(An) analysis of sensor attacks and its effect on open-sourced ADS

Abstract

The reliability of automated driving systems (ADS) heavily relies on the accuracy of sensor data. Despite the existence of known sensor attacks targeting the sensors used by ADS, previous research has lacked a comprehensive evaluation of ADS against such attacks. To address this research gap, we performed comprehensive evaluations on an ADS testbed, specifically targeting camera saturation attack, color stripe injection attack, and LiDAR saturation attack. Through these evaluations, we successfully identified four distinct vulnerable designs within the design of the ADS. Additionally, we proposed and implemented countermeasures to mitigate the impact of sensor attacks, emphasizing the importance of robust security measures to fortify the resilience of ADS in real-world situations. By combining the construction of the ADS testbed with the evaluation of sensor attacks, this study provides valuable insights into the vulnerabilities and potential mitigations for enhancing the security and reliability of ADS.