Abstract:
Recently, Software-Defined Networking (SDN) architecture has offered great benefits due to the separation between the control and network elements such as routers and swi...Show MoreMetadata
Abstract:
Recently, Software-Defined Networking (SDN) architecture has offered great benefits due to the separation between the control and network elements such as routers and switches. Unfortunately, the enormous growth of attacks hinders the wide adoption of SDN. Intrusion Detection Systems (IDSs) are used as significant tools to detect and mitigate network attacks of anomalous nature. Several Deep Learning (DL) models, such as Convolutional Neural Networks (CNNs), have been utilized for building IDSs in cyber security in recent years, because they achieve the desired detection results. Nonetheless, little work has been introduced on IDSs in SDN systems. Hence, this paper presents an efficient framework based on intelligent techniques for predicting attacks in SDN systems. This work focuses on the early detection of abnormal attacks existing in an SDN environment. When malicious traffic is identified in an SDN topology, the Artificial Intelligence (AI) module employs Machine Learning (ML) or DL models to identify and stop the attack source. The architecture presented in this research allows for comparing several ML and DL classification techniques that can be used to identify different sorts of network attacks. The proposed framework is tested on the InSDN dataset using several learning models like Random Forest (RF), AdaBoost (AB), K-Nearest Neighbor (KNN), Naïve Bayes (NB), Decision Tree (DT), and Logistic Regression (LR) classifiers. In addition, DL models such as Deep CNN and Long Short Term Memory (LSTM) are considered. The results demonstrate that the proposed Deep CNN model for multi-class attack data classification achieves the highest accuracy of 99.85% compared to RF, AB, KNN, NB, DT, LR, and LSTM classifiers with accuracy levels of 95%, 88 %, 97%, 93 %, 90 %, 98%, and 88.31 %, respectively.
Date of Conference: 07-08 October 2023
Date Added to IEEE Xplore: 21 November 2023
ISBN Information: