Measuring and modeling the BitTorrent content distribution system

Abstract

The paper reports on a detailed study of the BitTorrent content distribution system. We first present a measurement infrastructure designed to allow detailed, message-level capture and analysis of P2P traffic. An associated modeling methodology is presented as well. These tools have been used to measure and model the BitTorrent protocol, which is observed to exhibit exponential characteristics of session interarrival times. We also observe that session durations and sizes are modeled with a lognormal distribution.

Ego network models for Future Internet social networking environments

“…Nowadays, content-centric services are mostly supported by Content Delivery Networks and P2P solutions [54]. Several solutions continue to be proposed for optimising systems based on these paradigms, in particular in the P2P area both for fixed and mobile networks [55][56][57][58][59][60]. SPNs could be used as a natural substrate for implementing trusted P2P-like content-centric solutions tailored to, for example, disseminating data to interested users in a scalable way.…”

Experimental Analysis of Bittorrent Traffic based on Heavy-Tailed Probability Distributions

“…BitTorrent Protocol [9] is a peer-to-peer protocol, where work load and time consumed in accessing server for large files is reduced considerably. The content is shared in pieces simultaneously among the peers, thus enabling efficient distribution of network load.…”