Semantic group network learning for enterprise and internet of things intrusion detection

Sivakumaran Arumugam, Sumithra Mani, Anbarasu Rajabather, Rajesh Elangovan, Arshita Srivastava, Prabhakaran Rajkumar

Abstract


Conventional intrusion detection models learn directly from individual network flow attributes, making their performance susceptible to variations in traffic characteristics across enterprise and internet of things (IoT) environments. To address this limitation, this paper proposes a semantic group network-based intrusion detection system (SGNet-IDS), a semantic group network learning framework that models network behavior through structured semantic representations rather than isolated features. The framework organizes normalized flow attributes into five semantic behavioral groups and employs a semantic group interaction module (SGIM) to capture complementary relationships among communication characteristics, producing compact latent representations for intrusion classification. SGNet-IDS was evaluated on the CSE-CIC-IDS2018 and CIC-IoT-DIAD2024 datasets and compared with representative intrusion detection methods and stronger tabular learning baselines. The proposed framework achieved 99.71% and 99.00% accuracy, with micro-averaged F1-scores of 99.14% and 99.48%, respectively. Ablation analysis demonstrated that semantic grouping and interaction learning each contributed to the overall detection performance, while computational evaluation confirmed competitive inference efficiency with moderate model complexity. These results indicate that SGNet-IDS provides an effective and computationally efficient semantic representation framework for intrusion detection across heterogeneous enterprise and IoT network environments.

Keywords


Behavioral learning; Enterprise networks; Intrusion detection; Network security; Semantic feature grouping; Semantic representation learning

Full Text:

PDF


DOI: https://doi.org/10.11591/eei.v15i6.14022

Refbacks

  • There are currently no refbacks.


Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

Bulletin of EEI Statistics

Bulletin of Electrical Engineering and Informatics (BEEI)
ISSN: 2089-3191 , e-ISSN: 2302-9285
This journal is published by the Institute of Advanced Engineering and Science (IAES) .