CN111314171A - Method, device and medium for predicting and optimizing SDN routing performance - Google Patents

Method, device and medium for predicting and optimizing SDN routing performance Download PDF

Info

Publication number
CN111314171A
CN111314171A CN202010058591.3A CN202010058591A CN111314171A CN 111314171 A CN111314171 A CN 111314171A CN 202010058591 A CN202010058591 A CN 202010058591A CN 111314171 A CN111314171 A CN 111314171A
Authority
CN
China
Prior art keywords
node
link
path
network
routing
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
Application number
CN202010058591.3A
Other languages
Chinese (zh)
Other versions
CN111314171B (en
Inventor
车向北
康文倩
邓彬
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Power Supply Bureau Co Ltd
Original Assignee
Shenzhen Power Supply Bureau Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shenzhen Power Supply Bureau Co Ltd filed Critical Shenzhen Power Supply Bureau Co Ltd
Priority to CN202010058591.3A priority Critical patent/CN111314171B/en
Publication of CN111314171A publication Critical patent/CN111314171A/en
Application granted granted Critical
Publication of CN111314171B publication Critical patent/CN111314171B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00—Arrangements for monitoring or testing data switching networks
    • H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/08—Configuration management of networks or network elements
    • H04L41/0803—Configuration setting
    • H04L41/0823—Configuration setting characterised by the purposes of a change of settings, e.g. optimising configuration for enhancing reliability
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/12—Discovery or management of network topologies
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/14—Network analysis or design
    • H04L41/145—Network analysis or design involving simulating, designing, planning or modelling of a network
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D30/00—Reducing energy consumption in communication networks
    • Y02D30/50—Reducing energy consumption in communication networks in wire-line communication networks, e.g. low power modes or reduced link rate

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Environmental & Geological Engineering (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

The invention provides a method, equipment and medium for predicting and optimizing SDN routing performance. The method includes, step S1, determining a link node V based on the routing policy R and the computer network topology TlAnd path node VpGraph relationships between; step S2, capturing path node V by multi-layer aggregation methodpAnd a link node VlThe relationship between; in step S3, the entire parameters in the network are updated by predicting the error between the performance index and the true value. The method predicts the end-to-end performance indexes of each routing strategy path under the current condition when the communication network topology structure, the flow matrix and the routing strategy are changed, thereby assisting the SDN controller to establish the routing strategy meeting the upper-layer requirements; the method can accurately estimate the network performance under a given routing scheme in the software defined network and optimize the routing scheme.

Description

Method, device and medium for predicting and optimizing SDN routing performance
Technical Field
The present invention relates to the field of routing, and in particular, to a method, a device, and a medium for predicting and optimizing SDN routing performance.
Background
Routing policy planning is a core task of the control plane in Software Defined Networking (SDN). At present, a mainstream routing strategy is established according to a shortest path routing algorithm, which can generally obtain better performance when the network traffic is smaller, but when the traffic is increased, the problem of network congestion is often caused, and particularly, the problem is serious today when the information amount is increased sharply. The SDN technology separates a data plane from a control plane, so that an SDN controller can globally control equipment and link information of the data plane. It is a very complicated problem how to relate physical link information in a network to the communication performance of a routing policy path. In the traditional scheme, the relation between the two is usually modeled based on methods such as queuing theory and the like, and many assumptions are made for the network in the modeling process to solve, but the assumptions are often key factors influencing the network performance. The graph neural network can fully describe the mutual relation between things, and the relation between a physical link in the network and a path in a routing strategy can be well learned by means of the graph neural network, so that the SDN control layer is assisted to establish a reasonable routing strategy under the condition of corresponding network topology and flow.
Disclosure of Invention
The technical problem to be solved by the embodiments of the present invention is to provide a method, a device, and a medium for predicting and optimizing SDN routing performance, so as to solve the problem that when traffic increases and a network is congested, the SDN routing performance is not predicted and optimized reasonably and comprehensively.
In one aspect of the present invention, a method for predicting and optimizing SDN routing performance is provided, which includes the following steps:
step S1, determining link node V according to routing strategy R and computer network topology TlAnd path node VpGraph relationships between;
step S2, capturing path node V by multi-layer aggregation methodpAnd a link node VlThe relationship between;
in step S3, the entire parameters in the network are updated by predicting the error between the performance index and the true value.
Further, in the step S2, the specific process of capturing the relationship between the path node and the link node by the multi-layer aggregation method includes:
step S21, according to the link node VlAnd path node VpThe self-attribute of the node initializes the initial characteristics of the node;
step S22, for all path nodes VpPerforming information aggregation;
step S23, for all link nodes VlPerforming information aggregation;
in step S24, all feature vectors are normalized.
Further, in step S21, the initializing the initial feature of the node is specifically calculated according to the following formula:
Figure BDA0002371844500000021
wherein the link node VlAnd path node VpThe self-attributes of (a) specifically include,
the link attribute is used for representing various physical attributes of the link;
and the attribute on the path is used for expressing the end-to-end flow condition of the link.
Further, in the step S22, the path node VpThe information aggregation specifically comprises the following steps of extracting features by using a long and short memory neural network (LSTM) through the following formula:
Figure BDA0002371844500000022
wherein the content of the first and second substances,
Figure BDA0002371844500000023
for the operation of the long-short term memory network and the network weight, N (v) represents the node v's leading point set.
Further, in the step S22, the path node VpPerforming information aggregation specifically includes performing nonlinear transformation on the extracted feature result by using the following formula:
Figure BDA0002371844500000024
wherein the content of the first and second substances,
Figure BDA0002371844500000025
a vector is embedded for each node and,
Figure BDA0002371844500000026
for non-linearly varying parameters, CONCAT means that the correlation vectors are spliced, and σ means a non-linear function, such as a tanh function.
Further, in the step S23, the link node VlInformation aggregation specifically utilizes the following formula for aggregation:
Figure BDA0002371844500000027
Figure BDA0002371844500000028
wherein the content of the first and second substances,
Figure BDA0002371844500000029
a vector is embedded for each node and,
Figure BDA00023718445000000210
for non-linearly varying parameters, the SUM operation means a column-wise summation operation on the correlation vectors.
Further, in the step S24, the feature vectors are normalized according to the following formula:
Figure BDA00023718445000000211
wherein the content of the first and second substances,
Figure BDA00023718445000000212
embedding vectors for each node, | · | ceiling2The L2 norm of the vector is shown, and V is the combination of all nodes.
Further, in the step S3, the predicted performance index is an end-to-end performance index obtained by using a feedforward neural network, the embedded vector is mapped to a neural network parameter of the performance index,
specifically, the calculation is performed according to the following formula:
Figure BDA00023718445000000213
Figure BDA00023718445000000214
wherein the content of the first and second substances,
Figure BDA00023718445000000215
embedding vectors, V, for each nodelIs a set of link nodes, VpFor the path node set, DNN () represents that the vector is operated on by a fully-connected neural network, usually two layers of neural networks are used.
Accordingly, a further aspect of the present invention also provides a computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the following steps when executing the computer program:
determining a link node V according to a routing policy R and a computer network topology TlAnd path node VpGraph relationships between;
capturing path nodes V through a multi-layer aggregation methodpAnd a link node VlThe relationship between;
the entire parameter in the network is updated by predicting the error of the performance indicator from the true value.
Accordingly, a further aspect of the present invention also provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of:
determining a link node V according to a routing policy R and a computer network topology TlAnd path node VpGraph relationships between;
capturing path segments by multi-layer aggregation methodPoint VpAnd a link node VlThe relationship between;
the entire parameter in the network is updated by predicting the error of the performance indicator from the true value.
In summary, the embodiment of the invention has the following beneficial effects:
the method, the device and the medium for predicting and optimizing the routing performance of the SDN provided by the invention utilize the strong capability of the neural network to represent and model things, fully model the relation among the communication network topology, the traffic condition and the routing strategy, predict the end-to-end performance indexes of each routing strategy path under the current condition when the communication network topology structure, the traffic matrix and the routing strategy are changed, and assist the SDN controller to establish the routing strategy meeting the upper-layer requirements; the method can accurately estimate the network performance under a given routing scheme in the software defined network and optimize the routing scheme.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is within the scope of the present invention for a person skilled in the art to obtain other drawings based on the drawings without paying creative efforts.
Fig. 1 is a flowchart of a method for predicting and optimizing SDN routing performance according to the present invention;
fig. 2 is a schematic network topology diagram of an embodiment of a method for predicting and optimizing SDN routing performance provided in the present invention;
fig. 3 is one of network performance comparison diagrams of an embodiment of a method for predicting and optimizing SDN routing performance provided by the present invention.
Fig. 4 is one of network performance comparison diagrams of an embodiment of a method for predicting and optimizing SDN routing performance provided by the present invention.
Fig. 5 is one of network performance comparison diagrams of an embodiment of a method for predicting and optimizing SDN routing performance provided by the present invention.
Fig. 6 is one of network performance comparison diagrams of an embodiment of a method for predicting and optimizing SDN routing performance provided by the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings.
Fig. 1 is a schematic main flow chart of an embodiment of a method for predicting and optimizing SDN route performance according to the present invention. In this embodiment, the method comprises the steps of:
step S1, determining link node V according to routing strategy R and computer network topology TlAnd path node VpGraph relationships between;
in one embodiment, as shown in FIG. 2, topology information in a computer network may be summarized by the interaction of link nodes and path nodes. We can model the relationship between links and paths as a binary graph model, where both links and paths are represented as nodes of the graph; on the left side, the computer network topology and the paths in the network, on the right side, the relationship between the link nodes and the path nodes is described by using a graph model, and then, each link node l is represented by means of a graph SAGE graph neural networkiAnd path node piIntroducing embedded vectors to be learned
Figure BDA0002371844500000047
And
Figure BDA0002371844500000048
these embedded vectors should contain semantic information related to the modeled target such as latency, packet loss rate, link utilization, etc. With these embedded vectors, we can predict the end-to-end performance indicators.
Step S2, capturing path node V by multi-layer aggregation methodpAnd a link node VlThe relationship between;
in a specific embodiment, the specific process of capturing the relationship between the path node and the link node by the multi-layer aggregation method includes:
step S21, according to the link node VlAnd path node VpThe self-attribute of the node initializes the initial characteristics of the node;
step S22, for all path nodes VpInformation aggregation is carried out, and each path is obtained by sequentially connecting a plurality of links;
step S23, for all link nodes VlPerforming information aggregation;
in step S24, all feature vectors are normalized.
Specifically, the initialization of the initial feature of the node is specifically calculated according to the following formula:
Figure BDA0002371844500000041
wherein, in the step S21, the link node VlAnd path node VpThe self attribute of the link specifically comprises the attribute of the link, which is used for representing various physical attributes of the link; the attributes on the path are used for expressing the end-to-end traffic condition of the link.
Specifically, in the step S22, the path node VpThe information aggregation specifically comprises the following steps of extracting features by using a long and short memory neural network (LSTM) through the following formula:
Figure BDA0002371844500000042
wherein the content of the first and second substances,
Figure BDA0002371844500000043
for the operation of the long-short term memory network and the network weight, N (v) represents the node v's leading point set.
In particular, the path node VpThe information aggregation specifically comprises a nonlinear transformation of the extracted feature results using the following formula to increaseThe expression capacity is as follows:
Figure BDA0002371844500000044
wherein the content of the first and second substances,
Figure BDA0002371844500000045
a vector is embedded for each node and,
Figure BDA0002371844500000046
for non-linearly varying parameters, CONCAT means that the correlation vectors are spliced, and σ means a non-linear function, such as a tanh function.
Specifically, in step S23, the link node VlInformation aggregation specifically utilizes the following formula for aggregation:
Figure BDA0002371844500000051
Figure BDA0002371844500000052
wherein the content of the first and second substances,
Figure BDA0002371844500000053
a vector is embedded for each node and,
Figure BDA0002371844500000054
for the nonlinear variation parameter, SUM operation means that the related vectors are summed according to columns; since each link may participate in multiple paths, its neighbor path node characteristics are aggregated by column using a summation function.
Specifically, in step S24, the feature vectors are normalized according to the following formula:
Figure BDA0002371844500000055
wherein the content of the first and second substances,
Figure BDA0002371844500000056
embedding vectors for each node, | · | ceiling2The L2 norm of the vector is shown, and V is the combination of all nodes.
Step S3, updating the whole parameters in the network by predicting the error between the performance index and the true value;
in a specific embodiment, the predicted performance index is an end-to-end performance index obtained by using a feedforward neural network, such as delay time and jitter of a path, and the embedded vector is mapped to a neural network parameter of the performance index, and is specifically calculated according to the following formula:
Figure BDA0002371844500000057
Figure BDA0002371844500000058
wherein the content of the first and second substances,
Figure BDA0002371844500000059
embedding vectors, V, for each nodelIs a set of link nodes, VpFor the path node set, DNN () represents that the vector is operated on by a fully-connected neural network, usually two layers of neural networks are used.
For the present invention, the final objective to be optimized includes the coefficients of the individual layer length short term memory network weights
Figure BDA00023718445000000510
Non-linear variation parameter of each layer
Figure BDA00023718445000000511
The embedded vector is mapped to a neural network parameter of the performance indicator. Embedding vectors for each node in a particular link-path graph in a training set
Figure BDA00023718445000000512
Only as an auxiliary training tool in the training processAnd is not of particular concern. And finally, performing information aggregation and performance prediction on a brand-new topological structure or routing strategy in the same way after a graph model is established through various parameters in the trained network.
According to the prediction model, under a specific scene, a series of routing strategies (deep-first search) are randomly generated, then the performance of the routing strategies is pre-estimated by using the prediction model, and the routing strategy which meets the conditions and has the optimal performance is selected.
As shown in fig. 3 to fig. 6, compared to the conventional shortest path routing policy, the method of the present invention can achieve better performance when the network traffic is large.
In experiments, the average relative error of the predicted end-to-end path performance of the model is found to be not more than 4.1% when the model is tested, wherein the communication network environment (network topology, routing strategy and traffic matrix) used for generating data used for testing is never used in a training set. Compared with the traditional shortest path routing algorithm, the routing scheme obtained by the model has the advantages that the average delay and the average jitter are respectively reduced by 19.8% and 33.52% under the condition of high traffic intensity, and the maximum delay and the maximum jitter are respectively reduced by 36.18% and 35.45%.
Accordingly, another aspect of the present invention also provides a computer device including a processor, a memory, and a network interface connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a method of SDN route performance prediction and optimization.
It will be appreciated by those skilled in the art that the above-described computer apparatus is only part of the structure associated with the present application and does not constitute a limitation on the computer apparatus to which the present application is applied, and a particular computer apparatus may comprise more or less components than those described above, or some components may be combined, or have a different arrangement of components.
In one embodiment, there is provided a computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing the following steps when executing the computer program:
determining a link node V according to a routing policy R and a computer network topology TlAnd path node VpGraph relationships between;
capturing path nodes V through a multi-layer aggregation methodpAnd a link node VlThe relationship between;
the entire parameter in the network is updated by predicting the error of the performance indicator from the true value.
Accordingly, a further aspect of the present invention also provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of;
determining a link node V according to a routing policy R and a computer network topology TlAnd path node VpGraph relationships between;
capturing path nodes V through a multi-layer aggregation methodpAnd a link node VlThe relationship between;
the entire parameter in the network is updated by predicting the error of the performance indicator from the true value.
It is understood that, for more details of the above-mentioned computer device and the steps involved in the computer readable storage medium, reference may be made to the aforementioned limitations on the SDN route performance prediction and optimization method, and details are not repeated here.
Any reference to memory, storage, databases, or other media used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
In summary, the embodiment of the invention has the following beneficial effects:
the method, the device and the medium for predicting and optimizing the routing performance of the SDN provided by the invention utilize the strong capability of the neural network to represent and model things, fully model the relation among the communication network topology, the traffic condition and the routing strategy, predict the end-to-end performance indexes of each routing strategy path under the current condition when the communication network topology structure, the traffic matrix and the routing strategy are changed, and assist the SDN controller to establish the routing strategy meeting the upper-layer requirements; the method can accurately estimate the network performance under a given routing scheme in the software defined network and optimize the routing scheme.
The method comprises the steps of establishing a plurality of different topology communication networks by utilizing OMNeT + + simulation software or equipment, setting a plurality of routing strategies for each network on the basis, transforming a flow matrix in the communication network under each routing strategy, carrying out simulation test, recording current network conditions (network topology, routing strategies and flow matrix) and various performance indexes in each path in the corresponding routing strategy, and taking the current network conditions, the routing strategies and the flow matrix as training data of a model. The model provided by the invention is trained by utilizing the training data, so that the model can accurately predict the routing performance under different network conditions.
While the invention has been described in connection with what is presently considered to be the most practical and preferred embodiment, it is to be understood that the invention is not to be limited to the disclosed embodiment, but on the contrary, is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.

Claims (10)

1. A method for predicting and optimizing SDN routing performance is characterized by comprising the following steps:
step S1, determining link node V according to routing strategy R and computer network topology TlAnd path node VpGraph relationships between;
step S2, capturing path node V by multi-layer aggregation methodpAnd a link node VlThe relationship between;
in step S3, the entire parameters in the network are updated by predicting the error between the performance index and the true value.
2. The method according to claim 1, wherein in the step S2, the specific process of capturing the relationship between the path node and the link node by the multi-layer aggregation method comprises:
step S21, according to the link node VlAnd path node VpInitializing the initial characteristics of the nodes by the self-attributes;
step S22, for all path nodes VpPerforming information aggregation;
step S23, for all link nodes VlPerforming information aggregation;
in step S24, all feature vectors are normalized.
3. The method according to claim 2, wherein in step S21, the initializing of the initial feature of the node is calculated according to the following formula:
Figure FDA0002371844490000011
wherein the link node VlAnd path node VpThe self-attributes of (a) specifically include,
the link attribute is used for representing various physical attributes of the link;
and the attribute on the path is used for expressing the end-to-end flow condition of the link.
4. The method according to claim 2, characterized in that in said step S22, said path node VpThe information aggregation specifically comprises the following steps of extracting features by using a long and short memory neural network (LSTM) through the following formula:
Figure 1
wherein the content of the first and second substances,
Figure FDA0002371844490000013
for the operation of the long-short term memory network and the network weight, N (v) represents the node v's leading point set.
5. The method according to claim 4, wherein in said step S22, said path node VpPerforming information aggregation specifically includes performing nonlinear transformation on the extracted feature result by using the following formula:
Figure FDA0002371844490000014
wherein the content of the first and second substances,
Figure FDA0002371844490000015
a vector is embedded for each node and,
Figure FDA0002371844490000016
for non-linearly varying parameters, CONCAT denotes the concatenation of the correlation vectors and σ denotes a non-linear function, such as a tanh function.
6. The method according to claim 2, wherein in step S23, the link node VlInformation aggregation specifically utilizes the following formula for aggregation:
Figure FDA0002371844490000021
Figure FDA0002371844490000022
wherein the content of the first and second substances,
Figure FDA0002371844490000023
a vector is embedded for each node and,
Figure FDA0002371844490000024
for non-linearly varying parameters, the SUM operation means a column-wise summation operation on the correlation vectors.
7. The method according to claim 2, wherein in step S24, the all feature vectors are normalized according to the following formula:
Figure FDA0002371844490000025
wherein the content of the first and second substances,
Figure FDA0002371844490000026
embedding vectors for each node, | · | ceiling2The L2 norm of the vector is shown, V being the union of all nodes.
8. The method of claim 1, wherein in step S3, the predicted performance indicator is an end-to-end performance indicator obtained by using a feedforward neural network, and the embedded vector is mapped to a neural network parameter of the performance indicator, and is calculated according to the following formula:
Figure FDA0002371844490000027
Figure FDA0002371844490000028
wherein the content of the first and second substances,
Figure FDA0002371844490000029
embedding vectors, V, for each nodelIs a set of link nodes, VpFor the path node set, DNN () represents that the vector is operated on by a fully-connected neural network, usually two layers of neural networks are used.
9. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the steps of the method of any of claims 1 to 8 are implemented when the computer program is executed by the processor.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 8.
CN202010058591.3A 2020-01-17 2020-01-17 SDN routing performance prediction and optimization method, equipment and medium Active CN111314171B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010058591.3A CN111314171B (en) 2020-01-17 2020-01-17 SDN routing performance prediction and optimization method, equipment and medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010058591.3A CN111314171B (en) 2020-01-17 2020-01-17 SDN routing performance prediction and optimization method, equipment and medium

Publications (2)

Publication Number Publication Date
CN111314171A true CN111314171A (en) 2020-06-19
CN111314171B CN111314171B (en) 2023-06-30

Family

ID=71161490

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010058591.3A Active CN111314171B (en) 2020-01-17 2020-01-17 SDN routing performance prediction and optimization method, equipment and medium

Country Status (1)

Country Link
CN (1) CN111314171B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2022101378A1 (en) * 2020-11-16 2022-05-19 Interdigital Ce Patent Holdings, Sas Method and system for predicting a network characteristic
CN114710427A (en) * 2022-06-06 2022-07-05 山东卓朗检测股份有限公司 Local area network link performance testing method based on Ethernet technology
WO2024037136A1 (en) * 2022-08-15 2024-02-22 南京邮电大学 Graph structure feature-based routing optimization method and system

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108667734A (en) * 2018-05-18 2018-10-16 南京邮电大学 It is a kind of that the through street with LSTM neural networks is learnt by decision making algorithm based on Q
CN108900419A (en) * 2018-08-17 2018-11-27 北京邮电大学 Route decision method and device based on deeply study under SDN framework
CN108989207A (en) * 2018-10-26 2018-12-11 北京邮电大学 Route decision method, device and SDN equipment
US20190349287A1 (en) * 2018-05-10 2019-11-14 Dell Products L. P. System and method to learn and prescribe optimal network path for sdn

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190349287A1 (en) * 2018-05-10 2019-11-14 Dell Products L. P. System and method to learn and prescribe optimal network path for sdn
CN108667734A (en) * 2018-05-18 2018-10-16 南京邮电大学 It is a kind of that the through street with LSTM neural networks is learnt by decision making algorithm based on Q
CN108900419A (en) * 2018-08-17 2018-11-27 北京邮电大学 Route decision method and device based on deeply study under SDN framework
CN108989207A (en) * 2018-10-26 2018-12-11 北京邮电大学 Route decision method, device and SDN equipment

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2022101378A1 (en) * 2020-11-16 2022-05-19 Interdigital Ce Patent Holdings, Sas Method and system for predicting a network characteristic
CN114710427A (en) * 2022-06-06 2022-07-05 山东卓朗检测股份有限公司 Local area network link performance testing method based on Ethernet technology
WO2024037136A1 (en) * 2022-08-15 2024-02-22 南京邮电大学 Graph structure feature-based routing optimization method and system

Also Published As

Publication number Publication date
CN111314171B (en) 2023-06-30

Similar Documents

Publication Publication Date Title
CN110866190B (en) Method and device for training neural network model for representing knowledge graph
CN110995520B (en) Network flow prediction method and device, computer equipment and readable storage medium
WO2021082681A1 (en) Method and device for multi-party joint training of graph neural network
CN111314171A (en) Method, device and medium for predicting and optimizing SDN routing performance
US20200125957A1 (en) Multi-agent cooperation decision-making and training method
Blanco et al. A genetic algorithm to obtain the optimal recurrent neural network
CN110929047A (en) Knowledge graph reasoning method and device concerning neighbor entities
CN106897254B (en) Network representation learning method
CN113988464A (en) Network link attribute relation prediction method and equipment based on graph neural network
CN113489654B (en) Routing method, device, electronic equipment and storage medium
KR20220064866A (en) Method for co-design of hardware and neural network architectures using coarse-to-fine search, two-phased block distillation and neural hardware predictor
Sabbeh et al. Performance prediction of software defined network using an artificial neural network
CN111967271A (en) Analysis result generation method, device, equipment and readable storage medium
CN108365969B (en) Self-adaptive service combination method based on wireless sensor network
CN116894469B (en) DNN collaborative reasoning acceleration method, device and medium in end-edge cloud computing environment
TWI452529B (en) Combined with the system equivalent model of the system and its computer program products
CN116644804A (en) Distributed training system, neural network model training method, device and medium
CN114205251B (en) Switch link resource prediction method based on space-time characteristics
Li et al. A graph-based reinforcement learning method with converged state exploration and exploitation
CN116055406A (en) Training method and device for congestion window prediction model
Rouhani et al. Nonlinear neural network congestion control based on genetic algorithm for TCP/IP networks
CN113239272A (en) Intention prediction method and intention prediction device of network management and control system
Bagherian et al. A swarm based method for solving transit network design problem
JP7294450B2 (en) Apparatus, method and program for generating network topology
Vasilakos et al. On routing and performance comparison of techniques for packet-switched networks using learning automata

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