Li, Zihao (Texas A&M University), Zhou, Yang (Texas A&M University), Kulkarni, Mihir (Texas A&M University), Jiang, Jiwan (University of Wisconsin-Madison), Zhang, Yunlong (Texas A&M University)

Probabilistic Safety Analysis of Sensor Attacks on Adaptive Cruise Control in Mixed Traffic

Scheduled for presentation during the Regular Session "Vehicle control" (FrBT4), Friday, September 27, 2024, 14:30−14:50, Salon 7

2024 IEEE 27th International Conference on Intelligent Transportation Systems (ITSC), September 24- 27, 2024, Edmonton, Canada

This information is tentative and subject to change. Compiled on October 14, 2024

Abstract

Adaptive Cruise Control (ACC) systems enhance vehicle safety by maintaining a safe following distance using onboard sensors, but the ACC system is vulnerable to sensing attacks. While previous research has focused on vehicle-to-vehicle communication attacks, the impact of sensing attacks on traffic safety remains under-investigated. This study reviews existing sensing attack models and designs two typical sensing attack scenarios considering human intervention and attack effectiveness: jamming and spoofing attacks. Then, we adopt the Generalized Force Model (GFM) and high-order linear feedback controller to generate the car-following behavior for human-driven vehicles and ACC, respectively. Rather than using default homogenous parameters, the calibrated car-following parameter distributions based on the NGSIM dataset and ACC field tests are utilized to capture the stochastic driving behavior. Finally, scenario-based analysis and sensitivity analyses reflect the safety impact of sensing attacks realistically but also analyze the critical factors (i.e., stochastic driving behavior, ACC penetration rate, and attack effectiveness) on safety risk under sensing attacks.