Improved event-triggered finite-time H ∞ control for neural networks subject to mixed-type communication attacks

• Published in: International journal of control Vol. 96; no. 8; pp. 2068 - 2079
• Main Authors: Xu, Yao, Lu, Hongqian, Song, Xingxing, Zhou, Wuneng
• Format: Journal Article
• Language: English
• Published: Abingdon Taylor & Francis 03.08.2023; Taylor & Francis Ltd
• Subjects: Co-design; Control systems design; Cybersecurity; Denial of service attacks; dynamic event-triggered scheme; Finite-time H; H-infinity control; Independent variables; Linear matrix inequalities; mixed-type communication attacks; Neural networks; stabilisation
• ISSN: 0020-7179; 1366-5820
• DOI: 10.1080/00207179.2022.2081259

This paper discusses the problem of finite-time stabilisation for neural networks (NNs) subject to mixed-type communication attacks via an improved dynamic event-triggered scheme (DETS). The complex cyber-attacks considered consist of three common types of attacks: replay attacks, deception attacks, and denial-of-service (DoS) attacks. Different from most articles which use independent Bernoulli variables to model the cyber-attacks, this paper considers these attacks into a unified Markovian jump framework for modelling. In order to save the limited network communication resources, the improved DETS is adopted. An appropriate Lyapunov-Krasovskii functional (LKF) containing the proposed improved DETS condition is constructed, and sufficient conditions are obtained to guarantee finite-time stabilisation of the system. Then, according to a set of feasible linear matrix inequalities (LMIs), the co-design of event-trigger and controller is given. Finally, two numerical examples are provided to demonstrate the effectiveness of our method.