The paper contributes to the development of a framework concerning the teletraffic modeling and analysis
of peer-to-peer systems that are based on a mesh-pull architecture. We look at the packet transmission
processes induced by a swarm of peers disseminating multimedia objects. We provide a survey on some
statistical techniques that are used to characterize the observed hierarchical popularity structure of a peer
population. Using collected flow data at a single peer, we show how Generalized Pareto and exponential
models can be applied to classify the behavior of the feeding peers. Our approach is illustrated by real
packet data of P2P sessions generated by BitTorrent clients in a mobile WiMAX testbed in Korea. The
stated techniques can be used to cope with an efficiency adaptation of P2P dissemination protocols to
mobile environments.
