Comparative Analysis of Deep Learning Techniques for Network Traffic Classification
JOIV : International Journal on Informatics Visualization · 31.05.2026 · 10 (3) · 1346
DOI: 10.62527/joiv.10.3.3402 ↗Kurz gefasst
Fachartikel in der Zeitschrift JOIV : International Journal on Informatics Visualization, online erschienen am 31.05.2026. Verfasst von Muhammad Alif Ikhwan Zainuddin und 3 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-SA 4.0).
Abstract (Originalsprache)
This research presents a comparative analysis of deep learning models, Artificial Neural Networks (ANNs), Convolutional Neural Networks (CNNs), and Recurrent Neural Networks (RNNs) for network traffic classification in today's growingly complex digital environments. The research focuses on identifying which model offers the best trade-off between accuracy, computational cost, and responsiveness to dynamic network conditions. The study utilizes actual Quality of Service (QoS) measurements, i.e., Bandwidth and Jitter, collected from Jabatan Data Negara Malaysia (JDN), as benign and malicious traffic under static and dynamic environments. The models were implemented using Python in a manner that optimized the ANN for fixed traffic streams, the CNN for spatial feature extraction processes, and the RNN, with the integration of Long Short-Term Memory (LSTM) layers, for handling temporal sequences. Preprocessing entailed data cleaning, normalization, and feature engineering to aid in effective model training. The performances of the models were observed using accuracy, precision, recall, and F1-score as metrics. Results show that the data obtained a validation accuracy of 99.73% under consistent conditions, CNN achieved 61.25% for the detection of spatial anomalies, and RNN achieved 100% accuracy under varying traffic conditions due to its advantage in the learning of sequences. These findings show that RNNs are most useful for real-time adaptive traffic control, while ANNs suit normal, predictable networks. This work contributes to the development of intelligent, QoS aware network management and provides the foundation for further investigation on hybrid deep learning models that can respond under varying network conditions.
Zitieren
APA
Zainuddin, M. A. I., Isa, M. R. M., Ali, F., & Kamarudin, N. D. (2026). Comparative Analysis of Deep Learning Techniques for Network Traffic Classification. JOIV : International Journal on Informatics Visualization, 10(3), 1346. https://doi.org/10.62527/joiv.10.3.3402
Vancouver
Zainuddin MAI, Isa MRM, Ali F, Kamarudin ND. Comparative Analysis of Deep Learning Techniques for Network Traffic Classification. JOIV : International Journal on Informatics Visualization. 2026;10(3):1346. doi:10.62527/joiv.10.3.3402
BibTeX
@article{Zainuddin2026,
author = {Zainuddin, Muhammad Alif Ikhwan and Isa, Mohd Rizal Mohd and Ali, Fazilatulaili and Kamarudin, Nur Diyana},
title = {Comparative Analysis of Deep Learning Techniques for Network Traffic Classification},
journal = {JOIV : International Journal on Informatics Visualization},
year = {2026},
volume = {10},
number = {3},
pages = {1346},
doi = {10.62527/joiv.10.3.3402}
}
EUVIMED