Segment Detection Algorithm: CAN bus intrusion detection based on Bit Constraint

Abstract

With the rapid development of Internet of Vehicles and autonomous driving technologies, car manufacturers provide more comfortable and safe driving experience while gradually exposing their vehicles to the background of cyber-attacks. As the car’s interior communicates through the CAN bus, the intrusion detection for CAN bus becomes crucial. Some studies use bus data characteristics, machine learning algorithms, or information theory algorithms to perform intrusion detection on the CAN bus, but they have problems such as low detection accuracy, high performance requirements, and insufficient detection granularity. This paper innovatively proposes a lightweight detection algorithm—Segment Detection Algorithm (SDA), which calculates the bit flip rate by segment, discovers the variation relationship between bits within each segment, and utilizes multiple inter-message features to achieve the detection of abnormal traffic. Experiments show that compared with existing research, the algorithm has effectively improved the detection accuracy, especially the detection of replay attacks. In addition, the algorithm has extremely low time complexity, can adapt to the limited resources in the vehicle environment, and achieve high-precision real-time detection of abnormal traffic.