CN104579454A - Multi-objective optimization satellite flow control method based on software defined network - Google Patents

Abstract

The invention discloses a multi-objective optimization satellite flow control method based on a software defined network. Due to the fact that a multi-objective optimization framework is introduced, fitting calculation is conducted on a flow with multiple QoS requirements in real time through a genetic algorithm so as to obtain a plurality of non-inferior solution sets, the solution sets are transmitted to a controller, appropriate flow tables are generated for corresponding switching equipment, and corresponding logical links are established. The method has the advantages that by the adoption of centralized control, compared with a traditional network distributed decision system, routing convergence is faster, and the method can adapt to a satellite network changing quickly; foreseeable geographical location information and link bandwidth information of the satellite network can be introduced into a multi-objective optimization system, and routing efficiency can be effectively improved; route planning is conducted by adopting the multi-objective optimization algorithm, complex QoS requirements can be met, and the probability that no route can be found under a complex QoS circumstance is reduced.

Description

Based on the satellite streams control method of the multiple-objection optimization of software defined network

Technical field

The present invention relates to Satellite Networking technical problem, particularly relate to a kind of satellite streams control method of the multiple-objection optimization based on software defined network.

Background technology

Satellite communication system has broad covered area, networking flexibility, the advantage such as easy to use, can be the communication service such as audio frequency and video, data that Global Subscriber provides Large Copacity, remote and maneuverability; Meanwhile, satellite communication can provide not by the communication service of geographical environment, weather conditions restriction, is convenient to build without interrupting Global coverage mobile communications network.Therefore, realize the observing and controlling in global range, navigation and group-net communication by satellite in orbit, one of important research direction having become next generation network.2006, former Commission of Science, Technology and Industry for National Defence deputy director, manned astro-engineering vice president commander, Aero-Space institute of incumbent Zhejiang University president, space flight measurement and control expert Shen Rongjun academician proposed the conception of building Incorporate network.2012, succeed in sending up lift-off along with " sky chain 03 star ", Chinese first generation relay satellite system has carried out global group net operation by three satellites, and Xiang Zhong, low orbit satellite and manned vehicle provide data retransmission and relay services.On December 27th, 2012, " Big Dipper " satellite navigation system be made up of 14 satellites formally opens to the Asian-Pacific area, and this system has tentatively possessed the short message function of network service.Along with the development of China Aerospace cause, satellite launch is further frequent, and satellite in orbit quantity gets more and more, and between star, the research of networking technology is extremely urgent.

The concept standard OpenFlow of software defined network combined proposition in 2008 by colleges and universities such as Stanford University, University of Washington, Massachusetts Polytechnics, University of California Berkeley, Princeton University, Washington University in St Louis.The support by increasing in common Ethernet switch OpenFlow agreement is wished in initial stage software defined network plan, to realize traditional network switch control plane to be separated with datum plane, experimental network protocol testing function [1,2] is supported in the small scale network of campus.According to OpenFlow 1.0 standard, typical OpenFlow switch comprises an inner stream table and standardized external interface increases or deletes stream list item.Meanwhile, OpenFlow switch is connected with controller by the escape way of an encryption, receives the order of self-controller, reports the ruuning situation [3] of switch to controller simultaneously.Along with the development of SDN technology, in OpenFlow 1.3 version, add multithread table serial process mechanism to tackle complicated current control demand [4].

List of references:

[1]N.McKeown,T.Anderson,et al.,OpenFlow:Enabling Innovation in CampusNetworks,ACM SIGCOMM Computer Communication Review,38(2):69-74,2008.

[2]N.Gude,T.Koponen and et al.,NOX:towards an operating system for networks,ACM SIGCOMM Computer Communication Review,38(3):105-110,2008

[3]OpenFlow Switch Specification,version 1.0.0.[EB/OL],http://www.openflow.org/documents/openflow-spec-v1.0.0.pdf,2014/02/21.

[4]OpenFlow Switch Specification,version 1.3.0.[EB/OL],https://www.opennetworking.org/images/stories/downloads/specification/openflow-spec-v1.3.0.pdf,2014/02/21.

Summary of the invention

The object of this invention is to provide a kind of satellite streams control method of the multiple-objection optimization based on software defined network.

A kind of step of satellite streams control method of the multiple-objection optimization based on software defined network is as follows:

1) LEO passing of satelline GEO or MEO satellite notice its concrete latitude and longitude information of controller;

2) controller is according to the advertised information of each LEO satellite, forms satellite link topological structure in real time;

3) ICBM SHF satellite terminal sends packet by LEO satellite to object node;

4) LEO satellite inspection self stream table, when the destination of satellite link exists this locality, is then forwarded to down hop link by packet; When the destination of satellite link does not exist this locality, then packet is passed through GEO or MEO satellite transmission to controller;

5) source address of controller record data bag, and check that whether the destination address of packet is known, when destination address is unknown, then abandon packet; When destination address is known, then transmit source address and destination address and present satellites topology to multiple-objection optimization device, plan non-minor path by genetic algorithm.

Described genetic algorithm plans that the method for non-minor path is as follows:

First be defined as follows: for two aritrary decision variable x _a, x _b∈ X _f, X _ffor feasible solution set,

1) and if only if $\&ForAll;i=\{\mathrm{1,2},...,k\}:f_i\left(x_A\right)<f_i$ Time, claim A to be dominant in B;

2) and if only if $\&ForAll;i=\{\mathrm{1,2},...,k\}:f_i\left(x_A\right)\&le;f_i$ And $\&Exists;i=\{\mathrm{1,2},...,k\}:f_i\left(x_A\right)<f_i$ Time, claim that A is weak is dominant in B;

3) and if only if, and A is not dominant in B, and B is not dominant when A, claims A and B indifference.

Genetic algorithm mainly comprises hierarchical algorithm and crowding comparison algorithm.

Described hierarchical algorithm is:

1) i=1 is established;

2) for all j=1,2, n and j ≠ i, according to the more individual x of above definition _iwith individual x _jbetween domination and non-dominant relation;

3) if there is no any one individual x _jbe better than x _i, then x _ibe labeled as non-dominant individuality;

4) make i=i+1, forward step (2) to, until find all non-dominant individual;

The non-dominant individuality collection obtained by hierarchical algorithm is the first order non-dominant layer of population, then, ignore these non-dominant marked individual, namely these individualities no longer carry out next round and compare, follow procedures 1 again)-step 4), will obtain second level non-dominant layer, the rest may be inferred, until whole population is layered;

Described crowding comparison algorithm is:

The crowding i of each point _dbe set to 0;

For each target, non-dominated ranking is carried out to population, make two of border individual crowdings be infinite, i.e. o _d=l _d=∞;

Other individualities are carried out to the calculating of crowding:

$i_d=\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}\left(\right|{f_j}^{i+1}-{f_j}^{i-1}\left|\right)$

Wherein, i _drepresent the crowding of i point, f _j ⁱ⁺¹represent a jth target function value of i+1 point, f _j ^i-1represent a jth target function value of i-1 point;

Crowding refers to the density of surrounding's individuality of set point in population, and startup crowding comparison algorithm is calculated its crowding to each individuality and sorts by multiple-objection optimization device completing hierarchical algorithm after at every turn;

Every layer is sorted by multiple-objection optimization device successively, sort by crowding with in layer, the half that superseded fitness is lower, after the mode of half higher for fitness being carried out the variation of bit reversal and step-by-step hybridization produces the new individuality of same number, add in the initial population of next round optimization calculating, start to start next round optimization to calculate, when multiple-objection optimization device cannot produce path in the step number of acquiescence, then abandon this packet; When multiple-objection optimization device creates corresponding path in the step number of acquiescence, then notification controller upgrades the stream table of junction associated, forms logical links.

The beneficial effect that the present invention compared with prior art has:

1, the advantage that the invention discloses the satellite streams control method of the multiple-objection optimization based on software defined network is: multiple-objection optimization ground centralized planning routed path, can process complicated QoS demand.

QoS algorithm common at present adopts the processing mode of better simply multi-stage stepwise usually, process complicated QoS demand time possibly cannot obtain optimal path and easily failure.And multi-objective optimization algorithm of the present invention carries out centralized planning routed path, the intelligent algorithms such as genetic algorithm can be adopted to carry out complicated QoS planning.Even if there is mutual afoul demand in the QoS of complexity, also suitable Noninferior Solution Set can be found.Therefore, centralized multi-objective optimization algorithm is adopted to contribute to processing day also complicated route QoS demand.

2, the advantage that the invention discloses the satellite streams control method of the multiple-objection optimization based on software defined network is: merged the orbit information of satellite, accelerated the route planning time.

Satellite is in high-speed cruising state, be engraved in during the satellite network of versus high motion and change, traditional network router strategy based thereon does not consider that satellite-orbit information restrains at short notice by being difficult to, and causes the route planning after restraining no longer to be applicable to the new topology network architecture of change at a high speed.Invention introduces satellite-orbit information, contribute to accelerating multi-objective optimization algorithm and search optimal path.Therefore, this method can provide quicker, real-time, suitable route planning service.

3, the advantage that the invention discloses the satellite streams control method of the multiple-objection optimization based on software defined network is: can adjust routing policy in real time after broadcasting TV programs by satellite.

Satellite streams control method of the present invention adopts software defined network technology, and the networking technology traditional relative to other, this networks trategy simplifies the decision-making capability on star.Satellite hardware only needs to understand comparatively simple path and forwards instruction, without the need to understanding complicated upper-layer protocol.The process of upper-layer protocol is processed by the control centre concentrated, and is decomposed into after simple path forwards instruction and is issued to satellite hardware.Therefore, adopt satellite network of the present invention can not modify to hardware after broadcasting TV programs by satellite and can complete the adjustment of routing policy in real time, be conducive to adapting to new route need.

Accompanying drawing explanation

Below in conjunction with drawings and Examples, the present invention will be further described.

Fig. 1 is that conventional satellite network and software definition satellite network contrast schematic diagram;

Fig. 2 is satellite network controller bay composition;

Fig. 3 is embodiment Top and result figure;

Fig. 4 is embodiment flow chart.

Embodiment

A kind of step of satellite streams control method of the multiple-objection optimization based on software defined network is as follows:

1) LEO passing of satelline GEO or MEO satellite notice its concrete latitude and longitude information of controller;

2) controller is according to the advertised information of each LEO satellite, forms satellite link topological structure in real time;

3) ICBM SHF satellite terminal sends packet by LEO satellite to object node;

4) LEO satellite inspection self stream table, when the destination of satellite link exists this locality, is then forwarded to down hop link by packet; When the destination of satellite link does not exist this locality, then packet is passed through GEO or MEO satellite transmission to controller;

5) source address of controller record data bag, and check that whether the destination address of packet is known, when destination address is unknown, then abandon packet; When destination address is known, then transmit source address and destination address and present satellites topology to multiple-objection optimization device, plan non-minor path by genetic algorithm.

Described genetic algorithm plans that the method for non-minor path is as follows:

First be defined as follows: for two aritrary decision variable x _a, x _b∈ X _f, X _ffor feasible solution set,

1) and if only if $\&ForAll;i=\{\mathrm{1,2},...,k\}:f_i\left(x_A\right)<f_i$ Time, claim A to be dominant in B;

2) and if only if $\&ForAll;i=\{\mathrm{1,2},...,k\}:f_i\left(x_A\right)\&le;f_i$ And $\&Exists;i=\{\mathrm{1,2},...,k\}:f_i\left(x_A\right)<f_i$ Time, claim that A is weak is dominant in B;

3) and if only if, and A is not dominant in B, and B is not dominant when A, claims A and B indifference.

Genetic algorithm mainly comprises hierarchical algorithm and crowding comparison algorithm.

Described hierarchical algorithm is:

1) i=1 is established;

2) for all j=1,2, n and j ≠ i, according to the more individual x of above definition _iwith individual x _jbetween domination and non-dominant relation;

3) if there is no any one individual x _jbe better than x _i, then x _ibe labeled as non-dominant individuality;

4) make i=i+1, forward step (2) to, until find all non-dominant individual;

The non-dominant individuality collection obtained by hierarchical algorithm is the first order non-dominant layer of population, then, ignore these non-dominant marked individual, namely these individualities no longer carry out next round and compare, follow procedures 1 again)-step 4), will obtain second level non-dominant layer, the rest may be inferred, until whole population is layered;

Described crowding comparison algorithm is:

The crowding i of each point _dbe set to 0;

For each target, non-dominated ranking is carried out to population, make two of border individual crowdings be infinite, i.e. o _d=l _d=∞;

Other individualities are carried out to the calculating of crowding:

$i_d=\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}\left(\right|{f_j}^{i+1}-{f_j}^{i-1}\left|\right)$

Wherein, i _drepresent the crowding of i point, f _j ⁱ⁺¹represent a jth target function value of i+1 point, f _j ^i-1represent a jth target function value of i-1 point;

Crowding refers to the density of surrounding's individuality of set point in population, and startup crowding comparison algorithm is calculated its crowding to each individuality and sorts by multiple-objection optimization device completing hierarchical algorithm after at every turn;

Every layer is sorted by multiple-objection optimization device successively, sort by crowding with in layer, the half that superseded fitness is lower, after the mode of half higher for fitness being carried out the variation of bit reversal and step-by-step hybridization produces the new individuality of same number, add in the initial population of next round optimization calculating, start to start next round optimization to calculate, when multiple-objection optimization device cannot produce path in the step number of acquiescence, then abandon this packet; When multiple-objection optimization device creates corresponding path in the step number of acquiescence, then notification controller upgrades the stream table of junction associated, forms logical links.

Embodiment

As shown in Figure 3,4, the step based on the satellite streams control method of the multiple-objection optimization of software defined network is as follows:

1.S0-S2, S11-S13, S22-S24, S33-S35 are part LEO satellite, and their real-time position information passes through GEO satellite transmission not shown in FIG. to ground controller;

2. controller is according to the advertised information of each LEO satellite, and form satellite link topological structure in real time, Fig. 4 is a part for topological diagram;

3. ICBM SHF satellite terminal H2 sends ping packet by LEO satellite S24 to object node H1;

4.S24 satellite inspection self stream table, there is not this locality in the destination H1 of this link, then this packet is passed through GEO satellite transmission to controller;

5. the source address of this packet of controller record, and check that whether the destination address of packet is known.Destination address H1 is unknown, abandons this packet.

6. subsequent time, ICBM SHF satellite terminal H1 sends ping packet by S0 satellite to object node H2;

7.S0 satellite inspection self stream table, there is not this locality in the destination H2 of this link, then this packet is passed through GEO satellite transmission to controller;

8. the source address of this packet of controller record, and check that whether the destination address of packet is known.Destination address is known, then transmission source address H1 and destination address H2 and present satellites topology give multiple-objection optimization device, plan non-minor path by the genetic algorithm described in claim 2, as S0-S11-S12-S23-S24, and this path is reached down on relevant all satellites successively;

The packet that 9.H1 sends transmits to H2 along S0-S11-S12-S23-S24, and vice versa.

Claims (4)

1., based on a satellite streams control method for the multiple-objection optimization of software defined network, it is characterized in that its step is as follows:

1) LEO passing of satelline GEO or MEO satellite notice its concrete latitude and longitude information of controller;

2) controller is according to the advertised information of each LEO satellite, forms satellite link topological structure in real time;

3) ICBM SHF satellite terminal sends packet by LEO satellite to object node;

4) LEO satellite inspection self stream table, when the destination of satellite link exists this locality, is then forwarded to down hop link by packet; When the destination of satellite link does not exist this locality, then packet is passed through GEO or MEO satellite transmission to controller;

5) source address of controller record data bag, and check that whether the destination address of packet is known, when destination address is unknown, then abandon packet; When destination address is known, then transmit source address and destination address and present satellites topology to multiple-objection optimization device, plan non-minor path by genetic algorithm.

2. the satellite streams control method of a kind of multiple-objection optimization based on software defined network as claimed in claim 1, is characterized in that described genetic algorithm plans that the method for non-minor path is as follows:

First be defined as follows: for two aritrary decision variable x _a, x _b∈ X _f, X _ffor feasible solution set,

1) and if only if time, claim A to be dominant in B;

2) and if only if and time, claim that A is weak is dominant in B;

3) and if only if, and A is not dominant in B, and B is not dominant when A, and claim A and B indifference, genetic algorithm mainly comprises hierarchical algorithm and crowding comparison algorithm.

3. the satellite streams control method of a kind of multiple-objection optimization based on software defined network as claimed in claim 2, is characterized in that described hierarchical algorithm is:

1) i=1 is established;

2) for all j=1,2, n and j ≠ i, according to the more individual x of above definition _iwith individual x _jbetween domination and non-dominant relation;

3) if there is no any one individual x _jbe better than x _i, then x _ibe labeled as non-dominant individuality;

4) make i=i+1, forward step (2) to, until find all non-dominant individual;

The non-dominant individuality collection obtained by hierarchical algorithm is the first order non-dominant layer of population, then, ignore these non-dominant marked individual, namely these individualities no longer carry out next round and compare, follow procedures 1 again)-step 4), will obtain second level non-dominant layer, the rest may be inferred, until whole population is layered.

4. the satellite streams control method of a kind of multiple-objection optimization based on software defined network as claimed in claim 2, is characterized in that described crowding comparison algorithm is:

The crowding i of each point _dbe set to 0;

For each target, non-dominated ranking is carried out to population, make two of border individual crowdings be infinite, i.e. o _d=l _d=∞;

Other individualities are carried out to the calculating of crowding:

Wherein, i _drepresent the crowding of i point, represent a jth target function value of i+1 point, represent a jth target function value of i-1 point;

Crowding refers to the density of surrounding's individuality of set point in population, and startup crowding comparison algorithm is calculated its crowding to each individuality and sorts by multiple-objection optimization device completing hierarchical algorithm after at every turn;

Every layer is sorted by multiple-objection optimization device successively, sort by crowding with in layer, the half that superseded fitness is lower, after the mode of half higher for fitness being carried out the variation of bit reversal and step-by-step hybridization produces the new individuality of same number, add in the initial population of next round optimization calculating, start to start next round optimization to calculate, when multiple-objection optimization device cannot produce path in the step number of acquiescence, then abandon this packet; When multiple-objection optimization device creates corresponding path in the step number of acquiescence, then notification controller upgrades the stream table of junction associated, forms logical links.