A Measurement Study Of The Emule Network: An Active Crawling Approach

In the past decades, P2P file-sharing has become a major approach to share contentsamong millions of users on the Internet. Previous measurement studies have shown thatusers exchange contents ranging from the legal exchange of open source Linuxdistributions to the illegal exchange of copyrighted songs, movies, TV shows, software,and books. Internet pornography and terrorist materials are also commonly available inP2P file-sharing networks. It is crucial to study the monitoring strategies and measurementmethods of P2P networks to achieve a safe Internet. In this thesis, we conducted acomprehensive measurement study of a large-scale P2P file sharing network, eMule.To design and implement the eMule measurement system, we first examined thedetails of the eMule protocol, analyze its signaling packet formats and the interactionprocess between eMule peers. Then, we design and implement a flexible eMule crawlingsystem, which can deploy various crawling strategies and interactive measurementprotocols according to particular crawling demands. In our measurement system, thecrawler behaves like a normal peer, joins the eMule network and then communicates withother ordinary peers. Through comprehensive measurement experiments, thismeasurement platform has demonstrated to crawl the data in a stable and efficient manner.Utilizing our measurement platform, we conducted a systematic measurement studyto harvest three-tier information in the eMule network: peer-level information (peerdynamics), file-level information (file description and distribution) and content-levelinformation. We further analyzed the harvested data set, and revealed the networktopology, peer behaviors, peer distribution and content distribution in eMule. Ourmeasurement results depict the most recent snapshots of the current eMule network. Ourmeasurement practice may bring forth insights into achieving a measurable, controllableand manageable Internet.