Trusted Gossip for Peer-to-Peer Information Sharing

August 20, 2008 -- TRLabs' researchers are proposing a trusted gossip protocol for rumor resistant information sharing in peer-to-peer networks.

The filtering module of the trusted gossip at each nodeBenefits

Background

In the last few years, the Internet has witnessed a rapid increase in the popularity of peer-to-peer (P2P) file sharing applications. A primary reason being that these systems are able to avoid single points of failure, overload, and liability using features inherent to P2P networks. Very recently, information sharing systems like blogs, wiki, and social networks [1] have revolutionized the human connection to the Internet. Following these systems, the logical step towards the next "new thing" would be to merge them to build more far-reaching and scalable systems. The potential of the P2P model to alleviate censorship and bias for information dissemination is one such effort.

Ideally, a successful P2P information dissemination protocol should possess the following properties: scalable and efficient implementation, censorship immunity, fault tolerance, and trustworthiness. Recent attempts [2,3] to build such a system using epidemic gossip protocols [4,5] fall short of incorporating trustworthiness to do message dissemination. Trust issues in such systems have been substituted by forging offline relationships among the participating entities. Obviously, this solution does not scale with dynamic networks where nodes can join and leave.  Trusted gossip that supports dynamic networks was proposed in [6]. The core idea of this proposal is to inline a message filtering process into the dissemination protocol such that the spread of trusted messages is maximized while checking the spread of untrusted messages. As messages spread, each node makes an accept, reject, and forward decisions based on a credibility measure which is computed from the estimates of the origin node's trust and the particular message's trust.

While trusted gossip significantly improved upon the normal gossip by disseminating messages according to the preferences of the participating nodes, it incurred significant overhead in terms of number of messages and reviews. To address this concern, in [7], the researchers also propose a targeted messaging scheme for the trusted gossip protocol where message originators can pinpoint potential target nodes which are more likely to accept their messages. The idea is to cut-down redundant gossip messages while allowing trustworthy messages to spread through the network with greater accuracy and speed. Using real traces collected from social network sites, the researchers were able to establish the performance preeminence of the trusted gossip scheme for information dissemination in a P2P network.

Login to see more detailed information on this ETB

See more technology bulletins »