Scholarlink

<p>Integrated Sensing and Communication (ISAC) networks made possible by unmanned aerial vehicles (UAVs) are becoming an essential part of 6G smart cities and surveillance systems. However, they are extremely susceptible to sophisticated cyberattacks, including replay, spoofing, jamming, flooding, and fuzzy assaults, due to their open wireless architecture, dominance of Line-of-Sight communication, and resource-constrained onboard computing. High false alarm rates, limited integration with secure communication protocols, poor scalability in multi-UAV scenarios, and high computing complexity are all problems with current intrusion detection systems. This research suggests a deep learning-based Collaborative Intrusion Detection System (CIDS) for ISAC networks enabled by UAVs that combines intelligent multi-attack detection with lightweight encryption in order to overcome these difficulties. While a Temporal Convolution Recurrent Network (TCRN) model examines encrypted traffic patterns to identify malicious activity, a Flexible Lightweight Symmetric Block Cryptography (FlexLSBCrypt) technique protects UAV to base station communication with lower encryption overhead. The UAVCAN dataset is used to validate the suggested framework in simulated multi-UAV communication scenarios and under various attack scenarios. The processing cost, delay, encryption overhead, false positive rate (FPR), false negative rate (FNR) and the detection accuracy are considered in the experimental evaluation. In comparison to CNN, RNN, and LSTM-based IDS models, the results demonstrate that the suggested model achieves 98.7% accuracy, enhances detection performance by 5.2%, lowers detection latency by 4.8%, and decreases FPR by 6.1%. Moreover, the encryption system provides an encryption time that is 67 times faster and a cost of computation that is 12% lower, and this indicates that the encryption mechanism can be adopted in the real-time implementation of multiple UAVs. On the whole, the proposed collaborative IDS provides a reliable, scalable, and computationally efficient architecture to protect ISAC networks with the assistance of UAVs to ensure sensing reliability, integrity and enhanced situational awareness will be deployed in next-generation 6G applications.</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

A Deep Learning-Based Collaborative Intrusion Detection System for UAV-Enabled Integrated Sensing and Communication

  • 期刊论文

  • 发表日期：

    2026年7月16日

查看全文
  • Siyyadula Venkata Rama Vara Prasad,
  • Pabitra Mohan Khilar

摘要

Integrated Sensing and Communication (ISAC) networks made possible by unmanned aerial vehicles (UAVs) are becoming an essential part of 6G smart cities and surveillance systems. However, they are extremely susceptible to sophisticated cyberattacks, including replay, spoofing, jamming, flooding, and fuzzy assaults, due to their open wireless architecture, dominance of Line-of-Sight communication, and resource-constrained onboard computing. High false alarm rates, limited integration with secure communication protocols, poor scalability in multi-UAV scenarios, and high computing complexity are all problems with current intrusion detection systems. This research suggests a deep learning-based Collaborative Intrusion Detection System (CIDS) for ISAC networks enabled by UAVs that combines intelligent multi-attack detection with lightweight encryption in order to overcome these difficulties. While a Temporal Convolution Recurrent Network (TCRN) model examines encrypted traffic patterns to identify malicious activity, a Flexible Lightweight Symmetric Block Cryptography (FlexLSBCrypt) technique protects UAV to base station communication with lower encryption overhead. The UAVCAN dataset is used to validate the suggested framework in simulated multi-UAV communication scenarios and under various attack scenarios. The processing cost, delay, encryption overhead, false positive rate (FPR), false negative rate (FNR) and the detection accuracy are considered in the experimental evaluation. In comparison to CNN, RNN, and LSTM-based IDS models, the results demonstrate that the suggested model achieves 98.7% accuracy, enhances detection performance by 5.2%, lowers detection latency by 4.8%, and decreases FPR by 6.1%. Moreover, the encryption system provides an encryption time that is 67 times faster and a cost of computation that is 12% lower, and this indicates that the encryption mechanism can be adopted in the real-time implementation of multiple UAVs. On the whole, the proposed collaborative IDS provides a reliable, scalable, and computationally efficient architecture to protect ISAC networks with the assistance of UAVs to ensure sensing reliability, integrity and enhanced situational awareness will be deployed in next-generation 6G applications.