Scholarlink

<p>The rapid development of smart cities, fueled by the growth of the Internet of Things (IoT) and interconnected systems, has greatly enhanced urban infrastructure, especially in transportation and energy management. However, this increased connectivity also raises the risk of cyberattacks, threatening service availability, financial stability, and public safety. This study introduces a resilient cybersecurity framework designed to detect and classify various cyber threats, including DoS, DDoS, Reconnaissance, Sybil, Replay, and Spoofing attacks, targeting critical transportation systems such as the Internet of Vehicles (IoV), electric vehicle (EV) charging networks, and Vehicular Ad hoc Networks (VANETs). By combining machine learning with Federated Learning (FL), the framework effectively tackles key challenges like high computational costs, dependence on centralized data, and scalability across different IoT systems. FL improves data privacy by keeping sensitive information on edge devices, reducing concerns over centralized data storage. Moreover, TreeSHAP, an interpretability technique, is utilized to provide transparency and deeper insights into attack detection. The proposed system achieves high F1 scores of 0.980, 0.982, and 0.99 on the CICIoV2024, CICEVSE2024, and VeReMi Extension datasets, respectively, demonstrating its effectiveness on multiple IoT security datasets relevant to smart city transportation and energy systems. while safeguarding user privacy.</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

FeXAI: Federated and Explainable AI for cyber threat detection in IoT-enabled smart transportation systems

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月9日

查看全文
  • Abirami Gurushanker,
  • A. Jeffrey Rufus,
  • C. Christopher Columbus,
  • C. K. Aravind

摘要

The rapid development of smart cities, fueled by the growth of the Internet of Things (IoT) and interconnected systems, has greatly enhanced urban infrastructure, especially in transportation and energy management. However, this increased connectivity also raises the risk of cyberattacks, threatening service availability, financial stability, and public safety. This study introduces a resilient cybersecurity framework designed to detect and classify various cyber threats, including DoS, DDoS, Reconnaissance, Sybil, Replay, and Spoofing attacks, targeting critical transportation systems such as the Internet of Vehicles (IoV), electric vehicle (EV) charging networks, and Vehicular Ad hoc Networks (VANETs). By combining machine learning with Federated Learning (FL), the framework effectively tackles key challenges like high computational costs, dependence on centralized data, and scalability across different IoT systems. FL improves data privacy by keeping sensitive information on edge devices, reducing concerns over centralized data storage. Moreover, TreeSHAP, an interpretability technique, is utilized to provide transparency and deeper insights into attack detection. The proposed system achieves high F1 scores of 0.980, 0.982, and 0.99 on the CICIoV2024, CICEVSE2024, and VeReMi Extension datasets, respectively, demonstrating its effectiveness on multiple IoT security datasets relevant to smart city transportation and energy systems. while safeguarding user privacy.