Resilient MPC of Networked Autonomous Vehicles Against FDI Attacks Based on Mixed-Integer Dynamic Tube

Networked Autonomous Vehicles (NAVs) face dual threats from cyber and physical domains during operation, presenting substantial challenges to the safety and security systems. This paper presents a resilient control approach to mitigate False Data Injection (FDI) attacks and scene uncertainties. To enhance the robustness of the control system, Tube-Based Robust Model Predictive Control (TRMPC) is employed to reduce localization data deviations resulting from FDI attacks. Additionally, dynamic scene information is integrated as tube constraints to address collision risks stemming from scene uncertainties. Considering scene constraints constitute a non-convex set, a mixed-integer method transforms dynamic tube constraints into convex constraints. Finally, a scenario-based case study is conducted to validate the effectiveness and performance of the proposed resilient control method based on mixed-integer dynamic tubes. The results demonstrate that the proposed method effectively mitigates the impacts of FDI attacks while preventing collision incidents at the physical layer, thereby ensuring the operational safety and stability of NAV systems.