【摘要】：With the rapid developing and popularizing of the Internet, The technologies of P2P (Peer to Peer) and CDN (Content Distribution Network) are widely applied as the tools of the network content (file) sharing. Due to the complementation differentiation between P2P and CDN when they provide content service, the technology integrating P2P with CDN is becoming a hotspot research field.
As the scale of P2P CDN is growing larger continuously, more problems related to the file sharing among the peers and to the file providing by surrogate are coming. As the requirement to the equilibrium of the network resource utilization and the limitation of time expending for the response of file request by the peer, a peer can only achieve files from the peers the number of whose connective relays with the peer are small. As the number of peers a surrogate should provide services to and the number of files a surrogate should provide with are become bigger and bigger, files should be transfer between the surrogates and the central server. Consequently, not only the time expending of the response to the file request will increase, but also the expenditure of the hardware and network resource will be enhanced.
By the analysis of the social attributes of users, the relationship between the social attributes of users and the file requirement of users are researched and will be applied to the technologies of P2P CDN in this paper. The research points of this paper are listed as follows:
(1) In social network, the things a person needing and possessing are accorded with the social attributes of him/her. Accordingly, in P2P network, the user of a peer is a person, and the files for a peer needing and possessing are accorded with the social attributes of the peer. The information for the file searching request and the file searching response which has the information of file requirement or file possession of the peer can be used to analyze the social attributes of the peer. By the six degrees of separation theory, if a person wants to find a stranger, the person can search the stranger with help of others one by one, and he/she can find the stranger within six persons averagely. The precondition of the six degree of separation is that every person is familiar with a certain number of other persons, and he/she can select proper person to help him/her to find the stranger. In this paper, the six degree of separation is applied to searching peers in P2P network. By file searching request and response among the peers, every peer can achieve a certain amount of information about a certain number of other peers. By P2P model based on the six degree of separation, a peer can find the file it requires by small number of searching relays.
(2) The intelligent recommendation is a technology for the prediction of user requirement which can predict the probable requirements of a user by the past requirements of the user and the similarity of the user's social attributes. In P2P CDN, it is significant to research the similarity and the prediction of the peer's file requirement. In this paper, by the analysis of the peers'file requirement and the requiring degree to files that are shown with the activities of the peers in P2P CDN, the intelligent recommendation methods based on the file classification and the overlap demand theory are proposed, which can be used to the similarity and prediction of the peer's file requirement.
(3) In P2P CDN with a large scale number of peers, to any peer, the peers which can have files a certain peer requiring are the peers that have similar file requirement with the peer, and the ratio of the peers in all peers is relatively small. In the paper, the P2P CDN model based on the intelligent recommendation are proposed by the intelligent recommendation technologies that are suitable for achieving the similarity and prediction of the peer's file requirement. In the P2P CDN model, the peers are placed with smaller number of connective relays if they are more similar in file requirement. As a file of a peer requiring is more probable to be found in the peers that have more similar file requirement with the peer and is less probable to found in other peers, more replicas can be found in this proposed organization when a file request is sent, and the efficiency of the file search and file distribution is enhanced. The prediction technology of the peer's file requirement can be applied to the file placement method in the surrogate. As the predictive file requirement of a peer is probable the file requirement of the peer in reality, the requested files of a peer can be place in the surrogate beforehand when the surrogate has spare capacity. Owe to the scale of the files needing by peers with similar file requirement is relatively smaller, it is more probable that the surrogate has storage capacity to place the files for the peers. Consequently, the P2P CDN model can not only reduce the response and distribution time expending of the file request, but also implement the equilibrium utilization of the surrogates.
(4) In the society, the acquaintance between persons needs a time procedure. Similarly, the information achieving about the file requirement of peers will require a time procedure in P2P CDN too. Along with the number of the activities of the peer increases, the acquaintance to the peer can be more integrative and precise. On the other hand, the user attributes are changing continuously. Consequently, the P2P CDN model based on the similarity and prediction of the peers'file requirement should be conformed to the variation. The evolution strategies are proposed in the paper which can employ distinct methods according to the variation of the requirement of the peers.