Machine Learning-Based Detection of False Traffic Information Attacks in Connected Vehicle Network

Connected vehicle networks rely on cooperative information sharing to improve traffic efficiency and road safety. However, this reliance makes such networks vulnerable to false traffic information attacks, in which authenticated insider vehicles disseminate misleading alerts such as fake congestion, accidents, or road blockages. Since these messages are syntactically valid, traditional security mechanisms based on authentication and message integrity are insufficient to detect such attacks. In this paper, a machine learning-based detection framework is proposed to identify false traffic information attacks in connected vehicle networks. The proposed approach evaluates the credibility of received traffic messages by jointly analyzing vehicle dynamics, cooperative information from neighboring vehicles, and historical traffic behavior, while exploiting spatiotemporal consistency among vehicles. An anomaly detection strategy based on the Isolation Forest algorithm is employed due to its efficiency and ability to detect abnormal patterns without relying on labeled data or predefined attack signatures. The effectiveness of the proposed method is evaluated through simulation-based experiments in a realistic connected vehicle environment. The results demonstrate that the proposed approach achieves a detection accuracy of ۹۲.۵%, with a precision of ۹۰%, recall of ۹۱.۳%, and an F۱-score of ۹۰.۶%. These findings confirm that the proposed framework can effectively enhance the reliability and security of traffic information in connected vehicle networks.

کلیدواژه ها:

Connected Vehicle Networks ، False Traffic Information Attack ، Anomaly Detection ، Machine Learning ، Isolation Forest ، Vehicular Security ، Spatiotemporal Consistency