Automatic Discovery of Network Applications: A hybrid Approach

Abstract

Automatic discovery of network applications is a very challenging task which has received a lot of attentions due to its importance in many areas such as network security, QoS provisioning, and network management. In this paper, we propose an online hybrid mechanism for the classification of network flows, in which we employ a signature-based classifier in the first level, and then using the weighted unigram model we improve the performance of the system by labeling the unknown portion. Our evaluation on two real networks shows between 5% and 9% performance improvement applying the genetic algorithm based scheme to find the appropriate weights for the unigram model.