Interest-based connections in peer-to-peer networks
AssigneeHewlett-Packard Development Company, L.P.
InventorsMurali Krishna Ramanathan, Vasiliki Kalogeraki, James Christopher Pruyne
Life of this patent
Estimating the term…
Loading patent…
AssigneeHewlett-Packard Development Company, L.P.
InventorsMurali Krishna Ramanathan, Vasiliki Kalogeraki, James Christopher Pruyne
Estimating the term…
Peers (p) monitor responses received from other peers (s, q) to requests for information and rate the responding peers on their ability to provide information of interest. When a responding peer (s) is discovered to frequently provide good results, the requesting peer (p) attempts to move closer to the responding peer (s) by creating a direct connection (p-s) with that peer and thereby promote the “good” peer to an “immediate” peer. If such a promotion would result in too many direct connections, the least important immediate peer is demoted to an “indirect” peer. The criteria (Imp) used for evaluating the relative importance of at least the immediate peers is preferably a time weighted average (Imp*(t)=α·Imp(t)+β·Imp*(t−1)) that also measures consistency and reliability and preferably includes factors not only representative of the peer's ability to provide requested information (Hits), but also of its proximity to the source of that information (1/Hops) so that it can prove that information efficiently. This leads to clusters of peers with similar interests, and in turn reduces the depth of searches typically required to achieve good results.
A method for reconfiguring a node (p) in a peer-to-peer network, comprising:
A method for dynamically measuring the current importance of a directly connected peer in a peer-to-peer network, comprising:
A dynamically reconfigurable peer-to-peer network comprising a plurality of physically connected addressable nodes, each said node comprising:
Dependencies are inferred from the claim text.
Drawings are temporarily unavailable. Open the PDF at USPTO
From the US filing history. Foreign family members aren’t listed.
Backward citations are indexed for grants only.
Implemented based on peer-to-peer networks, e.g. gnutella: management specially adapted to peer-to-peer storage networks
Computing and data processing
Term end not estimated
The present invention relates generally to networks having distributed storage, and more particularly to direct exchange of information between peers in a network.
Decentralized sharing of information and data through direct exchange may be used to distribute storage capacity and load across a network of peers. In fully decentralized peer-to-peer networks there is no need for a central coordinator; rather, communication may be handled individually by each peer.
In a peer-to-peer network, a node becomes a member of the network by establishing (possibly on a purely ad-hoc basis, or possibly in response to a message from a remote coordinator) a connection with at least one peer currently in the network. Each node typically maintains a small number of connections with its peers and messages are sent from one peer to another with each peer responding to queries for information that it may have stored locally. For example, to search for a file, a node broadcasts a search request to its directly connected peers, those directly connected peers propagate the requests to their own peers, and so on. Each such hop from one peer to the next contributes to an increase in the bandwidth on the communication links and furthermore to the time required to get results for the queries. The network bandwidth utilized for a search request is proportional to the number of messages, which in turn is dependent upon the number of peers that must process the request before the requested data has been located.
FIG. 1 depicts certain logical connections in an exemplary peer-to-peer network.
FIG. 2 depicts certain physical subsystems associated with an exemplary node of the peer-to-peer network of FIG. 1, including a File Repository, a Scores Repository, and connections to the peer-to-peer network.
FIG. 3 depicts the format of an exemplary peer-to-peer message that may be used by the node of FIG. 2.
A distributed decentralized online database (e.g., an encyclopedia) may be organized as a logical network of peers, each maintaining a set of files (articles). There is no centralized directory that holds (or points to) all the files, the files are each dynamically maintained at the users' machines. The files can be updated very frequently as in the case of news bulletins or can be more static as in the case of historical information. Also, the files are preferably not necessarily unique; duplicate copies of static files (i.e., popular articles) may be locally stored or temporarily cached by many users and even dynamically changing files may be cached and synchronized in known manner at many different locations. In order to facilitate dynamic reconfiguration of the peer-to-peer network under local control using locally available information, this decentralized logical network preferably operates under one or more of the following assumptions:
Each active peer has a sufficiently large number of active connections to its peers that the network remains highly connected.
Multiple paths normally exist between any two active peers.