CN105471997A - Method for controlling flow in P2P file sharing network based on price mechanism - Google Patents

Abstract

The invention relates to a method for controlling flow in a P2P file sharing network based on a price mechanism, which comprises the following steps: the price charged by the initialization of the uploading link of the resource provider and the downloading link of the resource requester respectively; adjusting the price at the next moment according to the total flow passing through the link; the resource provider initializes a download rate for each resource requester; the resource requester obtains the price to be paid according to the total flow obtained by the resource requester; the resource provider adjusts the rate distributed for the resource requester at the next moment according to the price paid by the resource requester, the price charged by the download link of the resource requester and the price charged by the upload link of the resource provider; and the link end and the user end iterate according to the steps until an optimal point is reached, namely the optimal flow distribution strategy of each user. The invention has the advantages of reasonable flow distribution, simple and accurate algorithm and the like.

Description

A kind of based on flow control methods in the P2P file sharing network of price mechanism

Technical field

The present invention relates to technical field of the computer network, particularly relate to a kind of flow control methods of Peer-to-Peer Network P2P file shared network.

Background technology

In recent years, P2P network application obtained develop rapidly, had on the one hand enriched the content of the Internet, on the other hand its flow explosive growth and to without restriction the taking of bandwidth, bring enormous impact to the Internet infrastructure.According to statistics, P2P applied business accounts for more than 80% of all wideband data throughputs altogether, and this wherein to take network bandwidth situation with P2P content share system even more serious, common P2P content share system comprises file download system Bittorrent, eDonkey, Gnutella, search and retrieval Bearshare, contents distribution, the network storage and peer broadcast Peercasting etc.

These P2P content share systems are universal and development rapidly, they unrestrictedly consume the network bandwidth, take Internet resources, become the maximum consumption person of Internet resources, exceed the data traffic of Web, E-mail, FTP etc. to a great extent and become the main burden of network, even can cause network congestion, thus reduce and affect the performance of other business.Therefore, effective control and management must be carried out to P2P network traffics, thus realize the controlled of P2P network traffics and can manage, improve the network congestion problem that P2P flow brings.

At present, the traditional control method based on " blocking up " is mainly still adopted in P2P flow control, comprise the strategies such as speed limit, Differentiated Services, obstruction, although the applied business of some keys in enterprise network can be ensured to a certain extent, save certain Internet resources, but these strategies reduce the service quality of P2P application, make user satisfaction decline, have impact on some rational P2P application.In order to solve the problem, need to propose new flow control policy to flow control technique, P2P communication flows is controlled in Intranet as far as possible, thus reduce the pressure of provider backbone network, while the satisfaction of Logistics networks user, improve network performance.

Summary of the invention

The object of the invention is to provide that a kind of assignment of traffic is reasonable, method is simple and effective based on flow control methods in the P2P file sharing network of price mechanism.

For achieving the above object, have employed following technical scheme: the inventive method mainly comprises Peer-to-Peer Network P2P, resource requestor s and resource provider p.

In Peer-to-Peer Network P2P, the access link of each user is fixed a price, according to the traffic conditions dynamic conditioning link price on link, and resource provider p is when providing file download service for resource requestor s, the price that the price provided according to resource requestor s and access link are collected, the reasonable distribution of dynamic conditioning flow between each resource requestor, finally realizes optimum allocation and the control of network;

The step of described control method is as follows:

Step 1, in the shared file system of Peer-to-Peer Network P2P, resource provider p uploads the price μ that link initialization collects _p(t), the price v that the download link initialization of resource requestor s is collected _st (), resource provider p is each resource requestor s initialization downloading rate x _sp(t);

Step 2, the total flow y that resource requestor s obtains according to it _st () obtains its price λ paid _s(t), and give its file provider by network notifications;

Step 3, the price λ that resource provider p pays according to resource requestor s _st price v that the download link initialization of (), resource requestor s is collected _st () and resource provider p upload the price μ that link initialization collects _pt (), adjusts the speed x that it distributes for resource requestor s _sp(t+1);

Step 4, the link of uploading of resource provider p upgrades it and new collects price μ _p(t+1); Meanwhile, the download link of resource requestor s upgrades it and new collects price v _s(t+1);

Step 5, resource provider p and resource requestor s according to above-mentioned steps iteration until reach optimum point, i.e. the optimal flow assignment strategy of each resource requestor s;

Step 6, if having new resource provider or resource requestor to add or original resource provider or resource requestor exit, so above-mentioned iterative process re-starts to reach new optimum point.

In step 2, the total flow y that obtains according to it of resource requestor s _st () obtains its price λ paid _s(t),

${\mathrm{\λ}}_s\left(t\right)=\frac{w_s}{max\{\mathrm{\ξ},y_s^{\mathrm{\α}}\left(t\right)\}},y_s\left(t\right)=\underset{p:p\∈P\left(s\right)}{\Σ}{x}_{sp}\left(t\right)$

In formula, p is resource provider; S is resource requestor; P is resource provider set; S is resource requestor set; y _st () is the total flow that resource requestor s obtains; ξ is a positive number being greater than zero, and object guarantees as total flow y _swhen () is too small t, price λ _st () is unlikely to excessive; x _spt () is resource provider p is the initialized downloading rate of each resource requestor s; w _sit is the expense that resource requestor s is ready to pay; P (s) is for resource requestor s provides the resource provider set of file download service; Parameter alpha >0 is fairness index, if α=1, then realizes the proportional fairness of Resourse Distribute between user, if α=2, then realize the harmonic average fairness of Resourse Distribute between user, if α → ∞, then realize the max-min fairness of Resourse Distribute between user.

In step 3, resource provider p adjustresources requestor s distributes speed x _sp(t+1) algorithm is as follows:

$x_{sp}(t+1)={\left(\right(1-\mathrm{\θ}\left)x_{sp}\right(t)+\mathrm{\θ}{\tilde{x}}_{sp}(t)+{\mathrm{\κx}}_{sp}(t\left)\right({\mathrm{\λ}}_s\left(t\right)-v_s\left(t\right)-{\mathrm{\μ}}_p\left(t\right)\left)\right)}_{x_{sp}\left(t\right)}^{+},$

${\tilde{x}}_{sp}(t+1)=(1-\mathrm{\θ}){\tilde{x}}_{sp}\left(t\right)+{\mathrm{\θx}}_{sp}\left(t\right)$

In formula, x _spt () is resource provider p is the initialized downloading rate of each resource requestor s; θ is the low-pass filtering factor, and 0< θ <1, effectively can eliminate the unique and algorithm fluctuation problem that causes of optimum point; to present rate x _spthe valuation of (t); be to resource provider p be resource requestor s distribute speed x _sp(t+1) valuation; κ is algorithm iteration step-length, and κ >0; λ _st () is the total flow y that resource requestor s obtains according to it _st () obtains its price paid; v _st () is the price that the download link initialization of resource requestor s is collected; μ _p(t) be resource provider p upload the price that link initialization collects;

$x_{sp}(t+1)={\left(\right(1-\mathrm{\&theta;}\left)x_{sp}\right(t)+\mathrm{\&theta;}{\tilde{x}}_{sp}(t)+{\mathrm{\&kappa;x}}_{sp}(t\left)\right({\mathrm{\&lambda;}}_s\left(t\right)-v_s\left(t\right)-{\mathrm{\&mu;}}_p\left(t\right)\left)\right)}_{x_{sp}\left(t\right)}^{+}$ Mean,

If x _sp(t) >0, then $x_{sp}(t+1)=(1-\mathrm{\&theta;})x_{sp}\left(t\right)+\mathrm{\&theta;}{\tilde{x}}_{sp}\left(t\right)+{\mathrm{\&kappa;x}}_{sp}\left(t\right)\left({\mathrm{\&lambda;}}_s\right(t)-v_s(t)-{\mathrm{\&mu;}}_p(t\left)\right)$

If x _sp(t)=0, then $x_{sp}(t+1)=max\{0,(1-\mathrm{\&theta;})x_{sp}\left(t\right)+\mathrm{\&theta;}{\tilde{x}}_{sp}\left(t\right)+{\mathrm{\&kappa;x}}_{sp}\left(t\right)\left({\mathrm{\&lambda;}}_s\right(t)-v_s(t)-{\mathrm{\&mu;}}_p(t\left)\right)\}.$

In step 4, the following algorithm of link utilization of uploading of resource provider p upgrades its new price μ collected _p(t+1)

${\mathrm{\&mu;}}_p(t+1)={\left({\mathrm{\&mu;}}_p\right(t)+\mathrm{\&tau;}(z_p\left(t\right)-C_p^u\left)\right)}_{{\mathrm{\&mu;}}_p\left(t\right)}^{+},z_p\left(t\right)=\underset{s:s\&Element;S\left(p\right)}{\&Sigma;}{x}_{sp}\left(t\right)$

In formula, p is resource provider; S is resource requestor; P is resource provider set; S is resource requestor set; μ _p(t) be resource provider p upload the price that link initialization collects; x _spt () is resource provider p is the initialized downloading rate of each resource requestor s; S (p) accepts the resource requestor set that resource provider p provides file download service; be resource provider p upload link bandwidth; z _pt () is the link of uploading of resource provider p is the initialized total uploading rate of resource requestor; τ is algorithm iteration step-length, and τ >0;

${\mathrm{\&mu;}}_p(t+1)={\left({\mathrm{\&mu;}}_p\right(t)+\mathrm{\&tau;}(z_p\left(t\right)-C_p^u\left)\right)}_{{\mathrm{\&mu;}}_p\left(t\right)}^{+}$ Mean,

If μ _p(t) >0, then ${\mathrm{\&mu;}}_p(t+1)={\mathrm{\&mu;}}_p\left(t\right)+\mathrm{\&tau;}\left(z_p\right(t)-C_p^u)$

If μ _p(t)=0, then ${\mathrm{\&mu;}}_p(t+1)=max\{0,{\mathrm{\&mu;}}_p\left(t\right)+\mathrm{\&tau;}\left(z_p\right(t)-C_p^u)\}.$

The download link of resource requestor s utilizes following algorithm to upgrade its new price v collected _s(t+1)

$v_s(t+1)={\left(v_s\right(t)+\mathrm{\&tau;}(y_s\left(t\right)-C_s^d\left)\right)}_{v_s\left(t\right)}^{+},y_s\left(t\right)=\underset{p:p\&Element;P\left(s\right)}{\&Sigma;}{x}_{sp}\left(t\right)$

In formula, p is resource provider; S is resource requestor; P is resource provider set; S is resource requestor set; x _spt () is resource provider p is the initialized downloading rate of each resource requestor s; v _st () is the price that the download link initialization of resource requestor s is collected; y _st () is the total flow that resource requestor s obtains; P (s) is for resource requestor s provides the resource provider set of file download service; it is the download link bandwidth of resource requestor s;

$v_s(t+1)={\left(v_s\right(t)+\mathrm{\&tau;}(y_s\left(t\right)-C_s^d\left)\right)}_{v_s\left(t\right)}^{+}$ Mean,

If v _s(t) >0, then $v_s(t+1)=v_s\left(t\right)+\mathrm{\&tau;}\left(y_s\right(t)-C_s^d)$

If v _s(t)=0, then $v_s(t+1)=max\{0,v_s\left(t\right)+\mathrm{\&tau;}\left(y_s\right(t)-C_s^d)\}.$

In step 5 and 6, the optimum point that algorithm iteration arrives is exactly the globe optimum of flow control Optimized model, and namely in peer-to-peer network, resource provider p is the optimal flux that each resource requestor s distributes.

Compared with prior art, tool of the present invention has the following advantages: assignment of traffic and control problem are summarized as network utility optimization problem, by by the reasonable distribution of the uploading bandwidth of resources in network supplier between resource requestor, while the satisfaction of Logistics networks service requester, realize reasonable distribution and the optimal control of flow between the network user, the pressure of final reduction provider backbone network, ensures that the health of network is run, improves P2P network performance.

Accompanying drawing explanation

Fig. 1 is 6 user network topology diagrams of the inventive method.

Fig. 2 is the optimal flux figure that in Fig. 1, resource requestor S1 obtains.

Fig. 3 is the optimal flux figure that in Fig. 1, resource requestor S2 obtains.

Fig. 4 is the optimal flux figure that in Fig. 1, resource requestor S3 obtains.

Fig. 5 is the optimal flux figure that in Fig. 1, resource requestor S4 obtains.

Fig. 6 is value of utility and the network polymerization value of utility of each resource requestor in Fig. 1.

Fig. 7 is the jitter conditions schematic diagram of each resource requestor in Fig. 1.

Fig. 8 is the optimal flux figure that in Fig. 1, under network fluctuation situation, resource requestor S1 obtains.

Fig. 9 is the optimal flux figure that in Fig. 1, under network fluctuation situation, resource requestor S2 obtains.

Figure 10 is the optimal flux figure that in Fig. 1, under network fluctuation situation, resource requestor S3 obtains.

Figure 11 is the optimal flux figure that in Fig. 1, under network fluctuation situation, resource requestor S4 obtains.

Drawing reference numeral: p1 is resource provider 1, p2 is resource provider 2, s1 is resource requestor 1, s2 is resource requestor 2, s3 is resource requestor 3, s4 is resource requestor 4.

Embodiment

Satisfaction when the present invention obtains service according to the network user chooses utility function, then set up flow control optimal model in P2P file sharing network, the flow control problems by P2P file sharing network is summed up as P2P file sharing network optimal utility model.The present invention designs a kind of flow control algorithm based on price mechanism, and this algorithm can converge to the optimum point of flow control model effectively, i.e. the optimal flow assignment of the network user.

Flow control problems describes:

In P2P network, each use is be connected to the Internet by access link per family.The technology such as optical fiber communication are taked due to backbone network in existing the Internet, it is generally acknowledged and extensive congestion phenomenon is less likely to occur, and so the access link of user has just become the bottleneck affecting user bandwidth, are also that network traffics are distributed and the key controlled.At present, what take during a lot of user Internet access is upload the situation that link and download link be separated, and as ADSL, and a user is when providing file download service, its uploading bandwidth is exactly resource for other network users, therefore will produce the competition of this resource between user.

In order to distinguish supplier and the requestor of file sharing network file, introduce resource provider set P and resource requestor S set.As long as a user provides file download service, so this user just belongs to P.And a user both can belong to P, also can belong to S, this depends on whether this user provides file service or demand file service.Being defined as resource requestor s ∈ S provides the resource provider set of file download service to be P (s); Resource provider p ∈ P provides all resource requestor set of download service to be S (p).Notice p ∈ P (s) and if only if s ∈ S (p).

If resource provider p for the downloading rate that resource requestor s provides be x _sp, then because a user can download identical file from multiple resource provider, total speed that therefore resource requestor s obtains is and should not exceed the download bandwidth of this user namely meanwhile, a resource provider can also simultaneously for multiple user provides download service, and therefore the upper transfer rate of this resource provider p is and be no more than the uploading bandwidth of this user namely

User utility function:

The satisfaction of user when obtaining service can describe by utility function, selects following utility function for this reason,

$U_s\left(y_s\right)=\left\{\begin{array}{c}{w}_s\log y_s,\mathrm{\&alpha;}=1\\ w_s\frac{y_s^{1-\mathrm{\&alpha;}}}{1-\mathrm{\&alpha;}},\mathrm{\&alpha;}\&NotEqual;1\end{array}\right.$

Wherein, w _sdescribe the expense that resource requestor s is ready to pay, and parameter alpha>=0 describes fairness index, if α=1, then realize the proportional fairness of Resourse Distribute between user, if α=2, then realize the harmonic average fairness of Resourse Distribute between user, if α → ∞, then realize the max-min fairness of Resourse Distribute between user.

Flow control model:

The flow control problems of P2P file sharing network can be summed up as following network utility optimization problem

$\begin{array}{cc}\max & \underset{s:s\&Element;S}{\&Sigma;}{U}_s\left(y_s\right)\\ subjectto& \underset{p:p\&Element;P\left(s\right)}{\&Sigma;}{x}_{sp}=y_s,s\&Element;S\\ & \underset{p:p\&Element;P\left(s\right)}{\&Sigma;}{x}_{sp}\&le;C_s^d,s\&Element;S\\ & \underset{p:p\&Element;P\left(s\right)}{\&Sigma;}{x}_{sp}\&le;C_p^u,p\&Element;P\\ over& x_{sp}\&GreaterEqual;0,s\&Element;S,p\&Element;P\end{array}$

Utilize Nonlinear Programming Theory, this Optimal Utility is convex optimization problem, to variable y _sbe strictly concave function, there is globally optimal solution and be unique, but this problem is to variable x _spbut not strict concave function, therefore optimal solution not unique.That is, total speed that each resource requestor obtains is unique, but owing to there is multiple resource provider, so multiple concrete assignment of traffic form can be there is.

Set up the Lagrangian of above-mentioned optimization problem

$\begin{array}{c}L(x,y;\mathrm{\&lambda;},\mathrm{\&mu;})=\underset{s:s\&Element;S}{\&Sigma;}\left(U_s\right(y_s)+{\mathrm{\&lambda;}}_s(\underset{p:p\&Element;P\left(s\right)}{\&Sigma;}{x}_{sp}-y_s)+v_s(C_s^d-\underset{p:p\&Element;P\left(s\right)}{\&Sigma;}{x}_{sp}-{\mathrm{\&delta;}}_s^2))\\ +\underset{p:p\&Element;P}{\&Sigma;}{\mathrm{\&mu;}}_p(C_p^u-\underset{s:s\&Element;S\left(p\right)}{\&Sigma;}{x}_{sp}-{\mathrm{\&epsiv;}}_p^2)\end{array}$

Wherein λ _s>=0 is the price of unit bandwidth that resource requestor s pays, v _s>=0 is the price of the unit bandwidth that the download link of resource requestor s is collected, μ _p>=0 is the price uploading the unit bandwidth that link is collected of resource provider p, with be slack variable, represent the download link of resource requestor s and the remaining bandwidth uploading link of resource provider p respectively.

The inventive method mainly comprises Peer-to-Peer Network P2P, resource requestor r and resource provider p, in Peer-to-Peer Network P2P, the access link of each user is fixed a price, according to the traffic conditions dynamic conditioning link price on link, and resource provider p is when providing file download service for resource requestor s, the price that the price provided according to resource requestor s and access link are collected, the reasonable distribution of dynamic conditioning flow between each resource requestor, finally realizes optimum allocation and the control of network;

The step of described control method is as follows:

Step 1, in the shared file system of Peer-to-Peer Network P2P, resource provider p uploads the price μ that link initialization collects _p(t), the price v that the download link initialization of resource requestor s is collected _st (), resource provider p is each resource requestor s initialization downloading rate x _sp(t);

Step 2, the total flow y that resource requestor s obtains according to it _st () obtains its price λ paid _s(t), and give its file provider by network notifications;

${\mathrm{\&lambda;}}_s\left(t\right)=\frac{w_s}{max\{\mathrm{\&xi;},y_s^{\mathrm{\&alpha;}}\left(t\right)\}},y_s\left(t\right)=\underset{p:p\&Element;P\left(s\right)}{\&Sigma;}{x}_{sp}\left(t\right)$

In formula, p is resource provider; S is resource requestor; P is resource provider set; S is resource requestor set; y _st () is the total flow that resource requestor s obtains; ξ is a positive number being greater than zero, and object guarantees as total flow y _swhen () is too small t, price λ _st () is unlikely to excessive; x _spt () is resource provider p is the initialized downloading rate of each resource requestor s; w _sit is the expense being ready to pay; P (s) is for resource requestor s provides the resource provider set of file download service; Parameter alpha >0 is fairness index, if α=1, then realizes the proportional fairness of Resourse Distribute between user, if α=2, then realize the harmonic average fairness of Resourse Distribute between user, if α → ∞, then realize the max-min fairness of Resourse Distribute between user.

Step 3, the price λ that resource provider p pays according to resource requestor s _st price v that the download link initialization of (), resource requestor s is collected _st () and resource provider p upload the price μ that link initialization collects _pt (), adjusts the speed x that it distributes for resource requestor s _sp(t+1);

$\begin{array}{c}{x}_{sp}(t+1)={\left(\right(1-\mathrm{\&theta;}\left)x_{sp}\right(t)+\mathrm{\&theta;}{\tilde{x}}_{sp}(t)+{\mathrm{\&kappa;x}}_{sp}(t\left)\right({\mathrm{\&lambda;}}_s\left(t\right)-v_s\left(t\right)-{\mathrm{\&mu;}}_p\left(t\right)\left)\right)}_{x_{sp}\left(t\right)}^{+}\\ {\tilde{x}}_{sp}\left(t+1\right)=(1-\mathrm{\&theta;}){\tilde{x}}_{sp}\left(t\right)+{\mathrm{\&theta;x}}_{sp}\left(t\right)\end{array},$

In formula, x _spt () is resource provider p is the initialized downloading rate of each resource requestor s; θ is the low-pass filtering factor, and 0< θ <1, effectively can eliminate the unique and algorithm fluctuation problem that causes of optimum point; to present rate x _spthe valuation of (t); be to resource provider p be resource requestor s distribute speed x _sp(t+1) valuation; κ is algorithm iteration step-length, and κ >0; λ _st () is the total flow y that resource requestor s obtains according to it _st () obtains its price paid; v _st () is the price that the download link initialization of resource requestor s is collected; μ _p(t) be resource provider p upload the price that link initialization collects;

$x_{sp}(t+1)={\left(\right(1-\mathrm{\&theta;}\left)x_{sp}\right(t)+\mathrm{\&theta;}{\tilde{x}}_{sp}(t)+{\mathrm{\&kappa;x}}_{sp}(t\left)\right({\mathrm{\&lambda;}}_s\left(t\right)-v_s\left(t\right)-{\mathrm{\&mu;}}_p\left(t\right)\left)\right)}_{x_{sp}\left(t\right)}^{+}$ Mean,

If x _sp(t) >0, then $x_{sp}(t+1)=(1-\mathrm{\&theta;})x_{sp}\left(t\right)+\mathrm{\&theta;}{\tilde{x}}_{sp}\left(t\right)+{\mathrm{\&kappa;x}}_{sp}\left(t\right)\left({\mathrm{\&lambda;}}_s\right(t)-{\mathrm{\&nu;}}_s\left(t\right)-{\mathrm{\&mu;}}_p\left(t\right))$

If x _sp(t)=0, then $x_{sp}(t+1)=max\{0,(1-\mathrm{\&theta;})x_{sp}\left(t\right)+\mathrm{\&theta;}{\tilde{x}}_{sp}\left(t\right)+{\mathrm{\&kappa;x}}_{sp}\left(t\right)\left({\mathrm{\&lambda;}}_s\right(t)-v_s(t)-{\mathrm{\&mu;}}_p(t\left)\right)\}.$

Step 4, the link of uploading of resource provider p upgrades it and new collects price μ _p(t+1);

${\mathrm{\&mu;}}_p(t+1)={\left({\mathrm{\&mu;}}_p\right(t)+\mathrm{\&tau;}(z_p\left(t\right)-C_p^u\left)\right)}_{{\mathrm{\&mu;}}_p\left(t\right)}^{+},z_p\left(t\right)=\underset{s:s\&Element;S\left(p\right)}{\&Sigma;}{x}_{sp}\left(t\right)$

In formula, p is resource provider; S is resource requestor; P is resource provider set; S is resource requestor set; μ _p(t) be resource provider p upload the price that link initialization collects; x _spt () is resource provider p is the initialized downloading rate of each resource requestor s; S (p) accepts the resource requestor set that resource provider p provides file download service; be resource provider p upload link bandwidth; z _pt () is the link of uploading of resource provider p is the initialized total uploading rate of resource requestor; τ is algorithm iteration step-length, and τ >0;

${\mathrm{\&mu;}}_p(t+1)={\left({\mathrm{\&mu;}}_p\right(t)+\mathrm{\&tau;}(z_p\left(t\right)-C_p^u\left)\right)}_{{\mathrm{\&mu;}}_p\left(t\right)}^{+}$ Mean,

If μ _p(t) >0, then ${\mathrm{\&mu;}}_p(t+1)={\mathrm{\&mu;}}_p\left(t\right)+\mathrm{\&tau;}\left(z_p\right(t)-C_p^u)$

If μ _p(t)=0, then ${\mathrm{\&mu;}}_p(t+1)=max\{0,{\mathrm{\&mu;}}_p\left(t\right)+\mathrm{\&tau;}\left(z_p\right(t)-C_p^u)\}.$

The download link of resource requestor s utilizes following algorithm to upgrade its new price v collected _s(t+1)

$v_s(t+1)={\left(v_s\right(t)+\mathrm{\&tau;}(y_s\left(t\right)-C_s^d\left)\right)}_{v_s\left(t\right)}^{+},y_s\left(t\right)=\underset{p:p\&Element;P\left(s\right)}{\&Sigma;}{x}_{sp}\left(t\right)$

In formula, p is resource provider; S is resource requestor; P is resource provider set; S is resource requestor set; x _spt () is resource provider p is the initialized downloading rate of each resource requestor s; v _st () is the price that the download link initialization of resource requestor s is collected; y _st () is the total flow that resource requestor s obtains; P (s) is for resource requestor s provides the resource provider set of file download service; it is the download link bandwidth of resource requestor s;

$v_s(t+1)={\left(v_s\right(t)+\mathrm{\&tau;}(y_s\left(t\right)-C_s^d\left)\right)}_{v_s\left(t\right)}^{+}$ Mean,

If v _s(t) >0, then $v_s(t+1)=v_s\left(t\right)+\mathrm{\&tau;}\left(y_s\right(t)-C_s^d)$

If v _s(t)=0, then $v_s(t+1)=max\{0,v_s\left(t\right)+\mathrm{\&tau;}\left(y_s\right(t)-C_s^d)\}.$

Step 5, resource provider p and resource requestor s according to above-mentioned steps iteration until reach optimum point, i.e. the optimal flow assignment strategy of each resource requestor s;

Step 6, if having new resource provider or resource requestor to add or original resource provider or resource requestor exit, so above-mentioned iterative process re-starts to reach new optimum point.

Convergence:

Convergence is the important indicator of measure algorithm performance.Contemplated by the invention 2 file service suppliers, the situation of 4 file server requests persons, as shown in Figure 1.Wherein, the link bandwidth of uploading of resource provider is $C^u=(C_1^u,C_2^u)=(12,20)Mbps,$ The download link bandwidth of resource requestor is $C^d=(C_1^d,C_2^d,C_3^d,C_4^d)=(15,$ $10,8,6)Mbps.$ The equitable proportion of uploading bandwidth between resource requestor that the present invention analyzes resource provider distributes (i.e. α=1), each simulation result is as shown in Fig. 2,3,4,5,6, as shown in Figure 2, the optimal flux that resource requestor 1 obtains from resource provider is therefore, algorithm effectively can converge to the optimum point of flow control model in limited iterations, realizes the reasonable distribution of flow between each user and control.The present invention considers P2P network jitter (Churn) to the impact of algorithm simultaneously, and jitter conditions as shown in Figure 7.Simulation result is as shown in Fig. 8,9,10,11, and as shown in Figure 8, stage 1:t=0 → 400, resources in network requestor only has user 1, and its optimal flux obtained is stage 2:t=401 → 800, resource requestor has user 1,2 and 3, and the optimal flux that now resource requestor 1 obtains is stage 3:t=801 → 1200, resource requestor has user 1,3 and 4, and the optimal flux that now resource requestor 1 obtains is even be not difficult to find out to there is the optimum point that fluctuation situation algorithm still effectively can converge to model.

Above-described embodiment is only be described the preferred embodiment of the present invention; not scope of the present invention is limited; under not departing from the present invention and designing the prerequisite of spirit; the various distortion that those of ordinary skill in the art make technical scheme of the present invention and improvement, all should fall in protection range that claims of the present invention determines.

Claims (5)

1. one kind based on flow control methods in the P2P file sharing network of price mechanism, mainly comprise Peer-to-Peer Network P2P, resource requestor s and resource provider p, it is characterized in that, in Peer-to-Peer Network P2P, the access link of each user is fixed a price, according to the traffic conditions dynamic conditioning link price on link, and resource provider p is when providing file download service for resource requestor s, the price that the price provided according to resource requestor s and access link are collected, the reasonable distribution of dynamic conditioning flow between each resource requestor, finally realize optimum allocation and the control of network,

The step of described control method is as follows:

Step 1, in the shared file system of Peer-to-Peer Network P2P, resource provider p uploads the price μ that link initialization collects _p(t), the price v that the download link initialization of resource requestor s is collected _st (), resource provider p is each resource requestor s initialization downloading rate x _sp(t);

Step 2, the total flow y that resource requestor s obtains according to it _st () obtains its price λ paid _s(t), and give its file provider by network notifications;

Step 3, the price λ that resource provider p pays according to resource requestor s _st price v that the download link initialization of (), resource requestor s is collected _st () and resource provider p upload the price μ that link initialization collects _pt (), adjusts the speed x that it distributes for resource requestor s _sp(t+1);

Step 4, the link of uploading of resource provider p upgrades it and new collects price μ _p(t+1); Meanwhile, the download link of resource requestor s upgrades it and new collects price v _s(t+1);

Step 5, resource provider p and resource requestor s according to above-mentioned steps iteration until reach optimum point, i.e. the optimal flow assignment strategy of each resource requestor s;

Step 6, if having new resource provider or resource requestor to add or original resource provider or resource requestor exit, so above-mentioned iterative process re-starts to reach new optimum point.

2. according to claim 1 a kind of based on flow control methods in the P2P file sharing network of price mechanism, it is characterized in that: in step 2, the total flow y that resource requestor s obtains according to it _st () obtains its price λ paid _s(t),

${\mathrm{\&lambda;}}_s\left(t\right)=\frac{w_s}{max\{\mathrm{\&xi;},y_s^{\mathrm{\&alpha;}}\left(t\right)\}},$ $y_s\left(t\right)=\underset{p:p\&Element;P\left(s\right)}{\&Sigma;}{x}_{sp}\left(t\right)$

In formula, p is resource provider; S is resource requestor; P is resource provider set; S is resource requestor set; y _st () is the total flow that resource requestor s obtains; ξ is a positive number being greater than zero, and object guarantees as total flow y _swhen () is too small t, price λ _st () is unlikely to excessive; x _spt () is resource provider p is the initialized downloading rate of each resource requestor s; w _sit is the expense that resource requestor s is ready to pay; P (s) is for resource requestor s provides the resource provider set of file download service; Parameter alpha >0 is fairness index, if α=1, then realizes the proportional fairness of Resourse Distribute between user, if α=2, then realize the harmonic average fairness of Resourse Distribute between user, if α → ∞, then realize the max-min fairness of Resourse Distribute between user.

3. according to claim 1ly a kind ofly to it is characterized in that based on flow control methods in the P2P file sharing network of price mechanism, in step 3, resource provider p adjustresources requestor s distributes speed x _sp(t+1) algorithm is as follows:

$x_{sp}(t+1)={\left(\right(1-\mathrm{\&theta;}\left)x_{sp}\right(t)+\mathrm{\&theta;}{\tilde{x}}_{sp}(t)+{\mathrm{\&kappa;x}}_{sp}(t\left)\right({\mathrm{\&lambda;}}_s\left(t\right)-v_s\left(t\right)-{\mathrm{\&mu;}}_p\left(t\right)\left)\right)}_{x_{sp}\left(t\right)}^{+},$

${\tilde{x}}_{sp}(t+1)=(1-\mathrm{\&theta;}){\tilde{x}}_{sp}\left(t\right)+{\mathrm{\&theta;x}}_{sp}\left(t\right)$

In formula, x _spt () is resource provider p is the initialized downloading rate of each resource requestor s; θ is the low-pass filtering factor, and 0< θ <1, effectively can eliminate the unique and algorithm fluctuation problem that causes of optimum point; to present rate x _spthe valuation of (t); be to resource provider p be resource requestor s distribute speed x _sp(t+1) valuation; κ is algorithm iteration step-length, and κ >0; λ _st () is the total flow y that resource requestor s obtains according to it _st () obtains its price paid; v _st () is the price that the download link initialization of resource requestor s is collected; μ _p(t) be resource provider p upload the price that link initialization collects;

$x_{sp}(t+1)={\left(\right(1-\mathrm{\&theta;}\left)x_{sp}\right(t)+\mathrm{\&theta;}{\tilde{x}}_{sp}(t)+{\mathrm{\&kappa;x}}_{sp}(t\left)\right({\mathrm{\&lambda;}}_s\left(t\right)-v_s\left(t\right)-{\mathrm{\&mu;}}_p\left(t\right)\left)\right)}_{x_{sp}\left(t\right)}^{+}$ Mean,

If x _sp(t) >0, then $x_{sp}(t+1)=(1-\mathrm{\&theta;})x_{sp}\left(t\right)+\mathrm{\&theta;}{\tilde{x}}_{sp}\left(t\right)+{\mathrm{\&kappa;x}}_{sp}\left(t\right)\left({\mathrm{\&lambda;}}_s\right(t)-v_s(t)-{\mathrm{\&mu;}}_p(t\left)\right)$

If x _sp(t)=0, then $x_{sp}(t+1)=max\{0,(1-\mathrm{\&theta;})x_{sp}\left(t\right)+\mathrm{\&theta;}{\tilde{x}}_{sp}\left(t\right)+{\mathrm{\&kappa;x}}_{sp}\left(t\right)\left({\mathrm{\&lambda;}}_s\right(t)-v_s(t)-{\mathrm{\&mu;}}_p(t\left)\right)\}.$

4. according to claim 1ly a kind ofly to it is characterized in that: in step 4 based on flow control methods in the P2P file sharing network of price mechanism, the following algorithm of link utilization of uploading of resource provider p upgrades its new price μ collected _p(t+1)

${\mathrm{\&mu;}}_p(t+1)={\left({\mathrm{\&mu;}}_p\right(t)+\mathrm{\&tau;}(z_p\left(t\right)-C_p^u\left)\right)}_{{\mathrm{\&mu;}}_p\left(t\right)}^{+},$ $z_p\left(t\right)=\underset{s:s\&Element;S\left(p\right)}{\&Sigma;}{x}_{sp}\left(t\right)$

In formula, p is resource provider; S is resource requestor; P is resource provider set; S is resource requestor set; μ _p(t) be resource provider p upload the price that link initialization collects; x _spt () is resource provider p is the initialized downloading rate of each resource requestor s; S (p) accepts the resource requestor set that resource provider p provides file download service; be resource provider p upload link bandwidth; z _pt () is the link of uploading of resource provider p is the initialized total uploading rate of resource requestor; τ is algorithm iteration step-length, and τ >0;

${\mathrm{\&mu;}}_p(t+1)={\left({\mathrm{\&mu;}}_p\right(t)+\mathrm{\&tau;}(z_p\left(t\right)-C_p^u\left)\right)}_{{\mathrm{\&mu;}}_p\left(t\right)}^{+}$ Mean,

If μ _p(t) >0, then ${\mathrm{\&mu;}}_p(t+1)={\mathrm{\&mu;}}_p\left(t\right)+\mathrm{\&tau;}\left(z_p\right(t)-C_p^u)$

If μ _p(t)=0, then ${\mathrm{\&mu;}}_p(t+1)=max\{0,{\mathrm{\&mu;}}_p\left(t\right)+\mathrm{\&tau;}\left(z_p\right(t)-C_p^u)\}.$

The download link of resource requestor s utilizes following algorithm to upgrade its new price v collected _s(t+1)

$v_s(t+1)={\left(v_s\right(t)+\mathrm{\&tau;}(y_s\left(t\right)-C_s^d\left)\right)}_{v_s\left(t\right)}^{+},$ $y_s\left(t\right)=\underset{p:p\&Element;P\left(s\right)}{\&Sigma;}{x}_{sp}\left(t\right)$

In formula, p is resource provider; S is resource requestor; P is resource provider set; S is resource requestor set; x _spt () is resource provider p is the initialized downloading rate of each resource requestor s; v _st () is the price that the download link initialization of resource requestor s is collected; y _st () is the total flow that resource requestor s obtains; P (s) is for resource requestor s provides the resource provider set of file download service; it is the download link bandwidth of resource requestor s;

$v_s(t+1)={\left(v_s\right(t)+\mathrm{\&tau;}(y_s\left(t\right)-C_s^d\left)\right)}_{v_s\left(t\right)}^{+}$ Mean,

If v _s(t) >0, then $v_s(t+1)=v_s\left(t\right)+\mathrm{\&tau;}\left(y_s\right(t)-C_s^d)$

If v _s(t)=0, then $v_s(t+1)=max\{0,v_s\left(t\right)+\mathrm{\&tau;}\left(y_s\right(t)-C_s^d)\}.$

5. according to claim 1 a kind of based on flow control methods in the P2P file sharing network of price mechanism, it is characterized in that: in step 5 and 6, the optimum point that algorithm iteration arrives is exactly the globe optimum of flow control Optimized model, and namely in peer-to-peer network, resource provider p is the optimal flux that each resource requestor s distributes.