Event-Triggered Data-Driven Adaptive Iterative Learning Containment Control for Multi-Agent Systems via Fading Channels Under Cyber-Attacks

Abstract

This article investigates the problem of model-free adaptive iterative learning containment control for nonlinear multi-agent systems (MASs) with fading measurements against periodic denial-of-service (DoS) attacks. The stochastic process is utilized to model periodic DoS attacks. Based on the historical data, a compensation algorithm is devised to minimize the adverse effects caused by such attacks. The design of the distributed controller only utilizes the data from neighboring nodes of each agent after undergoing random attenuation. Additionally, a dynamic event-triggering mechanism is incorporated along the iterative axis to make communication load less and exhibits adaptability more. Building upon this foundation, a novel model-free adaptive iterative learning containment control method is proposed, which enables all followers to converge within the convex hull formed by the leader even in the presence of DoS attacks. Finally, two simulations are introduced to depict the availability of the provided design approach.