Quantum optimization for fast CAN bus intrusion detection

Abstract

Modern vehicles, nowadays, come loaded with hundreds of different sensors generating a huge amount of data. This data is shared and processed among different Electronic Control Units (ECUs) through an in-vehicle network, such as the CAN bus, to improve the driver's experience and safety. However, the implementation of new features increases exposure to cyber-attacks. The CAN bus, which is designed to grant reliable communication, has many security weaknesses that might be exploited by an attacker. The need for highly accurate real-time intrusion detection systems (IDSs) for the automotive industry is limited to classical machine learning techniques, which are usually time-consuming and have hardware limitations. In this work, we analyze an optimized and efficient version of a network-based IDS for CAN bus attack detection based on Quantum Annealing. The models were tested on two different CAN bus datasets. The results show that the Quantum Annealing algorithm outperforms a classical classification algorithm in terms of time performance, which is important in the identification of attacks in the automotive sector, and achieves similar results for detection accuracy.