A Review of Machine Learning Approaches for DDoS Attack Detection and Mitigation in Software-Defined Networking

Software-defined networking (SDN) has changed the way networks are designed and controlled by separating the control plane from the data plane. This separation gives administrators a more programmable, centralized, and flexible way to manage traffic. At the same time, the SDN controller becomes a critical point of exposure. Distributed denial-of-service (DDoS) attacks can overload the controller, switches, links, or servers and can prevent legitimate users from accessing network services. The reviewed paper examines DDoS threats in SDN by discussing attack categories, public datasets, detection methods, and mitigation frameworks. It emphasizes the importance of open datasets for reproducible experiments and highlights that many available datasets are either outdated, not SDN-specific, or limited in attack diversity. The study also shows that machine learning and deep learning models have become central tools for detection, although their real-world usefulness depends on dataset quality, controller overhead, scalability, and false-positive control.

Hali tarjima qilinmagan