Real-time traffic decoding and classification for mobile applications

OData support
Dr. Orosz Péter
Department of Telecommunications and Media Informatics

Nowadays media applications and within that real-time streaming applications are getting more and more popular. Providing these services raises a new challenge for the network operators since certain QoE expectations must be fullfilled for good user experience. Specifying these expectations, defining intervals and creating a system around them is a very complex project. Recently, numerous experiments were performed and systems were made to categorize Internet traffic. While these systems separated the traffic by categories with the expansion of media traffic within the global Internet traffic more precise solutions are necessary to develop. Future systems must be able to detect and indentify media traffic at application level, or even deeper: in use case level. Based on this detections they must constantly control the traffic by distributing the resources of the network among flows and flow groups.

Firstly, this document presents the process of apllication profiling through examples from a popular mobile application (Spotify), which creates the opportunity to make a regulation system based on the results. Then describes the planning phase of this regulation system and gives a detailed view about the realization of the protocol decoder part and the categorization its output into TCP streams, DNS and SSL information. The realization described in this document was made in Python and Java high-level programming languages.


Please sign in to download the files of this thesis.