Hardware Design of Intrusion Detection System for Automotive CAN Bus Using Random Forest

Abstract

Controller area network (CAN) bus is a reliable protocol that connects electronic control units in a vehicle to send and receive control data for driving a vehicle. However, CAN-bus is quite weak against external attacks since it was not designed considering cybersecurity. In this paper, we propose an intrusion detection system (IDS). It monitors data frames on CAN bus, and detects spoofing attack using random forest, one of the machine learning techniques. When combined with node exclusion system (NES) in the previous works, IDS can effectively detect the hacked node and NES can exclude it from the CAN bus. The proposed IDS and NES have been designed and verified with Modelsim simulator.