Real-Time Low-Rate Ddos Attack Detection In Sdn Environments Using Liquid Neural Networks With Ryu Controller
International Journal of Drug Delivery Technology · 18.05.2026 · 16 (28s)
DOI: 10.25258/ijddt.16.28s.57 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Drug Delivery Technology, online erschienen am 18.05.2026. Verfasst von D. Rajani Kumar und 1 weiteren Autor:innen.
Abstract (Originalsprache)
The proliferation of software-defined networking (sdn) has revolutionized network management through centralized control and programmability. However, this architectural paradigm introduces new security vulnerabilities, particularly low-rate distributed denial of service (lddos) attacks that exploit tcp's congestion control mechanisms while maintaining low average bandwidth consumption. Traditional detection approaches struggle to identify these stealthy attacks due to their periodic burst patterns that mimic legitimate traffic fluctuations. This paper presents a novel detection framework that integrates liquid neural networks (lnns) with the ryu sdn controller for real-time lddos attack detection. Unlike conventional recurrent architectures, lnns employ continuous-time dynamics with time-varying synaptic connections, enabling superior capture of temporal patterns in network traffic. The proposed architecture processes openflow statistics in real-time, extracting 23 flow features including packet counts, byte rates, inter-arrival times, and spectral characteristics. The system implements an adaptive learning mechanism with continuous weight updates, preventing catastrophic forgetting while adapting to evolving attack patterns. Experimental validation on the cic-ids2017 dataset and custom-generated lddos traffic demonstrates that the proposed framework achieves 98.7% detection accuracy with 2.3% false positive rate and 12ms detection latency— significantly outperforming existing approaches including lstm (96.3%), cnn (95.1%), and svm (91.2%). The lnn-based detector identifies attacks within 2-3 burst cycles, enabling proactive mitigation before throughput degradation occurs. This work establishes the first integration of liquid neural networks with sdn controllers for cybersecurity applications, achieving a 76% improvement in detection latency compared to state-of-the-art lstm-based methods.
Zitieren
APA
Kumar, D. R., & Sudha, P. V. (2026). Real-Time Low-Rate Ddos Attack Detection In Sdn Environments Using Liquid Neural Networks With Ryu Controller. International Journal of Drug Delivery Technology, 16(28s). https://doi.org/10.25258/ijddt.16.28s.57
Vancouver
Kumar DR, Sudha PV. Real-Time Low-Rate Ddos Attack Detection In Sdn Environments Using Liquid Neural Networks With Ryu Controller. International Journal of Drug Delivery Technology. 2026;16(28s). doi:10.25258/ijddt.16.28s.57
BibTeX
@article{Kumar2026,
author = {Kumar, D. Rajani and Sudha, P. V.},
title = {Real-Time Low-Rate Ddos Attack Detection In Sdn Environments Using Liquid Neural Networks With Ryu Controller},
journal = {International Journal of Drug Delivery Technology},
year = {2026},
volume = {16},
number = {28s},
doi = {10.25258/ijddt.16.28s.57}
}
EUVIMED