Optimal Security Control for Networked Switched Systems with a State-Prediction-based Anti-Dos Mechanism

Yiwen Qi,Honglin Geng,Ning Xing,Jie Huang,Yanwei Huang
DOI: https://doi.org/10.1002/rnc.6860
IF: 3.8973
2023-01-01
International Journal of Robust and Nonlinear Control
Abstract:The openness of networks renders them vulnerable to various forms of attacks. When networked switched systems (NSS) suffer from denial of service (DoS) attacks and delays, the real-time property of the dataset decreases, greatly affecting the control performance. To address this issue, this article proposes a switched adaptive dynamic programming (ADP) predictive control method. An event-triggered mechanism is designed to reduce unnecessary waste of network transmission resources. A predictive mechanism is designed to accurately reconstruct the missing system state and switching signal under DoS network attacks. Then, the reconstructed data are applied to train the actor and critic neural networks, which are used to approximate the optimal control policy and performance index function (PIF) of the NSS, respectively. Furthermore, the iterative convergence of the switched ADP algorithm is proved. Finally, a numerical example is provided to verify the effectiveness of the proposed method.
What problem does this paper attempt to address?