CN106059947B - A kind of method of dynamic optimization overlay network performance - Google Patents
A kind of method of dynamic optimization overlay network performance Download PDFInfo
- Publication number
- CN106059947B CN106059947B CN201610365510.8A CN201610365510A CN106059947B CN 106059947 B CN106059947 B CN 106059947B CN 201610365510 A CN201610365510 A CN 201610365510A CN 106059947 B CN106059947 B CN 106059947B
- Authority
- CN
- China
- Prior art keywords
- matrix
- physical
- network
- covering
- overlay
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/10—Flow control; Congestion control
- H04L47/12—Avoiding congestion; Recovering from congestion
- H04L47/125—Avoiding congestion; Recovering from congestion by balancing the load, e.g. traffic engineering
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
A kind of method of dynamic optimization overlay network performance, including following operative step: (1) setting initial value to the flow proportional matrix X of physical network, set initial value to the assignment of traffic matrix Y of each overlay network;(2) physical network updates X;(3) each overlay network updates Y;(4) step 2 and step 3 are repeated, until X and Y reach stable convergence;Or until X is globally optimal solution;(5) each overlay network carries out coverage water according to Y;Physical network carries out Physical Network routing according to X;(6) physical network and overlay network repeat step 1 according to the uniform period of setting and arrive step 5, to realize dynamic optimization overlay network performance.Method of the invention optimizes the network routing of nerve of a covering, improves network performance and stability.
Description
Technical field
The present invention relates to a kind of methods of dynamic optimization overlay network performance, belong to technical field of the computer network, especially
It is to belong to coverage water optimisation technique field.
Background technique
Overlay network is a kind of virtual network being deployed on physical network, does not need to change network structure just on a large scale
More preferable and more reliable network service can be provided.Overlay network is by being distributed overlay node in a network and connecting their logic
Link composition, wherein overlay node is router, host or the application server in physical network, has and routes, at data
The functions such as reason and data preservation, and corresponding one or more physical link of logical links, that is, covering link.Overlay network possesses one
Cover the routing mechanism that route independently of physical network layer, can application layer according to the strategy of oneself carry out the routing of data packet with
Forwarding.In the past few years, a large amount of different types of overlay networks are by SPs (Service Providers, service provider)
It is deployed in internet, for providing different services, for example, CDN (Content DeliveryNetwork, content delivery network
Network), P2P (Peer-to-Peer, peer-to-peer network) and RON (Resilient OverlayNetwork, elastic overlay network).
Although these overlay networks improve the routing performance of conventional physical network, largely occur covering in a network
The routings such as the routing interaction between routing interaction, overlay network between cover net network and physical network mix interaction problems, this
Routing mixing interaction problems lead to unstable networks and performance loss.The reason is as follows that: on the one hand, overlay network is generally according to one
Specific performance indicator optimizes their routing, such as minimizes time delay cost, and physical network generallys use TE
(Traffic Engineering, traffic engineering) carrys out the cost of peak optimizating network, such as minimizes network congestion cost.Covering
For network according to the Target Assignment flow of themselves on logical layer, it is no longer optimal for leading to the existing TE routing policy of bottom
Strategy, then TE, which can be triggered, readjusts routing, and new routing will affect the performance of overlay network above, therefore mesh again in turn
Target is inconsistent to cause covering routing and TE to interact, and then affects the stability and optimality of overall network.Another party
Face, when multiple overlay networks are deployed on the same physical network simultaneously, their overlay path may be overlapped, because
A plurality of overlay path may use simultaneously arrived same physical link, some overlay network optimize its routing when may
The performance of other overlay networks is impacted, therefore the competition of Internet resources results in the interaction between overlay network, in turn
Also the stability and optimality of overlay network are affected.
There are problems that largely routing the optimization of network performance under mixing interaction scenario becomes mesh in nerve of a covering and physical network
One, preceding Computer Network Project field technical problem urgently to be resolved.
Summary of the invention
In view of this, both having considered between each overlay network and physical network the purpose of the present invention is inventing a kind of method
Routing interaction, it is also considered that between overlay network routing interaction, be deployed in same physical network to solve multiple overlay networks
On caused routing mix interaction problems, the final performance optimization for realizing overlay network.
In order to achieve the above object, the invention proposes a kind of method of dynamic optimization overlay network performance, the methods
Including following operative step:
(1) initial value is set to the flow proportional matrix X of physical network, to the assignment of traffic matrix Y of each overlay network
Set initial value;
The flow proportional matrix X of the physical network refers in all nerve of a covering assignment of traffic to different physical pathways
The matrix that proportionality coefficient is constituted, the matrix are the matrixes of R row N column, and R indicates the item of all physical pathways in physical network
Number, N indicate the total number of the covering link of all overlay networks;The physical meaning of i-th row jth column element of the matrix refers to
Ratio in the nerve of a covering assignment of traffic of j covering link to i-th physical pathway, i are greater than oneself for being less than or equal to R equal to 1
So number, j are greater than the natural number for being less than or equal to N equal to 1;
The assignment of traffic matrix Y of the overlay network refers to that every covering flow point in the overlay network is fitted on the covering
The matrix that flow on the different overlay paths of net is constituted, the matrix are the matrixes of P row Q column, and P indicates the overlay network
Overlay path item number, Q indicate the overlay network covering stream item number;The physics of the n-th column element of m row of the matrix contains
Justice refers to that nth covering flow point is fitted on the flow on the m articles overlay path, and m is greater than the natural number for being less than or equal to P equal to 1, n
It is greater than and is equal to 1 natural number for being less than or equal to Q;
(2) according to the current value of each overlay network assignment of traffic matrix Y, physical network updates its flow proportional matrix X;
(3) according to the current value of physical network flow proportional matrix X, each overlay network updates its assignment of traffic matrix Y;
(4) step (2) and step (3) are repeated, until the flow proportional matrix X and each overlay network of physical network
Assignment of traffic matrix Y reach stable convergence, no longer change;Or until the flow proportional matrix X of physical network be it is global most
It is minimum that the selection of excellent solution, i.e. physical network executes the congestion cost that the flow proportional matrix X is obtained;
(5) each overlay network carries out coverage water according to its current assignment of traffic matrix Y;Physical network is according to it
Current flow proportional matrix X carries out Physical Network routing;
(6) physical network and overlay network repeat step (1) according to the uniform period of setting and arrive step (5), with reality
Existing dynamic optimization overlay network performance.
The particular content of the step (2) be include following operation sub-step:
(201) physical network sets its congestion metric function f (X), which is equal to the congestion cost of all physical links
The sum of, the function is specific as follows:
F (X)=δTO(L)
In above formula.
δ is that length is E, and element value is all 1 vector, and E indicates the item number of all physical links in physical network;
O (L) indicates the congestion cost function vector of all physical links, defines O (L)=(o1(l1) ..., oe(le) ...,
oE(lE))T, wherein oe(le) indicating the congestion cost function of physical link e, the function is specific as follows:
ceIndicate the capacity of physical link e;
leIndicate the flow value on physical link e;
L indicates the flow value vector of all physical links, defines L=(l1..., le..., lE)T, the calculating of L is as follows:
L=AXH
A indicates whether the physical pathway of physical network passes through the identity matrix of physical link, which is an E row R column
Matrix, E indicate physical network in all physical links item number, R indicate physical network in all physical pathways item number;It should
The physical meaning of the s row t column element of matrix refers to whether the t articles physical pathway passes through the s articles physical link, should when passing through
Element value is 1, without when the element value be 0, wherein s is greater than the natural number for being less than or equal to E equal to 1, and t is greater than equal to 1
Natural number less than or equal to R;
The flow proportional matrix of X expression physical network;
H is the matrix of a N row 1 column, and specific formula for calculation is as follows:
In above formula,
For each nerve of a covering, B indicates whether the overlay path in the overlay network passes through covering in the overlay network
The identity matrix of lid link, matrix B are the matrixes of W row P column, and W indicates the item number of the covering link of the overlay network, P table
Show the item number of the overlay path of the overlay network;The physical meaning of the u row v column element of the matrix refers to the v articles covering road
Whether diameter passes through the u articles covering link, when passing through the element value be 1, without when the element value be 0, wherein u, which is greater than, is equal to
1 is less than or equal to the natural number of W, and v is greater than the natural number for being less than or equal to P equal to 1;
Indicate nonzero element all in the assignment of traffic matrix Y of overlay network according to its column sequence successively
Carry out the matrix that the P row 1 that column arrangement is constituted arranges;
The number order according to all nerves of a covering is indicated, all nerves of a coveringSuccessively carry out column arrangement institute's structure
At N row 1 arrange matrix;
(202) according to the calculation method of setting, physical network carries out at optimization the congestion metric function f (X)
Reason, that is, calculate so that congestion metric function f (X) obtain minimum value when flow proportional matrix X;
The particular content of the step (3) be include following operation sub-step:
(301) each overlay network sets its time delay target function g (Y), which is equal to all of the overlay network
The sum of time delay cost of covering link, the function are specific as follows:
G (Y)=(AX Ω)TD(L)
In above formula,
A indicates whether the physical pathway of physical network passes through the identity matrix of physical link;
The flow proportional matrix of X expression physical network;
Ω is the matrix of a N row 1 column, and specific formula for calculation is as follows:
In above formula
B indicates whether the overlay path of overlay network passes through the identity matrix of covering link;
Indicate nonzero element all in the assignment of traffic matrix Y of the nerve of a covering according to its column sequence according to
The secondary matrix for carrying out the P row 1 that column arrangement is constituted and arranging;
The number order according to all nerves of a covering is indicated, by the nerve of a coveringIt is extended to and matrix H size
The matrix that consistent N row 1 arranges, and the element in the matrix in addition to the nerve of a covering corresponding position is all set to zero;
D (L) indicates the time delay cost function vector of all physical links, defines D (L)=(d1(l1) ..., de(le) ...,
dE(lE))T, wherein de(le) indicating the time delay cost function of physical link e, the function is specific as follows:
de(le(the c of)=1/e-le)+p
The function is equal to the sum of queuing delay and propagation delay, ceIndicate the capacity of physical link e, 1/ (ce-le) indicate M/
M/1 queuing delay, constant p indicate propagation delay;
(302) according to the calculation method of setting, overlay network carries out optimization processing to its time delay target function g (Y), i.e.,
Calculate so that the overlay network time delay target function g (Y) obtain minimum value when assignment of traffic matrix Y.
In the step (202) or the step (302), the calculation method of the setting is genetic algorithm.
The beneficial effects of the present invention are: method of the invention has comprehensively considered the road between overlay network and physical network
By interact and overlay network between routing interaction etc. routings mixing interaction various situations, and combine physical network and covering
Different situations of the network on network status are realized and are optimized to the routing of the network of nerve of a covering, improve network performance and
Stability.
Detailed description of the invention
Fig. 1 is a kind of flow chart of the method for dynamic optimization overlay network performance proposed by the present invention.
Fig. 2 is the network diagram of the embodiment of the present invention.
Specific embodiment
To make the object, technical solutions and advantages of the present invention clearer, the present invention is made with reference to the accompanying drawing further
Detailed description.
Referring to Fig. 1, the method for introducing a kind of dynamic optimization overlay network performance proposed by the present invention, the method includes under
Column operating procedure:
(1) initial value is set to the flow proportional matrix X of physical network, to the assignment of traffic matrix Y of each overlay network
Set initial value;
The flow proportional matrix X of the physical network refers in all nerve of a covering assignment of traffic to different physical pathways
The matrix that proportionality coefficient is constituted, the matrix are the matrixes of R row N column, and R indicates the item of all physical pathways in physical network
Number, N indicate the total number of the covering link of all overlay networks (including the corresponding overlay network generated of background stream);The matrix
The physical meaning of i-th row jth column element refers in the nerve of a covering assignment of traffic to i-th physical pathway of j-th strip covering link
Ratio, i are greater than the natural number for being less than or equal to R equal to 1, and j is greater than the natural number for being less than or equal to N equal to 1;
Referring to fig. 2, what Fig. 2 was provided is the network of a physical network and three overlay networks signal in the embodiment of the present invention
Figure.
Overlay network 1 has 5 overlay node G1, E1, F1, H1, I1With 12 covering links: G1-E1, E1-I1, G1-F1, F1-
I1, G1-H1, H1-I1, E1-G1, I1-E1, F1-G1, I1-F1, H1-G1, I1-H1;Have in overlay network 1 one from overlay node G1It arrives
I1Demand be 1Mbps covering stream, by G1-E1-I1, G1-F1-I1And G1-H1-I1Three overlay path transmission;
Overlay network 2 has 4 overlay node A2, C2, F2, D2With 8 covering links: A2-C2, C2-D2, A2-F2, F2-D2,
C2-A2, D2-C2, F2-A2, D2-F2;Have in overlay network 2 one from overlay node A2To D2Demand be 2Mbps covering stream,
By A2-C2-D2And A2-F2-D2Two overlay path transmission;
In the present invention, the flow removed except the flow that overlay network generates in physical network is called background by we
Stream, in order to calculate unification, the present invention corresponds to background stream and handles at an overlay network.In the embodiment shown in Figure 2, it covers
Cover net network 3 is overlay network corresponding to background stream.Having a demand from physical node A to B in physical network is 3Mbps
Background stream;Then the corresponding overlay network 3 generated of background stream has 2 overlay node A3, B3With 2 covering links: A3-B3, B3-
A3;
Physical network has 16 links: A-B, B-C, C-D, A-E, C-E, E-D, A-G, A-F, E-I, D-I, G-E, G-F, F-
D, F-I, G-H and H-I, in order respectively marked as 1 to 16.
Table 1
What table 1 provided is the physics road of all covering links mapping in overlay network 1, overlay network 2 and overlay network 3
Diameter collection table.In the present embodiment, because of covering link E in overlay network 11-G1, I1-E1, F1-G1, I1-F1, H1-G1, I1-
H1, covering link C in overlay network 22-A2, D2-C2, F2-A2, D2-F2With the covering link B in overlay network 33-A3On do not have
Flow is covered, and the physical pathway collection that they are mapped is combined into empty (i.e. there is no path is reachable), so just not listing them in table 1
The physical pathway set of mapping.
In physical network, the flow in every covering link will bring a physical streams, the physics to physical network
The source of stream, destination node are the corresponding physical nodes of covering link both ends overlay node, and the demand of the physical streams is equal to this
Covering flow in covering link.For example, if covering link G1-E1On have the flow of 1/3Mbps, it gives physical network band
The physical streams that a demand from physical node G to E is 1/3Mbps are carried out.
In the embodiment shown in Figure 2, because of the total item of the covering link of overlay network 1, overlay network 2 and overlay network 3
Number is 22, and the corresponding physical pathway total number of 22 covering links is 15 (15 in table 1), then the flow of physical network
Scaling matrices X is 15 row, 22 column matrix.
Only having in 11 covering links in embodiment has covering flow: G1-E1, E1-I1, G1-F1, F1-I1, G1-H1, H1-I1,
A2-C2, C2-D2, A2-F2, F2-D2, A3-B3, therefore, in order respectively marked as 1 to 11, other covering links are sequentially marked for we
Numbers 12 to 22.Again because this 11 covering links have mapped 15 physical pathways: G-E, E-D-I, E-I, G-F, F-D-I, F-I,
G-H, H-I, A-B-C, C-D, C-E-D, A-F, A-G-F, F-D and A-B, therefore we are also in order respectively marked as 1 to 15.
Therefore, the flow proportional matrix X of physical network can be expressed as
Wherein xijIndicate the ratio in the nerve of a covering assignment of traffic to i-th physical pathway of j-th strip covering link, x11=
1, x22+x32=1, x43=1, x54+x64=1, x75=1, x86=1, x97=1, x10,8+x11,8=1, x12,9+x13,9=1, x14,10
=1, x15,11=1, indicate that physics flow point is fitted on the sum of flow proportional in all physical pathways equal to 1.Because of G1To E1、G1It arrives
F1、G1To H1、H1To I1、A2To C2、F2To D2、A3To B3Corresponding stream only 1, so x11, x43, x75, x86, x97, x14,10,
x15,11Value be 1.Therefore, the initial value of X can be set are as follows:
The assignment of traffic matrix Y of the overlay network refers to that every covering flow point in the overlay network is fitted on the covering
The matrix that flow on the different overlay paths of net is constituted, the matrix are the matrixes of P row Q column, and P indicates the overlay network
Overlay path item number, Q indicate the overlay network covering stream item number;The physics of the n-th column element of m row of the matrix contains
Justice refers to that nth covering flow point is fitted on the flow on the m articles overlay path, and m is greater than the natural number for being less than or equal to P equal to 1, n
It is greater than and is equal to 1 natural number for being less than or equal to Q;
Referring to fig. 2, have in overlay network 1 one from overlay node G1To I1Demand be 1Mbps covering stream, by three
The transmission of overlay path, then the assignment of traffic matrix Y of overlay network 1 is the matrix of 3 rows 1 column,Wherein y11Indicating should
Covering flow point is fitted on overlay path G1-E1-I1On covering flow, y21Indicate that the covering flow point is fitted on overlay path G1-F1-I1On
Flow, y31Indicate that the covering flow point is fitted on overlay path G1-H1-I1On flow, and y11+y21+y31=1, indicate that this is covered
The sum of the covering flow that lid flow point is fitted on three overlay paths is equal to the 1Mbps demand of the covering stream.Here gives
Set initial value as
Have in overlay network 2 one from overlay node A2To D2Demand be 2Mbps covering stream, by two covering roads
Diameter transmission, then the assignment of traffic matrix Y of overlay network 2 is the matrix of 2 rows 1 column,Wherein y11Indicate the covering flow point
It is fitted on overlay path A2-C2-D2On covering flow, y21Indicate that the covering flow point is fitted on overlay path A2-F2-D2On flow,
And y11+y21=2, indicate that the sum of the covering flow that the covering flow point is fitted on two overlay paths is equal to the covering stream
2Mbps demand.Here, it givesSet initial value as
Have in overlay network 3 one from overlay node A3To B3Demand be 3Mbps covering stream, by a covering road
Diameter transmission, then the assignment of traffic matrix Y in overlay network 3 is the matrix of 1 row 1 column, Y=[y11], wherein y11Indicate the covering stream
It is assigned to overlay path A3-B3On covering flow, y11=3.Therefore, Y=[y11] initial value be Y=[3];
(2) according to the current value of each overlay network assignment of traffic matrix Y, physical network updates its flow proportional matrix X;
(3) according to the current value of physical network flow proportional matrix X, each overlay network updates its assignment of traffic matrix Y;
(4) step (2) and step (3) are repeated, until the flow proportional matrix X and each overlay network of physical network
Assignment of traffic matrix Y reach stable convergence, no longer change;Or until the flow proportional matrix X of physical network be it is global most
It is minimum that the selection of excellent solution, i.e. physical network executes the congestion cost that the flow proportional matrix X is obtained;
(5) each overlay network carries out coverage water according to its current assignment of traffic matrix Y;Physical network is according to it
Current flow proportional matrix X carries out Physical Network routing;
(6) physical network and overlay network repeat step (1) according to the uniform period of setting and arrive step (5), with reality
Existing dynamic optimization overlay network performance.
The particular content of the step (2) be include following operation sub-step:
(201) physical network sets its congestion metric function f (X), which is equal to the congestion cost of all physical links
The sum of, the function is specific as follows:
F (X)=δTO(L)
In above formula,
δ is that length is E, and element value is all 1 vector, and E indicates the item number of all physical links in physical network;
Referring to Fig. 2, physical network has 16 physical links, and all δ are the matrix of 16 rows 1 column, then δ are as follows:
δ=[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]T
O (L) indicates the congestion cost function vector of all physical links, defines O (L)=(o1(l1) ..., oe(le) ...,
oE(lE))T, wherein oe(le) indicating the congestion cost function of physical link e, the function is specific as follows:
ceIndicate the capacity of physical link e;
leIndicate the flow value on physical link e;
L indicates the flow value vector of all physical links, defines L=(l1..., le..., lE)T, the calculating of L is as follows:
L=AXH
A indicates whether the physical pathway of physical network passes through the identity matrix of physical link, which is an E row R column
Matrix, E indicate physical network in all physical links item number, R indicate physical network in all physical pathways item number;It should
The physical meaning of the s row t column element of matrix refers to whether the t articles physical pathway passes through the s articles physical link, should when passing through
Element value is 1, without when the element value be 0, wherein s is greater than the natural number for being less than or equal to E equal to 1, and t is greater than equal to 1
Natural number less than or equal to R;
Referring to Fig. 2, there are 16 physical links and 15 physical pathways in physical network, then A is the matrix of 16 rows 15 column, then
A is
The flow proportional matrix of X expression physical network;
H is the matrix of a N row 1 column, and specific formula for calculation is as follows:
In above formula,
For each nerve of a covering, B indicates whether the overlay path in the overlay network passes through covering in the overlay network
The identity matrix of lid link, matrix B are the matrixes of W row P column, and W indicates the item number of the covering link of the overlay network, P table
Show the item number of the overlay path of the overlay network;The physical meaning of the u row v column element of the matrix refers to the v articles covering road
Whether diameter passes through the u articles covering link, when passing through the element value be 1, without when the element value be 0, wherein u, which is greater than, is equal to
1 is less than or equal to the natural number of W, and v is greater than the natural number for being less than or equal to P equal to 1;
Indicate nonzero element all in the assignment of traffic matrix Y of overlay network according to its column sequence successively
Carry out the matrix that the P row 1 that column arrangement is constituted arranges;
The number order according to all nerves of a covering is indicated, all nerves of a coveringSuccessively carry out column arrangement institute's structure
At N row 1 arrange matrix;
Referring to fig. 2, there are 12 covering links and three overlay paths in overlay network 1, then the identity matrix of overlay network 1
B is the matrix of 12 rows 3 column, then B is
Because the initial value of overlay network 1 isSo according toDefinition,ThenFor
There are 8 covering links and 2 overlay paths in overlay network 2, then the identity matrix B of overlay network 2 is 8 rows 2 column
Matrix, then B be
Because the initial value of overlay network 2 isSo according toDefinition,Then
There are 1 covering link and 1 overlay path in overlay network 3, then the identity matrix B of overlay network 3 is 1 row 1 column
Matrix, then B be
B=[1]
Because the initial value of overlay network 3 is Y=[3], according toAbout define,Then
According to the formula of H, H is the matrix of 22 rows 1 column, then H is
Here, the label sequence of all covering links must keep one with the label sequence of covering links all in X in H
It causes.
(202) according to the calculation method of setting, physical network carries out at optimization the congestion metric function f (X)
Reason, that is, calculate so that congestion metric function f (X) obtain minimum value when flow proportional matrix X;
The particular content of the step (3) be include following operation sub-step:
(301) each overlay network sets its time delay target function g (Y), which is equal to all of the overlay network
The sum of time delay cost of covering link, the function are specific as follows:
G (Y)=(AX Ω)TD(L)
In above formula,
A indicates whether the physical pathway of physical network passes through the identity matrix of physical link;
The flow proportional matrix of X expression physical network;
Ω is the matrix of a N row 1 column, and specific formula for calculation is as follows:
In above formula,
B indicates whether the overlay path of overlay network passes through the identity matrix of covering link;
Indicate nonzero element all in the assignment of traffic matrix Y of overlay network according to its column sequence successively
Carry out the matrix that the P row 1 that column arrangement is constituted arranges;
The number order according to all nerves of a covering is indicated, by the nerve of a coveringIt is extended to and matrix H size
The matrix that consistent N row 1 arranges, and the element in the matrix in addition to the nerve of a covering corresponding position is all set to zero;
Overlay network 1According to's
Definition, Ω are the matrix of 22 rows 1 column, then Ω are as follows:
Overlay network 2According toDefinition, Ω 22
The matrix that row 1 arranges, then Ω are as follows:
Ω=[0 00000111100000000000 0]T
Overlay network 3According toDefinition, Ω is the matrix of 22 rows 1 column, then Ω are as follows:
Ω=[0 00000000030000000000 0]T
D (L) indicates the time delay cost function vector of all physical links, defines D (L)=(d1(l1) ..., de(le) ...,
dE(lE))T, wherein de(le) indicating the time delay cost function of physical link e, the function is specific as follows:
de(le(the c of)=1/e-le)+p
The function is equal to the sum of queuing delay and propagation delay, ceIndicate the capacity of physical link e, 1/ (ce-le) indicate M/
M/1 queuing delay, constant p indicate propagation delay (value of p generally between two end node of physical link at a distance from it is directly proportional, imitative
In true experiment, we are uniformly set as 1 second);
(302) according to the calculation method of setting, overlay network carries out optimization processing to its time delay target function g (Y), i.e.,
Calculate so that the overlay network time delay target function g (Y) obtain minimum value when assignment of traffic matrix Y.
In the step (202) or the step (302), the calculation method of the setting is genetic algorithm.
In genetic algorithm, the chromosome of overlay network is expressed as Y, and the chromosome of physical network is expressed as X, chromosome
Number is 50, and tolerable error is set as 0.001, and crossover probability is set as 0.8, and mutation probability is set as 0.5.In the initial rank of genetic algorithm
Section, initialize 50 feasible chromosomes, then successively using evaluation, selection, intersect and variation operation successively iteration until
The congestion metric function f (X) (or time delay target function g (Y) of overlay network) of physical network obtains minimum value.
Finally, when the flow proportional matrix X of physical network and the assignment of traffic matrix Y of each overlay network reach stable
As a result as follows:
Overlay network 1Overlay network 2
Inventor has carried out a large amount of emulation experiments to method of the invention, and the interaction of simulation results show mixed logic dynamic leads to net
The oscillation of network performance, after method of the invention, system has converged to a stable shape after the interaction of multiple mixed logic dynamic
State obtains the optimal performance of physical network and overlay network.
Claims (5)
1. a kind of method of dynamic optimization overlay network performance, it is characterised in that: the method includes following operative steps:
(1) initial value is set to the flow proportional matrix X of physical network, is set to the assignment of traffic matrix Y of each overlay network
Initial value;
The flow proportional matrix X of the physical network refers to the ratio in all nerve of a covering assignment of traffic to different physical pathways
The matrix that coefficient is constituted, the matrix are the matrixes of R row N column, and R indicates the item number of all physical pathways in physical network, N
Indicate the total number of the covering link of all overlay networks;The physical meaning of i-th row jth column element of the matrix refers to j-th strip
Ratio in the nerve of a covering assignment of traffic of covering link to i-th physical pathway, i are greater than the nature for being less than or equal to R equal to 1
Number, j are greater than the natural number for being less than or equal to N equal to 1;
The assignment of traffic matrix Y of the overlay network refers to that every covering flow point in the overlay network is fitted on the nerve of a covering
The matrix that flow on different overlay paths is constituted, the matrix are the matrixes of P row Q column, and P indicates covering for the overlay network
The item number in lid path, Q indicate the item number of the covering stream of the overlay network;The physical meaning of the n-th column element of m row of the matrix is
Refer to that nth covering flow point is fitted on the flow on the m articles overlay path, m is greater than the natural number for being less than or equal to P equal to 1, and n is big
In the natural number for being less than or equal to Q equal to 1;
(2) according to the current value of each overlay network assignment of traffic matrix Y, physical network updates its flow proportional matrix X;
(3) according to the current value of physical network flow proportional matrix X, each overlay network updates its assignment of traffic matrix Y;
(4) step (2) and step (3) are repeated, until the flow proportional matrix X of physical network and the stream of each overlay network
Amount allocation matrix Y reaches stable convergence, no longer changes;Or until the flow proportional matrix X of physical network is globally optimal solution,
I.e. it is minimum to execute the congestion cost that the flow proportional matrix X is obtained for physical network selection;
(5) each overlay network carries out coverage water according to its current assignment of traffic matrix Y;Physical network is current according to it
Flow proportional matrix X carry out Physical Network routing;
(6) physical network and overlay network repeat step (1) according to the uniform period of setting and arrive step (5), dynamic to realize
State optimizes overlay network performance.
2. a kind of method of dynamic optimization overlay network performance according to claim 1, it is characterised in that: the step
(2) particular content be include following operation sub-step:
(201) physical network sets its congestion metric function f (X), which is equal to the sum of the congestion cost of all physical links,
The function is specific as follows:
F (X)=δTO(L)
In above formula,
δ is that length is E, and element value is all 1 vector, and E indicates the item number of all physical links in physical network;
O (L) indicates the congestion cost function vector of all physical links, defines O (L)=(o1(l1),…,oe(le),…,oE
(lE))T, wherein oe(le) indicating the congestion cost function of physical link e, the function is specific as follows:
ccIndicate the capacity of physical link e;
leIndicate the flow value on physical link e;
L indicates the flow value vector of all physical links, defines L=(l1,…,le,…,lE)T, the calculating of L is as follows:
L=AXH
A indicates whether the physical pathway of physical network passes through the identity matrix of physical link, which is the square of E row R column
Battle array, E indicate that the item number of all physical links in physical network, R indicate the item number of all physical pathways in physical network;The matrix
The physical meaning of s row t column element refer to whether the t articles physical pathway passes through the s articles physical link, element when passing through
Value be 1, without when the element value be 0, wherein s be greater than equal to 1 be less than or equal to E natural number, t is greater than to be less than equal to 1
Natural number equal to R;
The flow proportional matrix of X expression physical network;
H is the matrix of a N row 1 column, and specific formula for calculation is as follows:
In above formula,
For each nerve of a covering, B indicates whether the overlay path in the overlay network passes through the covering chain in the overlay network
The identity matrix on road, matrix B are the matrixes of W row P column, and W indicates the item number of the covering link of the overlay network, and P is indicated should
The item number of the overlay path of overlay network;The physical meaning of the u row v column element of the matrix refers to that the v articles overlay path is
It is no pass through the u articles covering link, when passing through the element value be 1, without when the element value be 0, wherein u is greater than small equal to 1
In the natural number for being equal to W, v is greater than the natural number for being less than or equal to P equal to 1;
Expression successively carries out nonzero element all in the assignment of traffic matrix Y of overlay network according to the sequence of its column
Column arrange the matrix that constituted P row 1 arranges;
The number order according to all nerves of a covering is indicated, all nerves of a coveringSuccessively carry out the N that column arrangement is constituted
The matrix that row 1 arranges;
(202) according to the calculation method of setting, physical network carries out optimization processing to the congestion metric function f (X), i.e.,
Calculate so that congestion metric function f (X) obtain minimum value when flow proportional matrix X.
3. a kind of method of dynamic optimization overlay network performance according to claim 1 or 2, it is characterised in that: the step
Suddenly the particular content of (3) be include following operation sub-step:
(301) each overlay network sets its time delay target function g (Y), which is equal to all coverings of the overlay network
The sum of time delay cost of link, the function are specific as follows:
G (Y)=(AX Ω)TD(L)
In above formula,
A indicates whether the physical pathway of physical network passes through the identity matrix of physical link;
The flow proportional matrix of X expression physical network;
Ω is the matrix of a N row 1 column, and specific formula for calculation is as follows:
In above formula,
B indicates whether the overlay path of overlay network passes through the identity matrix of covering link;
Indicate nonzero element all in the assignment of traffic matrix Y of the nerve of a covering according to its column sequence successively into
Ranks arrange the matrix that constituted P row 1 arranges;
The number order according to all nerves of a covering is indicated, by the nerve of a coveringIt is extended in the same size with matrix H
The matrix that arranges of N row 1, and the element in the matrix in addition to the nerve of a covering corresponding position is all set to zero;
D (L) indicates the time delay cost function vector of all physical links, defines D (L)=(d1(l1),…,de(le),…,dE
(lE))T, wherein de(le) indicating the time delay cost function of physical link e, the function is specific as follows:
de(le(the c of)=1/e-le)+p
The function is equal to the sum of queuing delay and propagation delay, ceIndicate the capacity of physical link e, 1/ (ce-le) indicate M/M/1
Queuing delay, constant p indicate propagation delay;
(302) according to the calculation method of setting, overlay network carries out optimization processing to its time delay target function g (Y), that is, calculates
Out so that the overlay network time delay target function g (Y) obtain minimum value when assignment of traffic matrix Y.
4. a kind of method of dynamic optimization overlay network performance according to claim 2, it is characterised in that: the step
(202), the calculation method of the setting is genetic algorithm.
5. a kind of method of dynamic optimization overlay network performance according to claim 3, it is characterised in that: the step
(302) in, the calculation method of the setting is genetic algorithm.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610365510.8A CN106059947B (en) | 2016-05-27 | 2016-05-27 | A kind of method of dynamic optimization overlay network performance |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610365510.8A CN106059947B (en) | 2016-05-27 | 2016-05-27 | A kind of method of dynamic optimization overlay network performance |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106059947A CN106059947A (en) | 2016-10-26 |
CN106059947B true CN106059947B (en) | 2019-05-03 |
Family
ID=57176107
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610365510.8A Active CN106059947B (en) | 2016-05-27 | 2016-05-27 | A kind of method of dynamic optimization overlay network performance |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106059947B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107707379B (en) * | 2017-06-29 | 2019-04-02 | 贵州白山云科技股份有限公司 | A kind of CDN covering scheme generation method and device |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101252533A (en) * | 2008-03-26 | 2008-08-27 | 中国科学院计算技术研究所 | Covering network system and route selecting method |
CN101547347A (en) * | 2009-04-30 | 2009-09-30 | 上海大学 | Overlay network layered multicast resource optimum allocation method for scalable video stream |
CN101626336A (en) * | 2008-07-09 | 2010-01-13 | 同济大学 | Method for improving P2P overlay network interaction performance by using trust mechanism |
CN101917338A (en) * | 2010-08-13 | 2010-12-15 | 浙江大学 | Path correlation based optimal path group selecting method of P2P (peer-to-peer) overlay network |
EP1481510B1 (en) * | 2002-03-01 | 2016-04-27 | Cisco Technology, Inc. | Method and system for constraint-based traffic flow optimisation |
-
2016
- 2016-05-27 CN CN201610365510.8A patent/CN106059947B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP1481510B1 (en) * | 2002-03-01 | 2016-04-27 | Cisco Technology, Inc. | Method and system for constraint-based traffic flow optimisation |
CN101252533A (en) * | 2008-03-26 | 2008-08-27 | 中国科学院计算技术研究所 | Covering network system and route selecting method |
CN101626336A (en) * | 2008-07-09 | 2010-01-13 | 同济大学 | Method for improving P2P overlay network interaction performance by using trust mechanism |
CN101547347A (en) * | 2009-04-30 | 2009-09-30 | 上海大学 | Overlay network layered multicast resource optimum allocation method for scalable video stream |
CN101917338A (en) * | 2010-08-13 | 2010-12-15 | 浙江大学 | Path correlation based optimal path group selecting method of P2P (peer-to-peer) overlay network |
Non-Patent Citations (1)
Title |
---|
Shan Jiang;Jianxin Liao;Jun Gong;Jingyu Wang;Tonghong Li.《Competitive equilibrium and stable coalition in overlay environments》.《2015 IEEE 40th Conference on Local Computer Networks (LCN)》.2015, |
Also Published As
Publication number | Publication date |
---|---|
CN106059947A (en) | 2016-10-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN101478802B (en) | Self-organizing QoS routing method based on ant colony algorithm | |
CN108476208A (en) | Multi-path transmission designs | |
CN107888496A (en) | Multiple path computings for label switched path | |
Kumar et al. | Exploring genetic algorithm for shortest path optimization in data networks | |
CN106105130A (en) | Carry the source routing of entropy head | |
Yeung et al. | Competition for shortest paths on sparse graphs | |
Wang et al. | Virtual network mapping with traffic matrices | |
Thar et al. | Online caching and cooperative forwarding in information centric networking | |
Tajiki et al. | CECT: computationally efficient congestion-avoidance and traffic engineering in software-defined cloud data centers | |
CN101478803B (en) | Self-organizing QoS routing method based on ant colony algorithm | |
Fujiwara et al. | Skywalk: A topology for HPC networks with low-delay switches | |
CN108833285A (en) | Network moving target defence method, electronic equipment, storage medium and system | |
CN106130931A (en) | Based on multiobject route frequency gap distribution method of evolving in elastic optical network | |
CN109962774A (en) | Quantum cryptography networks key relays dynamic routing method | |
Hamdan et al. | DPLBAnt: Improved load balancing technique based on detection and rerouting of elephant flows in software-defined networks | |
Smith et al. | Target assignment for robotic networks: Asymptotic performance under limited communication | |
RU2636665C1 (en) | Method of multipath routing using data traffic flow splitting | |
CN106059947B (en) | A kind of method of dynamic optimization overlay network performance | |
CN109922161A (en) | Content distribution method, system, equipment and the medium of dynamic cloud content distributing network | |
Fajjari et al. | Cloud networking: An overview of virtual network embedding strategies | |
CN105847144B (en) | A kind of method of cooperation optimization overlay network performance | |
Alqahtani et al. | Ernie: scalable load-balanced multicast source routing for cloud data centers | |
CN110519166A (en) | A kind of safe multi-path route method for network function virtualization NFV framework | |
Wu et al. | Optimization algorithms for proxy server placement in content distribution networks | |
Thongpook | Load balancing of adaptive zone routing in ad hoc networks |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |