Research on Attack Detection Method for V2V System Based on Multidimensional Taylor Series Expansion Framework

Strongly stealthy cyberattacks in networked vehicle-to-vehicle (V2V) network environments often manifest themselves as small changes to data, which makes it difficult for existing methods to effectively detect such attacks. Although the initial impact of these attacks is minimal, they are capable of significantly disrupting the system over time. To address this issue, this article proposes a new data representation model for attack detection, aiming to overcome the limitations of existing detection methods. For subtle and persistent attack features, we present a multilevel data representation framework, which enables effective comparison of attack features at different scales by extracting features at various levels. Through hierarchical processing of features, these subtle changes are revealed at the microlevel, thereby enhancing the system’s sensitivity to stealthy attacks. The experimental results show that the method in this article can effectively detect covert false data injection (FDI) attacks. Meanwhile, we compare the proposed method with the most recent competing methods, and the results demonstrate that our method has higher detection accuracy and stronger robustness compared with the existing methods.