TY - GEN
T1 - Tree-IDS
T2 - 48th IEEE Conference on Local Computer Networks , LCN 2023
AU - Zhao, Yuxin
AU - Bi, Zixiang
AU - Xu, Guosheng
AU - Wang, Chenyu
AU - Xu, Guoai
N1 - Publisher Copyright:
© 2023 IEEE.
PY - 2023
Y1 - 2023
N2 - The rapid development of Internet of Vehicles technology has led to the continuous upgrading of the functions of connected vehicles. While connected vehicles bring convenience to people’s life, there are also many security threats. Connected vehicles not only have intra-vehicle networks communication, but also communicate with the external network. The diversity of communication methods makes the attack surface wider, and some new attacks are constantly emerging. In order to ensure vehicle security, This paper focuses on the attacks that are vulnerable to vehicles, and proposes an incremental intrusion detection system, which can not only detect attacks, but also incrementally learn new types of attacks. Experimental results illustrate that the proposed system can incrementally learn new attacks and avoid catastrophic forgetting problems, and can detect various types of known attacks with 99.99% accuracy on the Car-Hacking Dataset and 99.37% accuracy on the CICIDS2017.
AB - The rapid development of Internet of Vehicles technology has led to the continuous upgrading of the functions of connected vehicles. While connected vehicles bring convenience to people’s life, there are also many security threats. Connected vehicles not only have intra-vehicle networks communication, but also communicate with the external network. The diversity of communication methods makes the attack surface wider, and some new attacks are constantly emerging. In order to ensure vehicle security, This paper focuses on the attacks that are vulnerable to vehicles, and proposes an incremental intrusion detection system, which can not only detect attacks, but also incrementally learn new types of attacks. Experimental results illustrate that the proposed system can incrementally learn new attacks and avoid catastrophic forgetting problems, and can detect various types of known attacks with 99.99% accuracy on the Car-Hacking Dataset and 99.37% accuracy on the CICIDS2017.
KW - Incremental Learning
KW - Intrusion Detection
UR - https://www.scopus.com/pages/publications/85182947585
U2 - 10.1109/LCN58197.2023.10223397
DO - 10.1109/LCN58197.2023.10223397
M3 - 会议稿件
AN - SCOPUS:85182947585
T3 - Proceedings - Conference on Local Computer Networks, LCN
BT - Proceedings of the 48th IEEE Conference on Local Computer Networks , LCN 2023
A2 - Bulut, Eyuphan
A2 - Tschorsch, Florian
A2 - Thilakarathna, Kanchana
PB - IEEE Computer Society
Y2 - 2 October 2023 through 5 October 2023
ER -