Offensive‐Defensive Integration in Autonomous Vehicles: A Novel Data‐Aware Attack and Resilient Control Framework

ABSTRACT

The cybersecurity of autonomous vehicles, particularly under stealthy data‐aware attacks, presents a critical challenge that conventional control systems are ill‐equipped to handle. While existing attack schemes lack efficiency or precision, this paper proposes a novel integrated offensive‐defensive framework to address this gap. First, a sophisticated data‐aware attack scheme is designed, which intelligently selects and corrupts the most critical measurement components in realtime based on a dynamic importance index and a dual‐threshold mechanism. This strategy ensures high destructiveness while maintaining strong concealment. Second, to counter such sophisticated attacks, an attack‐target‐related switching controller is developed that dynamically adjusts its gains from eight pre‐designed modes based on real‐time attack identification. The exponential stability and a prescribed disturbance attenuation performance of the closed‐loop system are rigorously guaranteed via Lyapunov theory. Finally, extensive simulations demonstrate that the proposed attack strategy can significantly degrade system performance and robustness, and more importantly, the proposed defense controller effectively maintains stability and ensures safe operation under malicious attacks, outperforming conventional controllers. This work establishes a foundational offensive‐defensive co‐design paradigm for securing autonomous vehicles against sophisticated cyber‐physical threats.