Scholarlink

<p>As intelligent transport systems advance, security becomes a more pressing concern. Researchers studying the security of vehicular ad hoc networks (VANETs) have made significant strides. The intrusion detection system attracts the most attention of all. In addition to position falsification, other types of attacks (such as DoS and DDoS) may be considered. This research presents a novel Attention-Enhanced HierarOptiCNN-LSTM architecture for threat detection and security enhancement in VANETs. Z-Score Normalization is first applied for data preprocessing, and SMOTE is then used to balance the data further. To efficiently extract deep features from VANETs based on a unique Hierarchical stacked convolutional neural network architecture. The optimal feature subset for intrusion detection is selected using Golden Jackal Optimization. Lastly, a bidirectional long short-term memory model with attention is presented for VANETs with longer temporal patterns, reducing false positives and improving intrusion detection. Lastly, attacks in VANETs are classified using a fully connected layer. The results show that the proposed method achieves high accuracy, precision, recall, and F1-score, and its detection rate is compared with the baseline works. Overall, the proposed method improves the security by precisely identifying and categorizing attacks in IDS-based VANETs.</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

Vanet security reinvented: hierarchical stacked CNNS and attention-driven BILSTM optimization

  • 期刊论文

  • 发表日期：

    2026年6月17日

查看全文
  • Saleha Saudagar,
  • Gayatri Jagnade,
  • Geetika Narang,
  • Rupali Maske,
  • Sneha Tirth

摘要

As intelligent transport systems advance, security becomes a more pressing concern. Researchers studying the security of vehicular ad hoc networks (VANETs) have made significant strides. The intrusion detection system attracts the most attention of all. In addition to position falsification, other types of attacks (such as DoS and DDoS) may be considered. This research presents a novel Attention-Enhanced HierarOptiCNN-LSTM architecture for threat detection and security enhancement in VANETs. Z-Score Normalization is first applied for data preprocessing, and SMOTE is then used to balance the data further. To efficiently extract deep features from VANETs based on a unique Hierarchical stacked convolutional neural network architecture. The optimal feature subset for intrusion detection is selected using Golden Jackal Optimization. Lastly, a bidirectional long short-term memory model with attention is presented for VANETs with longer temporal patterns, reducing false positives and improving intrusion detection. Lastly, attacks in VANETs are classified using a fully connected layer. The results show that the proposed method achieves high accuracy, precision, recall, and F1-score, and its detection rate is compared with the baseline works. Overall, the proposed method improves the security by precisely identifying and categorizing attacks in IDS-based VANETs.