子科技大学网络空间安全学院,杭州310018 [2]浙江工业大学信息工程学院,杭州310023

出　　处：《自动化学报》2023年第2期425-436,共12页Acta Automatica Sinica

基　　金：国家自然科学基金(61803135,61873239,62073109);浙江省公益技术应用研究项目(LGF21F020011)资助。

摘　　要：针对通信网络遭受欺骗攻击的离散时间多智能体系统,研究其均值趋同和隐私保护问题.首先,考虑链路信道存在窃听者的情形,提出一种基于状态分解思想的分布式网络节点值重构方法,以阻止系统初始信息的泄露.其次,针对所构建的欺骗攻击模型,利用重构后节点状态信息并结合现有的安全接受广播算法,提出一种适用于无向通信网络的多智能体系统均值趋同控制方法.理论分析表明,该方法能够有效保护节点初始状态信息的隐私,并能消除链路中欺骗攻击的影响,实现分布式系统中所有节点以初始值均值趋同.最后,通过数值仿真实验验证了该方法的有效性.This paper investigates the average consensus and privacy-preserving problem for discrete-time multiagent systems with deception attacks.First,considering the situation where there are eavesdroppers on the link channel,a distributed node value reconstruction method based on the idea of state decomposition is proposed to prevent the leakage of the initial information of the system.Then,under the constructed deception attack model,a novel average consensus control method for multi-agent systems with undirected communication networks is proposed,which uses the reconstructed node status information and combines with the secure acceptance and broadcast algorithm.Theoretical analysis shows that the proposed method can effectively protect the privacy of the initial state information of the nodes,and eliminate the influence of deception attacks in the links,and reach a consensus.Finally,numerical simulation experiments verify the effectiveness of the proposed method.

关 键 词：多智能体系统 均值趋同 欺骗攻击 隐私保护 网络安全 

分 类 号：TP309[自动化与计算机技术—计算机系统结构] TP273[自动化与计算机技术—计算机科学与技术]