Scholarlink

Under the background of the big era of Internet live broadcast, providing low latency and high bandwidth network performance is key to sustainable development. Quick User Datagram Protocol Internet Connection Protocol (QUIC), as the underlying protocol of Hypertext Transfer Protocol Version 3 (HTTP/3), offers more efficient, reliable, and secure transmission features compared to HTTP/2, making HTTP/3 increasingly a research and application hotspot. However, some mechanisms of QUIC are easily exploited to launch network attacks, posing a serious threat to the smoothness of video streams. This article proposes a QUIC network anomaly traffic detection method based on CEEMD-CC-PE-KFD. Firstly, the noise-dominated intrinsic mode functions (IMFs) are determined through the combination of Complementary Ensemble Empirical Mode Decomposition (CEEMD), Correlation Coefficient (CC), and Permutation Entropy (PE). These IMFs are then denoised using the Kalman Filter Denoising (KFD) principle. Finally, all low-frequency IMFs are reconstructed, and the Hurst exponent is calculated to achieve attack warning. The results show that this method can effectively remove noise and preserve signal features while improving adaptive ability and accuracy, making it suitable for QUIC network traffic anomaly detection.

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

Complementary Ensemble Empirical Mode Decomposition and Permutation Entropy-Empowered Network Traffic Anomaly Detection for Secure QUIC Communications

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Liang Shan,
  • Ming Jiang,
  • Yuehua Fan,
  • Jinquan Nie,
  • Junyi Wu,
  • Yuanlong Cao

摘要

Under the background of the big era of Internet live broadcast, providing low latency and high bandwidth network performance is key to sustainable development. Quick User Datagram Protocol Internet Connection Protocol (QUIC), as the underlying protocol of Hypertext Transfer Protocol Version 3 (HTTP/3), offers more efficient, reliable, and secure transmission features compared to HTTP/2, making HTTP/3 increasingly a research and application hotspot. However, some mechanisms of QUIC are easily exploited to launch network attacks, posing a serious threat to the smoothness of video streams. This article proposes a QUIC network anomaly traffic detection method based on CEEMD-CC-PE-KFD. Firstly, the noise-dominated intrinsic mode functions (IMFs) are determined through the combination of Complementary Ensemble Empirical Mode Decomposition (CEEMD), Correlation Coefficient (CC), and Permutation Entropy (PE). These IMFs are then denoised using the Kalman Filter Denoising (KFD) principle. Finally, all low-frequency IMFs are reconstructed, and the Hurst exponent is calculated to achieve attack warning. The results show that this method can effectively remove noise and preserve signal features while improving adaptive ability and accuracy, making it suitable for QUIC network traffic anomaly detection.