Evaluating Machine Learning and Deep Learning Techniques for Effective DDoS Detection in Software-Defined Networks

EasyChair Preprint 15494

Abstract

Software-Defined Networking (SDN) enhances network management and adaptability by decoupling control and data planes. However, its centralized architecture makes it vulnerable to Distributed Denial-of-Service (DDoS) attacks. Machine Learning (ML) and Deep Learning (DL) algorithms have emerged as promising solutions for anomaly detection in SDN environments. This paper systematically compares ML and DL approaches for detecting DDoS attacks in SDN. We evaluate various architectures, datasets, and evaluation metrics to understand their strengths and limitations. Experimental results show that DL models outperform traditional ML approaches in terms of accuracy and scalability, but at the cost of higher computational requirements.

Keyphrases: Software Define Network, deep learning, machine learning, model, network

@booklet{EasyChair:15494,
  author    = {Li Wang and Maria Kin and James Rajez and Elvard John and Samul Tick and Mehmmet Amin},
  title     = {Evaluating Machine Learning and Deep Learning Techniques for Effective DDoS Detection in Software-Defined Networks},
  howpublished = {EasyChair Preprint 15494},
  year      = {EasyChair, 2024}}