p2p news / p2pnet: Dutch researchers have developed a network for a new generation of p2p file sharing through which, they say, faster down- and uploading and live video streaming will become a reality.

“Instead of the lone hacker, for the first time, it was written by a team of more than a dozen scientists,” Johan Pouwelse, one of the group, told p2pnet.

Called Tribler, it falls under the I-Share project of the Delft University of Technology’s Faculty of Electrical Engineering, Mathematics and Computer Science and will introduce, “the person behind the computer,” says Pavel Garbacki, who’s worked on I-Share’s Tribler component for the past three years.

The current favourite, BitTorrent, doesn’t save information and users remain anonymous, he says . Tribler however saves users’ historie and, “just like with MSN, with Tribler people can create groups of friends.

“So you’ll know who you download files from. These are people you can trust, and therefore you won’t get viruses, and they’re also people who have the same interests as you.”

Nor are these the only differences.

“With Tribler, the costs are much lower and downloading times twice as fast,” Garbacki promises.

The group is also working on adding tag-based navigation for Tribler, and reducing protocol overhead to enable HTML seeding.

“The past decade has shown that metadata are often unavailable and searching is best done using the text itself rather than relying on metadata,” says the development zone.

“In general, metadata run short in quality and quantity. Tags can address these deficiencies. Websites such as Flickr.com, Del.icio.us, and CiteUlike.org have shown the popularity of tags for search and attracted a combined user base of millions. The key to these websites is their use of volunteers to augment content with tags. Every visitor to such websites can participate in this collaborative categorization, thus creating a limited two-way web. Different websites may have different underlying reasons why people tag, but the altruistic effect is helping other users in exploring the content because the system performs better if more users participate.

“We developed software to generate tags from Wikipedia in cooperation with researchers from HIIT, Finland. The challenge is not only generating tags, but also organizing them into top-tags, sub-tags, subsub-tags, and adding weights. We implemente d the generic GenerateTopTags function to generate tags. This function can generate both top-tags, sub-tags, subsub-tags, and handle the AND operator. It increases the freedom users have in searching and exploring content, and this bootstrap should stimulate more users to generate more tags for their own content.”

Below is a brief outline of Tribler. Go here for the full paper. >>>>>>>>>>>>>>>>>>>>>>>

P2P-TV
Peer-to-Peer content sharing using social networking, content recommendation, and cooperative downloads
J.Pouwelse, P.Garbacki, J.Wang, A.Bakker, J.Yang, A.Iosup, D.Epema, M.Reinders, M. van Steen, H.Sips

Introduction - Most current Peer-to-Peer (P2P) file sharing systems treat their users as anonymous, unrelated entities, and completely disregard any social relationships between them. However, social phenomena such as friendship and the existence of communities of users with similar tastes may be well exploited in such systems, to increase their usability and performance. We present a novel social-based P2P file-sharing paradigm that exploits social phenomena by maintaining social networks and using these in content discovery, content recommendation, and content downloading.

P2P-TV system architecture - We have implemented a P2P system based on the ABC open-source BitTorrent client. By extending this popular client (78.000 downloads per week, Nov 2005) we aim to have a large user base in a relative short time, besides having a tested code base for our implementation.

P2P Social groups - The prime social phenomenon that we exploit is that ‘kinship fosters cooperation’.

• Simulate altruistic behavior
• Friends cooperate and share resources
• Distinguish friend, foe, and newcomer
• De-anonymize peers and facilitate social groups formation
• Use nicknames to de-anonymize
• Import from existing networks (MSN, GMail)

To the right, statistics of the friendster.com social network

Taste buddy-based-content discovery - Locating content is critical for P2P systems. Current solutions are based on one or a combination of: query flooding, distributed hash-tables, and semantic clustering. We provide a novel solution based on taste buddies.

Automatically generate profile from download behavior

• Use epidemic protocols to spread profiles
• Connect to peers with similar profiles, your taste buddies
• Taste buddies provide an efficient mechanism for content discovery
• New content can spread in seconds when peers keep thousands of taste buddies

P2P Cooperative downloading - Members of a social group who trust each other cooperate to improve their download performance.

• Friends cooperate in download
• Peers donate idle bandwidth to their friends
• Significant download performance improvement
• Peers with asymmetric links (eg, ADSL) download up to 6 times faster

Also See:
Johan Pouwelse - BitTorrent: chapter and verse, December 14, 2004

This entry was posted on Tuesday, February 21st, 2006 at 11:36 am and is filed under Uncategorized. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

HOME

Viewed 3673 times