Data‐Driven Fully Distributed Adaptive Event‐Triggered Consensus Control for Multi‐Agent Systems under DoS Attacks

ABSTRACT This paper considers the leader–follower consensus of continuous‐time linear multi‐agent systems (MASs) with unknown dynamics on undirected topologies vulnerable to denial‐of‐service (DoS) attacks. First, to mitigate limitations on communication bandwidth, a fully distributed adaptive dynamic event‐triggered (ADET) control strategy is proposed. This strategy relies solely on local information, avoids the use of any global network parameters, and ensures system stability while achieving leader–follower consensus. Second, aiming to resolve the previously open problem of consensus for MASs with unknown dynamics under DoS attacks, we design a distributed adaptive law incorporating exponential terms and double damping coupling weights. By constructing an appropriate Lyapunov function, we rigorously prove that consensus can still be attained in the presence of such attacks. Furthermore, an online data‐driven learning algorithm is employed to obtain a system representation, and the stability conditions are reformulated into a data‐based linear matrix inequality (LMI) framework. This approach eliminates the dependence on an accurate system model and offers a feasible research pathway for unknown dynamical systems under DoS attacks. Finally, numerical simulations are conducted to validate the effectiveness of the proposed control scheme and supporting theoretical results.