A Naive Bayes Multi-class Weighted Classifier of Internet packet flows over a MPLS network

Abstract

Our simulation is based first, on a qualitative approach for the classification of flows and traffic, and next on an experimental approach for the management of data volume on the other hand. The adopted approaches allowed us to get an idea on a NBWM (Naive Bayes Weighted Multi-class) classifier capable to output differentiated service classes in MPLS (Multiple Protocol Label Switching) networks. The classifiers we compared to our benchmark model were thoroughly processed. The accuracy rate of the proposed NBWM (Naïve Bayes Weighted Multiclass) classifier is about 68.75%, which puts it ahead of the other models encountered.