Resilience Analysis of Closed-Loop Multiagent Systems Under Replay Attacks in the Context of Consensus Tracking
针对重放攻击下严格反馈非线性多智能体系统的韧性分析难题，提出一种基于分布式自适应一致性控制框架的动态稳定性分析方法，通过设计韧性指标和迭代算法计算跟踪误差上界，并证明攻击间隔满足条件时可维持系统韧性。
At present, extensive research efforts have been devoted to investigating network attacks on control systems. However, comparatively limited attention has been directed toward the resilience of multiagent systems (MASs) under such attacks, particularly in the case of replay attacks. Addressing the challenge of resilience analysis in strict-feedback nonlinear MASs under replay attacks, this article proposes a dynamic stability analysis method based on a classical distributed adaptive consensus control framework. To evaluate the resilience of the MASs in the context of aperiodic replay attacks, a dynamic compact set model is designed as a resilience metric. An iterative algorithm is then developed to compute the upper bound of tracking error jump at the beginning and end of the attack. In the scenario where the control signals under replay attacks cannot be explicitly modeled, this study derives an upper bound on the variation of the tracking error during the attack period using the Lyapunov stability analysis. It is proven that resilience can be maintained when the resting time between two consecutive replay attacks satisfies a given sufficient condition. Finally, simulation results illustrate the effectiveness of the proposed analysis method.