Automotive Cybersecurity Scheme for Intrusion Detection in CAN-Driven Artificial Intelligence of Things

Abstract

The Artificial Intelligence of Things (AIoT) is applicable for various domains, that is, smart healthcare, smart cities, industrial sectors, transportation systems, and many more. Controller area network (CAN) facilitates the integration of the sensing devices, thus enables them to send their data for analysis to various artificial intelligence (AI) algorithms. CAN also provides reliability and fault tolerance to the AIoT applications, as it has been designed to deal with noisy environments. Thus, CAN-driven AIoT improves the efficiency, reliability, and functionalities of the devices and systems, which is very much needed for various AIoT applications. However, it is vulnerable to various cyber-attacks like message replay, modification attack, fuzzy attack, denial of service, and spoofing the RPM gauge or drive gear. Therefore, an intrusion detection system (IDS) is required to detect attacks on the CAN bus. In this paper, we propose a lightweight and efficient intrusion detection system which successfully detects multiple intrusions based on the type of attack on CAN bus without causing additional traffic overhead to the ongoing communications (in short, ACID-CAN). The presented mechanism is very much needed for the CAN-driven AIoT applications. Experimental results show that the proposed ACID-CAN successfully detects intrusions even when the amount of intrusion data is reduced to of normal data. The obtained results were compared with those of previous studies in the field of CANs intrusion detection, and it has been noted that the proposed ACID-CAN offers comparable and better results.