A Systematic Comparison on Prevailing Intrusion Detection Models
Modern vehicles have become connected via On-Board Units (OBUs) involving many complex embedded and networked devices with steadily increasing processing and communication resources. Those devices exchange information through intra-vehicle networks to implement various functionalities and perform actions. Vehicles’ connectivity has also been extended to external networks through vehicle-to-everything technologies, enabling communications with other vehicles, infrastructures, and smart devices. In parallel to the significant increase in quality of service, the connectivity of modern vehicles raises their vulnerabilities to cyber-attacks targeting both intra-vehicle and external networks. To secure communications in vehicular networks, there has been a consistent effort to develop intrusion detection systems based on machine learning techniques to detect and ultimately react to malicious cyber-attacks. In this article, we study several machine learning algorithms, deep learning models, and hyper-parameter optimization techniques to detect vulnerability attacks on vehicular networks. Experimental results on well-known data sets such as CICIDS2017, NSL-KDD, IoTID20, KDDCup99, and UNSW-NB15 indicate that learning-based algorithms can detect various types of intrusion detection attacks with significant performance.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
First Page Number
Last Page Number
Liu, Jianxuan; Xue, Haotian; Wang, Junfeng; Hong, Shuo; Fu, Hao; and Dib, Omar, "A Systematic Comparison on Prevailing Intrusion Detection Models" (2023). Kean Publications. 372.