Skype traffic detection and characterization

No Thumbnail Available
Journal Title
Journal ISSN
Volume Title
Master's thesis
Date
2007
Major/Subject
Networking Technology
Tietoverkkotekniikka
Mcode
S-38
Degree programme
Tietoliikennetekniikan tutkinto-ohjelma
Language
en
Pages
vii, 72
Series
Abstract
Skype is a very popular VoIP software which has recently attracted the attention of the research community and network operators; furthermore Skype uses a proprietary signalling design and its source code is unavailable. This makes its analysis really important since the classification of IP flows becomes increasingly crucial in modern network management platforms. Traditional classification systems based on packet headers are rapidly becoming ineffective. In this work after a general analysis of Skype protocol and traffic in both time and frequency domain, a new classification method is presented. It is based on statical classification of the flow, using only three basic properties of IP packets: their size, interarrival time and order of arrival. The whole process is based on a new quantity called Protocol Fingerprint. Its aim is to express these quantities in an efficient way. An important part in the classification process is taken by a Gaussian filter that smooths the protocol fingerprints avoiding misclassifications caused by any kind of noise generated in the network. Even if this technique is at an early stage of development and requires more work, it is quite promising.
Description
Supervisor
Virtamo, Jorma; Prof.
Thesis advisor
;
Keywords
Skype, traffic, statistical classification, fingerprint
Other note
Citation
Permanent link to this item
https://urn.fi/urn:nbn:fi:tkk-009990