EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Network Intrusion Detection With Spatial and Temporal Knowledge Graphs and Semi-Supervised Learning

Asanka Kavinda Mananayaka, Sun Sunnie Chung

IEEE Access · 2026 · 14 · 95955-95968

DOI: 10.1109/access.2026.3706431 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Asanka Kavinda Mananayaka und 1 weiteren Autor:innen. Die Arbeit nennt 67 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

As the Internet of Things (IoT), Industrial Internet of Things (IIoT), and Cloud Computing continue to drive exponential growth in network connectivity, cyber-attacks targeting these networks are also increasing at an alarming rate. Early and accurate intrusion detection is therefore critical for safeguarding modern network infrastructure. Conventional deep learning methods are limited by their reliance on linear packet sequences to capture spatial and temporal features, preventing them from effectively leveraging the underlying topological structure of network communications. In addition, these methods often require large amounts of labeled data, which are increasingly difficult to obtain as new attacks emerge frequently. To address these challenges, we propose a novel Network Intrusion Detection framework based on Knowledge Graphs (KGs) and Graph Representation Learning (GRL). We construct two complementary KGs: a Spatial KG to capture topology-aware communication relationships and a Temporal KG to capture temporal flow interactions. The learned node embeddings are then used as expressive input features to an ensemble classifier. We also evaluate the semi-supervised learning capability of the proposed framework under reduced labeled data conditions. The proposed method was evaluated using three widely used intrusion detection datasets representing both IoT and Enterprise networks. Experimental results achieved an average detection-rate of 0.9832 and a false-alarm-rate of 0.0035, outperforming several existing graph-based approaches and demonstrating the effectiveness of combining graph-based learning with ensemble learning for intrusion detection applications.

Zitieren

APA

Mananayaka, A. K., & Chung, S. S. (2026). Network Intrusion Detection With Spatial and Temporal Knowledge Graphs and Semi-Supervised Learning. IEEE Access, 14, 95955-95968. https://doi.org/10.1109/access.2026.3706431

Vancouver

Mananayaka AK, Chung SS. Network Intrusion Detection With Spatial and Temporal Knowledge Graphs and Semi-Supervised Learning. IEEE Access. 2026;14:95955-95968. doi:10.1109/access.2026.3706431

BibTeX
@article{Mananayaka2026,
  author = {Mananayaka, Asanka Kavinda and Chung, Sun Sunnie},
  title = {Network Intrusion Detection With Spatial and Temporal Knowledge Graphs and Semi-Supervised Learning},
  journal = {IEEE Access},
  year = {2026},
  volume = {14},
  pages = {95955--95968},
  doi = {10.1109/access.2026.3706431}
}
Kostenlos starten →