Data‐driven resilient platooning control for vehicular platooning systems with denial‐of‐service attacks

Abstract

This article addresses the data‐driven resilient platooning control problem for nonlinear vehicular platooning systems with denial‐of‐service attacks. First, the dynamic linearization technique is used for transforming the nonlinear vehicular platooning systems into an equivalent linear data model with robustness. Then, an observer is designed to present the estimation of the pseudo‐partial derivative parameter and a novel model‐free adaptive platooning control (MFAPC) framework is presented by defining a novel vehicular platooning system output. Based on this framework, a novel MFAPC algorithm is designed with an attack compensation mechanism to achieve vehicular platooning control objectives. At last, the effectiveness of the MFAPC algorithm is proved by an example with comparisons.

Resilient time‐varying output formation tracking of heterogeneous linear multiagent systems under malicious false data injection attacks and denial‐of‐service attacks over digraphs

“…The key technical challenges arising from such a scenario include: (i) how to design a distributed resilient observer to estimate and reconstruct the leader's output information for each follower under the impact of DoS attacks; and (ii) how to design a distributed observer‐based controller to guarantee attack‐resilient time‐varying output formation tracking under the influence of FDI and DoS attacks. Moreover, this work differs from these latest related works in References 25‐30 from the following aspects: (a) the control methods in References 25‐30 tackle either the FDI attacks or DoS attacks, whereas our proposed control method does not only deal with the impact of DoS attacks, but also compensate for the influence of FDI attacks; (b) the FDI attacks in References 25‐30 are handled as disturbances, noises or faults that have to be bounded with certain known knowledge, while in contrast, they are allowed to be malicious and unbounded with unknown knowledge; and (c) the obtained result in this work ensures an exponential convergence, whereas the UUB convergence results are obtained in References 25‐30.…”

“…Further, a novel resilient distributed output feedback control algorithm is developed under FDI sensor attacks to ensure an exponential convergence. In contrast to designs in References 16‐18 in which resilient function calculation and consensus were studied under the constraints on the number of malicious agents or certain special structure of attacks, those requirements are not required in this article. Moreover, in comparison with existing related works in References 19‐26 to deal with attacks as disturbances, noises or faults that have to be bounded with known knowledge, the studied FDI attacks are malicious, unknown and unbounded, which is practical for real‐world formation tracking applications. In addition, unlike 19‐26,28‐30 to obtain the UUB convergence results, the zero‐error global exponential convergence can be achieved using the proposed resilient distributed control law, which does not need information on FDI sensor attacks. Moreover, the information interaction among agents is subject to DoS attacks in which the digraphs are allowed to be disconnected, which is less mild than fixed undirected or directed graphs in References 19‐26. The novel resilient distributed output feedback control architecture enables the global exponential stability of the system for time‐varying output containment‐formation tracking with multiple leaders, which covers 26 as a special case.…”

“…Those designs in References 19‐26 to handle FDI attacks as disturbances, noises or faults, heavily depend on certain assumptions that these FDI attacks either satisfy some special structures or are upper bounded with certain known knowledge. Moreover, the authors in Reference 28‐30 consider DoS attacks to study the resilient platooning, formation tracking and flocking problems, respectively, where the proposed designs ensure the convergence to a small bound. The existing methods in References 19‐30 deal with either the FDI attacks or DoS attacks.…”

Dynamic event‐triggered model‐free adaptive security tracking control for constrained subway train system

“…From (13), we can obtain that ē(k) = ℏ(k + 1)e(k + 1). Then according to (14), if e(k + 1) is bounded, one can get ē(k + 1) is bounded. So, we need to prove the boundedness of e(k + 1) first.…”

“…There are two main types of attacks: denial‐of‐service (DoS) attacks and false data injection (FDI) attacks 7,8 . Research on DoS attacks has been studied for many years 9‐16 . DoS attacks are usually handled in three ways: detection attacks, resistance attacks, and compensation attacks.…”