Traffic Qualification in SD-WMN using Ensemble based Machine Learning Approach

Résumé

Abstract The traffic classification problem formulation is NP-hard and has known several resolution approaches where the emerging one is the machine learning approach. However, these approaches have primarily focused on traditional wired and wireless networks and rarely on Software-Defined Wireless Mesh Networks (SD-WMNs). A Software-Defined Network (SDN) makes network monitoring easier by separating the control plane of the network from the data plane. This paper discusses the limits of traffic classification in the network and proposes an approach based on supervised ensemble machine learning adapted to SD-WMN to classifier traffic efficiently in three stages: (a) a traffic-monitoring phase, (b) an IP flow collection phase and, (c) a traffic classification phase by the ensemble supervised machine learning. Ensemble methods are techniques that aim at improving the accuracy of results in models by combining multiple models instead of using a single model. The combined models significantly increase the accuracy of results. We performed experiments on Mininet-wifi emulation platform as data plane with Ryu as SDN controller in control plane. The supervised ensemble learning yields: (a) for the Bagging algorithm with the Random Forest algorithm, an accuracy of 99.90%, with an F1 score of 99.90% and, (b) for Boosting with the XGBoost algorithm, an accuracy of 99.97% with an F1 score of 99.96%. XGBoost appears as the best traffic classification model.

Citer ce document

Accès au document

Ce document est hébergé sur son dépôt institutionnel d'origine.

Auteur(s)

Statistiques

Consultations : 1

Téléchargements : 0