CN103841000A - Virtual link mapping algorithm based on minimum cost flow model - Google Patents

Abstract

The invention discloses a virtual link mapping algorithm based on a minimum cost flow model. The virtual link mapping algorithm based on the minimum cost flow model sequentially comprises the following steps that the minimum cost flow model and a virtual network multilink mapping algorithm are established. According to the minimum cost flow model and the algorithm of virtual network multilink mapping, a cost parameter of unit network flow is set, a bottom layer network link is mapped through a single virtual link with the minimum cost and the minimum load, system benefits and the virtual network receiving rate can be improved, and the time complexity is reduced.

Description

A kind of virtual link mapping algorithm based on least cost flow model

[technical field]

The present invention relates to the technical field of virtual link mapping algorithm, particularly the technical field of the virtual link mapping algorithm based on least cost flow model.

[background technology]

Network virtualization is the important technology of Future Internet, cloud computing and software defined network.Multiple virtual networks can be shared same bottom physical network resource.Intel Virtualization Technology is cut apart, integration networks infrastructure resources, makes in existing network situation, to dispose the new network architecture, agreement and application and become possibility not affecting.

Along with the development of network virtualization technology, multipath virtual link maps becomes the important technology of network virtualization.Virtual network link maps based on many Commodity Flows consumes minimum mode with bottom-layer network aggregate resource and shines upon virtual link, obtains good system benefit.But that multipath link maps algorithm time complexity based on many Commodity Flows is affected by virtual network and bottom-layer network scale is larger, be difficult to meet the requirement of real-time of online virtual network mapping; And virtual network request is a dynamic changing process.Therefore study virtual network mapping dynamic process, design effective virtual network multipath link maps algorithm significant for the real-time that ensures online virtual network mapping.

[summary of the invention]

Object of the present invention solves the problems of the prior art exactly, a kind of virtual link mapping algorithm based on least cost flow model is proposed, find the dynamic phenomena of inversion of virtual network mapping cost income, the minimum cost flow model and algorithm of virtual network multipath link maps is proposed, the cost parameters of setting unit network traffics, make wall scroll virtual link with minimum cost, minimum load mapping bottom-layer network link, can improve system benefit, virtual network acceptance rate, and reduce time complexity.

For achieving the above object, the present invention proposes a kind of virtual link mapping algorithm based on least cost flow model, comprise the following steps successively:

A) set up least cost flow model:

A1) undirected network: undirected network NG=(V, A, C), wherein V is node set, A is nonoriented edge set, and C is edge capacity set, for every limit (i, j) ∈ A, to there being an edge capacity c (i, j)>=0, edge capacity is abbreviated as c _ij;

A2) undirected network flow and flow: in NG, specifying 1 s is source point, and specifying another t is meeting point, and remaining point is called intermediate point,

the f that meets following condition is called the undirected network flow of s to t:

(1) capacity limit condition: $\∀(i,j)\∈A,0\≤f_{\mathrm{ij}}\≤c_{\mathrm{ij}};$

(2) direction condition:

feasible flow has directivity;

(3) equilibrium condition: for intermediate point, discharge equals influx, each i (i ≠ s, t), has $\underset{(i,i)\∈A}{\mathrm{\Σ}}{f}_{\mathrm{ij}}=0;$ For s and t, $\underset{(s,p)\∈A}{\mathrm{\Σ}}{f}_{\mathrm{sp}}=\underset{(q,t)\∈A}{\mathrm{\Σ}}{f}_{\mathrm{qt}}=v\left(f\right),$ V (f) is called the flow of f;

A3) least cost flow model: given NG, on each limit (i, j) ∈ A, expense b (i, j)>=0 of a given unit discharge, notes by abridging as b _ij; Be exactly the given s of minimum cost flow problem, t and flow m, the stream f obtaining from s to t meets flow v (f)=m, total and make stream transportation expenses

get minimum value, obtain least cost flow model: $\mathrm{Minimize}:z(s,t)=\underset{(i,j)\∈A}{\mathrm{\Σ}}{b}_{\mathrm{ij}}{f}_{\mathrm{ij}};$

B) virtual network multilink mapping algorithm:

B1) input virtual network VN and bottom-layer network SN, calls least cost flow model and finds the minimum cost flow mapping of a virtual link, until complete all virtual link mappings;

B2) find a unmapped virtual link lv, take out both link ends point v1, v2 and link flow vbw;

B3) find out two bottom node s1 and the s2 that lv shines upon;

B4) create undirected network NG=(V, A, C), the unit discharge expense on every limit is set;

B5) call least cost flow model, if find the minimum cost flow f that s1 is vbw to s2 bandwidth traffic, distribute f to lv;

B6) output virtual network link maps result.

As preferably, described step b1) in virtual network VN formed by dummy node and virtual link, bottom-layer network SN is made up of bottom layer node and bottom link, virtual network mapping refers to dummy node and link maps, to the bottom layer node and the link that meet virtual resource demand, to be divided into node mapping and link maps.

As preferably, described step b4) in different NG parameter and unit discharge expenses, produce different mapping algorithms, be respectively: UNMCF-S and UMMCF-L; NG is set described UNMCF-S and unit discharge expense is as follows: capacity c (i, j)=bw (i, j), if c (i, j)==0 represents that node i and j do not exist link; If bw (i, j) > 0, unit discharge expense b (i, j)=1; If bw (i, j)==0, unit discharge expense b (i, j)=0; NG is set UMMCF-L and unit discharge expense is as follows: capacity c (i, j)=bw (i, j), if c (i, j)==0 represents that node i and j do not exist link; B (i, j)=bwl (i, j), wherein bw (i, j) represents the remaining total amount of bandwidth of bottom link l (i, j), bwl (i, j) represents the total amount of bandwidth that bottom link l (i, j) has shone upon.

Beneficial effect of the present invention: the present invention is by proposing the minimum cost flow model and algorithm of virtual network multipath link maps, the cost parameters of setting unit network traffics, make wall scroll virtual link with minimum cost, minimum load mapping bottom-layer network link, can improve system benefit, virtual network acceptance rate, and reduce time complexity.

A kind of virtual link mapping algorithm based on least cost flow model of the present invention, comprises the following steps successively:

A) set up least cost flow model:

A1) undirected network: undirected network NG=(V, A, C), wherein V is node set, A is nonoriented edge set, and C is edge capacity set, for every limit (i, j) ∈ A, to there being an edge capacity c (i, j)>=0, edge capacity is abbreviated as c _ij;

A2) undirected network flow and flow: in NG, specifying 1 s is source point, and specifying another t is meeting point, and remaining point is called intermediate point,

the f that meets following condition is called the undirected network flow of s to t:

(1) capacity limit condition: $\∀(i,j)\∈A,0\≤f_{\mathrm{ij}}\≤c_{\mathrm{ij}};$

(2) direction condition:

feasible flow has directivity;

(3) equilibrium condition: for intermediate point, discharge equals influx, each i (i ≠ s, t), has $\underset{(i,i)\∈A}{\mathrm{\Σ}}{f}_{\mathrm{ij}}=0;$ For s and t, $\underset{(s,p)\∈A}{\mathrm{\Σ}}{f}_{\mathrm{sp}}=\underset{(q,t)\∈A}{\mathrm{\Σ}}{f}_{\mathrm{qt}}=v\left(f\right),$ V (f) is called the flow of f;

A3) least cost flow model: given NG, on each limit (i, j) ∈ A, expense b (i, j)>=0 of a given unit discharge, notes by abridging as b _ij; Be exactly the given s of minimum cost flow problem, t and flow m, the stream f obtaining from s to t meets flow v (f)=m, total and make stream transportation expenses

get minimum value, obtain least cost flow model: $\mathrm{Minimize}:z(s,t)=\underset{(i,j)\∈A}{\mathrm{\Σ}}{b}_{\mathrm{ij}}{f}_{\mathrm{ij}};$

B) virtual network multilink mapping algorithm:

B1) input virtual network VN and bottom-layer network SN, calls least cost flow model and finds the minimum cost flow mapping of a virtual link, until complete all virtual link mappings;

B2) find a unmapped virtual link lv, take out both link ends point v1, v2 and link flow vbw;

B3) find out two bottom node s1 and the s2 that lv shines upon;

B4) create undirected network NG=(V, A, C), the unit discharge expense on every limit is set;

B5) call least cost flow model, if find the minimum cost flow f that s1 is vbw to s2 bandwidth traffic, distribute f to lv;

B6) output virtual network link maps result.

Described step b1) in virtual network VN formed by dummy node and virtual link, bottom-layer network SN is made up of bottom layer node and bottom link, virtual network mapping refers to dummy node and link maps to the bottom layer node and the link that meet virtual resource demand, be divided into node mapping and link maps, described step b4) in different NG parameter and unit discharge expenses, produce different mapping algorithms, be respectively: UNMCF-S and UMMCF-L; NG is set described UNMCF-S and unit discharge expense is as follows: capacity c (i, j)=bw (i, j), if c (i, j)==0 represents that node i and j do not exist link; If bw (i, j) > 0, unit discharge expense b (i, j)=1; If bw (i, j)==0, unit discharge expense b (i, j)=0; NG is set UMMCF-L and unit discharge expense is as follows: capacity c (i, j)=bw (i, j), if c (i, j)==0 represents that node i and j do not exist link; B (i, j)=bwl (i, j), wherein bw (i, j) represents the remaining total amount of bandwidth of bottom link l (i, j), bwl (i, j) represents the total amount of bandwidth that bottom link l (i, j) has shone upon.

Virtual link mapping algorithm based on least cost flow model has improved system benefit and virtual network acceptance rate, UNMCF-S, UNMCF-L and MCF comparison, and system benefit and the virtual network acceptance rate of algorithm that the present invention carries are improved.For example, in 1000 virtual networks of mapping, the system benefit of MCF algorithm is 10.73951, and UNMCF-S algorithm brings up to 11.69141.The virtual network acceptance rate of MCF algorithm is 0.30531 simultaneously, and the virtual network acceptance rate of UNMCF-S and UNMCF-L algorithm is 0.40, has improved 10% compared with MCF.This is that algorithm that the present invention carries can shine upon more small-scale virtual network, improves virtual network acceptance rate, thereby has improved system benefit due to virtual network mapping dynamic.

Above-described embodiment is to explanation of the present invention, is not limitation of the invention, any scheme after simple transformation of the present invention is all belonged to protection scope of the present invention.

Claims (3)

1. the virtual link mapping algorithm based on least cost flow model, comprises the following steps successively:

A) set up least cost flow model:

A1) undirected network: undirected network NG=(V, A, C), wherein V is node set, A is nonoriented edge set, and C is edge capacity set, for every limit (i, j) ∈ A, to there being an edge capacity c (i, j)>=0, edge capacity is abbreviated as c _ij;

A2) undirected network flow and flow: in NG, specifying 1 s is source point, and specifying another t is meeting point, and remaining point is called intermediate point,

the f that meets following condition is called the undirected network flow of s to t:

(1) capacity limit condition: $\∀(i,j)\∈A,0\≤f_{\mathrm{ij}}\≤c_{\mathrm{ij}};$

(2) direction condition:

feasible flow has directivity;

(3) equilibrium condition: for intermediate point, discharge equals influx, each i (i ≠ s, t), has

$\underset{(i,i)\∈A}{\mathrm{\Σ}}{f}_{\mathrm{ij}}=0;$ For s and t, $\underset{(s,p)\∈A}{\mathrm{\Σ}}{f}_{\mathrm{sp}}=\underset{(q,t)\∈A}{\mathrm{\Σ}}{f}_{\mathrm{qt}}=v\left(f\right),$ V (f) is called the flow of f;

A3) least cost flow model: given NG, on each limit (i, j) ∈ A, expense b (i, j)>=0 of a given unit discharge, notes by abridging as b _ij; Be exactly the given s of minimum cost flow problem, t and flow m, the stream f obtaining from s to t meets flow v (f)=m, total and make stream transportation expenses

get minimum value, obtain least cost flow model: $\mathrm{Minimize}:z(s,t)=\underset{(i,j)\∈A}{\mathrm{\Σ}}{b}_{\mathrm{ij}}{f}_{\mathrm{ij}};$

B) virtual network multilink mapping algorithm:

B1) input virtual network VN and bottom-layer network SN, calls least cost flow model and finds the minimum cost flow mapping of a virtual link, until complete all virtual link mappings;

B2) find a unmapped virtual link lv, take out both link ends point v1, v2 and link flow vbw;

B3) find out two bottom node s1 and the s2 that lv shines upon;

B4) create undirected network NG=(V, A, C), the unit discharge expense on every limit is set;

B5) call least cost flow model, if find the minimum cost flow f that s1 is vbw to s2 bandwidth traffic, distribute f to lv;

B6) output virtual network link maps result.

2. a kind of virtual link mapping algorithm based on least cost flow model as claimed in claim 1, it is characterized in that: described step b1) in virtual network VN formed by dummy node and virtual link, bottom-layer network SN is made up of bottom layer node and bottom link, virtual network mapping refers to dummy node and link maps, to the bottom layer node and the link that meet virtual resource demand, to be divided into node mapping and link maps.

3. a kind of virtual link mapping algorithm based on least cost flow model as claimed in claim 1, is characterized in that: described step b4) in different NG parameter and unit discharge expenses, produce different mapping algorithms, be respectively: UNMCF-S and UMMCF-L; NG is set described UNMCF-S and unit discharge expense is as follows: capacity c (i, j)=bw (i, j), if c (i, j)==0 represents that node i and j do not exist link; If bw (i, j) > 0, unit discharge expense b (i, j)=1; If bw (i, j)==0, unit discharge expense b (i, j)=0; NG is set UMMCF-L and unit discharge expense is as follows: capacity c (i, j)=bw (i, j), if c (i, j)==0 represents that node i and j do not exist link; B (i, j)=bwl (i, j), wherein bw (i, j) represents the remaining total amount of bandwidth of bottom link l (i, j), bwl (i, j) represents the total amount of bandwidth that bottom link l (i, j) has shone upon.