A Blockchain-Powered Malicious Node Detection in Internet of Autonomous Vehicles

• Published in: IEEE transactions on intelligent transportation systems 2024-11, Vol.25 (11), p.16277-16287
• Main Authors: Beni Prathiba, Sahaya, Murali, Pranav, Shenbaga Moorthy, Rajalakshmi, Kumar Anandhan, Deepak, Selvaraj, Arikumar K., Rodrigues, Joel J. P. C.
• Format: Article
• Language: English
• Subjects: Anomaly detection; Autonomous vehicles; blockchain; Blockchains; Feature extraction; Gray-scale; image classification; Internet of Autonomous Vehicles; Malware; malware detection; Security
• ISSN: 1524-9050; 1558-0016
• DOI: 10.1109/TITS.2024.3433480

The proliferation of Autonomous Vehicles (AVs) in recent times has opened up new possibilities for effective and secure transportation. However, with the increasing adoption of AVs, guaranteeing the accuracy and security of their sensory systems is becoming paramount. Specifically, the vulnerability of these systems to malware and sensor faults can pose significant risks to the dependable and secure operation of the vehicle. To identify and combat these issues we propose a stream-based Blockchain-powered Malicious Node Detection (BMND) method to analyze and report any malicious activity of the AV operating as a node on the Internet of Autonomous Vehicles (IoAV) network, wherein the existing solutions are at lower latency. BMND involves the detection of sensor anomalies and defects post-production of the AV. In the case that malware or any other malicious software is detected on the onboard compute unit it is isolated and contained, and the AV will be classified as malicious until appropriate remedial measures are taken to deter any sharing of erroneous or malicious data. When the AV is deemed safe from malware and defects in the system then a block is mined by the AV node and a unique ID is assigned to allow data transfers on the blockchain with other nodes. Only active nodes with assigned IDs and available blocks for transactions on the blockchain influence AV decision-making. BMND would allow modern AVs on the road to effectively communicate with reliable information. Experimental analysis shows that the malware detection in BMND is 4.4% more accurate with an F1-score of ~0.99 as compared to previous and other current state-of-the-art methods, and the communication capabilities of BMND are also better regarding security and latency concerning proposed vanilla blockchain methods.