SNIC
SUPR
SNIC SUPR
Deep Learning for Network Flow Classification
Dnr:

SNIC 2018/5-21

Type:

SNAC Small

Principal Investigator:

Johan Garcia

Affiliation:

Karlstads universitet

Start Date:

2018-02-21

End Date:

2019-03-01

Primary Classification:

20204: Telecommunications

Webpage:

Allocation

Abstract

Classification of flows on the Internet in general, and in Cellular networks in particular, are useful for performing traffic engineering to achieve high utilization of the network resources while ensuring high Quality of Experience for the users. By applying appropriate queuing of different types of traffic such as VoIP, Video streaming and Web browsing the delay sensitive traffic can be separated from non-delay sensitive traffic. Traffic classification is currently done through Deep-Packet Inspection, but this will not be sustainable as more and more of the traffic is becoming encrypted. In this project we will extend our previous work on encrypted flow classification to also examine deep learning techniques for flow classification.