Hyunjun Park, Insup Lee, Journal of Internet Computing and Services, Vol. 26, No. 4, pp. 1-8, Aug. 2025
10.7472/jksii.2025.26.4.1, Full Text:  HTML
Keywords: ddos detection, Deep Learning, Transfer Learning, traffic volume-based labeling

Abstract

As distributed denial of service (DDoS) attacks grow in scale and complexity, deep learning-based DDoS detection has gained significant attention. Deep learning models achieve high detection accuracy by effectively learning network traffic patterns from large-scale datasets. However, the detection of attack types with low traffic volumes or previously unseen variants remains challenging due to the limited availability of training data, which reduces detection performance. To address this limitation, this paper proposes a DDoS detection method that integrates domain knowledge-based labeling and transfer learning. This method reclassifies attack traffic into high-volume and low-rate DDoS categories based on traffic volume, applying a 1D CNN model with transfer learning to enhance detection performance.

Statistics
Cite this article