CN102571839A - P2P content delivery method based on social attribute of users and system adopting same - Google Patents

Abstract

The invention discloses P2P (peer-to-peer) content delivery network based on the social attribute of users. The P2P content delivery network comprises a central server, an edge server and a P2P sub-network, wherein the edge server and the central server are mutually connected through a special network or a high-speed network; the edge server and the P2P sub-network are mutually connected through the Internet; the P2P sub-network consists of a plurality of user terminals; the central server is used for acquiring the interaction between all users in the P2P sub-network and the content delivery network, tracking the users and computing the document demand and the demand degree of the users; and the demand similarity and demand forecast among the users are computed according to the document demand and the demand degree of the users, and demand evolution treatment is carried out according to the change of the social attribute and the unceasingly acquired user demand. The invention also discloses a P2P content delivery method based on the social attribute of the users. By adopting the invention, the content delivery efficiency is improved.

Description

P2P content distribution method and system based on user's social property

$\mathrm{dpredicton}\left(c\right)=\underset{p_i\∈\mathrm{sub}\left(P\right)}{\mathrm{\Σ}}\frac{\underset{p_j\∈P\left(c\right)}{\mathrm{\Σ}}{r}_{j,c}\×\mathrm{sim}{(p_i,p_j)}_{t{z}_h}}{\underset{p_j\∈P\left(c\right)}{\mathrm{\Σ}}\mathrm{sim}{(p_i,p_j)}_{{\mathrm{tz}}_h}}$

Wherein, sub (P) representes the set of the node that certain sub-net comprises, and dprediction (c) is the desirability of node to content c, and P (c) is all node set that content c had demand.

10, method according to claim 7 is characterized in that, said method also comprises:

When said user node application adds network, be user node allocations of edge server according to the similitude result of calculation after present node and the adding of each Edge Server by said central server; The global similarity property result of calculation that after user node intend to add certain Edge Server, causes this autonomy territory does not allow this user node to add during less than setting threshold, and said central server is that said user node is redistributed Edge Server; When said user node can not add any Edge Server, then integral body was turned down global similarity property threshold value, confirmed the Edge Server that said user node can add again again, and wherein, the unsteady of global similarity property threshold value do not exceed between the setting district;

Said central server periodically sends probe message to said Edge Server, and whether detect said Edge Server online; Said Edge Server is user node transmission probe message in the P2P subnet periodically, and whether detect said user node online;

Said Edge Server is after user node or other Edge Servers send the file polling request in the P2P subnet, and the affirmation message of returning in the stipulated time is efficient message, otherwise the inquiry failure.

P2P content distribution method and system based on user's social property

Technical field

The present invention relates to content distribution techniques, relate in particular to a kind of P2P (Peer toPeer) content distribution method and system based on user's social property.

Background technology

Content distributing network (CDN, Content Delivery Network) is also referred to as the content delivery network.Nowadays there have been a lot of enterprises and service provider to bring into use content distribution network CDN technology to improve network service quality at home and abroad, to satisfy user's demand.CDN is a kind of value-added network that is based upon on the existing IP network foundation structure, is one deck network architecture of disposing in application layer.The core of CDN is to utilize intelligent strategy technology; The content and the service at center are pushed to network edge; Make the user to obtain service with service place the best in quality nearest, its key technology comprises high-speed cache, load balancing, content route, content distribution, content stores and Content Management etc.The technological merit of CDN comprises: (1) global load-balancing, and access speed is fast, lets the user be connected to nearest server and gets on; (2) seamless integrated, highly reliable, availability and autgmentability; The content of (3) sending is protected, and content supplier can oneself determine the content of service local.

Based on above-mentioned advantage; The CDN technology can improve the utilance of broadband resource; Increase Streaming Media application popularization on the internet, these application comprise online playing, Music on Demand, live telecast, interactive game etc., have promoted the development of Internet application and service greatly.

And peer-to-peer network (P2P, Peer to Peer) technology be a kind ofly be used between different personal computers (PC, the Personal Computer) user, without the technology of direct swap data of trunking or service.P2P has broken traditional Client/Server pattern, and in peer-to-peer network, the status of each node all is identical, possesses the client and server double grading, can be simultaneously as service user and ISP.The core of P2P is to utilize user resources, carries out file transfer through Peer, and this is different from traditional client server computation model fully.The design of P2P through " decentralization " is machine-processed with multicast communication, has realized not reliance server and swap file apace.

At present, CDN is used to different fields with P2P, and CDN is mainly used in reliable Web and streaming medium content distribution, and P2P then is mainly used in content exchange.Obviously, the pluses and minuses of CDN and P2P technology have certain complementarity on using.If can combine reliability, the manageability of extended capability and the CDN of P2P, just can make up manageable, as can to carry a carrier class content application IPTV system platform.From amalgamation mode; CDN and P2P merge has two kinds of forms, and a kind of is that the Cache equipment of CDN is organized with the mode of P2P, utilizes directory service and the multicast communication ability of P2P; Realize the content exchange between the CDN Cache equipment, promote the content distribution capability of CDN; A kind of in addition is that administrative mechanism and the service ability of CDN are introduced the P2P network; Formation is reliable content core with CDN, is the framework of service edge with P2P, through this framework; Can when not increasing the CDN cost, effectively promote the CDN service ability; Many drawbacks of more effectively having avoided P2P to use, under this framework, the user need bring in the service of obtaining through the client of P2P.

The purpose of setting up the CDN network is to carry out content distribution, reduces the load of original contents server, reduces consumption of broadband, can keep client response time faster.Yet when the quantity of client sharply increased, bottleneck had just appearred in the performance of the Edge Server of CDN, and one of solution is exactly the quantity that increases the CDN Edge Server, but this can increase the construction cost of whole C DN network.The P2P-CDN system then reduces the load of Edge Server, can better network service be provided with lower input.

Yet P2P has also brought the problem that itself exists, and mainly concentrates on: like how lower offered load with than the short time delay of searching, improve the content search success rate; Under the CDN framework, how to accelerate the speed of content distribution; How to solve copy of content and distribute more after a little while, it is low that it is searched the probability of success, distributes more for a long time, is easy to generate the problem of unnecessary offered load again; How to improve stability of content distribution or the like.This need have comparatively accurately global information judges.

Fig. 1 is the structural representation of the system of P2P and CDN combination; As shown in Figure 1; Comprise central server, Edge Server in the system, wherein, central server and Edge Server are interconnected to form content distributing network; Through being interconnected to form a plurality of P2P subnets, each subnet links to each other with an Edge Server through a plurality of nodes the P2P node each other.Get up through high speed or dedicated Internet access between central server and the Edge Server, to realize content delivery at a high speed, simultaneously, they also are connected to client through Internet content service are provided.Client is mutual through Internet and central server, Edge Server and other clients.

Though network configuration shown in Figure 1 has realized the fusion of P2P and CDN; But because based on user's adding of P2P pattern and the randomness that withdraws from; Can cause the extremely unbalanced of Edge Server access customer number amount; Some Edge Server is owing to the user too much causes network congestion, and some Edge Server has a large amount of bandwidth idle.Moreover; When the user carries out the content download; Also exist because of distribution of content is uneven and cause the lower situation of speed of download, for example, when most of user's interest contents only are distributed in certain several Edge Server; Will certainly cause to certain several Edge Server, downloading associated documents between the Edge Server; Thereby cause the download efficiency between the Edge Server to reduce, be distributed to the user who is connected with self again, obviously can cause the reduction of user's download speed and download content from other Edge Servers.In addition; In the UNE of existing P 2P and CDN; Because Network Management Function concentrates on the central server basically; Though can realize the optimization to the Edge Server holistic management, Content Management and network management load that this has increased central server undoubtedly cause the content downloading efficiency low.

Summary of the invention

In view of this; Main purpose of the present invention is to provide a kind of P2P content distribution method and system based on user's social property; Can realize fastly the distribution of content, and, promoted content and distributed efficient through being published to Edge Server in will issue in advance and user node being divided into groups etc.

For achieving the above object, technical scheme of the present invention is achieved in that

A kind of P2P content distributing network based on user's social property; Comprise central server, Edge Server and with peer-to-peer network P2P subnet; Wherein, Said Edge Server is connected with dedicated network or express network with said central server, and said Edge Server and P2P subnet are through Internet connection, and said P2P subnet is made up of a plurality of user terminals; Wherein, said central server is used for obtaining the interbehavior of each user of P2P subnet and this content distributing network, and the user is followed the tracks of, and according to user's interbehavior and trace information, calculates user's document requirement and desirability; Document requirement and desirability according to the user further carry out the calculating of demand similitude, requirement forecasting calculating between the user, and carry out demand evolution processing according to the variation of user's social property and the user's request information of constantly obtaining; Wherein, user's social property comprises user's identity, occupation, preference information; Said central server also is used to store user's identity, occupation, the log-on message of hobby; And, confirm user's hobby, user identity and variation thereof, user's occupation and variation thereof according to the demand and the desirability of user to file, demand and the close user of needed length are made as neighbor node mutually; When confirming that demand that file is downloaded exceeds setting threshold, corresponding document is recommended to the big Edge Server of demand or recommends to the Edge Server that all do not store corresponding document;

Said Edge Server be used for being responsible for user node adding, leave the processing of P2P subnet; And after receiving user's file download request; To with P2P subnet that this Edge Server is connected in the user send the file polling request; After obtaining the response with inquiry file, the user who downloads for demand file provides the file distributing service; And, on this Edge Server the demand file of no institute and with P2P subnet that this Edge Server is connected in all users when also not having institute's demand file, send the file download request to other Edge Servers, from other Edge Server file in download and storage; When receiving the file download request of other Edge Servers, the Edge Server of downloading for demand file provides the file distributing service; Be used to be responsible for safeguarding this Edge Server listed files, add new file and deleted file; The listed files that responsible selection prestores, and select suitable opportunity, obtain the file that need prestore from other Edge Servers; And, obtain the evolution information of user from said central server to document requirement, self storage file is adjusted;

Said user terminal is used to provide the interactive environment with the user, the user through interactive environment carry out that fileinfo is browsed, file search, file are uploaded, file is downloaded and the document score operation; And provide and obtain user behavior and the context-aware environment of following the tracks of user behavior; Be used for collecting and the Maintenance Point data, said node data comprises that file hits effect matrix data and local similitude data; And, when adding the P2P subnet, safeguard self neighbor node according to network configuration as user node; And when neighbor node leaves the P2P subnet, upgrade neighbor node automatically and tabulate.

Preferably, said central server further comprises nodes ' behavior information acquisition unit, data processing unit and System Management Unit, wherein; Said nodes ' behavior information acquisition unit further comprises:

Interbehavior obtains subelement, is used to be responsible for when the user node interbehavior takes place, obtain the interbehavior of each user node that is produced; Wherein, the node interbehavior that is obtained comprises at least to be browsed, search for, downloads, uploads and mark;

Interbehavior is followed the tracks of subelement, is used for being responsible for each accessed user node interbehavior is followed the tracks of, and obtains the information relevant with this interbehavior;

Interbehavior analysis subelement is used to utilize interbehavior to obtain subelement and interbehavior is followed the tracks of the information that subelement is obtained separately, and the user node that the interbehavior of analysis user node reflects is to the demand and the desirability of file;

Said data processing unit further comprises:

Demand similitude computation subunit is used to utilize the demand and the desirability information of the user node that said nodes ' behavior information acquisition unit obtained, the demand similitude of content between the computing node; And, according to the demand of user node and Edge Server, the similitude between the computing node tissue, and send to node corresponding or Edge Server;

The requirement forecasting computation subunit; Be used for similitude data according to content between the user node; The document requirement of predictive user node and user node group will be recommended user node to the document requirement prediction of user node; To send to the respective edges server to the document requirement prediction of groups of nodes, for the file of Edge Server prepares to provide foundation;

Evolve to handle subelement, be used for, and the corresponding demand file is sent to node corresponding and Edge Server by period upgrade demand similitude data and requirement forecasting data.

Preferably, the demand similitude of content is specially between the said demand similitude computation subunit computing node,

Make C=(c ₁, c ₂... c _n) expression content c _iSet, P=(p ₁, p ₂... p _m) expression node p _iSet, use r _{I, j}Expression node p _iTo content c _jDesirability, r _{I, j}∈ (1,2 ... k) have the desirability of k different stage, then between the node similitude of content demand by computes:

$\mathrm{sim}(p_a,p_b)=\frac{{\mathrm{\Σ}}_{c\∈C_{\mathrm{ab}}}{r}_{a,i}{r}_{b,c}}{\sqrt{{\mathrm{\Σ}}_{c\∈C_{\mathrm{ab}}}{\left(r_{a,c}\right)}^2}\sqrt{{\mathrm{\Σ}}_{c\∈C_{\mathrm{ab}}}{\left(r_{b,c}\right)}^2}}$

Wherein, sim (p _a, p _b) expression node p _aAnd p _bBetween the demand similitude, C _AbBe node p _aAnd p _bShow the common factor of the properties collection of demand;

Similitude between said demand similitude computation subunit computing node tissue is specially, with p _XyExpression node p _xWith node p _yOverall similitude or local similar property, | sub (P) _i| expression node grouping sub (P) _iThe number of middle node, d (sub (P) _i, sub (P) _j) represent the similarity degree between two groupings, its computing formula is following:

$d(\mathrm{sub}{\left(P\right)}_i,\mathrm{sub}{\left(P\right)}_j)=\frac{\underset{p_x\∈\mathrm{sub}{\left(P\right)}_i}{\mathrm{\Σ}}\underset{p_y\∈\mathrm{sub}{\left(P\right)}_j}{\mathrm{\Σ}}{p}_{\mathrm{xy}}}{\left|\mathrm{sub}{\left(P\right)}_i\right|\×\left|\mathrm{sub}{\left(P\right)}_j\right|}$

Preferably, said requirement forecasting computation subunit is according to the similitude data of content between the user node, and the document requirement of predictive user node and user node group is specially,

$\mathrm{dpredicton}\left(c\right)=\underset{p_i\∈\mathrm{sub}\left(P\right)}{\mathrm{\Σ}}\frac{\underset{p_j\∈P\left(c\right)}{\mathrm{\Σ}}{r}_{j,c}\×\mathrm{sim}{(p_i,p_j)}_{{\mathrm{tz}}_h}}{\underset{p_j\∈P\left(c\right)}{\mathrm{\Σ}}\mathrm{sim}{(p_i,p_j)}_{{\mathrm{tz}}_h}}$

Wherein, sub (P) representes the set of the node that certain sub-net comprises, and dprediction (c) is the desirability of node to content c, and P (c) is all node set that content c had demand.

Preferably, when said user node application adds network, be user node allocations of edge server according to the similitude result of calculation after present node and the adding of each Edge Server by said central server; The global similarity property result of calculation that after user node intend to add certain Edge Server, causes this autonomy territory does not allow this user node to add during less than setting threshold, and said central server is that said user node is redistributed Edge Server; When said user node can not add any Edge Server, then integral body was turned down global similarity property threshold value, confirmed the Edge Server that said user node can add again again, and wherein, the unsteady of global similarity property threshold value do not exceed between the setting district.

Preferably, said central server periodically sends probe message to said Edge Server, and whether detect said Edge Server online; Said Edge Server is user node transmission probe message in the P2P subnet periodically, and whether detect said user node online;

Said Edge Server is after user node or other Edge Servers send the file polling request in the P2P subnet, and the affirmation message of returning in the stipulated time is efficient message, otherwise the inquiry failure.

A kind of P2P content distribution method based on user's social property; Be applied in the P2P content distributing network; Said P2P content distributing network comprises central server, Edge Server and with peer-to-peer network P2P subnet, wherein, said Edge Server is connected with dedicated network or express network with said central server; Said Edge Server and P2P subnet are through Internet connection, and said P2P subnet is made up of a plurality of user terminals; Said method also comprises:

Notify said central server after the request of user node adding P2P subnet; Said central server obtains the log-on message of said user node; Log-on message according to user node calculates the similitude result of calculation after said user node adds each Edge Server, and makes user node join the corresponding Edge Server of the maximum result of similitude;

User node is after Edge Server sends the file download request; Said Edge Server to P2P subnet that this Edge Server is connected in the user send the file polling request; After obtaining response with inquiry file; The user who downloads for demand file provides the file distributing service, and, on this Edge Server the demand file of no institute and with P2P subnet that this Edge Server is connected in all users when also not having institute's demand file; Send the file download request to other Edge Servers, from other Edge Server file in download and storage;

When said Edge Server received the file download request of other Edge Servers, the Edge Server of downloading for demand file provided the file distributing service;

Said central server obtains each user's and this content distributing network in the P2P subnet interbehavior, and the user is followed the tracks of, and according to user's interbehavior and trace information, calculates user's document requirement and desirability; Document requirement and desirability according to the user further carry out the calculating of demand similitude, requirement forecasting calculating between the user, and carry out demand evolution processing according to the variation of user's social property and the user's request information of constantly obtaining; Wherein, user's social property comprises user's identity, occupation, preference information; Said central server also is used to store user's identity, occupation, the log-on message of hobby; And, confirm user's hobby, user identity and variation thereof, user's occupation and variation thereof according to the demand and the desirability of user to file, demand and the close user of needed length are made as neighbor node mutually; When confirming that demand that file is downloaded exceeds setting threshold, corresponding document is recommended to the big Edge Server of demand or recommends to the Edge Server that all do not store corresponding document; The Edge Server that receives recommendation obtains the file and the storage that need prestore from other Edge Servers; Said central server further calculates the evolution information of user to document requirement, and the relevant Edge Server of notice, and the evolution information that Edge Server is notified according to said central server is adjusted self storage file.

Preferably, the demand similitude of content is specially between the said computing node,

Make C=(c ₁, c ₂... c _n) expression content c _iSet, P=(p ₁, p ₂... p _m) expression node p _iSet, use r _{I, j}Expression node p _iTo content c _jDesirability, r _{I, j}∈ (1,2 ... k) have the desirability of k different stage, then between the node similitude of content demand by computes:

$\mathrm{sim}(p_a,p_b)=\frac{{\mathrm{\Σ}}_{c\∈C_{\mathrm{ab}}}{r}_{a,i}{r}_{b,c}}{\sqrt{{\mathrm{\Σ}}_{c\∈C_{\mathrm{ab}}}{\left(r_{a,c}\right)}^2}\sqrt{{\mathrm{\Σ}}_{c\∈C_{\mathrm{ab}}}{\left(r_{b,c}\right)}^2}}$

Wherein, sim (p _a, p _b) expression node p _aAnd p _bBetween the demand similitude, C _AbBe node p _aAnd p _bShow the common factor of the properties collection of demand;

Similitude between said computing node tissue is specially, with p _XyExpression node p _xWith node p _yOverall similitude or local similar property, | sub (P) _i| expression node grouping sub (P) _iThe number of middle node, d (sub (P) _i, sub (P) _j) represent the similarity degree between two groupings, its computing formula is following:

$d(\mathrm{sub}{\left(P\right)}_i,\mathrm{sub}{\left(P\right)}_j)=\frac{\underset{p_x\∈\mathrm{sub}{\left(P\right)}_i}{\mathrm{\Σ}}\underset{p_y\∈\mathrm{sub}{\left(P\right)}_j}{\mathrm{\Σ}}{p}_{\mathrm{xy}}}{\left|\mathrm{sub}{\left(P\right)}_i\right|\×\left|\mathrm{sub}{\left(P\right)}_j\right|}$

Preferably, said similitude data according to content between the user node, the document requirement of predictive user node and user node group is specially:

$\mathrm{dpredicton}\left(c\right)=\underset{p_i\∈\mathrm{sub}\left(P\right)}{\mathrm{\Σ}}\frac{\underset{p_j\∈P\left(c\right)}{\mathrm{\Σ}}{r}_{j,c}\×\mathrm{sim}{(p_i,p_j)}_{{\mathrm{tz}}_h}}{\underset{p_j\∈P\left(c\right)}{\mathrm{\Σ}}\mathrm{sim}{(p_i,p_j)}_{{\mathrm{tz}}_h}}$

Wherein, sub (P) representes the set of the node that certain sub-net comprises, and dprediction (c) is the desirability of node to content c, and P (c) is all node set that content c had demand.

Preferably, said method also comprises:

When said user node application adds network, be user node allocations of edge server according to the similitude result of calculation after present node and the adding of each Edge Server by said central server; The global similarity property result of calculation that after user node intend to add certain Edge Server, causes this autonomy territory does not allow this user node to add during less than setting threshold, and said central server is that said user node is redistributed Edge Server; When said user node can not add any Edge Server, then integral body was turned down global similarity property threshold value, confirmed the Edge Server that said user node can add again again, and wherein, the unsteady of global similarity property threshold value do not exceed between the setting district;

Said central server periodically sends probe message to said Edge Server, and whether detect said Edge Server online; Said Edge Server is user node transmission probe message in the P2P subnet periodically, and whether detect said user node online;

Said Edge Server is after user node or other Edge Servers send the file polling request in the P2P subnet, and the affirmation message of returning in the stipulated time is efficient message, otherwise the inquiry failure.

Among the present invention; Because the identity information through the user etc. has been determined user's content demand and hobby etc. in advance; Can the user that the demand similitude is high be divided into one group; Simultaneously, when being user's allocations of edge server, come to confirm Edge Server to be added for user node through the global similarity property of calculating autonomous territory; That is to say that networking was carried out in user's request when the present invention distributed in network configuration, make networking structure of the present invention search in raising on the basis of success rate, the average content search offered load of native system has obvious reduction than conventional method.The present invention has also reduced average content search time delay; Often find a more than copy of content in the search; The content search time delay is meant to be initiated to from content search and finds a desired value that copy passes through the node number at least; Search in raising on the basis of success rate, the average content search time delay of native system has obvious reduction than conventional method.The present invention can also accelerate content distribution speed, improves the stability of content distribution; In the content search process; Can obtain the Distribution Statistics situation of current copy through server; Adjust the searching algorithm parameters neatly, can improve the less resource of copy and search probability of successful, the resource more to copy can reduce unnecessary offered load.

Description of drawings

Fig. 1 is the structural representation of the system of P2P and CDN combination;

Fig. 2 is the structural representation that the present invention is based on the P2P content distributing network of user's social property;

Fig. 3 is the structural representation of central server that the present invention is based on the P2P content distributing network of user's social property;

Fig. 4 is the structural representation of Edge Server that the present invention is based on the P2P content distributing network of user's social property;

Fig. 5 is the structural representation of user terminal that the present invention is based on the P2P content distributing network of user's social property;

Fig. 6 the present invention is based on interaction diagrams between each network element of P2P content distributing network of user's social property.

Embodiment

For making the object of the invention, technical scheme and advantage clearer, below lift embodiment and with reference to accompanying drawing, to further explain of the present invention.

Fig. 2 is the structural representation that the present invention is based on the P2P content distributing network of user's social property; As shown in Figure 2; The P2P content distributing network that the present invention is based on user's social property comprises central server, Edge Server and with peer-to-peer network P2P subnet, wherein, Edge Server is connected with dedicated network or express network with central server; Edge Server and P2P subnet are through Internet connection, and a plurality of user terminals are formed the P2P subnet with the P2P mode.

The calculating of central server responsible node demand similitude and node requirement forecasting, and the collection and the management of node demand similitude and node requirement forecasting calculating information needed.Edge Server links to each other with the node of some according to node demand similitude and node requirement forecasting, and file service is provided for these nodes.When needed, can obtain the demand similitude information and the forecast demand information of interdependent node, can search each other as required and distribution of document between the Edge Server from central server.Central server and Edge Server are interconnected to form content distributing network (CDN), and through being interconnected to form a plurality of P2P subnets, each P2P subnet links to each other with an Edge Server through a plurality of nodes the P2P node each other.

Node can be as required, adds and leave P2P CDN system, and node adds fashionable, according to node similitude information, gets into suitable position, sets up and the getting in touch of other nodes, and when node left, adjacent node was adjusted according to real-time condition accordingly.The node of entering system can send file request, searches required file, and obtains file from respective edges server and P2P node.

Below the concrete structure of central server, Edge Server and user terminal is described.

Fig. 3 is the structural representation of central server that the present invention is based on the P2P content distributing network of user's social property; As shown in Figure 3, the central server that the present invention is based on the P2P content distributing network of user's social property comprises nodes ' behavior acquiring unit, data processing unit and System Management Unit;

Wherein, the nodes ' behavior acquiring unit is responsible for obtaining the interbehavior of node (certain user terminal) and system, and the line trace of going forward side by side according to interbehavior and trace information, is calculated document requirement and desirability that these interbehaviors show then.The nodes ' behavior acquiring unit comprises three sub-cells again, is respectively that the node interbehavior obtains subelement, interbehavior follows the tracks of subelement and interbehavior is analyzed subelement.

Interbehavior obtains subelement and mainly is responsible for when the node interbehavior takes place, obtaining the node interbehavior of generation, and retrievable node interbehavior comprises: browse, search for, download, upload and mark.

Interbehavior is followed the tracks of subelement and mainly is responsible for each the node interbehavior that gets access to is followed the tracks of, and obtains the information relevant with this interbehavior.When browsing the recommended information of a file like certain node users, interbehavior is followed the tracks of subfunction, can obtain its browsing time through browsing behavior tracking.

Interbehavior is analyzed subelement and is utilized interbehavior to obtain the information of subfunction and the acquisition of interbehavior tracking subfunction, and the node that the interbehavior of analysis node reflects is to the demand and the desirability of file.

The calculating of the demand similitude of content, requirement forecasting are calculated between the main responsible node of data processing unit, and carry out demand evolution processing according to the variation of user's social property and the user's request information of constantly obtaining.Data processing unit comprises demand similitude computation subunit, requirement forecasting subelement and demand evolution and handles subelement.

The great deal of nodes user's that demand similitude computation subunit is obtained according to the nodes ' behavior acquiring unit interbehavior, and these node users document requirement and the desirability data that show, the demand similitude of content between the computing node.According to the demand of node and Edge Server, the similitude between the computing node tissue, and send to node corresponding or Edge Server.

The requirement forecasting computation subunit is according to the similitude data of content between the node; The document requirement of prediction node and groups of nodes; Will be to the document requirement prediction of node; Recommend node users on suitable opportunity, will send to the respective edges server to the document requirement prediction of groups of nodes, for the file of Edge Server prepares to provide foundation.

Wherein, the demand similitude of content is specially between the computing node,

Make C=(c ₁, c ₂... c _n) expression content set, P=(p ₁, p ₂... p _m) expression node set, use r _{I, j}Expression node p _iTo content c _jDesirability, r _{I, j}∈ (1,2 ... k) have the desirability of k different stage, then between the node similitude of content demand by computes:

$\mathrm{sim}(p_a,p_b)=\frac{{\mathrm{\Σ}}_{c\∈C_{\mathrm{ab}}}{r}_{a,i}{r}_{b,c}}{\sqrt{{\mathrm{\Σ}}_{c\∈C_{\mathrm{ab}}}{\left(r_{a,c}\right)}^2}\sqrt{{\mathrm{\Σ}}_{c\∈C_{\mathrm{ab}}}{\left(r_{b,c}\right)}^2}}$

Wherein, sim (p _a, p _b) expression node p _aAnd p _bBetween the demand similitude, C _AbBe node p _aAnd p _bShow the common factor of the properties collection of demand;

Similitude between said computing node tissue is specially, with p _XyExpression node p _xWith node p _yOverall similitude or local similar property, | sub (P) _i| expression node grouping sub (P) _iThe number of middle node, d (sub (P) _i, sub (P) _j) represent the similarity degree between two groupings, its computing formula is following:

$d(\mathrm{sub}{\left(P\right)}_i,\mathrm{sub}{\left(P\right)}_j)=\frac{\underset{p_x\∈\mathrm{sub}{\left(P\right)}_i}{\mathrm{\Σ}}\underset{p_y\∈\mathrm{sub}{\left(P\right)}_j}{\mathrm{\Σ}}{p}_{\mathrm{xy}}}{\left|\mathrm{sub}{\left(P\right)}_i\right|\×\left|\mathrm{sub}{\left(P\right)}_j\right|}$

According to the similitude data of content between the user node, the document requirement of predictive user node and user node group is specially,

$\mathrm{dpredicton}\left(c\right)=\underset{p_i\∈\mathrm{sub}\left(P\right)}{\mathrm{\Σ}}\frac{\underset{p_j\∈P\left(c\right)}{\mathrm{\Σ}}{r}_{j,c}\×\mathrm{sim}{(p_i,p_j)}_{{\mathrm{tz}}_h}}{\underset{p_j\∈P\left(c\right)}{\mathrm{\Σ}}\mathrm{sim}{(p_i,p_j)}_{{\mathrm{tz}}_h}}$

Wherein, sub (P) representes the set of the node that certain sub-net comprises, and dprediction (c) is the desirability of node to content c, and P (c) is all node set that content c had demand.

The subnet content needing forecasting method that local similar property matrix forms is following:

$\mathrm{dpredicton}\left(c\right)=\underset{p_i\∈\mathrm{sub}\left(P\right)}{\mathrm{\Σ}}\frac{\underset{p_j\∈P\left(c\right)}{\mathrm{\Σ}}{r}_{j,c}\×\mathrm{sim}{(p_i,p_j,t_x)}_{{\mathrm{tz}}_h}}{\underset{p_j\∈P\left(c\right)}{\mathrm{\Σ}}\mathrm{sim}{(p_i,p_j,t_x)}_{{\mathrm{tz}}_h}}$

Wherein content c belongs to type t _x

The set of P2P subnet desired content can be expressed as: C (sub (P))={ c|dprediction (c)>=d _Min, d wherein _MinBe a desirability value of choosing, work as d _MinIncrease, the element number of content requirements set can reduce, otherwise then can increase.

Evolve and handle the time period division methods of subelement,, bring in constant renewal in demand similitude data and requirement forecasting data, and send to node corresponding and Edge Server along with the passing of time period according to system.

System Management Unit mainly is in charge of system resource, comprises file, Edge Server, node and system parameters.Correspondingly, System Management Unit include file management subelement, edge service management subelement and system parameters management subelement.

The file management subelement mainly is in charge of the file in the system, preserves description information of files, handles operations such as file interpolation, deletion, and to the management of document classification.

Edge service management subelement mainly is responsible for the running status and the resource operating position of all Edge Servers of monitoring.

The node administration subelement is used to preserve the log-on message with Maintenance Point, the state of monitoring and record node.

System parameters management subelement is used for being responsible for the configuration and the maintenance of system parameters (like system time section, document classification index etc.).

Fig. 4 is the structural representation of Edge Server that the present invention is based on the P2P content distributing network of user's social property, and as shown in Figure 4, the Edge Server function comprises node transaction unit, file management unit and file distribution unit, wherein,

The main responsible node in node transaction unit adds, the relevant treatment node transaction unit when leaving with file request comprises node adding processing subelement, node leaves and handles subelement and node file request processing subelement.

Node add to be handled subelement, is used for when node adds and becomes this Edge Server and need file service be provided, carrying out the response and the relevant treatment of Edge Server.

Node leaves the processing subelement, is used for when this Edge Server provides the node of file service to leave, carrying out the response and the relevant treatment of Edge Server.

The node file request is handled subelement, when the node that is used for this Edge Server and provides file service sends file request, carries out the response and the relevant treatment of Edge Server.

File management unit mainly is responsible for safeguarding the file in the Edge Server, and is carried out the file associative operation as required.The file management unit include file is safeguarded subelement, file preliminary treatment subelement and is evolved and handle subelement.

The file maintenance subelement mainly is responsible for safeguarding the Edge Server listed files, adds new file and with different mode deleted files.

File prestores and handles the listed files that the mainly responsible selection of subelement prestores, and selects suitable opportunity, obtains the file that need prestore from other Edge Servers.

Evolve and handle subelement when the new time period arrives, obtain evolution information, the storage file of Edge Server is adjusted from central server.

The file distributing processing unit mainly is responsible for file distributing relevant treatment task, and the file distributing processing unit comprises file distributing subelement between node demand file distribution subelement and server.

Node demand file distribution subelement mainly is responsible for after node sends file request, for node provides the file distributing service.

File distributing subelement between server is used for receiving file that other Edge Server branches send and to other Edge Server distributions of document.

Fig. 5 is the structural representation of user terminal that the present invention is based on the P2P content distributing network of user's social property; As shown in Figure 5; The user terminal that the present invention is based on the P2P content distributing network of user's social property comprises user interface section, node data administrative unit and node transaction unit; Wherein

User interface section mainly is responsible for accomplishing with the mutual of user and user behavior information collecting environment is provided; User interface section comprises User Interface subelement and context-aware environment subelement; Wherein, The User Interface subelement is used to be responsible for to provide the interactive environment with the user, the user through interactive environment carry out that fileinfo is browsed, file search, file are uploaded, file is downloaded and the document score operation.Context-aware environment subelement is used to be responsible for providing the context-aware environment that obtains user behavior and tracking user behavior.

The node data administrative unit mainly be responsible for to be collected and the Maintenance Point data, and node data comprises that file hits data such as data such as effect matrix and local similitude.The node data administrative unit comprises node data collection subelement, node data safeguards that subelement and node neighbours adjust subelement; Wherein, node data collection subelement mainly is responsible for according to specific mode collector node related data.Node data safeguards that subelement is mainly responsible for specific mode Maintenance Point related data.The node neighbours adjust subelement, and are concrete, and neighbours adjust to node, have under three kinds of situations to adjust, concrete, and that node adds is fashionable, when node leaves and adjust symmetrical neighbours.Node adds and the neighbours when leaving adjust function and are responsible for by node transaction unit.The node neighbours of this place adjust subelement and are responsible for adjusting symmetrical neighbours.

It is M={S that above-mentioned file hits the effect matrix, U}, and it is to be used for describing its a certain adjacent node of node statistics of appointment to own ff (ff that comprises other nodes of forwarding) request responding result, wherein s is a vector { s ₁, s ₂, s ₃..., s _n} ^T, it each is to search the number of times of success corresponding to each component among the file type vector F.U also is a vector { u ₁, u ₂, u ₃..., u _n, it each is to search unsuccessful number of times corresponding to each component among the file type vector F.Hit the effect matrix according to file and can obtain the data of hitting the effect matrix correlation with this document.

The subnet content needing forecasting method that local similar property matrix forms is following:

$\mathrm{dpredicton}\left(c\right)=\underset{p_i\∈\mathrm{sub}\left(P\right)}{\mathrm{\Σ}}\frac{\underset{p_j\∈P\left(c\right)}{\mathrm{\Σ}}{r}_{j,c}\×\mathrm{sim}{(p_i,p_j,t_x)}_{{\mathrm{tz}}_h}}{\underset{p_j\∈P\left(c\right)}{\mathrm{\Σ}}\mathrm{sim}{(p_i,p_j,t_x)}_{{\mathrm{tz}}_h}}$

Wherein content c belongs to type t _xOne of ordinary skill in the art will readily recognize that according to local similar property matrix and can obtain local similar property data.

Node transaction unit is different from the node transaction unit in the Edge Server; At first; Their run location is different; Operate in client and Edge Server respectively, this node of node transaction subfunction processing affairs of next client reach the interacting transaction with other nodes, and the node transaction subfunction of Edge Server is responsible for the interacting transaction of processing node and Edge Server.Node transaction unit be used for to node add, other nodes leave, node sends or transmit a document search packet, node receives and the forward node ff is replied and handled, and file distributing affairs between responsible processing node and the node.

In the community network that the person to person forms, people can constantly understand other people social property through social interaction through direct or indirect mode, and set up contact according to social property, form the complicated social network.In community network, if a people searches certain things, will at first consider the people's related social property with this things, search this things through searching people then with this social property.

People's social property is corresponding with its possible demand; The people's of all ages and classes, sex, occupation, education background, life background demand is different, if the people is classified according to its social property, and people's demand is also classified; Has corresponding relation between the two; Like women's demand cosmetics, be engaged in the people of computer technology aspect occupation, the books of demand computer technology aspect.

Its things that possibly have of social property decision of people; And the relevant information that whether has such things; Be that the women possibly have cosmetics, and possibly be familiar with to obtain the place of cosmetics, be engaged in the people of computer technology aspect occupation; The books that possibly have the computer technology aspect, and possibly be familiar with obtaining the place of computer technology aspect books.

Thereby user's social property is same or similar, then the user have also similar with the things of demand, otherwise the user has similar with the things of demand, then user's social property is also same or similar.

People's social property and corresponding demand can be known; If a people often buys cosmetic product, can infer that this people possibly be the women, other things that meet women's demand possibly also be this people's demands; If a people often buys the books of computer technology aspect; Can infer it and possibly be engaged in the occupation of computer technology aspect that other meet the thing of people's demand of being engaged in the relevant occupation of computer technology, also possibly be this people's demands.

This shows, can pass through user's behavioural analysis user's social property, and then the demand of predictive user.For solving the several major issues that exist among the P2P-CDN; Thereby the P2P CDN system based on the analysis of user's social property has been proposed; The calculating of central server responsible node demand similitude and node requirement forecasting, and the collection and the management of node demand similitude and node requirement forecasting calculating information needed.

Certainly, As time goes on, people's social property can change to some extent, and for example, social property is student's people, after in a few years, graduates, and then possibly have some professional attributes, as become teacher or office worker; Perhaps, become father by the man again, the Ms becomes mother's attribute change.And along with people's social property changes, some demands also can significant change, for example, after the student becomes teacher, because professional demand can need some and teaching file associated, after the Ms becomes mother, possibly more pay close attention to the required file of some childs etc.The present invention utilizes people's social property just, user node is carried out calculating such as similitude, determines the approximate user node of demand; And be one group (for example neighborhood) each other with these node division, like this, because the demand of content is more approaching between these neighbor nodes; Therefore; Its required associated documents also have very large general character, thereby (between the groups of nodes) is when carrying out the file download between these neighbor nodes; Possibility mutual mutually is just very high, the network access path in the time of reducing the file download like this.

Among the present invention; After central server is determined some file and is become popular file; Will notify to the Edge Server of not storing these files; Make these Edge Servers download these files when the free time to other Edge Servers and store, to satisfy the file download demand of the user node these Edge Servers under.Certainly, when definite some file download extremely after a little while, can notify Edge Server or initiatively delete these files by Edge Server.

Edge Server links to each other with the node of some according to node demand similitude and node requirement forecasting, and file service is provided for these nodes.When needed, can obtain the demand similitude information and the forecast demand information of interdependent node, can search each other as required and distribution of document between the Edge Server from central server.

Node can be as required, adds and leave P2P CDN system, and node adds fashionable, according to node similitude information, gets into suitable position, sets up and the getting in touch of other nodes, and when node left, adjacent node was adjusted according to real-time condition accordingly.The node of entering system can send file request, searches required file, and obtains file from respective edges server and P2P node.

Among the present invention,, after after a while, can produce the sequence of operations behavior, such as browsing, search for, download, upload and scoring etc. when client user's node joins in the P2P-CDN network.And central server is collected user behavior information, and calculates the similitude of content between the node, and the user can obtain the neighbor node the highest with its similarity degree; When the user searches resource needed, can be to sending request, because the demand of these nodes is more similar with oneself with former the highest neighbor nodes of own similarity degree; So; The other side exists own resource needed possibility big, and this must improve the success rate and the minimizing of searching and time delay to a certain extent; The number of nodes of transmitting is few, therefore can reduce average Network Search load.Central server is according to the demand similitude; Can predict the demand of node and groups of nodes; Will be to the document requirement prediction of node; Recommend node users on suitable opportunity (when needing recommendation service), will send to the respective edges server to the document requirement prediction of groups of nodes, this inevitable distribution speed that has improved whole network greatly.Because the existence of Edge Server, the user can find the resource that exists in the network, has therefore guaranteed the stability of distribution.When user's demand constantly changed, central server also can constantly carry out similitude again and calculate, and user's total energy obtains and the highest neighbor node of own similarity degree, and this also is to guarantee to search a high major reason of success rate.The user can obtain the statistics of some current networks from central server, global resource is distributed make judgement, in search procedure, can in time regulate parameter; Like this, to scarce resource, improved and searched success rate; To the more resource that distributes, reduced unnecessary offered load.

Based on foregoing description, introduce the interactive mode between each network element in the network of the present invention again.

Fig. 6 the present invention is based on interaction diagrams between each network element of P2P content distributing network of user's social property, and is as shown in Figure 6, and the main interactive mode and the flow process that the present invention is based between each network element of P2P content distributing network of user's social property are following:

Step 1.1, the Edge Server application adds network;

Step 1.2, the central server response is for Edge Server distributes unique ID;

Step 2.1, central server detecting periodically Edge Server;

Step 2.2, the detection of Edge Server responsing center server;

Step 2.3 is not informed node that this Edge Server is had jurisdiction over if Edge Server has response, for these nodes are redistributed Edge Server;

Step 2.4, central server are informed current added other Edge Servers, have new Edge Server to add;

Step 3.1, node adds to the central server application;

Step 3.2, the information that the central server return node adds successfully or fails if success is then for node distributes an Edge Server, and is returned Edge Server information and the neighbor list that is distributed to node;

Step 3.3, central server are informed the Edge Server that this node distributes, and have node to add;

Step 4.1, Edge Server detecting periodically node;

Step 4.2, the node echo probe;

Step 4.3, if node does not have response and then informs central server, this node goes offline;

Step 5.1, node is initiated search to neighbor node and Edge Server;

Step 5.2, neighbor node are hit and are returned acknowledge message, and join in the Resources list of initiating search node;

Step 5.3 if neighbor node does not search resource needed, then continues to transmit to the neighbor node of oneself;

Step 5.4, Edge Server returns acknowledge message, and joins in the Resources list of initiating search node;

Step 5.5 if Edge Server does not search resource needed, is then sent query messages to other Edge Servers;

Step 5.6 is if other Edge Servers hit and then return acknowledge message;

Step 6.1, the node node demand file corresponding blocks in the Resources list;

Step 6.2, the node in the Resources list returns corresponding blocks to requesting node;

Step 7.1, Edge Server other Edge Server demand file corresponding blocks in the Resources list;

Step 7.2, other Edge Servers in the Resources list return corresponding blocks to the request Edge Server.

The present invention has also put down in writing a kind of P2P content distribution method based on user's social property; Be applied in the P2P content distributing network; Said P2P content distributing network comprises central server, Edge Server and with peer-to-peer network P2P subnet, wherein, said Edge Server is connected with dedicated network or express network with said central server; Said Edge Server and P2P subnet are through Internet connection, and said P2P subnet is made up of a plurality of user terminals; Said method also comprises:

Notify said central server after the request of user node adding P2P subnet; Said central server obtains the log-on message of said user node; Log-on message according to user node calculates the similitude result of calculation after said user node adds each Edge Server, and makes user node join the corresponding Edge Server of the maximum result of similitude;

User node is after Edge Server sends the file download request; Said Edge Server to P2P subnet that this Edge Server is connected in the user send the file polling request; After obtaining response with inquiry file; The user who downloads for demand file provides the file distributing service, and, on this Edge Server the demand file of no institute and with P2P subnet that this Edge Server is connected in all users when also not having institute's demand file; Send the file download request to other Edge Servers, from other Edge Server file in download and storage;

When said Edge Server received the file download request of other Edge Servers, the Edge Server of downloading for demand file provided the file distributing service;

Said central server obtains each user's and this content distributing network in the P2P subnet interbehavior, and the user is followed the tracks of, and according to user's interbehavior and trace information, calculates user's document requirement and desirability; Document requirement and desirability according to the user further carry out the calculating of demand similitude, requirement forecasting calculating between the user, and carry out demand evolution processing according to the variation of user's social property and the user's request information of constantly obtaining; Wherein, user's social property comprises user's identity, occupation, preference information; Said central server also is used to store user's identity, occupation, the log-on message of hobby; And, confirm user's hobby, user identity and variation thereof, user's occupation and variation thereof according to the demand and the desirability of user to file, demand and the close user of needed length are made as neighbor node mutually; When confirming that demand that file is downloaded exceeds setting threshold, corresponding document is recommended to the big Edge Server of demand or recommends to the Edge Server that all do not store corresponding document; The Edge Server that receives recommendation obtains the file and the storage that need prestore from other Edge Servers; Said central server further calculates the evolution information of user to document requirement, and the relevant Edge Server of notice, and the evolution information that Edge Server is notified according to said central server is adjusted self storage file.

Preferably, said method also comprises:

When said user node application adds network, be user node allocations of edge server according to the similitude result of calculation after present node and the adding of each Edge Server by said central server; The global similarity property result of calculation that after user node intend to add certain Edge Server, causes this autonomy territory does not allow this user node to add during less than setting threshold, and said central server is that said user node is redistributed Edge Server; When said user node can not add any Edge Server, then integral body was turned down global similarity property threshold value, confirmed the Edge Server that said user node can add again again, and wherein, the unsteady of global similarity property threshold value do not exceed between the setting district.

Preferably, said central server periodically sends probe message to said Edge Server, and whether detect said Edge Server online; Said Edge Server is user node transmission probe message in the P2P subnet periodically, and whether detect said user node online;

Said Edge Server is after user node or other Edge Servers send the file polling request in the P2P subnet, and the affirmation message of returning in the stipulated time is efficient message, otherwise the inquiry failure.

The above is merely preferred embodiment of the present invention, is not to be used to limit protection scope of the present invention.

Claims (10)

1. P2P content distributing network based on user's social property; Comprise central server, Edge Server and with peer-to-peer network P2P subnet; Wherein, Said Edge Server is connected with dedicated network or express network with said central server, and said Edge Server and P2P subnet are through Internet connection, and said P2P subnet is made up of a plurality of user terminals; It is characterized in that,

Said central server is used for obtaining the interbehavior of each user of P2P subnet and this content distributing network, and the user is followed the tracks of, and according to user's interbehavior and trace information, calculates user's document requirement and desirability; Document requirement and desirability according to the user further carry out the calculating of demand similitude, requirement forecasting calculating between the user, and carry out demand evolution processing according to the variation of user's social property and the user's request information of constantly obtaining; Wherein, user's social property comprises user's identity, occupation, preference information; Said central server also is used to store user's identity, occupation, the log-on message of hobby; And, confirm user's hobby, user identity and variation thereof, user's occupation and variation thereof according to the demand and the desirability of user to file, demand and the close user of needed length are made as neighbor node mutually; When confirming that demand that file is downloaded exceeds setting threshold, corresponding document is recommended to the big Edge Server of demand or recommends to the Edge Server that all do not store corresponding document;

Said Edge Server be used for being responsible for user node adding, leave the processing of P2P subnet; And after receiving user's file download request; To with P2P subnet that this Edge Server is connected in the user send the file polling request; After obtaining the response with inquiry file, the user who downloads for demand file provides the file distributing service; And, on this Edge Server the demand file of no institute and with P2P subnet that this Edge Server is connected in all users when also not having institute's demand file, send the file download request to other Edge Servers, from other Edge Server file in download and storage; When receiving the file download request of other Edge Servers, the Edge Server of downloading for demand file provides the file distributing service; Be used to be responsible for safeguarding this Edge Server listed files, add new file and deleted file; The listed files that responsible selection prestores, and select suitable opportunity, obtain the file that need prestore from other Edge Servers; And, obtain the evolution information of user from said central server to document requirement, self storage file is adjusted;

Said user terminal is used to provide the interactive environment with the user, the user through interactive environment carry out that fileinfo is browsed, file search, file are uploaded, file is downloaded and the document score operation; And provide and obtain user behavior and the context-aware environment of following the tracks of user behavior; Be used for collecting and the Maintenance Point data, said node data comprises that file hits effect matrix data and local similitude data; And, when adding the P2P subnet, safeguard self neighbor node according to network configuration as user node; And when neighbor node leaves the P2P subnet, upgrade neighbor node automatically and tabulate.

2. system according to claim 1 is characterized in that said central server further comprises nodes ' behavior information acquisition unit, data processing unit and System Management Unit, and wherein, said nodes ' behavior information acquisition unit further comprises:

Interbehavior obtains subelement, is used to be responsible for when the user node interbehavior takes place, obtain the interbehavior of each user node that is produced; Wherein, the node interbehavior that is obtained comprises at least to be browsed, search for, downloads, uploads and mark;

Interbehavior is followed the tracks of subelement, is used for being responsible for each accessed user node interbehavior is followed the tracks of, and obtains the information relevant with this interbehavior;

Interbehavior analysis subelement is used to utilize interbehavior to obtain subelement and interbehavior is followed the tracks of the information that subelement is obtained separately, and the user node that the interbehavior of analysis user node reflects is to the demand and the desirability of file;

Said data processing unit further comprises:

Demand similitude computation subunit is used to utilize the demand and the desirability information of the user node that said nodes ' behavior information acquisition unit obtained, the demand similitude of content between the computing node; And, according to the demand of user node and Edge Server, the similitude between the computing node tissue, and send to node corresponding or Edge Server;

The requirement forecasting computation subunit; Be used for similitude data according to content between the user node; The document requirement of predictive user node and user node group will be recommended user node to the document requirement prediction of user node; To send to the respective edges server to the document requirement prediction of groups of nodes, for the file of Edge Server prepares to provide foundation;

Evolve to handle subelement, be used for, and the corresponding demand file is sent to node corresponding and Edge Server by period upgrade demand similitude data and requirement forecasting data.

3. system according to claim 2 is characterized in that, the demand similitude of content is specially between the said demand similitude computation subunit computing node,

Make C=(c ₁, c ₂... c _n) expression content c _iSet, P=(p ₁, p ₂... p _m) expression node p _iSet, use r _{I, j}Expression node p _iTo content c _jDesirability, r _{I, j}∈ (1,2 ... k) have the desirability of k different stage, then between the node similitude of content demand by computes:

$\mathrm{sim}(p_a,p_b)=\frac{{\mathrm{\Σ}}_{c\∈C_{\mathrm{ab}}}{r}_{a,i}{r}_{b,c}}{\sqrt{{\mathrm{\Σ}}_{c\∈C_{\mathrm{ab}}}{\left(r_{a,c}\right)}^2}\sqrt{{\mathrm{\Σ}}_{c\∈C_{\mathrm{ab}}}{\left(r_{b,c}\right)}^2}}$

Wherein, sim (p _a, p _b) expression node p _aAnd p _bBetween the demand similitude, C _AbBe node p _aAnd p _bShow the common factor of the properties collection of demand;

Similitude between said demand similitude computation subunit computing node tissue is specially, with p _XyExpression node p _xWith node p _yOverall similitude or local similar property, | sub (P) _i| expression node grouping sub (P) _iThe number of middle node, d (sub (P) _i, sub (P) _j) represent the similarity degree between two groupings, its computing formula is following:

$d(\mathrm{sub}{\left(P\right)}_i,\mathrm{sub}{\left(P\right)}_j)=\frac{\underset{p_x\∈\mathrm{sub}{\left(P\right)}_i}{\mathrm{\Σ}}\underset{p_y\∈\mathrm{sub}{\left(P\right)}_j}{\mathrm{\Σ}}{p}_{\mathrm{xy}}}{\left|\mathrm{sub}{\left(P\right)}_i\right|\×\left|\mathrm{sub}{\left(P\right)}_j\right|}.$

4. system according to claim 2 is characterized in that, said requirement forecasting computation subunit is according to the similitude data of content between the user node, and the document requirement of predictive user node and user node group is specially,

$\mathrm{dpredicton}\left(c\right)=\underset{p_i\∈\mathrm{sub}\left(P\right)}{\mathrm{\Σ}}\frac{\underset{p_j\∈P\left(c\right)}{\mathrm{\Σ}}{r}_{j,c}\×\mathrm{sim}{(p_i,p_j)}_{{\mathrm{tz}}_h}}{\underset{p_j\∈P\left(c\right)}{\mathrm{\Σ}}\mathrm{sim}{(p_i,p_j)}_{{\mathrm{tz}}_h}}$

Wherein, sub (P) representes the set of the node that certain sub-net comprises, and dprediction (c) is the desirability of node to content c, and P (c) is all node set that content c had demand.

5. system according to claim 1 is characterized in that, when said user node application adds network, is user node allocations of edge server by said central server according to the similitude result of calculation after present node and the adding of each Edge Server; The global similarity property result of calculation that after user node intend to add certain Edge Server, causes this autonomy territory does not allow this user node to add during less than setting threshold, and said central server is that said user node is redistributed Edge Server; When said user node can not add any Edge Server, then integral body was turned down global similarity property threshold value, confirmed the Edge Server that said user node can add again again, and wherein, the unsteady of global similarity property threshold value do not exceed between the setting district.

6. system according to claim 1 is characterized in that said central server periodically sends probe message to said Edge Server, and whether detect said Edge Server online; Said Edge Server is user node transmission probe message in the P2P subnet periodically, and whether detect said user node online;

Said Edge Server is after user node or other Edge Servers send the file polling request in the P2P subnet, and the affirmation message of returning in the stipulated time is efficient message, otherwise the inquiry failure.

7. P2P content distribution method based on user's social property; Be applied in the P2P content distributing network; Said P2P content distributing network comprises central server, Edge Server and with peer-to-peer network P2P subnet, wherein, said Edge Server is connected with dedicated network or express network with said central server; Said Edge Server and P2P subnet are through Internet connection, and said P2P subnet is made up of a plurality of user terminals; It is characterized in that said method also comprises:

Notify said central server after the request of user node adding P2P subnet; Said central server obtains the log-on message of said user node; Log-on message according to user node calculates the similitude result of calculation after said user node adds each Edge Server, and makes user node join the corresponding Edge Server of the maximum result of similitude;

User node is after Edge Server sends the file download request; Said Edge Server to P2P subnet that this Edge Server is connected in the user send the file polling request; After obtaining response with inquiry file; The user who downloads for demand file provides the file distributing service, and, on this Edge Server the demand file of no institute and with P2P subnet that this Edge Server is connected in all users when also not having institute's demand file; Send the file download request to other Edge Servers, from other Edge Server file in download and storage;

When said Edge Server received the file download request of other Edge Servers, the Edge Server of downloading for demand file provided the file distributing service;

Said central server obtains each user's and this content distributing network in the P2P subnet interbehavior, and the user is followed the tracks of, and according to user's interbehavior and trace information, calculates user's document requirement and desirability; Document requirement and desirability according to the user further carry out the calculating of demand similitude, requirement forecasting calculating between the user, and carry out demand evolution processing according to the variation of user's social property and the user's request information of constantly obtaining; Wherein, user's social property comprises user's identity, occupation, preference information; Said central server also is used to store user's identity, occupation, the log-on message of hobby; And, confirm user's hobby, user identity and variation thereof, user's occupation and variation thereof according to the demand and the desirability of user to file, demand and the close user of needed length are made as neighbor node mutually; When confirming that demand that file is downloaded exceeds setting threshold, corresponding document is recommended to the big Edge Server of demand or recommends to the Edge Server that all do not store corresponding document; The Edge Server that receives recommendation obtains the file and the storage that need prestore from other Edge Servers; Said central server further calculates the evolution information of user to document requirement, and the relevant Edge Server of notice, and the evolution information that Edge Server is notified according to said central server is adjusted self storage file.

8. method according to claim 7 is characterized in that, the demand similitude of content is specially between the said computing node,

Make C=(c ₁, c ₂... c _n) expression content c _iSet, P=(p ₁, p ₂... p _m) expression node p _iSet, use r _{I, j}Expression node p _iTo content c _jDesirability, r _{I, j}∈ (1,2 ... k) have the desirability of k different stage, then between the node similitude of content demand by computes:

$\mathrm{sim}(p_a,p_b)=\frac{{\mathrm{\Σ}}_{c\∈C_{\mathrm{ab}}}{r}_{a,i}{r}_{b,c}}{\sqrt{{\mathrm{\Σ}}_{c\∈C_{\mathrm{ab}}}{\left(r_{a,c}\right)}^2}\sqrt{{\mathrm{\Σ}}_{c\∈C_{\mathrm{ab}}}{\left(r_{b,c}\right)}^2}}$

Wherein, sim (p _a, p _b) expression node p _aAnd p _bBetween the demand similitude, C _AbBe node p _aAnd p _bShow the common factor of the properties collection of demand;

Similitude between said computing node tissue is specially, with p _XyExpression node p _xWith node p _yOverall similitude or local similar property, | sub (P) _i| expression node grouping sub (P) _iThe number of middle node, d (sub (P) _i, sub (P) _j) represent the similarity degree between two groupings, its computing formula is following:

$d(\mathrm{sub}{\left(P\right)}_i,\mathrm{sub}{\left(P\right)}_j)=\frac{\underset{p_x\∈\mathrm{sub}{\left(P\right)}_i}{\mathrm{\Σ}}\underset{p_y\∈\mathrm{sub}{\left(P\right)}_j}{\mathrm{\Σ}}{p}_{\mathrm{xy}}}{\left|\mathrm{sub}{\left(P\right)}_i\right|\×\left|\mathrm{sub}{\left(P\right)}_j\right|}.$

9. method according to claim 7 is characterized in that, said similitude data according to content between the user node, and the document requirement of predictive user node and user node group is specially,

$\mathrm{dpredicton}\left(c\right)=\underset{p_i\∈\mathrm{sub}\left(P\right)}{\mathrm{\Σ}}\frac{\underset{p_j\∈P\left(c\right)}{\mathrm{\Σ}}{r}_{j,c}\×\mathrm{sim}{(p_i,p_j)}_{t{z}_h}}{\underset{p_j\∈P\left(c\right)}{\mathrm{\Σ}}\mathrm{sim}{(p_i,p_j)}_{{\mathrm{tz}}_h}}$

Wherein, sub (P) representes the set of the node that certain sub-net comprises, and dprediction (c) is the desirability of node to content c, and P (c) is all node set that content c had demand.

10. method according to claim 7 is characterized in that, said method also comprises:

When said user node application adds network, be user node allocations of edge server according to the similitude result of calculation after present node and the adding of each Edge Server by said central server; The global similarity property result of calculation that after user node intend to add certain Edge Server, causes this autonomy territory does not allow this user node to add during less than setting threshold, and said central server is that said user node is redistributed Edge Server; When said user node can not add any Edge Server, then integral body was turned down global similarity property threshold value, confirmed the Edge Server that said user node can add again again, and wherein, the unsteady of global similarity property threshold value do not exceed between the setting district;

Said central server periodically sends probe message to said Edge Server, and whether detect said Edge Server online; Said Edge Server is user node transmission probe message in the P2P subnet periodically, and whether detect said user node online;

Said Edge Server is after user node or other Edge Servers send the file polling request in the P2P subnet, and the affirmation message of returning in the stipulated time is efficient message, otherwise the inquiry failure.

Images