Internet traffic clustering using packet header information
Pekka S. Kumpulainen, Kimmo Hätönen, Olli Knuuti, Teemu Alapaholuoma · Common Library Network (Der Gemeinsame Bibliotheksverbund) · 2011
IP networks transfer huge amounts of data and information all over the word.The operator maintains the network and monitors it continuously to detect and eliminate any disturbances as soon as possible.Features and parameters included in the IP packet headers enable the operator to identify lots of information about the IP traffic and network users.The goal of this study is to analyze the measurements about the IP packet header information to support the operators in network management and optimization, trouble shooting, service creation and marketing.In this paper we propose multi-layer clustering that can reveal detailed description of traffic patterns and behaviour profiles of the network addresses.These descriptions can be enhanced by identifying abnormal traffic and analysing the reasons for such behaviour.Utilizing the measurement information included in the IP packet headers enables all these analysis tasks without violating privacy of the consumers.