Optimal Denial-of-Service Attack Sequence towards the Consensus of Multi-Agent Systems

This paper investigates the optimization of Denial-of-Service (DoS) attack strategies targeting a multi-agent system aiming to achieve consensus. The attack mode is a multi-channel DoS attack, which can target the communication edges of the system. By combining communication edges, various attack strategies can be obtained. Considering the system in discrete time, the attack strategy is determined at each system step based on the system state, and the combination of these strategies forms the attack sequence. The goal is to optimize the attack strategy to reduce the attacker's energy consumption while maintaining a certain attack effect. The energy of the attacker is characterized by the Frobenius norm of the Laplacian matrix of the graph formed by the attacked edges. And the degree of system degradation is characterized by the consensus error of the system states. The optimization objective function is formulated by combining these two metrics. Finally, a simulation of this attack method is implemented on MATLAB.