SPICE: Scalable P2P implicit group messaging
Computer Communications
Querying communities of interest in peer database networks
DBISP2P'05/06 Proceedings of the 2005/2006 international conference on Databases, information systems, and peer-to-peer computing
| Hi-index | 0.00 |
Peer-to-peer systems have the ability to harness vast amounts of resources from a scalable collection of autonomous peers and especially emphasize on de-centralization and lack of a central authority. As a result these systems are particularly attractive to everyday home computer users, who seem empowered by the potential to independently select and change their own policies, roles, and responsibilities. By allowing peers to share a portion of the authority, these systems also possess other interesting technical characteristics such as self-organization and adaptation. This dissertation introduces the notion of interest-based communities of peers and describes how these implicit structures can be used to naturally organize peer-to-peer systems for discriminative information dissemination, pruning the search space, and role-based trust. Communities are like interest groups, modeled after human communities. They are self-organizing, possibly overlapping structures involving peers that are actively engaged in the sharing, communication and promotion of common interests. Initially, the behavior of randomly created communities is investigated and modeled. Then the community formation and discovery algorithms are presented and their complexity studied. The experiments illustrate that these algorithms do not require extensive computation or communication on the part of individual peers. Subsequently, a novel push-pull gossiping technique is provided that improves decentralized information dissemination by communicating only amongst peers within a specified community of interest. The technique enables the maintenance of dynamically changing communities and consists of a distributed discovery algorithm followed by repeatable push-pull gossiping. Experiments show that pushing gossip information to only a small number of peers allows a large percentage of peer members of a community to obtain (pull) the information within just two hops. Peer communities help in pruning the search space and in content-based searches within the peer-to-peer network. This dissertation describes a community-based search query propagation scheme which provides more efficient searching by targeting one or more communities, irrespective of the current membership of the searching peer. The experiments demonstrate how a community-based search can reduce the number of messages and improve the quality of the search results when compared to other known peer-to-peer search algorithms. Finally, an approach for community-based trust management is discussed. An optimistic role-based model for trust amongst peers is employed and it is shown to be scalable, dynamic, revocable, secure and transitive. The solution permits asymmetric trust relationships that can be verified by any peer in the system through a simple, low-cost algorithm.