CN108540384A - Intelligent heavy route method and device based on congestion aware in software defined network - Google Patents
Intelligent heavy route method and device based on congestion aware in software defined network Download PDFInfo
- Publication number
- CN108540384A CN108540384A CN201810332577.0A CN201810332577A CN108540384A CN 108540384 A CN108540384 A CN 108540384A CN 201810332577 A CN201810332577 A CN 201810332577A CN 108540384 A CN108540384 A CN 108540384A
- Authority
- CN
- China
- Prior art keywords
- path
- congestion
- network
- data
- link
- 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.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/12—Shortest path evaluation
- H04L45/124—Shortest path evaluation using a combination of metrics
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/12—Shortest path evaluation
- H04L45/121—Shortest path evaluation by minimising delays
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/12—Shortest path evaluation
- H04L45/123—Evaluation of link metrics
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/12—Shortest path evaluation
- H04L45/125—Shortest path evaluation based on throughput or bandwidth
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/302—Route determination based on requested QoS
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/70—Routing based on monitoring results
Abstract
Intelligent heavy route method and device based on congestion aware in software defined network, the network monitoring module of control layer periodically collects and calculates the network state of data Layer after initialization, the congestion detection module of control layer has carried out fuzzy evaluation with duty ratio and link load change rate two indices to link, obtain the fuzzy evaluation value of current ink congestion quality, according to the average Congestion Level SPCC of all links on path, the congestion quality fuzzy evaluation value in this path is obtained;After the congestion detection module of control layer completes the congestion quality evaluation of all alternative paths, path selection module is established and is connected with network monitoring module with congestion detection module, is obtained Path selection parameter, is carried out data stream transmitting optimal route selection.The present invention provides a kind of intelligent Congestion Avoidance route selection method for the data flow in network, effectively improves performance and the QoS of customer experience of network.
Description
Technical field
The present invention relates to the communications fields, and in particular to the intelligent heavy-route based on congestion aware in a kind of software defined network
Method and apparatus.
Background technology
With continuing to bring out for the new networks such as Internet of Things, mobile Internet, system for cloud computing, stream medium data presents quick-fried
The growth of fried formula, network link congestion problems getting worse.The distributed architecture of traditional network lack to Internet resources and
The control of global information, therefore be difficult that Internet resources are efficiently utilized.Software defined network framework has broken traditional net
The control plane of centralization and distributed data surface are separated, the complete of control plane both may be implemented by the design concept of network system
Office's optimization and centralized control, and high performance forwarded ability may be implemented.Software defined network framework is mainly put down by data
Face, control plane and application plane three parts form.The equipment of data plane only has forwarding capability, does not have control function.
Control plane has the global information of bottom-layer network and central controlled ability.It is open to user using plane, for using
Network demand and research and innovation are changed in family.The centralized management and control of software defined network and programmable interface are the network management side of bringing
Just.
In recent years, network congestion problem is solved under software defined network framework have some relevant researchs.First, right
In terms of network congestion evaluation, existing evaluation method has threshold method and Fuzzy Evaluation Method based on switch port data.But
But lack the whole congestion condition evaluation to data transfer path.Secondly, the Congestion Managment master based on software defined network
It is divided into end side and network side two major classes.End side Congestion Managment is mainly the improvement to conventional TCP protocols, by right
Terminal sends window and receives the adjustment of window, the congestion condition of respite link, including SCCP agreement, SDTCP agreements etc..
But network congestion is adjusted based on end side, other a large amount of links are likely to be in idle state in network, and Internet resources obtain
Less than making full use of.Network side jamming control method is mainly to carry out heavy-route to the data on congestion path, by congested node
The data flow at place routes to the lower path of network link utilization rate, alleviates network congestion situation.It is more including ECMP method for routing
Alternate path method and congested node avoid selection method etc..Heavy route method can preferably utilize the resource of network, but again
The good and bad judgement of routed path is the difficult point that this method is faced.
With the development of artificial intelligence, it there is now research and solve Network route Problem using intensified learning method.Main packet
Include the intensified learning method for routing based on QoS of survice.Intensified learning method in real time can adjust decision according to the feedback of network,
It is more intelligent.
Generally speaking, from the current study, the problem of network congestion management and control includes mainly following two aspects is urgently
It solves:
(1) congestion aware ability
Traditional segment link congestion judgement is difficult to obtain the whole congestion effect of data transfer path, is selected for optimal path
Selecting makes troubles.
(2) congestion Restoration Mechanism
Traditional network side congestion Restoration Mechanism is short of the considerations of network resource optimization.Ignore the spy of network state variation
Point, it is excessively inflexible for the selection of backup path, and do not account for heavy-route path and bring network effect.
Invention content
The purpose of the present invention is to propose to the intelligent heavy route methods and device based on congestion aware in software defined network, should
Method can reduce the congestion path in network, improve the service quality of network.
To achieve the above object, the present invention adopts the following technical scheme that:
The intelligent heavy route method based on congestion aware, includes the following steps in software defined network:
(1) netinit;
(2) network monitor process:
After the completion of netinit, the network state of data Layer is periodically collected and calculated to the network monitoring module of control layer,
Including:Link duty ratioLink load change rateData transfer path time delay D elay, path end to end
The handling capacity TH of packet loss Loss and path, the evaluation parameter of the network state of data Layer as the path quality for calculating network;
When data stream transmitting, the network monitoring module statistical data layer data transmission path of control layer time delay end to end
The handling capacity TH of Delay, the packet loss Loss in path and path, and by data transfer path time delay D elay, path end to end
Packet loss Loss and path handling capacity TH alternately path QoS data, alternative path QoS data imitates for data transmission
The value of feedback of fruit calculates;
(3) congestion detection process:
Based on Fuzzy System Method, the congestion detection module of control layer is to link duty ratioChange with link load
RateTwo indices carry out fuzzy evaluation, obtain the fuzzy evaluation value rank of current ink congestion qualityij, according to institute on path
The average Congestion Level SPCC for having link obtains the congestion quality fuzzy evaluation value rank in this pathi;
(4) routing procedure:
After the congestion detection module of control layer completes the congestion quality evaluation of all alternative paths, path selection module with gather around
It fills in detection module and network monitoring module establishes connection, obtain Path selection parameter, carry out data stream transmitting optimal path choosing
It selects.
The present invention, which further improves, to be, initialization detailed process is as follows in step (1):User Access Layer transmission source section
Data stream transmitting is asked between point and destination node, and control layer is established with data Layer in network and connected, then the routing choosing of control layer
It is K shortest path of data-flow computation that module, which is selected, according to shortest path first, as the optional path set of steaming transfer, finally,
One shortest path selects it for newly arrived stream request control layer, and is handed down to data Layer and carries out steaming transfer.
The present invention, which further improves, to be, step (2) link duty ratioIt is calculated by formula (1):
In formula (1),For t moment link load,For t moment link total capacity;
Link load change rateIt is calculated by formula (2):
In formula,For t moment link load,For t-1 moment link loads.
The present invention, which further improves, to be, the detailed process of step (3) is as follows:
The congestion detection module of (3a) control layer is connect with network monitoring module, obtains path congestion quality evaluation parameter:
Link duty ratioWith link load change rate
(3b) is blurred:Using link with duty ratio membership to link duty ratioCarry out fuzzy evaluation;
Using link load change rate membership to link load change rateCarry out fuzzy evaluation;
(3c) fuzzy reasoning table maps:By link duty ratioWith link load change ratePass through fuzzy reasoning table
It is mapped, obtains link congestion degree RANKij;
(3d) ambiguity solution:Using being utmostly subordinate to method for defined fuzzy output value link congestion degree RANKij
Defuzzification is carried out, the evaluation of estimate rank of link congestion quality is obtainedij;
(3e) path congestion quality fuzzy evaluation:
For all path p between source node and destination nodei∈ P carry out path congestion quality evaluation, obtain on path
The average congestion quality evaluation value rank of all linksi:
In formula (3), rankiIndicate path p between a pair of of source node and destination nodeiCongestion quality evaluation value, rankij
(m) path p is indicatediThe congestion quality evaluation value for the link that upper m order passes through, N indicate path piAll links of upper process are total
Number.
The present invention, which further improves, to be, the detailed process of step (4) is as follows:
(4a) path selection module is connect with network monitoring module, obtains all alternative path QoS datas:Data transmission route
The handling capacity TH of diameter time delay D elay, the packet loss Loss in path and path end to end;
(4b) path selection module is connect with congestion detection module, obtains the congestion quality evaluation value of all alternative paths
ranki;
(4c) calculates the long-term financial value Rwd of network according to alternative path QoS data;
(4d) calculates source node to all alternative path p of this period between destination node according to intensified learning methodiRespectively
Quality evaluation value Qi(P,pi), according to this period all alternative path piRespective Quality evaluation value Qi(P,pi) more
The Q value tables of new alternative path;
(4e) is according to this period all alternative path piRespective Quality evaluation value Qi(P,pi), path selection module
Determine all alternative path p of next periodiSelected probability π (pi), next period all alternative path piSelected probability
π(pi) computational methods such as formula (6) shown in;
Wherein,
In formula (6) and (7):
π(pi) it is all alternative path p of next periodiSelected probability, K are between each pair of source node and destination node
Total number of paths;τnIt is a temperature parameter, which controls the random degree for selecting a certain path;T indicates convergence
Time;τ0And τTIndicate the final temperature of initial temperature and time T;
(4f) path selection module more new route, and flow table issuance is carried out data transmission to data Layer;Detection data stream is
The no end of transmission, return to step (2) if without the end of transmission carry out new periodic path quality evaluation and update, if transmission
End of data, then network monitor terminate.
The present invention, which further improves, to be, the long-term financial value Rwd of network is calculated by the following formula in step (4c):
Rwd=ω1·TH+ω2/Delay+ω3/Loss (5)
Wherein:ω1, ω2, ω3It is each evaluation index proportion shared in value of feedback.
The present invention, which further improves, to be, this period alternative path Quality evaluation value Qi(P,pi) pass through following formula meter
It calculates:
Qi(P,pi)←α×ranki×{Rwd+γ×maxy∈PQi(P,y)}+(1-α)×Qi(P,pi) (4)
Wherein, rankiIt is this period alternative path piCongestion quality evaluation value, α is learning rate, Qi(P,pi) it is this week
Phase alternative path piQuality evaluation value.
Intelligent heavy-route device based on congestion aware in a kind of software defined network, including User Access Layer, data Layer,
Control layer and application layer;User Access Layer is made of Web vector graphic user, for proposing data stream transmitting request;Data Layer by
Interchanger and network link composition, realize the forwarding capability of data flow;Control layer is made of software defined network controller, software
It includes network monitoring module, congestion detection module and routing selecting module to define network controller, and network monitoring module is for week
Each data transfer path end-to-end QoS data in phase property collection network link state and network, and it is transmitted to congestion detection mould
Block and routing selecting module, congestion detection module is used for can transmission path to data according to the data that network monitoring module is collected into
Congestion Level SPCC carry out fuzzy evaluation, routing selecting module be used for according to the path congestion status evaluation value of congestion detection module and
The path QoS data feedback of network monitoring module, using the comprehensive quality of intensified learning method overall merit alternative path, and with
It is routed based on the comprehensive quality of alternative path;Application layer changes network demand and research for user.
Compared with prior art, the device have the advantages that:The path congestion evaluation method of the present invention is fuzzy system
System, routing resource are intensified learning method.Using the link state of data Layer as congestion quality evaluation parameter, control layer is gathered around
It fills in detection module and carries out path congestion quality evaluation;It is to comment with path congestion quality evaluation value and the path QoS data of data Layer
Valence parameter, the path selection module of control layer carry out the Quality evaluation in path, and based on the comprehensive evaluation value with certain general
Rate selects optimal path, is handed down to data Layer and carries out forwarding data flow.The present invention first, passes through introducing illegibility systematic evaluation side
Method realizes the thoroughly evaluating to data flow transmission route Congestion Level SPCC in network, realizes perception of the network to congestion path
Ability reduces traditional congestion Restoration Mechanism to the degree of dependence of congestion detection, can have a premonition network path congestion situation
Generation, make a response in advance;And the sensitivity that network perceives congestion path is improved to a certain extent, to improve
The reliability of network.Secondly, this method utilizes intensified learning side using data flow transmission route QoS data as evaluation parameter
Method evaluates the comprehensive quality of data transfer path, to realize the intelligent of data flow transmission route selection and improve net
The performance and QoS of customer of network are experienced.
Description of the drawings
Fig. 1 is this Intelligent routing selection device structure chart.
Fig. 2 is this Intelligent routing selection device overall flow figure.
Fig. 3 layer congestion detection module flow diagrams in order to control.
Fig. 4 layer routing selecting module flow charts in order to control.
Fig. 5 is congestion detection module link duty ratio membership function figure.
Fig. 6 is congestion detection module link load changing rate membership function figure.
Fig. 7 is congestion detection module link congestion quality evaluation value membership function figure.
In figure, 101 be application layer, and 102 layers in order to control, 103 be data Layer, and 104 be User Access Layer.
Specific implementation mode
Carry out the content that the present invention will be described in detail below with reference to Figure of description and in conjunction with example.
In the example of the present invention:One, software defined network communication environment is necessary condition, and Fig. 1 is the specific of necessary condition
It embodies, Fig. 2 is the Data Stream Processing flow of Fig. 1;Two, path congestion quality evaluation is one of the features of the present invention, and Fig. 3 is path
The flow that congestion quality evaluation is realized, Fig. 5, Fig. 6 and Fig. 7 are the fuzzy relations of congestion quality assessment process evaluation parameter;Three, road
Diameter selection method is core of the invention, and Fig. 4 is the flow of routing resource.
The realization environment of the present invention is divided into User Access Layer 104, data Layer 103, control layer 102 and application layer 101.With
Family access layer 104 is made of Web vector graphic user, proposes data stream transmitting request;Data Layer 103 is by interchanger and network link
Composition, realizes the forwarding capability of data flow;Control layer 102 is made of software defined network controller, software definition in the present invention
Network controller includes mainly network monitoring module, congestion detection module and routing selecting module;Application layer 101 is used for user more
Change network demand and research and innovation.
Network monitoring module, it is end-to-end for each data transfer path in periodic harvest link in network state and network
QoS data.
Congestion detection module, data for being collected into according to network monitoring module to data can transmission path congestion journey
Degree carries out fuzzy evaluation.
Blurring:Based on the fuzzy relation in the present invention to the link duty ratio and chain of all links on alternative path
Road load changing rate carries out fuzzy evaluation.
Fuzzy rule maps:Based on the rule list in the present invention by Indistinct Input value:Link duty ratio, link load
Change rate and fuzzy output value:Link congestion degree is mapped.
Defuzzification:Defuzzification is carried out to link congestion degree based on maximum membership degree method, obtains link congestion matter
Measure evaluation of estimate.
Alternative path congestion condition is evaluated:By the average congestion quality of each link on alternative path, alternately path is gathered around
Fill in quality evaluation value.
Routing selecting module, for according to the path congestion quality evaluation value of congestion detection module and network monitoring module
Path QoS feedback data, using the comprehensive quality of intensified learning method overall merit alternative path.And it is carried out based on the value
Route Selection.
The path QoS value of feedback of Based Network Monitoring module calculates the long-term gain value in path.
The long-term gain value application intensified learning method of congestion quality evaluation value and path based on alternative path calculates
To the Quality evaluation value of alternative path.
Routing selecting module carries out optimal Route Selection with certain probability according to defined formula to data stream.
The intelligent congestion of the present invention avoids route selection method from being realized in control layer 102.Fig. 1 be the present invention realize must
Want environment.Specific implementation process is as follows:
(1) network status initialization
As shown in Figure 1, in the netinit stage, User Access Layer 104 sends data flow between source node and destination node
Transmission request, control layer 102 are established with data Layer 103 in network and are connected.Then, the routing selecting module of control layer 102 according to
Shortest path first is K shortest path of data-flow computation, the optional path set as steaming transfer.It is reached finally, for new
Stream request control layer 102 select it shortest path, and be handed down to data Layer 103 carry out steaming transfer.Initial
The workflow for changing phase Network is similar with traditional network.
(2) network monitor process
Referring to Fig. 2, after the completion of the netinit stage, but it is different from traditional network, control layer after start-up operation
102 network monitoring module periodically collects the network state of data Layer 103, including:Link duty ratioLink load becomes
RateThe handling capacity TH of data transfer path time delay D elay, the packet loss Loss in path and path end to end, data Layer
Evaluation parameter of 103 network state as the path quality for calculating network.
Wherein, if t moment link duty ratioLink load change rateIt is calculated according to formula (1) and (2).
In formula (1),For t moment link load,For t moment link total capacity.
If t moment link load change rateFor:
In formula,For t moment link load,For t-1 moment link loads.
Link load change rate describes current time link load relative to the increased ratio of previous moment link load.
It should be noted that:When link load change rate is positive value, indicate that the load of current time chain road is in increase state;Chain road
Load indicate that the load of current time chain road is in reduction state when being negative value.The absolute value of link load change rate is bigger,
Indicate that the load variations of chain road are more notable.
Meanwhile when data stream transmitting, 103 data transfer path of network monitoring module statistical data layer of control layer 102
The handling capacity TH of time delay D elay, the packet loss Loss in path and path end to end, and by the value alternately path QoS number
According to value of feedback of the alternative path QoS data for data transmission effect calculates.
(3) congestion detection process
Introduce term " fuzzy system " in the present invention first, fuzzy system refers to a kind of apish complex reasoning to being
The method that system carries out decision-making management.
Based on Fuzzy System Method, the congestion detection module of control layer 102 is to link duty ratioAnd link load
Change rateTwo indices carry out fuzzy evaluation, obtain the fuzzy evaluation value rank of current ink congestion qualityij, according to path
The average Congestion Level SPCC of upper all links can obtain the congestion quality fuzzy evaluation value rank in this pathi。
In data stream transmitting, own between the congestion detection module periodical evaluation source node and destination node of control layer 102
The congestion quality of alternative path, obtains alternative path piCongestion quality evaluation value ranki.Calculating process is as shown in figure 3, specific
It is as follows:
The congestion detection module of (3a) control layer 102 is connect with network monitoring module, obtains path congestion quality evaluation ginseng
Number:Link duty ratioWith link load change rate
(3b) is blurred:Referring to Fig. 5, using link with duty ratio membership to link duty ratioCarry out mould
Paste evaluation;Referring to Fig. 6, using link load change rate membership to link load change rateCarry out fuzzy evaluation.
(3c) fuzzy reasoning table maps:Referring to table 1, using " if ... so ... " regular ambiguity in definition rule list, for example,
First " if ... so ... " rule in fuzzy reasoning table is defined as follows:
If link duty ratioFuzzy evaluation value be low, link load change rateFuzzy evaluation value be
It is negative big, then link congestion degree RANKijFuzzy evaluation value be perfect.
The evaluation situation that 15 rules indicate link congestion degree under different network environments is defined in fuzzy reasoning table altogether, in detail
It is shown in Table 1.
By fuzzy input variable link duty ratioWith link load change rateIt is reflected by fuzzy reasoning table
It penetrates, obtains fuzzy output value link congestion degree RANKij。
1 congestion detection module fuzzy reasoning table of table
(3d) ambiguity solution:Referring to Fig. 7, using being utmostly subordinate to method for defined fuzzy output variable link congestion
Degree RANKijDefuzzification is carried out, the evaluation of estimate rank of link congestion quality is obtainedij。
(3e) path congestion quality fuzzy evaluation:
For all path p between source node and destination nodei∈ P carry out path congestion quality evaluation.Here road is used
The average congestion quality evaluation value rank of all links on diameteriAs the congestion quality evaluation value in path, computational methods such as formula
(3) shown in.
In formula (3), rankiIndicate path p between a pair of of source node and destination nodeiCongestion quality evaluation value, rankij
(m) path p is indicatediThe congestion quality evaluation value for the link that upper m order passes through, N indicate path piAll links of upper process are total
Number.
(4) routing procedure
After 102 congestion detection module of control layer completes the congestion quality evaluation of all alternative paths, path selection module with
Congestion detection module and network monitoring module establish connection, obtain Path selection parameter, carry out data stream transmitting optimal path choosing
It selects.Path selection module realizes process as shown in figure 4, specific as follows:
(4a) path selection module is connect with network monitoring module, obtains all alternative path QoS datas:Data transmission route
The handling capacity TH of diameter time delay D elay, the packet loss Loss in path and path end to end.
(4b) path selection module is connect with congestion detection module, obtains the congestion quality evaluation value of all alternative paths
ranki。
(4c) calculates the long-term financial value Rwd of network according to alternative path QoS data, shown in computational methods such as formula (5).
(4d) in intensified learning method as computation model, according to intensified learning method calculate to source node to destination node it
Between this period all alternative path piRespective Quality evaluation value Qi(P,pi), shown in computational methods such as formula (4).
Qi(P,pi)←α×ranki×{Rwd+γ×maxy∈PQi(P,y)}+(1-α) (4)
×Qi(P,pi)
Wherein,
Rwd=ω1·TH+ω2/Delay+ω3/Loss (5)
In formula (4) (5):
Rwd is that previous moment selects this path piThe network path long-term gain value brought afterwards, ω1, ω2, ω3It is respectively to comment
Valence index proportion shared in value of feedback;rankiIt is this period alternative path piCongestion quality evaluation value, α is learning rate,
Qi(P,pi) it is this period alternative path piQuality evaluation value.
(4e) path selection module is according to this period all alternative path piRespective Quality evaluation value Qi(P,pi) more
The Q value tables of new alternative path.
(4f) is according to according to this period all alternative path piRespective Quality evaluation value Qi(P,pi), Path selection
Module determines all alternative path p of next periodiSelected probability π (pi), next period all alternative path piIt is selected
Probability π (pi) computational methods such as formula (6) shown in.
Wherein,
In formula (6) and (7):
π(pi) it is all alternative path p of next periodiSelected probability, K are between each pair of source node and destination node
Total number of paths;τnIt is a temperature parameter, which controls the random degree for selecting a certain path;T indicates convergence
Time;τ0And τTIndicate the final temperature of initial temperature and time T.
The τ when training is initialnIt is larger, realize the more excellent exploration to alternative path;And when network environment tends towards stability, it should
Value is smaller, realizes optimal convergence.
(4g) path selection module more new route, and flow table issuance is carried out data transmission to data Layer 103.
Detection data stream whether the end of transmission, return to step (2) if without the end of transmission carry out new periodic path matter
Amount evaluation and update, if transmission data terminates, network monitor terminates.
Software defined network controller includes three modules, network monitoring module, congestion detection module and road in the present invention
By selecting module, each module is all realized in the controller.The periodic collection network link state of network monitoring module and path
End-to-end QoS data.Based on Fuzzy System Method, the congestion detection module application data are to each pair of source node and purpose in network
The congestion condition of all alternative paths carries out fuzzy evaluation between node, obtains alternative path current congestion quality evaluation value.Base
In path current congestion quality evaluation value, the QoS feedback data in path, routing selecting module application intensified learning method calculates
To the comprehensive evaluation value of alternative path, and optimal routing is selected for data flow based on the value.The present invention is the number in network
A kind of intelligent Congestion Avoidance route selection method is provided according to stream, effectively improves the performance and QoS of customer body of network
It tests.
By the way that the specific descriptions of embodiment, the intelligent congestion of the invention based on software defined network avoids path above
Selection method and device, professional and technical personnel in the field will be clearly understood that the acquisition that can reach from model to network state,
The effect that network congestion is avoided;Meanwhile the path selection device of Congestion Avoidance carries in software defined network of the invention
The support to software defined network path calculation method has been supplied, has contained the acquisition of data Layer network state, data layer data passes
The selection of the Congestion Level SPCC evaluation and data stream transmitting optimal path in defeated path.
Embodiment of above is the preferred embodiment of the present invention, and the present invention is not limited to the present embodiment, also includes characteristic range
Interior son invention and its method, thought and spirit.For the present invention and its main feature and thought, and the main spy of son invention
Thought of seeking peace should all be within protection domain.
Claims (8)
1. the intelligent heavy route method based on congestion aware in software defined network, which is characterized in that include the following steps:
(1) netinit;
(2) network monitor process:
After the completion of netinit, the network monitoring module of control layer (102) periodically collects and calculates the network of data Layer (103)
State, including:Link duty ratioLink load change rateData transfer path time delay D elay, road end to end
The packet loss Loss of the diameter and handling capacity TH in path, the network state of data Layer (103) is as the path quality for calculating network
Evaluation parameter;
When data stream transmitting, network monitoring module statistical data layer (103) data transfer path of control layer (102) is end-to-end
Time delay D elay, the packet loss Loss in path and the handling capacity TH in path, and by data transfer path time delay end to end
Alternately path QoS data, alternative path QoS data are used for the handling capacity TH of Delay, the packet loss Loss in path and path
The value of feedback of data transmission effect calculates;
(3) congestion detection process:
Based on Fuzzy System Method, the congestion detection module of control layer (102) is to link duty ratioBecome with link load
RateTwo indices carry out fuzzy evaluation, obtain the fuzzy evaluation value rank of current ink congestion qualityij, according on path
The average Congestion Level SPCC of all links obtains the congestion quality fuzzy evaluation value rank in this pathi;
(4) routing procedure:
After the congestion detection module of control layer (102) completes the congestion quality evaluation of all alternative paths, path selection module with
Congestion detection module is and network monitoring module establishes connection, obtains Path selection parameter, carries out data stream transmitting optimal path
Selection.
2. the intelligent heavy route method based on congestion aware, feature exist in software defined network according to claim 1
In initialization detailed process is as follows in step (1):User Access Layer (104) sends data between source node and destination node and spreads
Defeated request, control layer (102) are established with data Layer (103) in network and are connected, then the routing selecting module root of control layer (102)
It is K shortest path of data-flow computation according to shortest path first, as the optional path set of steaming transfer, finally, for newly arriving
The stream request control layer (102) reached selects it shortest path, and is handed down to data Layer (103) and carries out steaming transfer.
3. the intelligent heavy route method based on congestion aware, feature exist in software defined network according to claim 1
In step (2) link duty ratioIt is calculated by formula (1):
In formula (1),For t moment link load,For t moment link total capacity;
Link load change rateIt is calculated by formula (2):
In formula,For t moment link load,For t-1 moment link loads.
4. the intelligent heavy route method based on congestion aware, feature exist in software defined network according to claim 1
In the detailed process of step (3) is as follows:
The congestion detection module of (3a) control layer (102) is connect with network monitoring module, obtains path congestion quality evaluation parameter:
Link duty ratioWith link load change rate
(3b) is blurred:Using link with duty ratio membership to link duty ratioCarry out fuzzy evaluation;Using
Link load change rate membership is to link load change rateCarry out fuzzy evaluation;
(3c) fuzzy reasoning table maps:By link duty ratioWith link load change rateIt is carried out by fuzzy reasoning table
Mapping, obtains link congestion degree RANKij;
(3d) ambiguity solution:Using being utmostly subordinate to method for defined fuzzy output value link congestion degree RANKijIt carries out
Defuzzification obtains the evaluation of estimate rank of link congestion qualityij;
(3e) path congestion quality fuzzy evaluation:
For all path p between source node and destination nodei∈ P carry out path congestion quality evaluation, obtain all chains on path
The average congestion quality evaluation value rank on roadi:
In formula (3), rankiIndicate path p between a pair of of source node and destination nodeiCongestion quality evaluation value, rankij(m) it indicates
Path piThe congestion quality evaluation value for the link that upper m order passes through, N indicate path piAll links sum of upper process.
5. the intelligent heavy route method based on congestion aware, feature exist in software defined network according to claim 1
In the detailed process of step (4) is as follows:
(4a) path selection module is connect with network monitoring module, obtains all alternative path QoS datas:Data transfer path end
To the time delay D elay, the packet loss Loss in path and the handling capacity TH in path at end;
(4b) path selection module is connect with congestion detection module, obtains the congestion quality evaluation value rank of all alternative pathsi;
(4c) calculates the long-term financial value Rwd of network according to alternative path QoS data;
(4d) calculates source node to all alternative path p of this period between destination node according to intensified learning methodiRespective synthesis
Quality evaluation value Qi(P,pi), according to this period all alternative path piRespective Quality evaluation value Qi(P,pi) update is alternatively
The Q value tables in path;
(4e) is according to this period all alternative path piRespective Quality evaluation value Qi(P,pi), path selection module determines
Next period all alternative path piSelected probability π (pi), next period all alternative path piSelected probability π
(pi) computational methods such as formula (6) shown in;
Wherein,
In formula (6) and (7):
π(pi) it is all alternative path p of next periodiSelected probability, K are the roads between each pair of source node and destination node
Diameter sum;τnIt is a temperature parameter, which controls the random degree for selecting a certain path;T indicates convergence time;
τ0And τTIndicate the final temperature of initial temperature and time T;
(4f) path selection module more new route, and flow table issuance is carried out data transmission to data Layer (103);Detection data stream is
The no end of transmission, return to step (2) if without the end of transmission carry out new periodic path quality evaluation and update, if transmission
End of data, then network monitor terminate.
6. the intelligent heavy route method based on congestion aware, feature exist in software defined network according to claim 5
In the long-term financial value Rwd of network is calculated by the following formula in step (4c):
Rwd=ω1·TH+ω2/Delay+ω3/Loss (5)
Wherein:ω1, ω2, ω3It is each evaluation index proportion shared in value of feedback.
7. the intelligent heavy route method based on congestion aware, feature exist in software defined network according to claim 6
In this period alternative path Quality evaluation value Qi(P,pi) be calculate by the following formula:
Qi(P,pi)←α×ranki×{Rwd+γ×maxy∈PQi(P,y)}+(1-α)×Qi(P,pi) (4)
Wherein, rankiIt is this period alternative path piCongestion quality evaluation value, α is learning rate, Qi(P,pi) it is that this period is alternative
Path piQuality evaluation value.
8. the device that the intelligent heavy route method based on congestion aware uses in software defined network described in claim 1, special
Sign is, including User Access Layer (104), data Layer (103), control layer (102) and application layer (101);User Access Layer
(104) it is made of Web vector graphic user, for proposing data stream transmitting request;Data Layer (103) is by interchanger and network link
Composition, realizes the forwarding capability of data flow;Control layer (102) is made of software defined network controller, software defined network control
Device processed includes network monitoring module, congestion detection module and routing selecting module, and network monitoring module is used for periodic harvest net
Each data transfer path end-to-end QoS data in network link state and network, and it is transmitted to congestion detection module and routing choosing
Select module, the data that congestion detection module is used to be collected into according to network monitoring module to data can transmission path Congestion Level SPCC
Fuzzy evaluation is carried out, routing selecting module is used for path congestion status evaluation value and network monitor mould according to congestion detection module
The path QoS data feedback of block, using the comprehensive quality of intensified learning method overall merit alternative path, and with alternative path
It is routed based on comprehensive quality;Application layer (101) changes network demand and research for user.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810332577.0A CN108540384B (en) | 2018-04-13 | 2018-04-13 | Intelligent rerouting method and device based on congestion awareness in software defined network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810332577.0A CN108540384B (en) | 2018-04-13 | 2018-04-13 | Intelligent rerouting method and device based on congestion awareness in software defined network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108540384A true CN108540384A (en) | 2018-09-14 |
CN108540384B CN108540384B (en) | 2020-07-28 |
Family
ID=63480545
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810332577.0A Active CN108540384B (en) | 2018-04-13 | 2018-04-13 | Intelligent rerouting method and device based on congestion awareness in software defined network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108540384B (en) |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109039885A (en) * | 2018-08-27 | 2018-12-18 | 广州爱拍网络科技有限公司 | A kind of data transfer path selection method and device |
CN109246231A (en) * | 2018-09-29 | 2019-01-18 | 北京深度奇点科技有限公司 | A kind of method for intelligently routing and smart routing devices |
CN109361601A (en) * | 2018-10-31 | 2019-02-19 | 浙江工商大学 | A kind of SDN route planning method based on intensified learning |
CN109579861A (en) * | 2018-12-10 | 2019-04-05 | 华中科技大学 | A kind of method for path navigation and system based on intensified learning |
CN110708242A (en) * | 2019-10-28 | 2020-01-17 | 迈普通信技术股份有限公司 | Traffic scheduling method and device, electronic equipment and storage medium |
WO2020232184A1 (en) * | 2019-05-14 | 2020-11-19 | Vmware, Inc | Congestion avoidance in a slice-based network |
US10892994B2 (en) | 2019-05-14 | 2021-01-12 | Vmware, Inc. | Quality of service in virtual service networks |
CN112383482A (en) * | 2020-11-16 | 2021-02-19 | 北京邮电大学 | Dynamic Q value route calculation method and device based on data plane |
US10944647B2 (en) | 2019-01-24 | 2021-03-09 | Vmware, Inc. | Dynamic inter-cloud placement of virtual network functions for a slice |
US10958579B2 (en) | 2019-05-14 | 2021-03-23 | Vmware, Inc. | Congestion avoidance in a slice-based network |
US10979314B2 (en) | 2019-01-24 | 2021-04-13 | Vmware, Inc. | Dynamic inter-cloud placement of virtual network functions for a slice |
CN112671649A (en) * | 2020-12-22 | 2021-04-16 | 广州技象科技有限公司 | Path selection method and device based on Internet of things transmission fault detection |
US11012288B2 (en) | 2019-05-14 | 2021-05-18 | Vmware, Inc. | Congestion avoidance in a slice-based network |
US11588733B2 (en) | 2019-05-14 | 2023-02-21 | Vmware, Inc. | Slice-based routing |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20150062285A1 (en) * | 2013-08-30 | 2015-03-05 | Futurewei Technologies Inc. | Multicast tree packing for multi-party video conferencing under sdn environment |
CN104734954A (en) * | 2015-03-27 | 2015-06-24 | 华为技术有限公司 | Routing determination method and device used for software defined network (SDN) |
CN105227481A (en) * | 2015-09-02 | 2016-01-06 | 重庆邮电大学 | The SDN congestion control method for routing of cost minimization is dispatched based on path cost and stream |
CN106341346A (en) * | 2016-09-08 | 2017-01-18 | 重庆邮电大学 | Routing algorithm of guaranteeing QoS in data center network based on SDN |
-
2018
- 2018-04-13 CN CN201810332577.0A patent/CN108540384B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20150062285A1 (en) * | 2013-08-30 | 2015-03-05 | Futurewei Technologies Inc. | Multicast tree packing for multi-party video conferencing under sdn environment |
CN104734954A (en) * | 2015-03-27 | 2015-06-24 | 华为技术有限公司 | Routing determination method and device used for software defined network (SDN) |
CN105227481A (en) * | 2015-09-02 | 2016-01-06 | 重庆邮电大学 | The SDN congestion control method for routing of cost minimization is dispatched based on path cost and stream |
CN106341346A (en) * | 2016-09-08 | 2017-01-18 | 重庆邮电大学 | Routing algorithm of guaranteeing QoS in data center network based on SDN |
Non-Patent Citations (3)
Title |
---|
AKASH SRIKANTH: "Congestion Control Mechanism in Software Defined Networking by Traffic Rerouting", 《2018 SECOND INTERNATIONAL CONFERENCE ON COMPUTING METHODOLOGIES AND COMMUNICATION (ICCMC)》 * |
SHADI ATTARHA: "A load balanced congestion aware routing mechanism for Software Defined Networks", 《2017 IRANIAN CONFERENCE ON ELECTRICAL ENGINEERING (ICEE)》 * |
董晓林: "面向SDN的多路径调度算法研究", 《中国优秀硕士学位论文电子期刊》 * |
Cited By (25)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109039885A (en) * | 2018-08-27 | 2018-12-18 | 广州爱拍网络科技有限公司 | A kind of data transfer path selection method and device |
CN109039885B (en) * | 2018-08-27 | 2022-01-07 | 广州猎游信息科技有限公司 | Data transmission path selection method and device |
CN109246231A (en) * | 2018-09-29 | 2019-01-18 | 北京深度奇点科技有限公司 | A kind of method for intelligently routing and smart routing devices |
CN109246231B (en) * | 2018-09-29 | 2022-11-18 | 北京深度奇点科技有限公司 | Intelligent routing method and intelligent routing equipment |
CN109361601B (en) * | 2018-10-31 | 2021-03-30 | 浙江工商大学 | SDN route planning method based on reinforcement learning |
CN109361601A (en) * | 2018-10-31 | 2019-02-19 | 浙江工商大学 | A kind of SDN route planning method based on intensified learning |
CN109579861A (en) * | 2018-12-10 | 2019-04-05 | 华中科技大学 | A kind of method for path navigation and system based on intensified learning |
CN109579861B (en) * | 2018-12-10 | 2020-05-19 | 华中科技大学 | Path navigation method and system based on reinforcement learning |
US11356338B2 (en) | 2019-01-24 | 2022-06-07 | Vmware, Inc. | Dynamic inter-cloud placement of virtual network functions for a slice |
US10979314B2 (en) | 2019-01-24 | 2021-04-13 | Vmware, Inc. | Dynamic inter-cloud placement of virtual network functions for a slice |
US10944647B2 (en) | 2019-01-24 | 2021-03-09 | Vmware, Inc. | Dynamic inter-cloud placement of virtual network functions for a slice |
US11329901B2 (en) | 2019-01-24 | 2022-05-10 | Vmware, Inc. | Dynamic inter-cloud placement of virtual network functions for a slice |
WO2020232184A1 (en) * | 2019-05-14 | 2020-11-19 | Vmware, Inc | Congestion avoidance in a slice-based network |
US11012288B2 (en) | 2019-05-14 | 2021-05-18 | Vmware, Inc. | Congestion avoidance in a slice-based network |
US10897423B2 (en) | 2019-05-14 | 2021-01-19 | Vmware, Inc. | Congestion avoidance in a slice-based network |
US10958579B2 (en) | 2019-05-14 | 2021-03-23 | Vmware, Inc. | Congestion avoidance in a slice-based network |
US10892994B2 (en) | 2019-05-14 | 2021-01-12 | Vmware, Inc. | Quality of service in virtual service networks |
US11588733B2 (en) | 2019-05-14 | 2023-02-21 | Vmware, Inc. | Slice-based routing |
US11595315B2 (en) | 2019-05-14 | 2023-02-28 | Vmware, Inc. | Quality of service in virtual service networks |
US11902080B2 (en) | 2019-05-14 | 2024-02-13 | Vmware, Inc. | Congestion avoidance in a slice-based network |
CN110708242A (en) * | 2019-10-28 | 2020-01-17 | 迈普通信技术股份有限公司 | Traffic scheduling method and device, electronic equipment and storage medium |
CN110708242B (en) * | 2019-10-28 | 2021-12-17 | 迈普通信技术股份有限公司 | Traffic scheduling method and device, electronic equipment and storage medium |
CN112383482B (en) * | 2020-11-16 | 2021-10-08 | 北京邮电大学 | Dynamic Q value route calculation method and device based on data plane |
CN112383482A (en) * | 2020-11-16 | 2021-02-19 | 北京邮电大学 | Dynamic Q value route calculation method and device based on data plane |
CN112671649A (en) * | 2020-12-22 | 2021-04-16 | 广州技象科技有限公司 | Path selection method and device based on Internet of things transmission fault detection |
Also Published As
Publication number | Publication date |
---|---|
CN108540384B (en) | 2020-07-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108540384A (en) | Intelligent heavy route method and device based on congestion aware in software defined network | |
CN108401015A (en) | A kind of data center network method for routing based on deeply study | |
CN113328938B (en) | Network autonomous intelligent management and control method based on deep reinforcement learning | |
CN108600102A (en) | A kind of flexible data Transmission system in the contract network based on wisdom | |
CN110933157B (en) | Industrial Internet of things-oriented edge computing task unloading method | |
CN108667734A (en) | It is a kind of that the through street with LSTM neural networks is learnt by decision making algorithm based on Q | |
EP2515478B1 (en) | Method, apparatus and system for joint optimizations | |
CN114884895B (en) | Intelligent flow scheduling method based on deep reinforcement learning | |
CN109639575A (en) | Route planning method based on link congestion coefficient | |
CN104022951B (en) | A kind of method for building up and system in network service path | |
CN105262534B (en) | A kind of method for routing and device suitable for satellite communication network | |
CN110225418A (en) | A kind of HTTP video flowing QoE routing optimization method based on SDN | |
CN111010341B (en) | Overlay network routing decision method based on deep learning | |
CN108966053A (en) | A kind of cross-domain route computing method of multiple-domain network dynamic domain sequence and device | |
CN106302012A (en) | A kind of PTN network simulation-optimization method and system | |
Abdel-Kader | An improved discrete PSO with GA operators for efficient QoS-multicast routing | |
CN106817306A (en) | A kind of method and device for determining target route | |
CN106850859A (en) | A kind of user's request distribution method of the CDN based on SDN | |
CN105430538B (en) | A kind of inter-domain routing method based on optical-fiber network subtopology figure | |
CN116170370B (en) | SDN multipath routing method based on attention mechanism and deep reinforcement learning | |
CN107171957A (en) | A kind of adaptive DTN routing algorithms based under the conditions of resource-constrained | |
CN106714265A (en) | Network node prediction method and device based on DTN (Delay and Disruption-Tolerant Networking) algorithm | |
CN114205300B (en) | Flow scheduling method capable of guaranteeing coflow transmission deadline under condition of incomplete flow information | |
CN110139173A (en) | A kind of network dividing area method reducing optical transfer network end-to-end time delay | |
CN106850431A (en) | A kind of optimal route selection method of many attributes for being applied to low rail Information Network |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |