A robust control framework for multi-agent systems under Byzantine attacks using hybrid event-triggered techniques

Abstract

This paper addresses the shortcomings of existing leader-following consensus approaches in multi-agent systems, particularly their vulnerability to Byzantine attacks and inefficiencies in communication and control updates. Current methods often struggle to handle sensor and actuator disruptions from Byzantine agents and suffer from excessive controller updates, which lead to communication overhead and the potential for Zeno behavior. To tackle these challenges, we introduce the Hybrid Dynamic Event-Triggered Mechanism (HDETM), designed to minimize unnecessary agent interaction and reduce the frequency of controller updates while maintaining system robustness. By employing regularization techniques that ensure a positive inter-event time, our approach eliminates the risk of Zeno behavior. Additionally, the framework integrates auxiliary variables to capture both flow and jump dynamics, and through Lyapunov consensus analysis, we establish L2 stability with finite gain, ensuring system resilience under Byzantine attacks. This research demonstrates that HDETM optimizes communication efficiency and enhances the system's ability to withstand Byzantine disruptions. Extensive numerical simulations validate the effectiveness of the proposed methodology in achieving robust consensus, highlighting its superiority over existing approaches.

Keywords

• Multi-agent systems
• Consensus
• Byzantine attacks
• Hybrid dynamic event-triggered mechanism (HDETM)
• Interaction between agents
• System stability