<![CDATA[Software-Defined Networking (SDN) has emerged as a foundational paradigm for programmable, centrally-managed networks, but its logically centralised control plane is highly attractive to Distributed Denial of Service (DDoS) adversaries. Traditional signature- and threshold-based defences struggle against polymorphic and low-rate attack patterns, motivating a rapid migration toward Deep Learning (DL) based detection. This Systematic Literature Review (SLR), conducted in accordance with the PRISMA 2020 guideline and a PICOC framework, identifies, classifies, and analyses 62 primary studies published between January 2020 and February 2026 on DL-based DDoS detection in SDN. Three research questions are answered, covering publication venues, the most active researchers, and the architectures, datasets, and evaluation metrics employed. The findings reveal that Convolutional Neural Networks (38.7%), hybrid CNN-LSTM models (24.2%), and Transformer/Graph Neural Networks (14.5%) dominate recent designs, while the InSDN and CIC-DDoS2019 datasets are the de-facto benchmarks. Macro-averaged accuracy across high-quality studies exceeds 99%, yet real-time deployment, explainability, and cross-dataset generalisability remain open challenges. The review provides a consolidated knowledge map and an empirically grounded research agenda for the next generation of intelligent SDN defences]]>
Copyrights © 2026
