NARX Model-Based Data-Driven Secure Control for UMVs Under Asynchronous DoS Attacks

In this paper, based on nonlinear auto-regressive model with exogenous inputs (NARX) neural network, a resilient data-driven multi-step error feedback secure control method is designed for unmanned marine vehicles (UMVs). And the asynchronous DoS attacks are considered in Sensor-Controller channel and Controller-Sensor channel. Significantly, the success probability of these DoS attacks varies due to energy constraints across distinct channels. To mitigate the negative impart of DoS attacks, a NARX model-based predictor is devised. And the predict error are considered to design control scheme, enhancing the control effect for UMVs under asynchronous DoS attacks. What's more, a resilient partial form multi-step error feedback control (PF-MSEFC) method is designed by incorporating the proportional control term, the different control term, and the quadratic difference control term, which improves the utilization of historical tracking error information and enhance the flexibility of the control scheme. Finally, the effectiveness of the proposed resilient NARX predictive model-based PF-MSEFC scheme is demonstrated through simulation examples.