摘要
多智能体系统通过智能体间的信息交互与协调实现整体目标,其一致性研究不仅揭示了自组织与分布式控制的特点,也在计算机科学和人工智能等领域展现出重要的应用价值.智能体间的相互作用高度依赖网络通信条件,而数据传输过程中网络攻击的频繁发生对系统的安全稳定运行提出了严峻挑战.鉴于此,总结网络攻击下多智能体系统一致性安全与隐私保护的研究进展.首先,简要介绍攻击者对信息传输的保密性、完整性以及可用性产生的影响,揭示系统在窃听攻击、虚假数据注入攻击和DoS攻击这3类典型的攻击模式下的脆弱性;然后,针对不同网络攻击类型的特点,从防御者的角度对多智能体一致性研究中的安全控制与隐私保护问题进行讨论,并进一步梳理应对攻击常用的检测与防御保护机制;此外,综合讨论不同攻击在分布式控制相关研究分支的发展现状;最后针对多智能体系统一致性安全与隐私保护研究中存在的技术难题与挑战进行分析,并对未来研究方向进行展望.
Multi-agent systems achieve overall objectives through information interaction and coordination among agents.Research on consensus of multi-agent systems not only reveals the characteristics of self-organization and distributed control but also demonstrates significant practical application value in fields such as computer science and artificial intelligence.However,as interactions among agents heavily rely on network communication conditions,the frequent occurrence of cyber attacks during data transmission poses severe challenges to the secure and stable operation of systems.Therefore,this paper aims to summarize the research progress on consensus security and privacy protection of multi-agent systems under cyber attacks.First,the impact of attackers on the confidentiality,integrity,and availability of information transmission is briefly illustrated,highlighting the vulnerabilities of systems under three typical attack patterns:eavesdropping attacks,false data injection attacks,and denial-of-service(DoS)attacks.Subsequently,based on the characteristics of different types of cyber attacks,the paper discusses secure control and privacy protection issues in consensus research from the defender's perspective,while further summarizing commonly used detection and defense mechanisms.Additionally,the study comprehensively reviews the development status of distributed control-related research branches,focusing on secure control strategies under false data injection attacks and DoS attacks,as well as the design of consensus privacy protection algorithms under eavesdropping attacks.Finally,the technical challenges and issues in the research of consensus security and privacy protection for multi-agent systems are analyzed and research directions worth exploring are further outlined.
作者
卢剑权
邢梦平
张晶
LU Jian-quan;XING Meng-ping;ZHANG Jing(School of Mathematics,Southeast University,Nanjing 210096,China;College of Electronic Warfare,National University of Defense Technology,Hefei 230037,China;School of Mathematical Sciences,Nanjing Normal University,Nanjing 210023,China)
出处
《控制与决策》
北大核心
2025年第11期3201-3219,共19页
Control and Decision
基金
江苏省自然科学基金攀登项目(BK20240009)
国家自然科学基金面上项目(623731050)
江苏省应用数学科学研究中心项目(BK20233002)。
关键词
多智能体系统
网络攻击
隐私保护
安全控制
一致性
分布式控制
multi-agent systems
cyber attacks
privacy protection
secure control
consensus
distributed control
