Event-Driven Safe and Resilient Control of Automated and Human-Driven Vehicles under EU-FDI Attacks

A research team including Yi Zhang, Yichao Wang, and Wei Xiao has published a paper introducing a novel control framework designed to secure Connected and Automated Vehicles (CAVs) operating alongside Human-Driven Vehicles (HDVs). The work addresses a critical gap in automotive cybersecurity: protecting safety-critical control systems from Exponentially Unbounded False Data Injection (EU-FDI) attacks, which can maliciously alter sensor or control inputs like acceleration commands. Current strategies often trade off resilience for safety or vice versa, but this new Event-Driven Safe and Resilient (EDSR) framework aims to provide both simultaneously, specifically for complex scenarios like lane-changing in mixed traffic.

The technical core of the EDSR framework combines event-driven Control Barrier Functions (CBFs) for guaranteed collision avoidance with Control Lyapunov Functions (CLFs) for stable velocity regulation, all augmented with adaptive attack-resilient control logic. A key innovation is its data-driven estimation module that predicts the behavior of unpredictable human drivers, allowing the CAV to plan safe maneuvers even under adversarial conditions. The event-driven design triggers control updates only when necessary, significantly reducing computational load compared to continuous processing while maintaining real-time safety guarantees. Simulation results validate that the EDSR framework outperforms non-resilient, purely event-driven methods, successfully executing collision-free lane changes and maintaining stable operation even when acceleration inputs are compromised by sophisticated EU-FDI attacks.