Research on APT Attack Provenance and Detection Mechanism Driven Based on Graph Neural Network
(Kunlun Digital Technology Co., Ltd., Beijing 102206, China)
Abstract
Aiming at the challenges of high false alarm rate and poor scalability in large-scale network attack provenance, an Advanced Persistent Threat (APT) intrusion detection system based on provenance graph and Graph Neural Network is constructed. Firstly, a typical attack provenance graph is constructed through the system log. Secondly, the semantic encoder is used to capture the basic semantic attributes and the temporal sequence of events in the provenance graph. Thirdly, a context encoder based on Graph Neural Network is used to effectively encode local and global graph structures into node embedding. Finally, the node embeddings generated during the training phase are quickly classified by the classifier. The algorithm achieves efficient processing of large-scale provenance graphs through Graph Neural Network, and takes into account the efficiency of data processing, which can be used for real-time detection of Advanced Persistent Threat. Compared with existing intrusion detection systems, the algorithm achieves higher detection accuracy on public test datasets, and shows better alarm efficiency and scalability.
Keywords
provenance graph; Graph Neural Network; APT attack detection
References
- [1] Fu Yu, Li Hongcheng, Wu Xiaoping, et al. A review of research on APT attack detection based on big data analysis [J]. Journal of Communications, 2015, 36 (11): 1-14
- [2] Liu Haibo, Wu Tianbo, Shen Jing, et al. APT attack detection based on GAN-LSTM [J]. Computer Science, 2020, 47 (1): 281-286
- [3] ALSHAMRANI A,MYNENI S,CHOWDHARY A,et al. A Survey on Advanced Persistent Threats: Techniques, Solutions, Challenges, and Research Opportunities [J].IEEE Communications Surveys and Tutorials,2019,21(2):1851-1877.
- [4] WANG S,WANG Z L,ZHOU T,et al. Threatrace: Detecting and Tracing Host-based Threats in Node Level Through Provenance Graph Learning [J].IEEE Transactions on Information Forensics and Security,2022,17:3972-3987.
- [5] SU J L,AHMED M,LU Y,et al. RoFormer: Enhanced Transformer with Rotary Position Embedding [J/OL]. Neurocomputing,2024,568:127063[2024-12-26].https://doi.org/10.1016/j.neucom.2023.127063.
- [6] HASSAN W U,BATES A,MARINO D. Tactical Provenance Analysis for Endpoint Detection and Response Systems [C]//2020 IEEE Symposium on Security and Privacy (SP).San Francisco:IEEE,2020:1172-1189.
- [7] LI M F,TAO H Y,LIU M K,et al. Study on Enhanced Fault Diagnosis of Chiller Units in HVAC Systems Under the Imbalanced Data Environment Using GA-optimized LightGBM [J].Energy and Buildings,2025,330:115360.
- [8] MANZOOR E,MILAJERDI S M,AKOGLU L. Fast Memory-efficient Anomaly Detection in Streaming Heterogeneous Graphs [C]//KDD'16: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.New York:ACM,2016:1035-1044.
- [9] MILLER & SMITH. Miller & Smith Announces “Unicorn” Quick Delivery Homes at Birchwood at Brambleton: A Unique and Convenient Solution for 55+ Homebuyers [EB/OL].(2024-10-30)[2025-06-04].https://www.abnewswire.com/pressreleases/miller-smith-announces-unicorn-quick-delivery-homes-at-birchwood-at-brambleton-a-unique-and-convenient-solution-for-55-homebuyers_718950.html.
- [10] DING X,BU W J,LI Z L,et al. A Cyber-attack Behavior Detection Model Based on Log Activity Graph [C]//2023 IEEE International Conference on Sensors, Electronics and Computer Engineering (ICSECE).Jinzhou:IEEE,2023:165-169.
About the authors
Zhang Liang (1980-), male, Han nationality, from Zhuozhou, Hebei Province, engineer, master's degree, research direction: cloud computing applications, network security system management; Li Cheng (1989-), male, Han ethnicity, from Rongcheng, Hebei Province, engineer, bachelor's degree, research direction: cloud service management, network security system management; Corresponding author: Chen Xiaobo (1983-), male, Han ethnicity, from Jizhou, Hebei Province, China. He is a senior engineer and holds a PhD. His research interests include cloud computing architecture planning, computing infrastructure, and cloud security; Li Baoke (1990-), male, Han ethnicity, from Liaocheng, Shandong, engineer, PhD, research direction: network security and cloud technology applications; Liu Kexin (1985-), male, Han nationality, from Tai'an, Shandong, engineer, master's degree, research direction: network security and cloud technology applications.