Characterization of TVAnts: Scaling Analysis of a P2P IPTV Traffic

P2P live streaming applications like P2P IPTV are emerging on the Internet and will be massively used in the future. There is only a few studies about these kinds of applications. Hei et. al [1] gave some insights about PPLive, a very popular application on the Internet, while our previous works [2] compare the most popular P2P IPTV applications trying to highlight their similarities and differences. However, comprehensive characterization of P2P IPTV network traffic is still missing. It is therefore important to study P2P IPTV traffic and to characterize its properties. The characterization of P2P IPTV traffic will allow to understand its impact on the network. P2P IPTV applications have stringent QoS constraints (bandwidth, delay, jitter) and their traffic characterization will make possible to understand their exact needs in network resources. The knowledge of the traffic properties enable the development of synthetic traffic generation models that are key input parameters when modeling or simulating these systems. From a traffic engineering point of view, well understanding P2P IPTV traffic is essential for Internet service providers to forecast their internal traffic and enssure a good provisioning of their network. And last but not least, global knowledge of the traffic properties will highlight some drawbacks of the application and will make it possible to improve the design of new P2P IPTV architectures. During the FIFA World Cup 2006, we measured network traffic generated by several P2P IPTV applications. In this paper, we present a complete multiscale analysis of the structure of the traffic generated by a single application TVAnts [3]. In the rest of this extended abstract, we will present the P2P IPTV traffic measurement acquisition method and architecture and the TVAnts application (§ II). We will introduce our analysis methodology (§ III) and the multiscale analysis tool, LDestimate [4] (§ IV). Then, we will show our characterization of TVAnts traffic (§ V). Finally, we conclude the paper and give perspectives (§ VI).