Predictive Containment Control for Multi-Agent Systems Subject to Denial-of-Service Attacks

This paper is concerned with the H_∞ containment cooperative control problem for multi-agent systems (MASs) in a networked predictive control (NPC) framework, in which a recursive predictor is proposed to compensate for lost data caused by denial-of-service (DoS) attacks. Firstly, to facilitate system analysis, the DoS attacks are modeled by introducing a constraint on the attack duration. Secondly, to reproduce the lost states and fulfil containment task, we investigate a cooperative protocol based on prediction. This protocol uses states received from neighbor agents and the predictors. If the attacks are detected, the system control task is performed using predicted states; otherwise, true states are used. Thirdly, some sufficient conditions in terms of linear matrix inequalities are derived, which can guarantee that the error systems with proposed predicted-based protocol can achieve H_∞ performance. Furthermore, the proposed predicted-based strategy is investigated with dynamic event-triggering mechanisms involving an auxiliary dynamic variable (ADV-DETMs) under switching topologies. A controller is designed to achieve the H_∞ containment control for the MASs. Finally, we build an unmanned surface vehicles (USVs) platform as an application of our methods numerically, which demonstrates the effectiveness of the proposed predict-based cooperative control strategy.