CN112560204B - Optical network route optimization method based on LSTM deep learning and related device thereof - Google Patents

Optical network route optimization method based on LSTM deep learning and related device thereof Download PDF

Info

Publication number
CN112560204B
CN112560204B CN202011112574.XA CN202011112574A CN112560204B CN 112560204 B CN112560204 B CN 112560204B CN 202011112574 A CN202011112574 A CN 202011112574A CN 112560204 B CN112560204 B CN 112560204B
Authority
CN
China
Prior art keywords
service
spectrum
deployed
matrix
time
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202011112574.XA
Other languages
Chinese (zh)
Other versions
CN112560204A (en
Inventor
郁小松
李新阳
赵永利
张�杰
何玲
郭学让
汪洋
张庚
王亚男
高凯强
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.)
State Grid Corp of China SGCC
Beijing University of Posts and Telecommunications
China Electric Power Research Institute Co Ltd CEPRI
Electric Power Research Institute of State Grid Xinjiang Electric Power Co Ltd
Original Assignee
State Grid Corp of China SGCC
Beijing University of Posts and Telecommunications
China Electric Power Research Institute Co Ltd CEPRI
Electric Power Research Institute of State Grid Xinjiang Electric Power 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 State Grid Corp of China SGCC, Beijing University of Posts and Telecommunications, China Electric Power Research Institute Co Ltd CEPRI, Electric Power Research Institute of State Grid Xinjiang Electric Power Co Ltd filed Critical State Grid Corp of China SGCC
Priority to CN202011112574.XA priority Critical patent/CN112560204B/en
Publication of CN112560204A publication Critical patent/CN112560204A/en
Application granted granted Critical
Publication of CN112560204B publication Critical patent/CN112560204B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/18—Network design, e.g. design based on topological or interconnect aspects of utility systems, piping, heating ventilation air conditioning [HVAC] or cabling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Geometry (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Artificial Intelligence (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computer Hardware Design (AREA)
  • Health & Medical Sciences (AREA)
  • Computational Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)
  • Telephonic Communication Services (AREA)

Abstract

One or more embodiments of the present specification provide an LSTM deep learning-based optical network route optimization method and a related apparatus thereof, which apply an LSTM prediction model in deep learning, learn a mapping relationship between link utilization and service remaining time characteristics in different scenes and a routing policy, take link and service overall information in a network as input, and rapidly obtain a reconstruction threshold through calculation of several layers of neural networks to decide whether to perform a routing optimization policy. The method solves the defects that the algorithm in the prior art is generally limited in feature learning capacity, poor in expression capacity of complex functions, single in optimization target parameters, limited in switch resources and huge in network resource overhead, and meanwhile, the threshold value can be changed in real time and efficiently along with the continuous arrival of the service, so that the condition of service blocking caused by network flow fluctuation or excessive bearing service volume is relieved, and the adaptive routing optimization of the optical network service is realized.

Description

Optical network route optimization method based on LSTM deep learning and related device thereof
Technical Field
One or more embodiments of the present disclosure relate to the field of routing optimization technologies, and in particular, to an optical network routing optimization method based on LSTM deep learning and a related apparatus thereof.
Background
The traditional optical network route optimization method includes two modes of rerouting and spectrum shifting. Rerouting, i.e. rerouting some traffic on non-optimal paths to select new optimal paths for transmission to improve the overall performance of the network. When a network carries a service for transmission, many services are not allocated on the shortest or optimal path because the resource requirements of the service are not satisfied. When the service request on the shortest path is removed, the service on other non-optimal paths can be transferred to the shortest or optimal path in a rerouting way, so that the use of network resources is optimized, and the overall performance of the network is improved. The spectrum shifting refers to searching whether idle low-order frequency slot resources exist or not in an original path of a service, if so, directly shifting the service to the idle frequency slot, and if not, not operating the service. Due to frequent establishment and release of service connection requests, the fragmentation degree of spectrum resources is very high, spectrum moving of services can integrate the spectrum resources, and therefore more services can be deployed to achieve the purpose of network optimization.
Some research results of optical network route optimization methods combined with deep learning methods have been emerged in recent years, but the currently proposed optical network route optimization algorithm based on deep learning has some defects: firstly, only traditional machine learning or shallow artificial neural network algorithms are used at present, and the traditional algorithms generally have the problems of limited feature learning capability, poor expression capability on complex functions and the like. Secondly, most of the existing schemes achieve the purpose of indirectly optimizing the service route by predicting certain parameters in the network and then according to the network parameters, however, the optimization of the network service route is a complex problem, and it is difficult to obtain obvious route optimization performance improvement aiming at the optimization of a single target parameter. Thirdly, a machine learning algorithm needs to be deployed in each switch, and then the switches collect the information to perform routing computation, but in an actual scene, the resources of the switches are very limited, and the requirements of the machine learning algorithm on the computing resources are difficult to meet. Fourthly, the traffic matrix is used as an input of a deep learning algorithm, but huge network resource overhead is needed for accurate traffic matrix measurement in a real scene.
Disclosure of Invention
In view of this, one or more embodiments of the present disclosure provide an optical network route optimization method based on LSTM deep learning and a related device thereof, so as to solve the problems in the prior art that an algorithm generally has limited feature learning capability, relatively poor expression capability for complex functions, single optimization target parameter, limited switch resource and huge network resource overhead.
In view of the above, one or more embodiments of the present specification provide an LSTM deep learning based optical network route optimization method and related apparatus, where the method includes:
receiving a request of a service to be deployed, analyzing the attribute of the service to be deployed, and calculating the shortest path of the service to be deployed according to the attribute of the service to be deployed;
judging whether the spectrum on the shortest path is sufficient;
if not, acquiring the remaining time of the deployed service on the blocked path which meets the frequency spectrum quantity required by the service to be deployed, calculating the time difference between the duration of the service to be deployed and the remaining time of the deployed service, and searching a suboptimal path for the service to be deployed for deployment;
collecting the data of the whole path spectrum occupation condition in the optical network, inputting the collected data into a trained LSTM deep learning model, and obtaining a preset reconstruction threshold;
and judging whether the time difference reaches the preset reconstruction threshold value or not, and distributing paths according to the judgment result.
After the determining whether the spectrum on the shortest path is sufficient includes:
and if the shortest path spectrum is sufficient, selecting and allocating the wavelength spectrum of the shortest path of the service to be deployed, updating the spectrum state information on the shortest path and finishing allocation.
The determining whether the time difference reaches the preset reconfiguration threshold and performing allocation according to the determination result includes:
if the time difference is greater than the preset reconstruction threshold, the service to be deployed is reconstructed after the time of the time difference, the wavelength spectrum of the shortest path of the service to be deployed is selected and allocated, the spectrum state information on the path is updated, the allocation is finished, and if the time difference is less than or equal to the preset reconstruction threshold, the allocation is finished.
The training step of the LSTM deep learning model comprises the following steps:
acquiring historical network spectrum occupation condition information and service information in a network;
generating a frequency spectrum matrix S and a service matrix R according to the historical network frequency spectrum occupation condition information and the service information and the time sequence, wherein the frequency spectrum matrix S and the service matrix R are included in the time sequence t 1 、t 2 To t n Then, a corresponding spectrum matrix S is obtained 1 、S 2 To S n And a service matrix R 1 、R 2 To R n ；
According to the current t i The acquired spectrum matrix S i And said traffic matrix R i As input data inputInitial LSTM prediction model, t i Is represented in the time series t 1 、t 2 To t n At any one time point in (1), use t i Next time t of i+1 Of the spectrum matrix S i+1 And said traffic matrix R i+1 As label data, training the selection of the preset reconstruction threshold;
according to each time t i Inputting the frequency spectrum matrix and the service matrix, simulating a predicted reconstruction threshold obtained by output, scoring the predicted reconstruction threshold according to a network service blocking rate obtained by using the predicted reconstruction threshold, and scoring the t i The prediction reconstruction threshold value obtained at the moment is taken as input information to obtain the t i+1 Simulating, scoring and inputting the prediction reconstruction threshold value of the moment until the moment is finished;
and repeating the four steps, repeatedly training the LSTM model, selecting different prediction reconstruction thresholds by the LSTM model for simulation, grading according to the service blocking rate in the network, and selecting the prediction reconstruction threshold with the highest score as a preset reconstruction threshold.
The service attributes comprise a source node, a destination node, service starting time, service duration and required frequency spectrum number;
the algorithm for calculating the shortest service path is Dijkstra algorithm.
The method adopted for selecting and allocating the wavelength spectrum of the shortest path of the service to be deployed is a First-Fit algorithm.
The form of the spectrum matrix S is:
Figure BDA0002729080500000031
wherein in the spectral matrix S
Figure BDA0002729080500000032
Representing routing nodes x in a network i And x j The spectrum occupation of the link between them, if node x i And x j Not directly connected, then
Figure BDA0002729080500000033
Is 0, if node x i And x j Direct coupling and spectrum occupancy of 0, then
Figure BDA0002729080500000034
Is 1.
The form of the service matrix R is as follows:
Figure BDA0002729080500000035
wherein in the traffic matrix R
Figure BDA0002729080500000036
Respectively representing services R i A source node and a sink node;
Figure BDA0002729080500000037
representative service R i The required spectrum resources;
Figure BDA0002729080500000038
representative service R i The remaining time of (c); p i Representative service R i The traffic matrix is updated at any time as traffic arrives and departs.
Based on the same inventive concept, one or more embodiments of the present specification further provide an electronic device, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor, and the processor implements the method as described in any one of the above items when executing the program.
Based on the same inventive concept, one or more embodiments of the present specification also provide a non-transitory computer-readable storage medium storing computer instructions for causing the computer to perform the method as described in any one of the above.
It can be seen from the foregoing that, according to the optical network route optimization method based on LSTM deep learning and the related apparatus thereof provided in one or more embodiments of the present invention, by learning the complex mapping relationship between the link utilization rate and the service remaining time characteristics in different scenarios and the routing policy, the overall information of the links and the services in the network is used as input, and the reconstruction threshold can be quickly obtained through several layers of neural network calculations to determine whether to perform the routing optimization policy.
Drawings
In order to more clearly illustrate one or more embodiments or prior art solutions of the present disclosure, the drawings that are needed in the description of the embodiments or prior art will be briefly described below, and it is apparent that the drawings in the description below are only one or more embodiments of the present disclosure, and that other drawings may be obtained by those skilled in the art without inventive effort from these drawings.
Fig. 1 is a flow diagram of a method for optical network route optimization according to one or more embodiments of the present disclosure;
fig. 2 is a flowchart illustrating specific steps of a method for optimizing an optical network route according to one or more embodiments of the present disclosure;
FIG. 3 is a diagram of an optical network topology according to one or more embodiments of the present disclosure;
FIG. 4 is a network virtual topology diagram of one or more embodiments of the present description;
fig. 5 is a schematic structural diagram of an electronic device according to one or more embodiments of the present disclosure.
Detailed Description
For the purpose of promoting a better understanding of the objects, aspects and advantages of the present disclosure, reference is made to the following detailed description taken in conjunction with the accompanying drawings.
It is to be understood that unless otherwise defined, technical or scientific terms used in one or more embodiments of the present disclosure should have the ordinary meaning as understood by one of ordinary skill in the art to which this disclosure belongs.
As described in the background section, the routing optimization method in the prior art also has the problems of limited feature learning capability, relatively poor expression capability to complex functions, single optimization target parameter, limited switch resources and huge network resource overhead, and is difficult to meet the routing optimization requirement when the network traffic fluctuates or the bearer traffic is excessive. In the process of implementing the present disclosure, the applicant finds that rerouting traffic can alleviate the blocking situation of the network, but needs to fully consider the probability and frequency of rerouting, i.e., whether route reconfiguration is necessary or not. After some services are processed, resources are released, idle spectrum resources are left on paths, and services on non-shortest or optimal paths in a network need a reconstruction operation probability to judge whether reconstruction is needed. The choice of reconstruction frequency is one of the issues that needs to be considered heavily, and it needs to be set dynamically according to the specific performance of the network. If the frequency is too low, the too few reconstruction times have very little effect, the network performance cannot be improved, and the expected effect cannot be achieved; if the frequency is too high, the network can frequently perform service reconstruction, increase the complexity of arrangement, increase the operation load of the network, and cause that newly arrived services cannot select the shortest path, possibly causing the blocking of more services with low delay requirements.
In view of this, one or more embodiments of the present disclosure provide an optical network route optimization scheme based on LSTM deep learning, which applies an LSTM prediction model in deep learning, learns a mapping relationship between link utilization and service remaining time characteristics under different scenarios and routing policies, takes link and service overall information in a network as input, and determines whether to perform a routing optimization policy by calculating and rapidly obtaining a reconstruction threshold through several layers of neural networks. The scheme solves the defects that the algorithm in the prior art is generally limited in feature learning capacity, poor in expression capacity of complex functions, single in optimization target parameters, limited in switch resources and huge in network resource overhead, and meanwhile, the threshold value can be changed in real time and efficiently along with the continuous arrival of services, so that the condition of service blocking caused by network flow fluctuation or excessive bearing service volume is relieved, and the adaptive routing optimization of the optical network services is realized.
The technical solutions of one or more embodiments of the present specification are described in detail below with reference to specific embodiments.
First, one or more embodiments of the present specification provide a method for optimizing an optical network route based on LSTM deep learning, which, with reference to fig. 1, includes the following steps:
101: receiving a request of a service to be deployed, analyzing the attribute of the service to be deployed, and calculating the shortest path of the service to be deployed according to the attribute of the service to be deployed.
102: and judging whether the spectrum on the shortest path is sufficient.
103: if not, acquiring the residual time of the deployed service on the blocked path which meets the number of the frequency spectrums needed by the service to be deployed, calculating the time difference between the duration time of the service to be deployed and the residual time of the deployed service, and searching a suboptimal path for the service to be deployed for deployment.
104: and collecting the data of the whole path spectrum occupation condition in the optical network, and inputting the collected data into the trained LSTM deep learning model to obtain a preset reconstruction threshold value.
105: and judging whether the time difference reaches the preset reconstruction threshold value or not, and distributing paths according to the judgment result.
Therefore, by the method, the reconfiguration threshold can be dynamically optimized in real time and high efficiency along with the continuous arrival of the service, so that the service blockage caused by the fluctuation of network flow or the excessive load-bearing service volume is relieved.
In one or more embodiments of the present disclosure, an LSTM deep learning model for predicting a preset reconstruction threshold is applied. The LSTM deep learning model requires training to implement the function of predicting the preset reconstruction threshold. Accordingly, the steps of the training method of the LSTM prediction model are as follows:
step 1: collecting historical frequency spectrum occupation condition information and service information in a network;
step 2: generating a frequency spectrum matrix S and a service matrix R according to the collected historical network frequency spectrum occupation data and service information according to a time sequence, wherein the time sequence comprises the time sequence t 1 、t 2 To t n Next, a corresponding spectrum matrix S is obtained 1 、S 2 To S n And a traffic matrix R 1 、R 2 To R n ；
Wherein, the network routing nodes are x in sequence 1 ，x 2 To x n (ii) a Current time link spectrum occupancy rate matrix S:
Figure BDA0002729080500000061
any of them
Figure BDA0002729080500000062
Representing a routing node x in a network i And x j The spectrum occupation of the link between nodes, if node x i And x j Not directly connected, then
Figure BDA0002729080500000063
Is 0, if node x i And x j Direct coupling and spectrum occupancy of 0, then
Figure BDA0002729080500000064
Is 1; deployed traffic matrix R:
Figure BDA0002729080500000065
wherein in the traffic matrix R
Figure BDA0002729080500000071
Respectively representing services R i A source node and a sink node;
Figure BDA0002729080500000072
representative service R i Required spectrum resources;
Figure BDA0002729080500000073
representative service R i The remaining time of (c); p i Representative service R i The working path of (1). The traffic matrix is updated at any time as traffic arrives and departs.
And 3, step 3: using the current t i Collected historical spectrum matrix S i And a traffic matrix R i Input the LSTM prediction model as input data, t i Is represented in the time series t 1 、t 2 To t n At any one time, use t i Next time t of i+1 Spectrum matrix S i+1 And a traffic matrix R i+1 As label data, selection of a reconstruction time threshold is trained.
And 4, step 4: according to each time t i And inputting the frequency spectrum matrix and the service matrix, simulating the output predicted reconstruction threshold, and scoring the predicted reconstruction threshold according to the network service blocking rate obtained by using the predicted reconstruction threshold. And will t i The prediction reconstruction threshold value obtained at the moment is taken as input information to obtain t i+1 And simulating the prediction reconstruction threshold value of the moment, scoring and inputting until the moment is finished.
And 5: and (4) repeating the steps 1 to 4, continuously and repeatedly training the LSTM model, wherein although the historical data input each time are the same, the LSTM model selects different prediction reconstruction thresholds to simulate and score according to the service blocking rate in the network, and repeatedly testing and selecting the prediction reconstruction threshold with the highest score, so that the selected reconstruction threshold enables the blocking rate of the current network to be as small as possible, and simultaneously the difference between the input data predicted by the prediction network at the next moment and the label data to be as small as possible. Therefore, the LSTM model obtains the optimal prediction result of the preset next moment according to the frequency spectrum matrix and the service matrix information input at the current moment.
As shown in fig. 2, a flowchart of specific steps of a method for optimizing an optical network route according to an embodiment of the present invention includes the following steps:
201: and when the service request arrives, calculating an optimal path. The optical network service dynamically arrives, and the service attributes such as a source node, a sink node, service start time, service duration, required frequency spectrum number and the like are analyzed; meanwhile, calculating the optimal path of the service by using a Dijkstra algorithm according to the source and destination nodes;
202: and judging whether the link frequency spectrum is sufficient. Judging whether the spectrum resources on each section of link are sufficient along the optimal path, and dividing the judgment into two conditions according to the judgment result:
if the spectrum on the optimal path link is sufficient, skipping the following steps and executing step 208;
if the spectrum on a certain link of the optimal path is insufficient, execute step 203;
203: the time difference is obtained. Acquiring a service set of a blocked road section on an optimal path, screening deployed services meeting the frequency spectrum number required by the services to be deployed, then selecting the services with the minimum residual time, and acquiring the residual time T of the services end (ii) a Calculating the duration T of the service to be deployed hold And T end Time difference T of r ；
204: and deploying the service to be reconstructed. Constructing a virtual topology, disconnecting a blocked link with insufficient spectrum resources in the virtual topology, calculating a path for a service to be reconstructed by using a Dijkstra algorithm, and deploying the service on a suboptimal path;
205: the LSTM deep learning model is used. Firstly, collecting the whole link spectrum data in the current network, and using the link spectrum occupancy rate matrix S and the deployed service matrix R in the current network as input data. Then inputting the matrix S and the matrix R into a pre-trained LSTM deep learning model to obtain a preset reconstruction time threshold T at the next moment TH ；
206: and judging whether to perform service reconfiguration. Judging the time difference T r And reconstructing the time threshold T at the next moment TH And dividing the judgment result into two conditions:
if the time difference T r Not satisfying the reconstruction time threshold T TH Ending the process;
if time difference T r Satisfying a reconstruction time threshold T TH Step 207 is executed;
207: and (5) service reconfiguration. The remaining time T of the deployed service screened by step 203 in the suboptimal path for the service to be reconstructed end After time, the screened deployed service leaves, the link spectrum resource state information is updated, service reconstruction is carried out, and step 208 is executed; firstly, a screened deployed service leaves, updating link spectrum resource state information, and then selecting and distributing spectrum resources on an optimal path by using a First-First algorithm to realize service deployment to be reconstructed;
208: and (5) service deployment. And reasonably selecting and distributing link spectrum resources for the service by using a First-Fit algorithm, updating spectrum state information on the path, and ending the process.
Based on the above method, one or more embodiments of the present specification further provide a 6-node, 8-link topology network optimization method, including:
as shown in fig. 3, an optical network topology diagram according to an embodiment of the present invention includes the following specific steps:
firstly, training a deep learning model under a 6-node and 8-link topological network of fig. 3:
step 1: collecting historical spectrum occupation condition information and service information in a network, wherein parameters comprise the spectrum occupation rate of a link at each moment; the source node and the sink node of the deployed service, the required spectrum resource, the remaining time and the working path.
Step 2: obtaining a historical spectrum matrix and a historical service matrix at t i ，t i+1 ，…，t i+n Generating a 6 x 6 link spectrum occupancy rate matrix S at a time i ：
Figure BDA0002729080500000081
And a deployed traffic matrix R i ：
Figure BDA0002729080500000091
And step 3: at this point, the raw data for training the model, i.e., the time series t, is obtained i ，t i+1 ，…，t i+n Lower link spectrum occupancy matrix S i ，S i+1 ，…，S i+n And a deployed traffic matrix R i ，R i+1 ，…，R i+n . The raw data is input to the LSTM.
And 4, step 4: suppose the kth training, time t i ，t i+1 ，…，t i+n The obtained reconstruction threshold is
Figure BDA0002729080500000098
And respectively simulating, and grading according to the service blocking rate.
And 5: and (4) repeating the steps 1-4, and continuously and repeatedly training the LSTM model to enable the selected reconstruction threshold value to enable the blocking rate of the current network at the next moment to be as small as possible. And the LSTM model obtains an optimal prediction result of the preset next moment according to the frequency spectrum matrix and the service matrix information input at the current moment, and the LSTM model training is finished at the moment.
After the model pre-training is completed, the service is completed according to the following steps:
step 1: service request r arrives, the request attribute is resolved, assuming the service has a service duration T from node 1 to node 6 hold The number of desired spectra is N s Calculating the optimal path to be 1-2-4-6 by a Dijkstra algorithm;
step 2: judging whether the frequency spectrum number on the path 1-2-4-6 meets the frequency spectrum number N required by the service r s If the link 2-4 is large in traffic and the spectrum resource is not enough to deploy the service r, it needs to be determined whether the service r needs to be reconstructed; if the number of frequency spectrums on 1-2-4-6 meets the number of frequency spectrums N required by service r s Directly executing the step 8;
and 3, step 3: due to the blockage of the links 2-4,at the moment, the required frequency spectrum number N meeting the service r is screened out on the links 2-4 s Assuming that r exists on links 2-4 at this time a And r b The frequency spectrum number of the two services is more than N s The remaining time is
Figure BDA0002729080500000093
And
Figure BDA0002729080500000094
suppose that
Figure BDA0002729080500000095
Thus calculating the time difference
Figure BDA0002729080500000096
Figure BDA0002729080500000097
And 4, step 4: as shown in fig. 4, a virtual topology is constructed for the network virtual topology diagram of this embodiment, a blocked link with insufficient spectrum resources is disconnected in the virtual topology, and a Dijkstra algorithm is used to calculate a path for a service r, so that the optimal path is 1-2-3-5-6, and the path can be successfully deployed with the service r after being judged;
and 5: firstly, collecting integral link spectrum data in a current network, acquiring a link spectrum occupancy rate matrix and a deployed service matrix in the current network as input data, then inputting the matrixes S and R into a pre-trained LSTM deep learning model, and acquiring a preset reconstruction time threshold T at the next time TH ；
Step 6: judging the time difference T r With a predetermined reconstruction time threshold T TH Size of (c), if T r >T TH Go to step 7, if T r <T TH Ending the process;
and 7: service r elapsed time on path 1-2-3-5-6
Figure BDA0002729080500000101
Traffic r on links 2-4 a Leaving, releasing spectrum resources, carrying out service reconstruction on the service r, moving to the optimal path 1-2-4-6 on the suboptimal path 1-2-3-5-6, and executing the step 8;
and 8: and reasonably selecting and distributing link spectrum resources for the service r by using a First-Fit algorithm, updating spectrum state information on the path, and ending the process.
Based on the same inventive concept, corresponding to any of the above embodiments, one or more embodiments of the present specification further provide an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the program, the LSTM deep learning based optical network route optimization method according to any of the above embodiments is implemented.
Fig. 5 is a schematic diagram illustrating a more specific hardware structure of an electronic device according to this embodiment, where the electronic device may include: a processor 1010, a memory 1020, an input/output interface 1030, a communication interface 1040, and a bus 1050. Wherein the processor 1010, memory 1020, input/output interface 1030, and communication interface 1040 are communicatively coupled to each other within the device via bus 1050.
The processor 1010 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an Application Specific Integrated Circuit (ASIC), or one or more Integrated circuits, and is configured to execute related programs to implement the technical solutions provided in the embodiments of the present disclosure.
The Memory 1020 may be implemented in the form of a ROM (Read Only Memory), a RAM (Random Access Memory), a static storage device, a dynamic storage device, or the like. The memory 1020 may store an operating system and other application programs, and when the technical solution provided by the embodiments of the present specification is implemented by software or firmware, the relevant program codes are stored in the memory 1020 and called to be executed by the processor 1010.
The input/output interface 1030 is used for connecting an input/output module to input and output information. The i/o module may be configured as a component in a device (not shown) or may be external to the device to provide a corresponding function. The input devices may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output devices may include a display, a speaker, a vibrator, an indicator light, etc.
The communication interface 1040 is used for connecting a communication module (not shown in the drawings) to implement communication interaction between the present apparatus and other apparatuses. The communication module can realize communication in a wired mode (such as USB, network cable and the like) and also can realize communication in a wireless mode (such as mobile network, WIFI, bluetooth and the like).
Bus 1050 includes a path that transfers information between various components of the device, such as processor 1010, memory 1020, input/output interface 1030, and communication interface 1040.
It should be noted that although the above-mentioned device only shows the processor 1010, the memory 1020, the input/output interface 1030, the communication interface 1040 and the bus 1050, in a specific implementation, the device may also include other components necessary for normal operation. In addition, those skilled in the art will appreciate that the above-described apparatus may also include only those components necessary to implement the embodiments of the present description, and not necessarily all of the components shown in the figures.
The electronic device of the foregoing embodiment is used to implement the corresponding LSTM deep learning-based optical network route optimization method in any of the foregoing embodiments, and has the beneficial effects of the corresponding method embodiment, which are not described herein again.
Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, one or more embodiments of the present specification further provide a non-transitory computer-readable storage medium storing computer instructions for causing the computer to perform the LSTM deep learning based optical network route optimization method according to any of the above-mentioned embodiments.
Computer-readable media, including both permanent and non-permanent, removable and non-removable media, for storing information may be implemented in any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device.
The computer instructions stored in the storage medium of the above embodiment are used to enable the computer to execute the method for optimizing the optical network route based on LSTM deep learning according to any of the above embodiments, and have the beneficial effects of corresponding method embodiments, and are not described herein again.
From the above, it can be seen that the invention applies the LSTM prediction model in deep learning, updates the predicted optimal reconstruction threshold in real time based on spectrum state information and service information in an optical network link, does not optimize for a single target parameter, and does not need to use a traffic matrix as an input of a deep learning algorithm, thereby better conforming to the real scene condition. Meanwhile, the reconstruction threshold value is updated and changed in real time according to the arrival of dynamic services, so that the network can be in the current optimal condition at any time, the probability and the frequency of reconstruction are fully considered, and the network performance can be well changed.
The foregoing description of specific embodiments has been presented for purposes of illustration and description. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.
Those of ordinary skill in the art will understand that: the discussion of any embodiment above is meant to be exemplary only, and is not intended to intimate that the scope of the disclosure, including the claims, is limited to these examples; within the spirit of the present disclosure, features from the above embodiments or from different embodiments may also be combined, steps may be implemented in any order, and there are many other variations of different aspects of one or more embodiments of the present description as described above, which are not provided in detail for the sake of brevity.
It is intended that the one or more embodiments of the present specification embrace all such alternatives, modifications and variations as fall within the broad scope of the appended claims. Therefore, any omissions, modifications, substitutions, improvements, and the like that may be made without departing from the spirit and principles of one or more embodiments of the present disclosure are intended to be included within the scope of the present disclosure.

Claims (8)

1. An optical network route optimization method based on LSTM deep learning is characterized by comprising the following steps:
receiving a request of a service to be deployed, analyzing the attribute of the service to be deployed, and calculating the shortest path of the service to be deployed according to the attribute of the service to be deployed;
judging whether the spectrum on the shortest path is sufficient;
if not, acquiring the residual time of the deployed service on the blocked path which meets the number of frequency spectrums needed by the service to be deployed, calculating the time difference between the duration time of the service to be deployed and the residual time of the deployed service, and searching a suboptimal path for the service to be deployed for deployment;
collecting the data of the whole path spectrum occupation condition in the optical network, and inputting the collected data into a trained LSTM deep learning model to obtain a preset reconstruction threshold;
the training step of the LSTM deep learning model comprises the following steps:
the first step is as follows: acquiring historical network spectrum occupation condition information and service information in a network;
the second step is that: network spectrum occupation condition information according to the historyService information, generating a spectrum matrix S and a service matrix R according to the time sequence, including in the time sequence t 1 、t 2 Obtaining corresponding frequency spectrum matrixes S1, S2 to Sn and service matrixes R1, R2 to Rn when the frequency spectrum matrixes are tn;
wherein, the form of the service matrix R is as follows:
Figure FDA0003921506670000011
wherein in the traffic matrix R
Figure FDA0003921506670000012
Respectively representing services R i A source node and a sink node;
Figure FDA0003921506670000013
representative service R i The required spectrum resources;
Figure FDA0003921506670000014
representative service R i The remaining time of (c); p i Representative service R i The service matrix is updated at any time along with the arrival and the departure of the service;
the third step: according to the current t i The collected spectrum matrix S i And said traffic matrix R i Input the initial LSTM prediction model as input data, t i Is represented in the time series t 1 、t 2 To t n At any one time, use t i Next time t of i+1 Of the spectrum matrix S i+1 And said traffic matrix R i+1 As label data, training the selection of the preset reconstruction threshold;
the fourth step: according to each time t i Inputting the frequency spectrum matrix and the service matrix, simulating a predicted reconstruction threshold obtained by output, scoring the predicted reconstruction threshold according to a network service blocking rate obtained by using the predicted reconstruction threshold, and scoring the t i The prediction reconstruction threshold value obtained at the moment is taken as input information to obtain the t i+1 Simulating, scoring and inputting the prediction reconstruction threshold value of the moment until the moment is ended;
repeating the four steps, and repeatedly training the LSTM model, wherein the LSTM model can select different prediction reconstruction thresholds for simulation and carries out scoring according to the service blocking rate in the network, and the prediction reconstruction threshold with the highest score is selected as a preset reconstruction threshold;
and judging whether the time difference reaches the preset reconstruction threshold value or not, and distributing paths according to the judgment result.
2. The LSTM deep learning-based optical network route optimization method of claim 1, wherein after determining whether the spectrum on the shortest path is sufficient, the method comprises:
and if the shortest path is sufficient, selecting and allocating the wavelength spectrum of the shortest path of the service to be deployed, updating the spectrum state information on the shortest path and finishing allocation.
3. The method of claim 1, wherein the determining whether the time difference reaches the preset reconstruction threshold and performing the distribution according to the determination result comprises:
if the time difference is greater than the preset reconstruction threshold, the service to be deployed is reconstructed after the time of the time difference, the wavelength spectrum of the shortest path of the service to be deployed is selected and allocated, the spectrum state information on the path is updated, the allocation is finished, and if the time difference is less than or equal to the preset reconstruction threshold, the allocation is finished.
4. The LSTM deep learning-based optical network route optimization method of claim 1, wherein the service attributes include source node, sink node, service start time, service duration and required spectrum number;
the algorithm for calculating the shortest service path is Dijkstra algorithm.
5. The LSTM deep learning-based optical network route optimization method of claim 3, wherein the method adopted for selecting and allocating the wavelength spectrum with shortest path to be deployed is a First-First algorithm.
6. The LSTM deep learning-based optical network route optimization method of claim 1, wherein the spectrum matrix S is of the form:
Figure FDA0003921506670000031
wherein in the spectral matrix S
Figure FDA0003921506670000032
Representing a routing node x in a network i And x j The spectrum occupation of the link between them, if node x i And x j Not directly connected, then
Figure FDA0003921506670000033
Is 0, if node x i And x j Direct coupling and spectrum occupancy of 0, then
Figure FDA0003921506670000034
Is 1.
7. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the method according to any of claims 1 to 6 when executing the program.
8. A non-transitory computer readable storage medium storing computer instructions for causing a computer to perform the method of any one of claims 1 to 6.
CN202011112574.XA 2020-10-16 2020-10-16 Optical network route optimization method based on LSTM deep learning and related device thereof Active CN112560204B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011112574.XA CN112560204B (en) 2020-10-16 2020-10-16 Optical network route optimization method based on LSTM deep learning and related device thereof

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011112574.XA CN112560204B (en) 2020-10-16 2020-10-16 Optical network route optimization method based on LSTM deep learning and related device thereof

Publications (2)

Publication Number Publication Date
CN112560204A CN112560204A (en) 2021-03-26
CN112560204B true CN112560204B (en) 2023-01-10

Family

ID=75042823

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011112574.XA Active CN112560204B (en) 2020-10-16 2020-10-16 Optical network route optimization method based on LSTM deep learning and related device thereof

Country Status (1)

Country Link
CN (1) CN112560204B (en)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113435567B (en) * 2021-06-25 2023-07-07 广东技术师范大学 Intelligent topology reconstruction method based on flow prediction, electronic equipment and storage medium
CN113452629B (en) * 2021-07-15 2022-06-14 深圳市高德信通信股份有限公司 Route switching system
CN114745618B (en) * 2022-04-11 2024-03-22 重庆邮电大学 Free space optical data center network architecture, topology reconstruction system and method
CN116032835A (en) * 2022-12-28 2023-04-28 中电信数智科技有限公司 Recommended route selection calculation method and device for customer service route
CN116137630B (en) * 2023-04-19 2023-08-18 井芯微电子技术(天津)有限公司 Method and device for quantitatively processing network service demands

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103581006A (en) * 2012-08-10 2014-02-12 北京邮电大学 System architecture for global optimization of flexible grid optical networks and global optimization method thereof
EP3343815B1 (en) * 2016-12-29 2019-05-08 Xieon Networks S.à r.l. Method and system for assigning resources in optical transport networks
CN109450587B (en) * 2018-10-24 2020-01-24 北京邮电大学 Spectrum integration processing method, device and storage medium
CN109617752B (en) * 2019-02-01 2021-06-01 苏州大学 Spectrum resource transaction method and system for virtual optical network
CN111586502B (en) * 2020-03-26 2021-06-15 北京邮电大学 Resource allocation method and system in elastic optical network

Also Published As

Publication number Publication date
CN112560204A (en) 2021-03-26

Similar Documents

Publication Publication Date Title
CN112560204B (en) Optical network route optimization method based on LSTM deep learning and related device thereof
CN112738820B (en) Dynamic deployment method and device of service function chain and computer equipment
US11082290B2 (en) Method and apparatus for optimizing a software defined network configuration
US20180204110A1 (en) Compressed neural network system using sparse parameters and design method thereof
US9680665B2 (en) Apparatus and method for dynamic hybrid routing in SDN networks to avoid congestion and balance loads under changing traffic load
CN111147387A (en) Flow control method and device for hybrid SDN network
WO2017117951A1 (en) Virtual mapping method and device
KR20200113744A (en) Method and apparatus for partitioning deep neural networks
CN109905784B (en) Service reconstruction method and equipment for optical network wavelength allocation
CN114205317A (en) Service function chain SFC resource allocation method based on SDN and NFV and electronic equipment
CN111181792A (en) SDN controller deployment method and device based on network topology and electronic equipment
CN117707795B (en) Graph-based model partitioning side collaborative reasoning method and system
Nguyen et al. Efficient virtual network embedding with node ranking and intelligent link mapping
CN111641556A (en) Routing resource allocation method and device of optical network
CN103179035A (en) Optical transmission network and method and device for selecting fixed routes thereof
US9391875B2 (en) Resource oriented dependency graph for network configuration
Bensalem et al. Scaling Serverless Functions in Edge Networks: A Reinforcement Learning Approach
CN113422726A (en) Service chain deployment method and device, storage medium and electronic device
CN116389266A (en) Method and device for slicing digital twin network based on reinforcement learning
EP3157209B1 (en) Route search apparatus and route search method
CN105515714A (en) Network upgrading perception protection method
CN115190021A (en) Slice deployment method for deterministic delay service and related equipment
KR102481227B1 (en) Apparatus for controlling cell association using monte carlo tree search and oprating method thereof
US20240171516A1 (en) Distributed neural network encoder-decoder system and method for traffic engineering
CN109921934B (en) Virtual network mapping method based on path splitting and energy consumption perception

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