Encrypted Malicious Traffic Detection Fusing Cross-Attention with Lightweight Graph Convolution
International Journal of Advanced Computer Science and Applications · 2026 · 17 (10)
DOI: 10.14569/ijacsa.2026.0171001 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Advanced Computer Science and Applications, 2026. Verfasst von Xiaogang Yuan und 3 weiteren Autor:innen.
Abstract (Originalsprache)
With the widespread adoption of encryption technologies, traditional deep packet inspection methods fail to parse encrypted payloads, making it difficult to effectively identify malicious traffic. To address this, this paper proposes a Dual-Level Cross-Attention Lightweight Graph Convolutional Network (DCA-LGCN). This method employs node-level attention to extract intra-flow features and edge-level attention to model inter-flow relationships, introducing a cross-attention mechanism for adaptive fusion of node and edge information. Simultaneously, it incorporates depthwise separable graph convolution and feature bottleneck layers to reduce computational overhead while maintaining detection accuracy. Experiments on the USTC-TFC2016 and CIC-Bell-DNS-EXF-2021 datasets demonstrate that DCA-LGCN significantly outperforms mainstream baseline models in terms of accuracy, inference latency, and model size. These results validate its efficiency and robustness in encrypted malicious traffic detection, achieving an effective balance between precision and efficiency.
Zitieren
APA
Yuan, X., Ni, M., An, D., & Wan, J. (2026). Encrypted Malicious Traffic Detection Fusing Cross-Attention with Lightweight Graph Convolution. International Journal of Advanced Computer Science and Applications, 17(10). https://doi.org/10.14569/ijacsa.2026.0171001
Vancouver
Yuan X, Ni M, An D, Wan J. Encrypted Malicious Traffic Detection Fusing Cross-Attention with Lightweight Graph Convolution. International Journal of Advanced Computer Science and Applications. 2026;17(10). doi:10.14569/ijacsa.2026.0171001
BibTeX
@article{Yuan2026,
author = {Yuan, Xiaogang and Ni, Minghang and An, Dezhi and Wan, Jianxin},
title = {Encrypted Malicious Traffic Detection Fusing Cross-Attention with Lightweight Graph Convolution},
journal = {International Journal of Advanced Computer Science and Applications},
year = {2026},
volume = {17},
number = {10},
doi = {10.14569/ijacsa.2026.0171001}
}
EUVIMED