Scholarlink

<p>Autonomous Vehicles (AVs) rely heavily on intricate and interconnected communication systems, making them vulnerable to various cyberattacks such as Denial of Service (DoS), spoofing, and sniffing. These threats can compromise vehicle safety, disrupt communication, and affect overall system reliability. To address these challenges, this study presents a smart Intrusion Detection System (IDS) specifically developed for AV ecosystems. It employs a tree-based machine learning model to efficiently detect cyber threats while maintaining low computational requirements.The IDS is strategically designed to protect two primary communication layers: the internal Controller Area Network (CAN Bus) and the external Vehicle-to-Everything (V2X) interface. A Random Forest Classifier was employed as the baseline detection model, configured with 100 estimators and optimized through parallel processing. StandardScaler was used for feature normalization, and the model was trained on an 80:20 train-test split. Evaluation results demonstrated exceptional performance: for CAN Bus attacks, the IDS achieved 96.0% accuracy, 96.01% precision, 96.0% recall, a 96.0% F1-score, and a 98.94% ROC AUC score. For V2X threats, it attained 94.94% across all key metrics with a 98.56% ROC AUC.These outcomes validate the IDS’s capability to detect and mitigate cyber threats in real time, ensuring a lightweight, scalable solution suitable for onboard deployment in AV systems. This work reinforces the critical need for intelligent cybersecurity frameworks in modern transportation, contributing to the development of resilient, secure autonomous vehicle ecosystems.</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

Securing autonomous vehicles: a dual-domain intrusion detection system for intra-vehicle and external networks

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年3月25日

查看全文
  • Zerui Shen,
  • Jingyang Zuo,
  • Shreyash Srivastva,
  • Anurag Sinha,
  • Saroj Kumar Pandey,
  • Mohammad Nadeem Ahmed,
  • Mohammad Rashid Hussain,
  • Ashish Kumar Verma

摘要

Autonomous Vehicles (AVs) rely heavily on intricate and interconnected communication systems, making them vulnerable to various cyberattacks such as Denial of Service (DoS), spoofing, and sniffing. These threats can compromise vehicle safety, disrupt communication, and affect overall system reliability. To address these challenges, this study presents a smart Intrusion Detection System (IDS) specifically developed for AV ecosystems. It employs a tree-based machine learning model to efficiently detect cyber threats while maintaining low computational requirements.The IDS is strategically designed to protect two primary communication layers: the internal Controller Area Network (CAN Bus) and the external Vehicle-to-Everything (V2X) interface. A Random Forest Classifier was employed as the baseline detection model, configured with 100 estimators and optimized through parallel processing. StandardScaler was used for feature normalization, and the model was trained on an 80:20 train-test split. Evaluation results demonstrated exceptional performance: for CAN Bus attacks, the IDS achieved 96.0% accuracy, 96.01% precision, 96.0% recall, a 96.0% F1-score, and a 98.94% ROC AUC score. For V2X threats, it attained 94.94% across all key metrics with a 98.56% ROC AUC.These outcomes validate the IDS’s capability to detect and mitigate cyber threats in real time, ensuring a lightweight, scalable solution suitable for onboard deployment in AV systems. This work reinforces the critical need for intelligent cybersecurity frameworks in modern transportation, contributing to the development of resilient, secure autonomous vehicle ecosystems.