Scholarlink

<p>With the rapid advancement of vehicular network technology, its open-ended nature and complexity have introduced unprecedented security challenges, rendering traditional intrusion detection systems inadequate for identifying unknown attack patterns. To address this issue, this paper proposes an efficient anomaly detection framework for vehicular network systems. By integrating autoencoder structures with dynamic threshold adjustment strategies through deep learning technology, the framework can comprehensively analyze the intricate communication patterns and attack characteristics inherent in vehicular network, thereby enhancing their security and reliability. Two benchmark datasets, Car-hacking and CIC-IoV 2024, were employed to thoroughly validate the effectiveness of the proposed method. This work lays a solid theoretical foundation for advancing the security of future vehicular network technologies and provides significant technical support for addressing security challenges in dynamic and complex environments.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An efficient vehicular network anomaly detection framework based on encoder and dynamic threshold adjustment

  • 期刊论文

  • 发表日期：

    2025年8月12日

查看全文
  • Huibin Xu,
  • Long Fang,
  • Jingnan Dong,
  • Jishui Shi

摘要

With the rapid advancement of vehicular network technology, its open-ended nature and complexity have introduced unprecedented security challenges, rendering traditional intrusion detection systems inadequate for identifying unknown attack patterns. To address this issue, this paper proposes an efficient anomaly detection framework for vehicular network systems. By integrating autoencoder structures with dynamic threshold adjustment strategies through deep learning technology, the framework can comprehensively analyze the intricate communication patterns and attack characteristics inherent in vehicular network, thereby enhancing their security and reliability. Two benchmark datasets, Car-hacking and CIC-IoV 2024, were employed to thoroughly validate the effectiveness of the proposed method. This work lays a solid theoretical foundation for advancing the security of future vehicular network technologies and provides significant technical support for addressing security challenges in dynamic and complex environments.