TIE: A community-oriented traffic classification platform

Abstract

The research on network traffic classification has recently become very active. The research community, moved by increasing difficulties in the automated identification of network traffic, started to investigate classification approaches alternative to port-based and payload-based techniques. Despite the large quantity of works published in the past few years on this topic, very few implementations targeting alternative approaches have been made available to the community. Moreover, most approaches proposed in literature suffer of problems related to the ability of evaluating and comparing them. In this paper we present a novel community-oriented software for traffic classification called TIE, which aims at becoming a common tool for the fair evaluation and comparison of different techniques and at fostering the sharing of common implementations and data. Moreover, TIE supports the combination of more classification plugins in order to build multi-classifier systems, and its architecture is designed to allow online traffic classification. © Springer-Verlag Berlin Heidelberg 2009.