Deep Learning-Based Approach for Detecting DDoS Attack on Software-Defined Networking Controller

The rapid growth of cloud computing has led to the development of the Software-Defined Network (SDN), which is a network strategy that offers dynamic management and improved performance. However, security threats are a growing concern, particularly with the SDN controller becoming an attractive targ...

Full description

Saved in:
Bibliographic Details
Published in:Systems (Basel) 2023-06, Vol.11 (6), p.296
Main Authors: Mansoor, Amran, Anbar, Mohammed, Bahashwan, Abdullah, Alabsi, Basim, Rihan, Shaza
Format: Article
Language:English
Subjects:
Accuracy
Artificial intelligence
Cloud computing
Communication
Controllers
Cybersecurity
Datasets
Deep learning
Denial of service attacks
Distributed Denial of Service
Infrastructure
Machine learning
Methods
Neural networks
Recurrent Neural Networks
Software
Software-Defined Networking
Virtual private networks
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The rapid growth of cloud computing has led to the development of the Software-Defined Network (SDN), which is a network strategy that offers dynamic management and improved performance. However, security threats are a growing concern, particularly with the SDN controller becoming an attractive target for malicious actors and potential Distributed Denial of Service (DDoS) attacks. Many researchers have proposed different approaches to detecting DDoS attacks. However, those approaches suffer from high false positives, leading to low accuracy, and the main reason behind this is the use of non-qualified features and non-realistic datasets. Therefore, the deep learning (DL) algorithmic technique can be utilized to detect DDoS attacks on SDN controllers. Moreover, the proposed approach involves three stages, (1) data preprocessing, (2) cross-feature selection, which aims to identify important features for DDoS detection, and (3) detection using the Recurrent Neural Networks (RNNs) model. A benchmark dataset is employed to evaluate the proposed approach via standard evaluation metrics, including false positive rate and detection accuracy. The findings indicate that the recommended approach effectively detects DDoS attacks with average detection accuracy, average precision, average FPR, and average F1-measure of 94.186%, 92.146%, 8.114%, and 94.276%, respectively.
ISSN:2079-8954
2079-8954
DOI:10.3390/systems11060296