Neural Predictor-Based Formation Control for Unmanned Surface Vehicles Under Aperiodic DoS Attacks: A Noncooperative Game Approach

This paper investigates a noncooperative gamebased formation control strategy for Unmanned Surface Vehicles (USVs) under Denial-of-Service (DoS) attacks. An resilient Nash equilibrium (NE) estimator is proposed, which is capable of stably searching for and tracking the NE solution despite the presence of aperiodic DoS attacks. To enhance prediction accuracy, a neural network integrated with accelerated learning is introduced to efficiently approximate the unknown parameters of the USV model. Within the noncooperative game framework, a distributed control strategy based on NE strategy seeking is designed, enabling each USV to coordinate its movement with neighboring USVs using local information. Simulation results demonstrate that the proposed control method exhibits robust performance in harsh network environments, effectively mitigating the negative impact of DoS attacks and ensuring precise control of the USV formation and tracking