THESIS
2023
1 online resource (x, 48 pages) : illustrations (some color)
Abstract
Adaptive cruise control (ACC) is a critical function of the advanced driver assistance
system (ADAS) and plays an essential role in autonomous vehicles. In
detail, ACC is responsible for maintaining a safe distance from vehicles ahead
by controlling acceleration and braking. However, with the increasing number
of vehicles connected to the Internet, the reliability of ACC is facing more challenges
as it becomes more vulnerable to cyber attacks. Among the various types
of cyber attacks, false data injection attacks and replay attacks pose the most
significant threat to ACC. In this thesis, we propose a Kalman-filter-based detector
to defend against these attacks. The detector combines the Chi-squared
(χ
2) test and cumulative sum control chart (CUSUM) to distinguish abnormal
sensor data. T...[
Read more ]
Adaptive cruise control (ACC) is a critical function of the advanced driver assistance
system (ADAS) and plays an essential role in autonomous vehicles. In
detail, ACC is responsible for maintaining a safe distance from vehicles ahead
by controlling acceleration and braking. However, with the increasing number
of vehicles connected to the Internet, the reliability of ACC is facing more challenges
as it becomes more vulnerable to cyber attacks. Among the various types
of cyber attacks, false data injection attacks and replay attacks pose the most
significant threat to ACC. In this thesis, we propose a Kalman-filter-based detector
to defend against these attacks. The detector combines the Chi-squared
(χ
2) test and cumulative sum control chart (CUSUM) to distinguish abnormal
sensor data. To address the issue of the χ
2 detector’s limited sensitivity to
replay attacks, we propose a method integrating a velocity planner as a watermark
maker into ACC, while maintaining driving comfort. We evaluate the
effectiveness of the proposed detection method by implementing it in a real-time
simulation platform of autonomous vehicles under both types of attacks. The
results show that the method achieves a high level of detection performance.
Post a Comment