On the Validation of Traffic Classification Algorithms
Detailed knowledge of the traffic mixture is essential for network operators and administrators, as it is a key input for numerous network management activities. Traffic classification aims at identifying the traffic mixture in the network. Several differ
- PDF / 354,772 Bytes
- 10 Pages / 430 x 660 pts Page_size
- 8 Downloads / 197 Views
Detailed knowledge of the traffic mixture is essential for network operators and administrators, as it is a key input for numerous network management activities. Traffic classification aims at identifying the traffic mixture in the network. Several different classification approaches can be found in the literature. However, the validation of these methods is weak and ad hoc, because neither a reliable and widely accepted validation technique nor reference packet traces with well-defined content are available. In this paper, a novel validation method is proposed for characterizing the accuracy and completeness of traffic classification algorithms. The main advantages of the new method are that it is based on realistic traffic mixtures, and it enables a highly automated and reliable validation of traffic classification. As a proof-of-concept, it is examined how a state-of-the-art traffic classification method performs for the most common application types.
1
Introduction
The aim of traffic classification is to find out what type of applications are run by the end users, and what is the share of the traffic generated by the different applications in the total traffic mix. Research for better and better traffic classification methods is blooming with the constant increase of network capacity, the emerging application types, and common usage of traffic deceiving techniques. However, the objective comparison of these methods has not been possible yet due to several reasons. Firstly, there are no perfectly classified traffic traces available. Moreover, the validation is typically done with another specific classification method. This situation results in such anarchy that papers can state nearly anything about their introduced method as there is no chance to check it by others or verify with a commonly known and accepted reference test. In this paper we provide a validation method, which can reliably test the accuracy of traffic classification algorithms. In practice, the objective is typically to identify applications in passively observed traffic. We believe that such a classification method can be convincingly validated only by an active test, for which a number of requirements are fulfilled, such as: – It should be independent from classification methods, i.e. the validation of a classification method by another one must be avoided, M. Claypool and S. Uhlig (Eds.): PAM 2008, LNCS 4979, pp. 72–81, 2008. c Springer-Verlag Berlin Heidelberg 2008
On the Validation of Traffic Classification Algorithms
73
– About each packet the test should provide reference information that can be compared to the result of the classification method under study, – The test should be deterministic, meaning that it should not rely on any probabilistic decisions, – Feasibility: it should be possible to create large tests in a highly automated way, and – The environment where the active measurements are collected should be realistic. The paper is organized as follows: in Section 2 an overview of existing traffic classification methods is provided together with a discussion of the techniques
Data Loading...
