WO2022083549A1 - Traffic signal conversion method and apparatus, electronic device, and storage medium - Google Patents
Traffic signal conversion method and apparatus, electronic device, and storage medium Download PDFInfo
- Publication number
- WO2022083549A1 WO2022083549A1 PCT/CN2021/124448 CN2021124448W WO2022083549A1 WO 2022083549 A1 WO2022083549 A1 WO 2022083549A1 CN 2021124448 W CN2021124448 W CN 2021124448W WO 2022083549 A1 WO2022083549 A1 WO 2022083549A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- traffic
- traffic signal
- topology
- otn
- conversion method
- Prior art date
Links
- 238000006243 chemical reaction Methods 0.000 title claims abstract description 60
- 238000000034 method Methods 0.000 title claims abstract description 55
- 230000002123 temporal effect Effects 0.000 claims abstract description 39
- 230000006870 function Effects 0.000 claims description 17
- 230000004913 activation Effects 0.000 claims description 10
- 230000015654 memory Effects 0.000 claims description 10
- 238000013528 artificial neural network Methods 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 6
- 238000004891 communication Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 17
- 238000010801 machine learning Methods 0.000 description 9
- 230000003287 optical effect Effects 0.000 description 7
- 230000006855 networking Effects 0.000 description 5
- 238000012549 training Methods 0.000 description 5
- 238000013527 convolutional neural network Methods 0.000 description 4
- 230000005693 optoelectronics Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 230000008094 contradictory effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
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
- H04L43/0876—Network utilisation, e.g. volume of load or congestion level
- H04L43/0888—Throughput
-
- 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
-
- 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
-
- 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
- G06N3/08—Learning methods
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04J—MULTIPLEX COMMUNICATION
- H04J3/00—Time-division multiplex systems
- H04J3/16—Time-division multiplex systems in which the time allocation to individual channels within a transmission cycle is variable, e.g. to accommodate varying complexity of signals, to vary number of channels transmitted
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04J—MULTIPLEX COMMUNICATION
- H04J3/00—Time-division multiplex systems
- H04J3/16—Time-division multiplex systems in which the time allocation to individual channels within a transmission cycle is variable, e.g. to accommodate varying complexity of signals, to vary number of channels transmitted
- H04J3/1605—Fixed allocated frame structures
- H04J3/1652—Optical Transport Network [OTN]
-
- 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
- H04L43/04—Processing captured monitoring data, e.g. for logfile generation
Definitions
- the embodiments of the present application relate to the field of communication technologies, and in particular, to a traffic signal conversion method, apparatus, electronic device, and storage medium.
- OTN Optical Transport Network
- eMBB Optical Transport Network
- uRLLC ultra-100G optical transport network technology
- mMTC three major application scenarios
- the intelligent adjustment of BOD mode is an important function that OTN's SDON (software-defined optical network) management and control system needs to provide to operator users.
- the embodiment of the present application provides a traffic signal conversion method, comprising: acquiring traffic signals at several times of OTN; using a graph convolution network to convert the traffic signals into traffic signals with temporal and spatial characteristics.
- the embodiment of the present application also provides a traffic signal conversion device, including: an acquisition module, used for acquiring the traffic signals at several times of the OTN; and spatial characteristics of flow signals.
- An embodiment of the present application further provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a program that can be executed by the at least one processor instructions, the instructions are executed by the at least one processor, so that the at least one processor can execute the above-mentioned traffic signal conversion method.
- Embodiments of the present application further provide a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the above-mentioned traffic signal conversion method is implemented.
- FIG. 1 is a schematic flowchart of a flow signal conversion method provided by a first embodiment of the present application
- Figure 2 is a schematic diagram of a traffic signal in an OTN networking environment
- Figure 3(a) is a schematic diagram of convolution of Euclidean spatial data using CNN
- FIG. 3(b) is a schematic diagram of convolving a non-Euclidean space using a circle convolution network in the traffic signal conversion method provided by the first embodiment of the present application;
- FIG. 4 is a schematic flowchart of a flow signal conversion method provided by a second embodiment of the present application.
- FIG. 5 is a schematic diagram of a model of a traffic signal of an ODU electrical layer according to time and space dimensions in the traffic signal conversion method provided by the second embodiment of the present application;
- FIG. 6 is a schematic flowchart of a traffic signal conversion method provided by a third embodiment of the present application.
- FIG. 7 is a schematic diagram of the service flow of the ODU electrical layer topology in the OTN networking environment
- FIG. 8 is a schematic diagram of a model of a traffic signal of an ODU electrical layer service according to time and space dimensions in the traffic signal conversion method provided by the third embodiment of the present application;
- FIG. 10 is a schematic flowchart of a traffic signal conversion method provided by the fourth embodiment of the present application.
- FIG. 11 is a schematic diagram of a schematic diagram of a traffic signal conversion method provided by the fourth embodiment of the present application.
- FIG. 12 is a schematic structural diagram of a module of a flow signal conversion device provided by a fifth embodiment of the present application.
- FIG. 13 is a schematic structural diagram of an electronic device provided by a sixth embodiment of the present application.
- the main purpose of the embodiments of this application is to propose a traffic signal conversion method, device, electronic device, and storage medium, which can perform effective machine learning on OTN to realize traffic prediction of OTN.
- the first embodiment of the present application relates to a method for converting traffic signals.
- a graph convolution network is used to convert traffic signals at several times of OTN into traffic signals with temporal and spatial characteristics.
- the non-European spatial data of the OTN traffic signal which is a graph convolutional network, is converted into European spatial data. Effective machine learning can be performed on the OTN traffic according to the converted traffic signal, thereby realizing OTN traffic prediction.
- the execution body of the traffic signal conversion method may be a server, wherein the server may be implemented by a single server or a server cluster composed of multiple servers.
- the server may be implemented by a single server or a server cluster composed of multiple servers. The following will take the server as an example for description. .
- FIG. 1 The specific flow of the traffic signal conversion method provided by the embodiment of the present application is shown in FIG. 1 , and specifically includes the following steps:
- S101 Acquire traffic signals of the OTN at several times.
- acquiring the traffic signals of the OTN at several times may specifically be acquiring the traffic signals of the ODU (Optical Data Unit) electrical layer topology of the OTN at several times.
- ODU Optical Data Unit
- FIG. 2 is a schematic diagram of traffic signals in an OTN networking environment.
- the OTN network is composed of optoelectronic hybrid scheduling network element nodes, and each optoelectronic hybrid scheduling network element node is roughly divided into two parts: an optical cross-connect device and an ODU electrical layer cross-connect device.
- the ODU traffic signal on each optoelectronic hybrid scheduling network element node is scheduled by the ODU electrical layer cross-connect device of this optoelectronic hybrid scheduling network element.
- ODU electrical layer pass-through traffic There are two types of traffic scheduling: ODU electrical layer pass-through traffic and ODU electrical layer add/drop traffic , the ODU electrical-layer pass-through traffic mainly refers to the ODU electrical-layer service traffic that is dispatched by the ODU electrical-layer cross-connect device and is not terminated; Layer business traffic.
- S102 Convert the traffic signal into a traffic signal with temporal and spatial features using a graph convolutional network.
- the machine learning of OTN can be effectively performed only by extracting the information of these two dimensions.
- Graph Convolutional networks can effectively extract the multi-dimensional data features of the irregular graph topology of OTN, convert the OTN traffic signal into a traffic signal with temporal and spatial characteristics, and further improve the OTN traffic signal. Traffic conducts effective machine learning for OTN traffic prediction.
- FIG. 3(a) is a schematic diagram of using CNN to convolve Euclidean spatial data
- FIG. 3(b) is a flow conversion method provided by an embodiment of the present application. Schematic illustration of convolution of non-Euclidean spatial data using graph convolutional networks.
- the traffic signals of the OTN can be monitored and counted according to the time and space dimensions, and the statistical traffic can be used as the input of the graph convolution network to train the graph convolution network.
- the traffic signals at several times of OTN obtained in S101 are used as the input of the graph convolution network, and the trained graph convolution network can be used to convert these traffic signals into traffic signals with temporal and spatial characteristics.
- the converted signal can be input into the neural network for training, so as to obtain the traffic prediction model of OTN, and then realize the Traffic forecast for OTN.
- the traffic signal conversion method provided by the embodiment of the present application converts the traffic signals of OTN at several moments into traffic signals with temporal and spatial characteristics through a graph convolution network, and can convert the non-European spatial data of the OTN traffic signal into Euclidean spatial data, so that effective machine learning can be carried out according to the converted traffic signal, and then the traffic prediction of OTN can be realized.
- the second embodiment of the present application relates to a traffic signal conversion method.
- the second embodiment is roughly the same as the first embodiment, with the main difference: in this embodiment, the traffic signals at several times of the ODU electrical layer topology of the OTN are obtained as a graph
- the OTN traffic model is established according to the through traffic, the add-on traffic and the total traffic of the topology nodes in the ODU electrical layer topology to realize the conversion of OTN traffic signals.
- FIG. 4 The specific flow of the traffic signal conversion method provided by the implementation manner of the present application is shown in FIG. 4 , and specifically includes the following steps:
- S201 Acquire the total traffic of the topology node at several times, the total traffic includes the through traffic and the add and drop traffic, and the topology node is a node included in the ODU electrical layer topology.
- S202 Using a graph convolutional network according to Convert the traffic signal into a traffic signal with temporal and spatial characteristics, where y' is the converted traffic signal, ⁇ is the activation function, and ⁇ jt is the convolution of the j-th dimension traffic attribute of the topology node at the t-th time Kernel parameter, L is the normalized graph Laplacian operator, x jt is the j-th dimension traffic attribute of the topology node at the t-th time, k is the total dimension of the traffic attribute, M is the total number of times, The dimensions of the traffic attributes include total traffic, through traffic, and add and drop traffic.
- FIG. 5 is a schematic diagram of a model of a traffic signal of an ODU electrical layer according to time and space dimensions.
- G t (V t , E, W)
- G t is the ODU electrical layer topology (traffic network topology) at the t-th time
- V t is the ODU electrical layer at the t-th time
- the topology node set of the topology, the number of topology nodes is set to n
- E is the edge (ie link) set of the ODU electrical layer topology
- W ie w ij in the figure
- the traffic signal in the ODU electrical layer topology can be defined as: X ⁇ R M ⁇ n ⁇ k represents the k-dimensional traffic attribute of n topology nodes at M times, and x i ⁇ R M ⁇ k represents the ith topology
- the k-dimensional traffic attribute of the node at M times, x it ⁇ R k represents the k-dimensional traffic attribute of the i-th topology node at the t-th time.
- x it there are two ways to define it:
- the add/drop traffic of i topology nodes, the traffic unit can be Gbit/15min, and there are Uf itq represents the add/drop traffic of port q of the i-th topology node at the t-th time, m represents the number of add/drop ports of the i-th topology node at the t-th time, including the client-side port and the line-side port, and the two are The pair appears; Sfit represents
- the traffic signal conversion method by acquiring the traffic signals at several times in the ODU electrical layer topology of the OTN as the input of the graph convolution network, according to the through traffic of the topology nodes in the ODU electrical layer topology, the add and drop traffic and the total traffic to establish the OTN traffic model, which can effectively realize the conversion of OTN traffic signals.
- the third embodiment of the present application relates to a traffic signal conversion method.
- the third embodiment is roughly the same as the first embodiment, and the main difference is that: in this embodiment, the services included in the topology nodes in the ODU electrical layer topology are obtained in several
- the traffic signal at the moment is used as the input of the graph convolution network, and the OTN traffic model is established according to the total number of services included in the ODU electrical layer, so as to realize the conversion of the OTN traffic signal.
- FIG. 6 The specific flow of the traffic signal conversion method provided by the embodiment of the present application is shown in FIG. 6 , and specifically includes the following steps:
- S301 Acquire traffic signals of services included in a topology node at several times, where the topology node is a node included in the ODU electrical layer topology.
- S302 Use graph convolutional network according to Convert the traffic signal into a traffic signal with temporal and spatial characteristics, where y' is the converted traffic signal, ⁇ is the activation function, and ⁇ jt is the traffic attribute of the jth service of the topology node at the tth time.
- convolution kernel parameters L is the normalized graph Laplacian operator, x jt is the traffic attribute of the jth service of the topology node at the t-th time, r is the total number of services included in the topology node, M is the total number of moments.
- FIG. 7 is a schematic diagram of service flow of an ODU electrical layer topology in an OTN networking environment.
- Establishing an OTN traffic model for the scenario corresponding to the figure can realize the prediction of ODU service traffic with BOD (Bandwidth On Demand) feature of OTN after traffic signal conversion.
- BOD Bandwidth On Demand
- the traffic size of a single ODU service is determined by the actual size of the traffic added to and from the service at the head and end nodes. Therefore, the BOD traffic bandwidth adjustment request can be sent to the head node of the service through the SDON controller, and the service path passes through. Intermediate nodes are not directly related.
- FIG. 8 is a schematic diagram of a model of a traffic signal of an ODU electrical layer service according to time and space dimensions.
- G t (V t , E, W)
- G t is the ODU electrical layer topology (traffic network topology) at the t-th time
- V t is the ODU electrical layer at the t-th time
- the topology node set of the topology, the number of topology nodes is set to n
- E is the edge (ie link) set of the ODU electrical layer topology
- W (ie w ij in the figure) is the weight adjacency matrix of adjacent topology nodes and has W ⁇ R n ⁇ n
- r represents the total number of services included in the topology node.
- the traffic signal of the services included in the topology nodes in the ODU electrical layer topology can be defined as: X ⁇ R M ⁇ n ⁇ r represents the traffic attributes of r services of n topology nodes at M times, and x i ⁇ R M ⁇ r represents the traffic attributes of the r services of the ith topology node at M times, and x it ⁇ R r represents the traffic attributes of the r services of the ith topology node at the t time.
- x it is defined in the following way :
- x it [x it1 ,x it1 ,...x itp ,...x itr ], where x itp represents the add/drop traffic of the ODU electrical layer service whose service ID is p at the t-th moment of the i-th topology node.
- y' is the converted traffic signal
- ⁇ is the activation function
- ⁇ jt is the convolution kernel parameter of the traffic attribute of the j-th service of the topology node at the t-th time
- L is the normalized graph Lapla Si operator
- x jt is the traffic attribute of the j-th service of the topology node at the t-th time
- r is the total number of services included in the topology node
- M is the total number of times.
- FIG. 9 is another schematic flowchart of the traffic signal conversion method provided by the embodiment of the present application, which specifically includes the following steps:
- S301' Obtain the traffic signals of the services included in the topology node at several times, and the topology node is a node included in the ODU electrical layer topology.
- S302' Using a graph convolutional network according to Convert the traffic signal into a traffic signal with temporal and spatial characteristics, where y' is the converted traffic signal, ⁇ is the activation function, and ⁇ t is the convolution kernel of the add and drop traffic attributes of the topology node at the t-th time parameters, L is the normalized graph Laplacian operator, x t is the traffic attribute of the topological node at the t-th time, and M is the total number of times.
- x it can also be defined in the following way:
- Ufit represents the add/drop traffic of the i-th topology node at the t-th time
- m represents the number of add/drop ports of the i-th topology node at the t-th time, including only the ports on the client side
- Uf itj represents the traffic at the t-th time
- the add/drop traffic of the client-side port j of the ith topology node at the moment, the traffic unit is, for example, Gbit/15min.
- y' is the converted traffic signal
- the ⁇ is the activation function
- the ⁇ t is the convolution kernel parameter of the add-and-drop traffic attribute of the topology node at the t-th moment
- the L is the normalization
- the x t is the attribute of the add/drop traffic of the topology node at the t-th time
- the M is the total number at the time.
- the OTN is established according to the total number of services included in the ODU electrical layer by acquiring the services included in the topology nodes in the ODU electrical layer topology at several times as the input of the graph convolution network.
- the traffic model can effectively realize the conversion of OTN traffic signals.
- a fourth embodiment of the present application provides a traffic signal conversion method, which converts an OTN traffic signal into a traffic signal with temporal and spatial characteristics by using a graph convolutional network, and then predicts OTN according to the traffic signal with temporal and spatial characteristics traffic, which can effectively predict OTN traffic.
- FIG. 10 The specific flow of the traffic signal conversion method provided by the embodiment of the present application is shown in FIG. 10 , which specifically includes the following steps:
- S401 Acquire traffic signals of the OTN at several times.
- S402 Use a graph convolutional network to convert the traffic signal into a traffic signal with temporal and spatial features.
- S403 Predict the traffic of the OTN according to the traffic signal with temporal characteristics and spatial characteristics.
- S401-S402 are the same as S101-S102 in the first embodiment.
- S401-S402 are the same as S101-S102 in the first embodiment.
- S401-S402 are the same as S101-S102 in the first embodiment.
- S401-S402 are the same as S101-S102 in the first embodiment.
- S401-S402 are the same as S101-S102 in the first embodiment.
- S401-S402 are the same as S101-S102 in the first embodiment.
- S401-S402 are the same as S101-S102 in the first embodiment.
- a deep neural network may be used to predict the OTN traffic according to the traffic signal with temporal and spatial features.
- other neural networks can also be used to make predictions, such as a long short-term memory network (LSTM).
- LSTM long short-term memory network
- FIG. 11 is a schematic diagram of a schematic diagram of a traffic signal conversion method provided by an embodiment of the present application.
- the traffic signals of the ODU layer topology at different times of the OTN can be used as the input of the graph convolution network (GCN), and the temporal and spatial characteristics of the traffic signals of the ODU layer topology can be extracted through the graph convolution network.
- the extracted traffic signal with temporal and empty features is used as the input of DNN, and the graph convolutional network and DNN are trained by back-propagation algorithm in the form of supervised learning, and finally the prediction of OTN traffic is realized.
- a traffic prediction model can be formed according to FIG. 11 on the basis of the third embodiment.
- the traffic prediction model can be obtained by training and applied according to the following steps: 1.
- the service flow model of the ODU electrical layer topology in the network environment collect the flow signals X ⁇ R M ⁇ n ⁇ r of r services in the ODU electrical layer topology at M times; 2.
- the input of the DNN is used to extract the temporal and spatial features of the traffic signal through the graph convolution network; 3.
- the traffic signal with the temporal and spatial characteristics outputted by the graph convolutional network is used as the input of the DNN, and the M times later
- the service traffic signal of the actual ODU electrical layer is used as the output of the DNN, and the traffic prediction sample is marked, so as to train the supervised model of the graph convolution network and DNN; 4.
- Obtain the traffic prediction model according to the trained graph convolution network and DNN using the traffic prediction model to predict the service traffic signal of the ODU electrical layer; 5.
- the traffic signal conversion method provided by the embodiment of the present application converts the traffic signal of OTN into a traffic signal with temporal and spatial characteristics by using a graph convolution network, and then predicts the traffic of OTN according to the traffic signal with temporal and spatial characteristics, It can effectively realize the prediction of OTN traffic; further, through the traffic prediction of OTN, the topology nodes of the ODU electrical layer topology can be expanded and upgraded, or the services of the ODU electrical layer can be reconfigured to cope with the OTN traffic tide. network traffic congestion or unbalanced network service load caused by the effect.
- the fifth embodiment of the present application relates to a flow signal conversion device 500. As shown in FIG. 12, it includes an acquisition module 501 and a conversion module 502. The functions of each module are described in detail as follows:
- the acquisition module 501 is used to acquire the traffic signals of the OTN at several moments;
- the conversion module 502 is used to convert the traffic signal into a traffic signal with temporal features and spatial features by using a graph convolutional network.
- the acquisition module 501 is specifically used for:
- the acquisition module 501 is specifically used for:
- the total traffic includes the through traffic and the add and drop traffic
- the topology node is the node included in the ODU electrical layer topology.
- conversion module 502 is specifically used for:
- ⁇ is the activation function
- ⁇ jt is the convolution of the j-th dimension traffic attribute of the topology node at the t-th time Kernel parameter
- L is the normalized graph Laplacian operator
- x jt is the j-th dimension traffic attribute of the topology node at the t-th time
- k is the total dimension of the traffic attribute
- M is the total number of times
- the dimensions of the traffic attributes include total traffic, through traffic, and add and drop traffic.
- the acquisition module 501 is specifically used for:
- the traffic signals of the services included in the topology node at several times are obtained, and the topology node is a node included in the ODU electrical layer topology.
- conversion module 502 is specifically used for:
- ⁇ is the activation function
- ⁇ jt is the traffic attribute of the jth service of the topology node at the tth time.
- convolution kernel parameters L is the normalized graph Laplacian operator
- x jt is the traffic attribute of the jth service of the topology node at the t-th time
- r is the total number of services included in the topology node
- M is the total number of moments.
- conversion module 502 is specifically used for:
- ⁇ is the activation function
- ⁇ t is the convolution kernel of the add and drop traffic attributes of the topology node at the t-th time parameters
- L is the normalized graph Laplacian operator
- x t is the traffic attribute of the topological node at the t-th time
- M is the total number of times.
- the flow signal conversion device provided by the embodiment of the present application further includes a prediction module, wherein the prediction module is used for: predicting the flow of the OTN according to the flow signal with temporal characteristics and spatial characteristics.
- the prediction model is specifically used for: predicting the traffic of the OTN according to the traffic signal with temporal and spatial characteristics by using a deep neural network.
- this embodiment is a system example corresponding to the first, second, third and fourth embodiments, and this embodiment can be implemented in cooperation with the first, second, third and fourth embodiments .
- the related technical details mentioned in the first, second, third, and fourth embodiments are still valid in this embodiment, and are not repeated here in order to reduce repetition.
- the related technical details mentioned in this embodiment can also be applied to the first, second, third and fourth embodiments.
- each module involved in this embodiment is a logical module.
- a logical unit may be a physical unit, a part of a physical unit, or multiple physical units.
- a composite implementation of the unit in order to highlight the innovative part of the present application, this embodiment does not introduce units that are not closely related to solving the technical problem raised by the present application, but this does not mean that there are no other units in this embodiment.
- the sixth embodiment of the present application relates to an electronic device, as shown in FIG. 13 , including: at least one processor 601 ; and a memory 602 communicatively connected to the at least one processor 601 ; Instructions executed by one processor 601, the instructions are executed by at least one processor 601, so that at least one processor 601 can execute the above-mentioned traffic signal conversion method.
- the memory and the processor are connected by a bus, and the bus may include any number of interconnected buses and bridges, and the bus connects one or more processors and various circuits of the memory.
- the bus may also connect together various other circuits, such as peripherals, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein.
- the bus interface provides the interface between the bus and the transceiver.
- a transceiver may be a single element or multiple elements, such as multiple receivers and transmitters, providing a means for communicating with various other devices over a transmission medium.
- the data processed by the processor is transmitted on the wireless medium through the antenna, and further, the antenna also receives the data and transmits the data to the processor.
- the processor is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interface, voltage regulation, power management, and other control functions. Instead, memory may be used to store data used by the processor in performing operations.
- the seventh embodiment of the present application relates to a computer-readable storage medium storing a computer program.
- the above method embodiments are implemented when the computer program is executed by the processor.
- the aforementioned storage medium includes: U disk, mobile hard disk, Read-Only Memory (ROM, Read-Only Memory), Random Access Memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program codes .
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Environmental & Geological Engineering (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
Embodiments of the present invention relate to the technical field of communications. Disclosed is a traffic signal conversion method, comprising: acquiring traffic signals of an OTN at a plurality of time points; and using a graph convolutional network to convert the traffic signals into traffic signals having temporal features and spatial features. Also disclosed in the embodiments of the present invention are a traffic signal conversion apparatus, an electronic device, and a storage medium.
Description
交叉引用cross reference
本申请基于申请号为“202011127237.8”、申请日为2020年10月20日的中国专利申请提出,并要求该中国专利申请的优先权,该中国专利申请的全部内容在此以引入方式并入本申请。This application is based on the Chinese patent application with the application number "202011127237.8" and the application date is October 20, 2020, and claims the priority of the Chinese patent application. The entire content of the Chinese patent application is hereby incorporated by reference. Apply.
本申请实施例涉及通信技术领域,特别涉及一种流量信号转换方法、装置、电子设备及存储介质。The embodiments of the present application relate to the field of communication technologies, and in particular, to a traffic signal conversion method, apparatus, electronic device, and storage medium.
近年来,随着OTN(光传送网络)特别是超100G光传送网技术的快速发展,以满足5G时代下的三大应用场景(eMBB、uRLLC、mMTC)需求为导向,深挖5G中回传OTN组网场景中的AI技术应用,已被业界高度认可,具有极其重要的现实意义。In recent years, with the rapid development of OTN (Optical Transport Network), especially the ultra-100G optical transport network technology, to meet the needs of the three major application scenarios (eMBB, uRLLC, mMTC) in the 5G era, dig deep into the 5G backhaul. The application of AI technology in OTN networking scenarios has been highly recognized by the industry and has extremely important practical significance.
对OTN中ODU(光数据单元)层承载的二层业务流量进行预测,并根据预测结果对OTN组网中的网元设备节点进行扩容改造规划,或者对ODU层业务的交换粒度ODUk、ODUFlex进行BOD模式的智能化调整,是OTN的SDON(软件定义光网络)管控***需要提供给运营商用户的重要功能。Predict the layer 2 service traffic carried by the ODU (optical data unit) layer in the OTN, and plan the expansion and reconstruction of the network element equipment nodes in the OTN network according to the prediction results, or perform the switching granularity ODUk and ODUFlex of the ODU layer services. The intelligent adjustment of BOD mode is an important function that OTN's SDON (software-defined optical network) management and control system needs to provide to operator users.
然而,由于普通的神经网络只能处理欧式空间数据,而OTN涉及的数据为非欧式空间数据,因此普通的神经网络无法对OTN进行有效的机器学习,以实现OTN的流量预测。However, since ordinary neural networks can only process Euclidean spatial data, and the data involved in OTN is non-Euclidean spatial data, ordinary neural networks cannot perform effective machine learning on OTN to achieve OTN traffic prediction.
申请内容Application content
本申请实施例提供了一种流量信号转换方法,包括:获取OTN若干个时刻的流量信号;利用图卷积网络将所述流量信号转换为具有时间特征和空间特征 的流量信号。The embodiment of the present application provides a traffic signal conversion method, comprising: acquiring traffic signals at several times of OTN; using a graph convolution network to convert the traffic signals into traffic signals with temporal and spatial characteristics.
本申请实施例还提供了一种流量信号转换装置,包括:获取模块,用于获取OTN若干个时刻的流量信号;转换模块,用于利用图卷积网络将所述流量信号转换为具有时间特征和空间特征的流量信号。The embodiment of the present application also provides a traffic signal conversion device, including: an acquisition module, used for acquiring the traffic signals at several times of the OTN; and spatial characteristics of flow signals.
本申请实施例还提供了一种电子设备,包括:至少一个处理器;以及,与所述至少一个处理器通信连接的存储器;其中,所述存储器存储有可被所述至少一个处理器执行的指令,所述指令被所述至少一个处理器执行,以使所述至少一个处理器能够执行上述的流量信号转换方法。An embodiment of the present application further provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a program that can be executed by the at least one processor instructions, the instructions are executed by the at least one processor, so that the at least one processor can execute the above-mentioned traffic signal conversion method.
本申请实施例还提供了一种计算机可读存储介质,存储有计算机程序,所述计算机程序被处理器执行时实现上述的流量信号转换方法。Embodiments of the present application further provide a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the above-mentioned traffic signal conversion method is implemented.
一个或多个实施例通过与之对应的附图中的图片进行示例性说明,这些示例性说明并不构成对实施例的限定。One or more embodiments are exemplified by the pictures in the corresponding drawings, and these exemplified descriptions do not constitute limitations on the embodiments.
图1是本申请第一实施方式提供的流量信号转换方法的流程示意图;1 is a schematic flowchart of a flow signal conversion method provided by a first embodiment of the present application;
图2是OTN组网环境下的流量信号示意图;Figure 2 is a schematic diagram of a traffic signal in an OTN networking environment;
图3(a)为使用CNN对欧式空间数据进行卷积的示意图;Figure 3(a) is a schematic diagram of convolution of Euclidean spatial data using CNN;
图3(b)是本申请第一实施方式提供的流量信号转换方法中使用圈卷积网络对非欧式空间进行卷积的示意图;FIG. 3(b) is a schematic diagram of convolving a non-Euclidean space using a circle convolution network in the traffic signal conversion method provided by the first embodiment of the present application;
图4是本申请第二实施方式提供的流量信号转换方法的流程示意图;4 is a schematic flowchart of a flow signal conversion method provided by a second embodiment of the present application;
图5是本申请第二实施方式提供的流量信号转换方法中按照时间和空间维度的ODU电层的流量信号的模型示意图;5 is a schematic diagram of a model of a traffic signal of an ODU electrical layer according to time and space dimensions in the traffic signal conversion method provided by the second embodiment of the present application;
图6是本申请第三实施方式提供的流量信号转换方法的流程示意图;6 is a schematic flowchart of a traffic signal conversion method provided by a third embodiment of the present application;
图7是OTN组网环境下ODU电层拓扑的业务流量示意图;FIG. 7 is a schematic diagram of the service flow of the ODU electrical layer topology in the OTN networking environment;
图8是本申请第三实施方式提供的流量信号转换方法中按照时间和空间维度的ODU电层业务的流量信号的模型示意图;8 is a schematic diagram of a model of a traffic signal of an ODU electrical layer service according to time and space dimensions in the traffic signal conversion method provided by the third embodiment of the present application;
图9是本申请第三实施方式提供的流量信号转换方法的另一流程示意图;9 is another schematic flowchart of the traffic signal conversion method provided by the third embodiment of the present application;
图10是本申请第四实施方式提供的流量信号转换方法的流程示意图;10 is a schematic flowchart of a traffic signal conversion method provided by the fourth embodiment of the present application;
图11是本申请第四实施方式提供的流量信号转换方法的原理示例图;11 is a schematic diagram of a schematic diagram of a traffic signal conversion method provided by the fourth embodiment of the present application;
图12是本申请第五实施方式提供的流量信号转换装置的模块结构示意图;12 is a schematic structural diagram of a module of a flow signal conversion device provided by a fifth embodiment of the present application;
图13是本申请第六实施方式提供的电子设备的结构示意图。FIG. 13 is a schematic structural diagram of an electronic device provided by a sixth embodiment of the present application.
为使本申请实施例的目的、技术方案和优点更加清楚,下面将结合附图对本申请的各实施例进行详细的阐述。然而,本领域的普通技术人员可以理解,在本申请各实施例中,为了使读者更好地理解本申请而提出了许多技术细节。但是,即使没有这些技术细节和基于以下各实施例的种种变化和修改,也可以实现本申请所要求保护的技术方案。以下各个实施例的划分是为了描述方便,不应对本申请的具体实现方式构成任何限定,各个实施例在不矛盾的前提下可以相互结合相互引用。In order to make the objectives, technical solutions and advantages of the embodiments of the present application more clear, each embodiment of the present application will be described in detail below with reference to the accompanying drawings. However, those of ordinary skill in the art can understand that, in each embodiment of the present application, many technical details are provided for the reader to better understand the present application. However, even without these technical details and various changes and modifications based on the following embodiments, the technical solutions claimed in the present application can be realized. The following divisions of the various embodiments are for the convenience of description, and should not constitute any limitation on the specific implementation of the present application, and the various embodiments may be combined with each other and referred to each other on the premise of not contradicting each other.
本申请实施例的主要目的在于提出一种流量信号转换方法、装置、电子设备及存储介质,可以对OTN进行有效的机器学习,以实现OTN的流量预测。The main purpose of the embodiments of this application is to propose a traffic signal conversion method, device, electronic device, and storage medium, which can perform effective machine learning on OTN to realize traffic prediction of OTN.
本申请第一实施方式涉及一种流量信号转换方法,通过获取OTN若干个时刻的流量信号,利用图卷积网络将OTN若干个时刻的流量信号转换为具有时间特征和空间特征的流量信号。通过图卷积网络将OTN的流量信号这一非欧式空间数据转换为欧式空间数据,可以根据转换后的流量信号对OTN的流量进行有效的机器学习,进而实现OTN的流量预测。The first embodiment of the present application relates to a method for converting traffic signals. By acquiring traffic signals at several times of OTN, a graph convolution network is used to convert traffic signals at several times of OTN into traffic signals with temporal and spatial characteristics. The non-European spatial data of the OTN traffic signal, which is a graph convolutional network, is converted into European spatial data. Effective machine learning can be performed on the OTN traffic according to the converted traffic signal, thereby realizing OTN traffic prediction.
应当说明的是,本申请实施方式提供的流量信号转换方法的执行主体可以为服务端,其中,服务端可由单独的服务器或多个服务器组成的服务器集群来实现,以下以服务端为例进行说明。It should be noted that the execution body of the traffic signal conversion method provided by the embodiments of the present application may be a server, wherein the server may be implemented by a single server or a server cluster composed of multiple servers. The following will take the server as an example for description. .
本申请实施方式提供的流量信号转换方法的具体流程如图1所示,具体包括以下步骤:The specific flow of the traffic signal conversion method provided by the embodiment of the present application is shown in FIG. 1 , and specifically includes the following steps:
S101:获取OTN若干个时刻的流量信号。S101: Acquire traffic signals of the OTN at several times.
在一个具体的例子中,获取OTN若干个时刻的流量信号,具体可以是获取OTN的ODU(光数据单元)电层拓扑若干个时刻的流量信号。In a specific example, acquiring the traffic signals of the OTN at several times may specifically be acquiring the traffic signals of the ODU (Optical Data Unit) electrical layer topology of the OTN at several times.
请参考图2,其为OTN组网环境下的流量信号示意图。具体地,OTN网络由光电混合调度网元节点组建,每个光电混合调度网元节点大体分为光交叉设备与ODU电层交叉设备两部分。在各光电混合调度网元节点上的ODU的流量信号由本光电混合调度网元的ODU电层交叉设备进行调度,分为两种类型的流量调度:ODU电层穿通流量和ODU电层上下路流量,其中,ODU电层穿通流量主要是指经ODU电层交叉设备调度的、不被终结的ODU电层业务流量;ODU电层上下路流量主要是指在ODU电层交叉设备上下路的ODU电层业务流量。Please refer to FIG. 2 , which is a schematic diagram of traffic signals in an OTN networking environment. Specifically, the OTN network is composed of optoelectronic hybrid scheduling network element nodes, and each optoelectronic hybrid scheduling network element node is roughly divided into two parts: an optical cross-connect device and an ODU electrical layer cross-connect device. The ODU traffic signal on each optoelectronic hybrid scheduling network element node is scheduled by the ODU electrical layer cross-connect device of this optoelectronic hybrid scheduling network element. There are two types of traffic scheduling: ODU electrical layer pass-through traffic and ODU electrical layer add/drop traffic , the ODU electrical-layer pass-through traffic mainly refers to the ODU electrical-layer service traffic that is dispatched by the ODU electrical-layer cross-connect device and is not terminated; Layer business traffic.
S102:利用图卷积网络将流量信号转换为具有时间特征和空间特征的流量信号。S102: Convert the traffic signal into a traffic signal with temporal and spatial features using a graph convolutional network.
由于一般的机器学习方法(例如CNN(卷积神经网络))只能处理欧式空 间数据,而OTN为图拓扑数据,为非欧式空间数据,因此一般的机器学习方法无法对OTN进行有效的机器学习,进而实现OTN的流量预测。Since general machine learning methods (such as CNN (Convolutional Neural Network)) can only process Euclidean spatial data, and OTN is graph topology data, which is non-Euclidean spatial data, general machine learning methods cannot perform effective machine learning on OTN. , and then realize the traffic prediction of OTN.
由于OTN的图拓扑结构中不仅包含了拓扑节点的特征信息,还包含了拓扑节点与拓扑节点之间的拓扑结构信息,因此提取这两个维度的信息才能有效地对OTN进行机器学习。图卷积网络(Graph Convolutional networks,GCN)可以有效地提取OTN这一不规则的图拓扑结构的多维数据特征,将OTN的流量信号转换为具有时间特征和空间特征的流量信号,进而对OTN的流量进行有效的机器学习,以实现OTN的流量预测。Since the graph topology of OTN contains not only the feature information of the topological nodes, but also the topological structure information between the topological nodes and the topological nodes, the machine learning of OTN can be effectively performed only by extracting the information of these two dimensions. Graph Convolutional networks (GCN) can effectively extract the multi-dimensional data features of the irregular graph topology of OTN, convert the OTN traffic signal into a traffic signal with temporal and spatial characteristics, and further improve the OTN traffic signal. Traffic conducts effective machine learning for OTN traffic prediction.
请参考图3(a)和图3(b),其中,图3(a)为使用CNN对欧式空间数据进行卷积的示意图,图3(b)为本申请实施方式提供的流量转换方法中使用图卷积网络对非欧式空间数据进行卷积的示意图。Please refer to FIG. 3(a) and FIG. 3(b), wherein, FIG. 3(a) is a schematic diagram of using CNN to convolve Euclidean spatial data, and FIG. 3(b) is a flow conversion method provided by an embodiment of the present application. Schematic illustration of convolution of non-Euclidean spatial data using graph convolutional networks.
可选地,在训练图卷积网络时,可以对OTN的流量信号按照时间、空间维度做监控和统计,并将统计的流量作为图卷积网络的输入,进行图卷积网络的训练。在训练好之后,将S101获取的OTN若干个时刻的流量信号作为图卷积网络的输入,就可以利用训练好的图卷积网络将这些流量信号转换为具有时间特征和空间特征的流量信号。Optionally, when training the graph convolution network, the traffic signals of the OTN can be monitored and counted according to the time and space dimensions, and the statistical traffic can be used as the input of the graph convolution network to train the graph convolution network. After training, the traffic signals at several times of OTN obtained in S101 are used as the input of the graph convolution network, and the trained graph convolution network can be used to convert these traffic signals into traffic signals with temporal and spatial characteristics.
可以理解的是,利用图卷积网络将流量信号转换为具有时间特征和空间特征的流量信号之后,可以将转换后的信号输入至神经网络进行训练,从而得到OTN的流量预测模型,进而实现对OTN的流量预测。It can be understood that after using the graph convolutional network to convert the traffic signal into a traffic signal with temporal and spatial characteristics, the converted signal can be input into the neural network for training, so as to obtain the traffic prediction model of OTN, and then realize the Traffic forecast for OTN.
本申请实施方式提供的流量信号转换方法,通过图卷积网络将OTN若干个时刻的流量信号转换为具有时间特征和空间特征的流量信号,可以将OTN的流量信号这一非欧式空间数据转换为欧式空间数据,从而可以根据转换后的流量信号进行有效的机器学习,进而实现OTN的流量预测。The traffic signal conversion method provided by the embodiment of the present application converts the traffic signals of OTN at several moments into traffic signals with temporal and spatial characteristics through a graph convolution network, and can convert the non-European spatial data of the OTN traffic signal into Euclidean spatial data, so that effective machine learning can be carried out according to the converted traffic signal, and then the traffic prediction of OTN can be realized.
本申请第二实施方式涉及一种流量信号转换方法,第二实施方式与第一实施方式大致相同,主要区别在于:本实施方式中,获取OTN的ODU电层拓扑若干个时刻的流量信号作为图卷积网络的输入,根据ODU电层拓扑中拓扑节点的穿通流量和上下路流量及总流量来建立OTN的流量模型,实现OTN流量信号的转换。The second embodiment of the present application relates to a traffic signal conversion method. The second embodiment is roughly the same as the first embodiment, with the main difference: in this embodiment, the traffic signals at several times of the ODU electrical layer topology of the OTN are obtained as a graph For the input of the convolutional network, the OTN traffic model is established according to the through traffic, the add-on traffic and the total traffic of the topology nodes in the ODU electrical layer topology to realize the conversion of OTN traffic signals.
本申请实现方式提供的流量信号转换方法的具体流程如图4所示,具体包括以下步骤:The specific flow of the traffic signal conversion method provided by the implementation manner of the present application is shown in FIG. 4 , and specifically includes the following steps:
S201:获取拓扑节点若干个时刻的总流量,总流量包括穿通流量和上下路流量,拓扑节点为ODU电层拓扑所包含的节点。S201: Acquire the total traffic of the topology node at several times, the total traffic includes the through traffic and the add and drop traffic, and the topology node is a node included in the ODU electrical layer topology.
S202:利用图卷积网络根据
将流量信号转换为具有时间特征和空间特征的流量信号,其中,y'为转换后的流量信号,σ为激活函数,α
jt为拓扑节点在第t个时刻的第j维流量属性的卷积核参数,L为归一化的图拉普拉斯算子,x
jt为拓扑节点在第t个时刻的第j维流量属性,k为流量属性的总 维度,M为时刻的总个数,流量属性的维度包括总流量、穿通流量和上下路流量。
S202: Using a graph convolutional network according to Convert the traffic signal into a traffic signal with temporal and spatial characteristics, where y' is the converted traffic signal, σ is the activation function, and α jt is the convolution of the j-th dimension traffic attribute of the topology node at the t-th time Kernel parameter, L is the normalized graph Laplacian operator, x jt is the j-th dimension traffic attribute of the topology node at the t-th time, k is the total dimension of the traffic attribute, M is the total number of times, The dimensions of the traffic attributes include total traffic, through traffic, and add and drop traffic.
对于S201-S202,具体说明如下:For S201-S202, the specific instructions are as follows:
请参考图5,其为按照时间和空间维度的ODU电层的流量信号的模型示意图。具体地,对图5可以定义:G
t=(V
t,E,W),其中,G
t为第t个时刻ODU电层拓扑(流量网络拓扑),V
t为第t个时刻ODU电层拓扑的拓扑节点集合,拓扑节点的个数设为n,E为ODU电层拓扑的边(即链路)集合,W(即图中的w
ij)为相邻拓扑节点的权重邻接矩阵。
Please refer to FIG. 5 , which is a schematic diagram of a model of a traffic signal of an ODU electrical layer according to time and space dimensions. Specifically, for Figure 5, it can be defined: G t =(V t , E, W), where G t is the ODU electrical layer topology (traffic network topology) at the t-th time, and V t is the ODU electrical layer at the t-th time The topology node set of the topology, the number of topology nodes is set to n, E is the edge (ie link) set of the ODU electrical layer topology, and W (ie w ij in the figure) is the weight adjacency matrix of adjacent topology nodes.
根据上述定义,ODU电层拓扑中流量信号可定义为:X∈R
M×n×k表示n个拓扑节点在M个时刻的k维流量属性,x
i∈R
M×k表示第i个拓扑节点在M个时刻的k维流量属性,x
it∈R
k表示第i个拓扑节在第t个时刻的k维流量属性。针对x
it,可以有以下两种定义方式:
According to the above definition, the traffic signal in the ODU electrical layer topology can be defined as: X∈R M×n×k represents the k-dimensional traffic attribute of n topology nodes at M times, and x i ∈R M×k represents the ith topology The k-dimensional traffic attribute of the node at M times, x it ∈ R k represents the k-dimensional traffic attribute of the i-th topology node at the t-th time. For x it , there are two ways to define it:
方式一:x
it=[Sf
it,Tf
it,Uf
it],此时k=3,其中,Tf
it表示第t个时刻第i个拓扑节点的穿通流量,流量单位可取值Gbit/15min,且有
Tf
itp表示第t个时刻第i个拓扑节点的端口p的穿通流量,u表示第t个时刻第i个拓扑节点的穿通流量端口数且全部为线路侧端口;Uf
it表示第t个时刻第i个拓扑节点的上下路流量,流量单位可取值Gbit/15min,且有
Uf
itq表示第t个时刻第i个拓扑节点的端口q的上下路流量,m表示第t个时刻第i个拓扑节点的上下路端口数,包括客户侧端口和线路侧端口,且两者成对出现;Sf
it表示第t个时刻第i个拓扑节点的总流量,流量单位可取值Gbit/15min,且有Sf
it=Tf
it+Uf
it。
Mode 1: x it =[Sfit , Tfit , Ufit ], at this time k=3, where Tfit represents the through-flow of the i -th topology node at the t-th time, and the flow unit can take the value of Gbit/15min, and have Tf itp represents the through traffic of port p of the ith topology node at the t th time, u represents the number of through traffic ports of the ith topology node at the t th time, and all of them are line-side ports; Uf it represents the t th time The add/drop traffic of i topology nodes, the traffic unit can be Gbit/15min, and there are Uf itq represents the add/drop traffic of port q of the i-th topology node at the t-th time, m represents the number of add/drop ports of the i-th topology node at the t-th time, including the client-side port and the line-side port, and the two are The pair appears; Sfit represents the total flow of the i -th topology node at the t-th time, and the flow unit can take the value of Gbit/15min, and there is Sfit = Tfit + Ufit .
方式二:x
it=[Sf
it],此时k=1。
Mode 2: x it =[ Sfit ], at this time k=1.
依据图卷积公式:y=Θ*gx=σ(α(L)x),其中,y表示图卷积的输出,x表示图卷积的输入,*g表示谱方法的图卷积算子,表示图信号x(即第t个时刻各拓扑节点的流量)与卷积核函数Θ的乘积,则有
其中,y'为转换后的流量信号,σ为激活函数,α
jt为拓扑节点在第t个时刻的第j维流量属性的卷积核参数,L为归一化的图拉普拉斯算子,x
jt为拓扑节点在第t个时刻的第j维流量属性,k为流量属性的总维度,M为时刻的总个数,流量属性的维度包括总流量、穿通流量和上下路流量。
According to the graph convolution formula: y=Θ*gx=σ(α(L)x), where y represents the output of the graph convolution, x represents the input of the graph convolution, and *g represents the graph convolution operator of the spectral method , representing the product of the graph signal x (that is, the flow of each topology node at the t-th moment) and the convolution kernel function Θ, then we have Among them, y' is the converted traffic signal, σ is the activation function, α jt is the convolution kernel parameter of the j-th dimension traffic attribute of the topology node at the t-th time, and L is the normalized graph Laplacian algorithm , x jt is the j-th dimension traffic attribute of the topology node at the t-th time, k is the total dimension of the traffic attribute, M is the total number of times, and the dimension of the traffic attribute includes the total traffic, the through traffic and the add and drop traffic.
本申请实施方式提供的流量信号转换方法,通过获取OTN的ODU电层拓扑若干个时刻的流量信号作为图卷积网络的输入,根据ODU电层拓扑中拓扑节点的穿通流量和上下路流量及总流量来建立OTN的流量模型,可以有效实现对OTN流量信号的转换。In the traffic signal conversion method provided by the embodiment of the present application, by acquiring the traffic signals at several times in the ODU electrical layer topology of the OTN as the input of the graph convolution network, according to the through traffic of the topology nodes in the ODU electrical layer topology, the add and drop traffic and the total traffic to establish the OTN traffic model, which can effectively realize the conversion of OTN traffic signals.
本申请第三实施方式涉及一种流量信号转换方法,第三实施方式与第一实 施方式大致相同,主要区别在于:在本实施方式中,获取ODU电层拓扑中拓扑节点所包含业务在若干个时刻的流量信号作为图卷积网络的输入,根据ODU电层中所包含业务的总条数来建立OTN的流量模型,从而实现OTN的流量信号的转换。The third embodiment of the present application relates to a traffic signal conversion method. The third embodiment is roughly the same as the first embodiment, and the main difference is that: in this embodiment, the services included in the topology nodes in the ODU electrical layer topology are obtained in several The traffic signal at the moment is used as the input of the graph convolution network, and the OTN traffic model is established according to the total number of services included in the ODU electrical layer, so as to realize the conversion of the OTN traffic signal.
本申请实施方式提供的流量信号转换方法的具体流程如图6所示,具体包括以下步骤:The specific flow of the traffic signal conversion method provided by the embodiment of the present application is shown in FIG. 6 , and specifically includes the following steps:
S301:获取拓扑节点所包含业务在若干个时刻的流量信号,拓扑节点为ODU电层拓扑所包含的节点。S301: Acquire traffic signals of services included in a topology node at several times, where the topology node is a node included in the ODU electrical layer topology.
S302:利用图卷积网络根据
将流量信号转换为具有时间特征和空间特征的流量信号,其中,y'为转换后的流量信号,σ为激活函数,α
jt为拓扑节点在第t个时刻的第j条业务的流量属性的卷积核参数,L为归一化的图拉普拉斯算子,x
jt为拓扑节点在第t个时刻的第j条业务的流量属性,r为拓扑节点所包含业务的总条数,M为时刻的总个数。
S302: Use graph convolutional network according to Convert the traffic signal into a traffic signal with temporal and spatial characteristics, where y' is the converted traffic signal, σ is the activation function, and α jt is the traffic attribute of the jth service of the topology node at the tth time. convolution kernel parameters, L is the normalized graph Laplacian operator, x jt is the traffic attribute of the jth service of the topology node at the t-th time, r is the total number of services included in the topology node, M is the total number of moments.
对于S301-S302,详细说明如下:For S301-S302, the details are as follows:
请参考图7,其为OTN组网环境下ODU电层拓扑的业务流量示意图。针对该图对应的场景建立OTN的流量模型,可以在流量信号转换之后,实现针对OTN具备BOD(带宽按需分配)特征的ODU业务流量的预测。通常情况下,单条ODU业务的流量大小由该业务在首末节点的上下路流量的实际大小决定,因此BOD的流量带宽调整请求通过SDON控制器下发业务首节点即可,和业务路径经过的中间节点没有直接关系。Please refer to FIG. 7 , which is a schematic diagram of service flow of an ODU electrical layer topology in an OTN networking environment. Establishing an OTN traffic model for the scenario corresponding to the figure can realize the prediction of ODU service traffic with BOD (Bandwidth On Demand) feature of OTN after traffic signal conversion. Usually, the traffic size of a single ODU service is determined by the actual size of the traffic added to and from the service at the head and end nodes. Therefore, the BOD traffic bandwidth adjustment request can be sent to the head node of the service through the SDON controller, and the service path passes through. Intermediate nodes are not directly related.
请参考图8,其为按照时间和空间维度的ODU电层业务的流量信号的模型示意图。具体地,对图8可以定义:G
t=(V
t,E,W),其中,G
t为第t个时刻ODU电层拓扑(流量网络拓扑),V
t为第t个时刻ODU电层拓扑的拓扑节点集合,拓扑节点的个数设为n,E为ODU电层拓扑的边(即链路)集合,W(即图中的w
ij)为相邻拓扑节点的权重邻接矩阵且有W∈R
n×n,r表示拓扑节点所包含业务的总条数。
Please refer to FIG. 8 , which is a schematic diagram of a model of a traffic signal of an ODU electrical layer service according to time and space dimensions. Specifically, for Figure 8, it can be defined: G t =(V t , E, W), where G t is the ODU electrical layer topology (traffic network topology) at the t-th time, and V t is the ODU electrical layer at the t-th time The topology node set of the topology, the number of topology nodes is set to n, E is the edge (ie link) set of the ODU electrical layer topology, W (ie w ij in the figure) is the weight adjacency matrix of adjacent topology nodes and has W∈R n×n , r represents the total number of services included in the topology node.
根据上述定义,ODU电层拓扑中拓扑节点所包含业务的流量信号可定义为:X∈R
M×n×r表示n个拓扑节点在M个时刻的r条业务的流量属性,x
i∈R
M×r表示第i个拓扑节点在M个时刻的r条业务的流量属性,x
it∈R
r表示第i个拓扑节点在第t个时刻的r条业务的流量属性。针对x
it,考虑到各ODU电层拓扑中拓扑节点的上下路流量不尽相同,为保证ODU电层拓扑的拓扑节点基于业务流量信号的维数的统一性,采用以下方式对x
it进行定义:
According to the above definition, the traffic signal of the services included in the topology nodes in the ODU electrical layer topology can be defined as: X∈R M×n×r represents the traffic attributes of r services of n topology nodes at M times, and x i ∈ R M×r represents the traffic attributes of the r services of the ith topology node at M times, and x it ∈ R r represents the traffic attributes of the r services of the ith topology node at the t time. For x it , considering that the add/drop traffic of the topology nodes in each ODU electrical layer topology is not the same, in order to ensure the unity of the dimension of the service traffic signal based on the topology nodes of the ODU electrical layer topology, x it is defined in the following way :
x
it=[x
it1,x
it1,…x
itp,…x
itr],其中,x
itp表示业务ID为p的ODU电层业务在第i个拓扑节点的第t个时刻的上下路流量,流量单位可取值Gbit/15min,若第i个拓扑节点不是业务p的首末节点,即该业务不在第i个拓扑节点做上下路,则x
itp=0。
x it =[x it1 ,x it1 ,...x itp ,...x itr ], where x itp represents the add/drop traffic of the ODU electrical layer service whose service ID is p at the t-th moment of the i-th topology node. The unit can take the value of Gbit/15min. If the ith topology node is not the first and last node of the service p, that is, the service is not added or dropped on the ith topology node, then x itp =0.
依据图卷积公式:y=Θ*gx=σ(α(L)x),则有:According to the graph convolution formula: y=Θ*gx=σ(α(L)x), there are:
其中,y'为转换后的流量信号,σ为激活函数,α
jt为拓扑节点在第t个时刻的第j条业务的流量属性的卷积核参数,L为归一化的图拉普拉斯算子,x
jt为拓扑节点在第t个时刻的第j条业务的流量属性,r为拓扑节点所包含业务的总条数,M为时刻的总个数。
Among them, y' is the converted traffic signal, σ is the activation function, α jt is the convolution kernel parameter of the traffic attribute of the j-th service of the topology node at the t-th time, and L is the normalized graph Lapla Si operator, x jt is the traffic attribute of the j-th service of the topology node at the t-th time, r is the total number of services included in the topology node, and M is the total number of times.
在一个具体的例子中,请参考图9,其为本申请实施方式提供的流量信号转换方法的另一流程示意图,具体包括以下步骤:In a specific example, please refer to FIG. 9 , which is another schematic flowchart of the traffic signal conversion method provided by the embodiment of the present application, which specifically includes the following steps:
S301’:获取拓扑节点所包含业务在若干个时刻的流量信号,拓扑节点为ODU电层拓扑所包含的节点。S301': Obtain the traffic signals of the services included in the topology node at several times, and the topology node is a node included in the ODU electrical layer topology.
S302’:利用图卷积网络根据
将流量信号转换为具有时间特征和空间特征的流量信号,其中,y'为转换后的流量信号,σ为激活函数,α
t为拓扑节点在第t个时刻的上下路流量属性的卷积核参数,L为归一化的图拉普拉斯算子,x
t为拓扑节点在第t个时刻的上下路流量属性,M为时刻的总个数。
S302': Using a graph convolutional network according to Convert the traffic signal into a traffic signal with temporal and spatial characteristics, where y' is the converted traffic signal, σ is the activation function, and α t is the convolution kernel of the add and drop traffic attributes of the topology node at the t-th time parameters, L is the normalized graph Laplacian operator, x t is the traffic attribute of the topological node at the t-th time, and M is the total number of times.
即还可以采用以下方式对x
it进行定义:
That is, x it can also be defined in the following way:
x
it=[Uf
it]且有
其中,Uf
it表示第t个时刻第i个拓扑节点的上下路流量,m表示在第t个时刻第i个拓扑节点的上下路端口数,仅包括客户侧端口,Uf
itj表示在第t个时刻第i个拓扑节点的客户侧端口j的上下路流量,流量单位例如是Gbit/15min。
x it = [Uf it ] and there is Among them, Ufit represents the add/drop traffic of the i-th topology node at the t-th time, m represents the number of add/drop ports of the i-th topology node at the t-th time, including only the ports on the client side, and Uf itj represents the traffic at the t-th time The add/drop traffic of the client-side port j of the ith topology node at the moment, the traffic unit is, for example, Gbit/15min.
依据图卷积公式:y=Θ*gx=σ(α(L)x),则有:According to the graph convolution formula: y=Θ*gx=σ(α(L)x), there are:
其中,y'为转换后的流量信号,所述σ为激活函数,所述α
t为所述拓扑节点在第t个时刻的上下路流量属性的卷积核参数,所述L为归一化的图拉普拉斯算子,所述x
t为所述拓扑节点在第t个时刻的上下路流量属性,所述M为所述时刻的总个数。
Wherein, y' is the converted traffic signal, the σ is the activation function, the α t is the convolution kernel parameter of the add-and-drop traffic attribute of the topology node at the t-th moment, and the L is the normalization The graph Laplacian of , the x t is the attribute of the add/drop traffic of the topology node at the t-th time, and the M is the total number at the time.
本申请实施方式提供的流量信号转换方法,通过获取ODU电层拓扑中拓扑节点所包含业务在若干个时刻作为图卷积网络的输入,根据ODU电层中所包含业务的总条数来建立OTN的流量模型,可以有效实现对OTN流量信号的转换。In the traffic signal conversion method provided by the embodiment of the present application, the OTN is established according to the total number of services included in the ODU electrical layer by acquiring the services included in the topology nodes in the ODU electrical layer topology at several times as the input of the graph convolution network. The traffic model can effectively realize the conversion of OTN traffic signals.
本申请第四实施方式提供一种流量信号转换方法,通过利用图卷积网络将OTN的流量信号转换为具有时间特征和空间特征的流量信号,再根据具有时间特征和空间特征的流量信号预测OTN的流量,可以有效地实现对OTN流量的预测。A fourth embodiment of the present application provides a traffic signal conversion method, which converts an OTN traffic signal into a traffic signal with temporal and spatial characteristics by using a graph convolutional network, and then predicts OTN according to the traffic signal with temporal and spatial characteristics traffic, which can effectively predict OTN traffic.
本申请实施方式提供的流量信号转换方法的具体流程如图10所示,具体包括以下步骤:The specific flow of the traffic signal conversion method provided by the embodiment of the present application is shown in FIG. 10 , which specifically includes the following steps:
S401:获取OTN若干个时刻的流量信号。S401: Acquire traffic signals of the OTN at several times.
S402:利用图卷积网络将流量信号转换为具有时间特征和空间特征的流量信号。S402: Use a graph convolutional network to convert the traffic signal into a traffic signal with temporal and spatial features.
S403:根据具有时间特征和空间特征的流量信号预测OTN的流量。S403: Predict the traffic of the OTN according to the traffic signal with temporal characteristics and spatial characteristics.
其中,S401-S402与第一实施方式中的S101-S102相同,具体请参见第一实施方式中的相关描述,为了避免重复,这里不再赘述。Wherein, S401-S402 are the same as S101-S102 in the first embodiment. For details, please refer to the relevant description in the first embodiment. In order to avoid repetition, details are not repeated here.
S403中,在根据具有时间特征和空间特征的流量信号预测OTN的流量时,具体可以是利用深度神经网络(DNN)根据具有时间特征和空间特征的流量信号预测OTN的流量。当然,也可以使用其它的神经网络来进行预测,例如是长短期记忆网络(LSTM)等。In S403, when predicting the OTN traffic according to the traffic signal with temporal and spatial features, specifically, a deep neural network (DNN) may be used to predict the OTN traffic according to the traffic signal with temporal and spatial features. Of course, other neural networks can also be used to make predictions, such as a long short-term memory network (LSTM).
请参考图11,其为本申请实施方式提供的流量信号转换方法的原理示例图。在具体训练时,可以是将OTN不同时刻的ODU层拓扑的流量信号作为图卷积网络(GCN)的输入,通过图卷积网络提取ODU层拓扑的流量信号的时间特征和空间特征,再将提取后具有时间特征和空特征的流量信号作为DNN的输入,通过监督学习的形式对图卷积网络和DNN采用反向传播算法等进行模型训练,最终实现OTN流量的预测。Please refer to FIG. 11 , which is a schematic diagram of a schematic diagram of a traffic signal conversion method provided by an embodiment of the present application. During specific training, the traffic signals of the ODU layer topology at different times of the OTN can be used as the input of the graph convolution network (GCN), and the temporal and spatial characteristics of the traffic signals of the ODU layer topology can be extracted through the graph convolution network. The extracted traffic signal with temporal and empty features is used as the input of DNN, and the graph convolutional network and DNN are trained by back-propagation algorithm in the form of supervised learning, and finally the prediction of OTN traffic is realized.
实际应用中,可以在第三实施方式的基础上根据图11形成一个流量预测模型,具体可以根据以下步骤训练得到流量预测模型并应用流量预测模型:1、根据第三实施方式中定义的OTN组网环境下的ODU电层拓扑的业务流量模型,采集M个时刻ODU电层拓扑中r条业务的流量信号X∈R
M×n×r;2、将采集的业务流量信号作为图卷积网络的输入,通过图卷积网络对流量信号进行时间特征和空间特征的提取;3、将经过图卷积网络处理输出的具有时间特征和空间特征的流量信号作为DNN的输入,并用M个时刻后面时刻实际ODU电层的业务流量信号作为DNN的输出,进行流量预测样本标记,从而对图卷积网络和DNN的监督式模型训练;4、根据训练后的图卷积网络和DNN得到流量预测模型,利用流量预测模型对ODU电层的业务流量信号进行预测;5、根据流量预测结果,对OTN中ODU电层的业务带宽资源进行全局或者局部BOD优化带宽调整。
In practical applications, a traffic prediction model can be formed according to FIG. 11 on the basis of the third embodiment. Specifically, the traffic prediction model can be obtained by training and applied according to the following steps: 1. According to the OTN group defined in the third embodiment; The service flow model of the ODU electrical layer topology in the network environment, collect the flow signals X∈R M×n×r of r services in the ODU electrical layer topology at M times; 2. Use the collected service flow signals as a graph convolution network The input of the DNN is used to extract the temporal and spatial features of the traffic signal through the graph convolution network; 3. The traffic signal with the temporal and spatial characteristics outputted by the graph convolutional network is used as the input of the DNN, and the M times later The service traffic signal of the actual ODU electrical layer is used as the output of the DNN, and the traffic prediction sample is marked, so as to train the supervised model of the graph convolution network and DNN; 4. Obtain the traffic prediction model according to the trained graph convolution network and DNN , using the traffic prediction model to predict the service traffic signal of the ODU electrical layer; 5. According to the traffic prediction result, perform global or local BOD optimization bandwidth adjustment on the service bandwidth resources of the ODU electrical layer in the OTN.
本申请实施方式提供的流量信号转换方法,通过利用图卷积网络将OTN的流量信号转换为具有时间特征和空间特征的流量信号,再根据具有时间特征和空间特征的流量信号预测OTN的流量,可以有效地实现对OTN流量的预测;进一步地,通过对OTN的流量预测,可对ODU电层拓扑的拓扑节点进行扩容升级,或者对ODU电层的业务进行重构,以应对因OTN流量潮汐效应引起的网络流量拥塞或网络业务负载不均衡等问题。The traffic signal conversion method provided by the embodiment of the present application converts the traffic signal of OTN into a traffic signal with temporal and spatial characteristics by using a graph convolution network, and then predicts the traffic of OTN according to the traffic signal with temporal and spatial characteristics, It can effectively realize the prediction of OTN traffic; further, through the traffic prediction of OTN, the topology nodes of the ODU electrical layer topology can be expanded and upgraded, or the services of the ODU electrical layer can be reconfigured to cope with the OTN traffic tide. network traffic congestion or unbalanced network service load caused by the effect.
此外,本领域技术人员可以理解,上面各种方法的步骤划分,只是为了描述清楚,实现时可以合并为一个步骤或者对某些步骤进行拆分,分解为多个步骤,只要包括相同的逻辑关系,都在本专利的保护范围内;对算法中或者流程中添加无关紧要的修改或者引入无关紧要的设计,但不改变其算法和流程的核心设计都在该专利的保护范围内。In addition, those skilled in the art can understand that the division of steps of the various methods above is only for the purpose of describing clearly, and may be combined into one step or split into several steps during implementation, and decomposed into multiple steps, as long as the same logical relationship is included , are all within the protection scope of this patent; adding insignificant modifications to the algorithm or process or introducing insignificant designs, but not changing the core design of the algorithm and process are all within the protection scope of this patent.
本申请第五实施方式涉及一种流量信号转换装置500,如图12所示,包括获取模块501和转换模块502,各模块功能详细说明如下:The fifth embodiment of the present application relates to a flow signal conversion device 500. As shown in FIG. 12, it includes an acquisition module 501 and a conversion module 502. The functions of each module are described in detail as follows:
获取模块501,用于获取OTN若干个时刻的流量信号;The acquisition module 501 is used to acquire the traffic signals of the OTN at several moments;
转换模块502,用于利用图卷积网络将流量信号转换为具有时间特征和空间特征的流量信号。The conversion module 502 is used to convert the traffic signal into a traffic signal with temporal features and spatial features by using a graph convolutional network.
进一步地,获取模块501具体用于:Further, the acquisition module 501 is specifically used for:
获取OTN的ODU电层拓扑若干个时刻的流量信号。Obtain the traffic signals at several times in the ODU electrical layer topology of the OTN.
进一步地,获取模块501具体用于:Further, the acquisition module 501 is specifically used for:
获取拓扑节点若干个时刻的总流量,总流量包括穿通流量和上下路流量,拓扑节点为ODU电层拓扑所包含的节点。Obtain the total traffic of the topology node at several times, the total traffic includes the through traffic and the add and drop traffic, and the topology node is the node included in the ODU electrical layer topology.
进一步地,转换模块502具体用于:Further, the conversion module 502 is specifically used for:
利用图卷积网络根据
将流量信号转换为具有时间特征和空间特征的流量信号,其中,y'为转换后的流量信号,σ为激活函数,α
jt为拓扑节点在第t个时刻的第j维流量属性的卷积核参数,L为归一化的图拉普拉斯算子,x
jt为拓扑节点在第t个时刻的第j维流量属性,k为流量属性的总维度,M为时刻的总个数,流量属性的维度包括总流量、穿通流量和上下路流量。
Using a graph convolutional network according to Convert the traffic signal into a traffic signal with temporal and spatial characteristics, where y' is the converted traffic signal, σ is the activation function, and α jt is the convolution of the j-th dimension traffic attribute of the topology node at the t-th time Kernel parameter, L is the normalized graph Laplacian operator, x jt is the j-th dimension traffic attribute of the topology node at the t-th time, k is the total dimension of the traffic attribute, M is the total number of times, The dimensions of the traffic attributes include total traffic, through traffic, and add and drop traffic.
进一步地,获取模块501具体用于:Further, the acquisition module 501 is specifically used for:
获取拓扑节点所包含业务在若干个时刻的流量信号,拓扑节点为ODU电层拓扑所包含的节点。The traffic signals of the services included in the topology node at several times are obtained, and the topology node is a node included in the ODU electrical layer topology.
进一步地,转换模块502具体用于:Further, the conversion module 502 is specifically used for:
利用图卷积网络根据
将流量信号转换为具有时间特征和空间特征的流量信号,其中,y'为转换后的流量信号,σ为激活函数,α
jt为拓扑节点在第t个时刻的第j条业务的流量属性的卷积核参数,L为归一化的图拉普拉斯算子,x
jt为拓扑节点在第t个时刻的第j条业务的流量属性,r为拓扑节点所包含业务的总条数,M为时刻的总个数。
Using a graph convolutional network according to Convert the traffic signal into a traffic signal with temporal and spatial characteristics, where y' is the converted traffic signal, σ is the activation function, and α jt is the traffic attribute of the jth service of the topology node at the tth time. convolution kernel parameters, L is the normalized graph Laplacian operator, x jt is the traffic attribute of the jth service of the topology node at the t-th time, r is the total number of services included in the topology node, M is the total number of moments.
进一步地,转换模块502具体用于:Further, the conversion module 502 is specifically used for:
利用图卷积网络根据
将流量信号转换为具有时间特征和空间特征的流量信号,其中,y'为转换后的流量信号,σ为激活函数,α
t为拓扑节点在第t个时刻的上下路流量属性的卷积核参数,L为归一化的图拉普拉斯算子,x
t为拓扑节点在第t个时刻的上下路流量属性,M为时刻的总个数。
Using a graph convolutional network according to Convert the traffic signal into a traffic signal with temporal and spatial characteristics, where y' is the converted traffic signal, σ is the activation function, and α t is the convolution kernel of the add and drop traffic attributes of the topology node at the t-th time parameters, L is the normalized graph Laplacian operator, x t is the traffic attribute of the topological node at the t-th time, and M is the total number of times.
进一步地,本申请实施方式提供的流量信号转换装置还包括预测模块,其中,预测模块用于:根据具有时间特征和空间特征的流量信号预测OTN的流量。Further, the flow signal conversion device provided by the embodiment of the present application further includes a prediction module, wherein the prediction module is used for: predicting the flow of the OTN according to the flow signal with temporal characteristics and spatial characteristics.
进一步地,预测模型具体用于:利用深度神经网络根据具有时间特征和空间特征的流量信号预测OTN的流量。Further, the prediction model is specifically used for: predicting the traffic of the OTN according to the traffic signal with temporal and spatial characteristics by using a deep neural network.
不难发现,本实施方式为与第一、第二、第三、第四实施方式相对应的系 统实施例,本实施方式可与第一、第二、第三、第四实施方式互相配合实施。第一、第二、第三、第四实施方式中提到的相关技术细节在本实施方式中依然有效,为了减少重复,这里不再赘述。相应地,本实施方式中提到的相关技术细节也可应用在第一、第二、第三、第四实施方式中。It is not difficult to find that this embodiment is a system example corresponding to the first, second, third and fourth embodiments, and this embodiment can be implemented in cooperation with the first, second, third and fourth embodiments . The related technical details mentioned in the first, second, third, and fourth embodiments are still valid in this embodiment, and are not repeated here in order to reduce repetition. Correspondingly, the related technical details mentioned in this embodiment can also be applied to the first, second, third and fourth embodiments.
值得一提的是,本实施方式中所涉及到的各模块均为逻辑模块,在实际应用中,一个逻辑单元可以是一个物理单元,也可以是一个物理单元的一部分,还可以以多个物理单元的组合实现。此外,为了突出本申请的创新部分,本实施方式中并没有将与解决本申请所提出的技术问题关系不太密切的单元引入,但这并不表明本实施方式中不存在其它的单元。It is worth mentioning that each module involved in this embodiment is a logical module. In practical applications, a logical unit may be a physical unit, a part of a physical unit, or multiple physical units. A composite implementation of the unit. In addition, in order to highlight the innovative part of the present application, this embodiment does not introduce units that are not closely related to solving the technical problem raised by the present application, but this does not mean that there are no other units in this embodiment.
本申请的第六实施方式涉及一种电子设备,如图13所示,包括:至少一个处理器601;以及,与至少一个处理器601通信连接的存储器602;其中,存储器602存储有可被至少一个处理器601执行的指令,指令被至少一个处理器601执行,以使至少一个处理器601能够执行上述的流量信号转换方法。The sixth embodiment of the present application relates to an electronic device, as shown in FIG. 13 , including: at least one processor 601 ; and a memory 602 communicatively connected to the at least one processor 601 ; Instructions executed by one processor 601, the instructions are executed by at least one processor 601, so that at least one processor 601 can execute the above-mentioned traffic signal conversion method.
其中,存储器和处理器采用总线方式连接,总线可以包括任意数量的互联的总线和桥,总线将一个或多个处理器和存储器的各种电路连接在一起。总线还可以将诸如***设备、稳压器和功率管理电路等之类的各种其他电路连接在一起,这些都是本领域所公知的,因此,本文不再对其进行进一步描述。总线接口在总线和收发机之间提供接口。收发机可以是一个元件,也可以是多个元件,比如多个接收器和发送器,提供用于在传输介质上与各种其他装置通信的单元。经处理器处理的数据通过天线在无线介质上进行传输,进一步,天线还接收数据并将数据传送给处理器。The memory and the processor are connected by a bus, and the bus may include any number of interconnected buses and bridges, and the bus connects one or more processors and various circuits of the memory. The bus may also connect together various other circuits, such as peripherals, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. The bus interface provides the interface between the bus and the transceiver. A transceiver may be a single element or multiple elements, such as multiple receivers and transmitters, providing a means for communicating with various other devices over a transmission medium. The data processed by the processor is transmitted on the wireless medium through the antenna, and further, the antenna also receives the data and transmits the data to the processor.
处理器负责管理总线和通常的处理,还可以提供各种功能,包括定时,***接口,电压调节、电源管理以及其他控制功能。而存储器可以被用于存储处理器在执行操作时所使用的数据。The processor is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interface, voltage regulation, power management, and other control functions. Instead, memory may be used to store data used by the processor in performing operations.
本申请第七实施例涉及一种计算机可读存储介质,存储有计算机程序。计算机程序被处理器执行时实现上述方法实施例。The seventh embodiment of the present application relates to a computer-readable storage medium storing a computer program. The above method embodiments are implemented when the computer program is executed by the processor.
即,本领域技术人员可以理解,实现上述实施例方法中的全部或部分步骤是可以通过程序来指令相关的硬件来完成,该程序存储在一个存储介质中,包括若干指令用以使得一个设备(可以是单片机,芯片等)或处理器(processor)执行本申请各个实施例所述方法的全部或部分步骤。而前述的存储介质包括:U盘、移动硬盘、只读存储器(ROM,Read-Only Memory)、随机存取存储器(RAM,Random Access Memory)、磁碟或者光盘等各种可以存储程序代码的介质。That is, those skilled in the art can understand that all or part of the steps in the method for implementing the above embodiments can be completed by instructing the relevant hardware through a program, and the program is stored in a storage medium and includes several instructions to make a device ( It may be a single chip microcomputer, a chip, etc.) or a processor (processor) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, Read-Only Memory (ROM, Read-Only Memory), Random Access Memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program codes .
本领域的普通技术人员可以理解,上述各实施例是实现本申请的具体实施例,而在实际应用中,可以在形式上和细节上对其作各种改变,而不偏离本申请的精神和范围。Those of ordinary skill in the art can understand that the above-mentioned embodiments are specific embodiments for realizing the present application, and in practical applications, various changes in form and details can be made without departing from the spirit and the spirit of the present application. scope.
Claims (12)
- 一种流量信号转换方法,包括:A flow signal conversion method, comprising:获取OTN若干个时刻的流量信号;Obtain OTN traffic signals at several moments;利用图卷积网络将所述流量信号转换为具有时间特征和空间特征的流量信号。The traffic signal is converted into a traffic signal with temporal and spatial features using a graph convolutional network.
- 根据权利要求1所述的流量信号转换方法,其中,所述获取OTN若干个时刻的流量信号,具体为:The traffic signal conversion method according to claim 1, wherein the acquiring the traffic signals of the OTN at several times is specifically:获取所述OTN的ODU电层拓扑若干个时刻的流量信号。Acquire traffic signals at several times in the ODU electrical layer topology of the OTN.
- 根据权利要求2所述的流量信号转换方法,其中,所述获取所述OTN的ODU电层拓扑若干个时刻的流量信号,具体为:The traffic signal conversion method according to claim 2, wherein the acquiring the traffic signals of the ODU electrical layer topology of the OTN at several times is specifically:获取拓扑节点若干个时刻的总流量,所述总流量包括穿通流量和上下路流量,所述拓扑节点为所述ODU电层拓扑所包含的节点。Acquire the total traffic of a topology node at several times, where the total traffic includes pass-through traffic and add/drop traffic, and the topology node is a node included in the ODU electrical layer topology.
- 根据权利要求3所述的流量信号转换方法,其中,所述利用图卷积网络将所述流量信号转换为具有时间特征和空间特征的流量信号,包括:The traffic signal conversion method according to claim 3, wherein the converting the traffic signal into a traffic signal with temporal characteristics and spatial characteristics by using a graph convolutional network comprises:活函数,所述α jt为所述拓扑节点在第t个时刻的第j维流量属性的卷积核参数,所述L为归一化的图拉普拉斯算子,所述x jt为所述拓扑节点在第t个时刻的第j维流量属性,所述k为流量属性的总维度,所述M为所述时刻的总个数,所述流量属性的维度包括所述总流量、所述穿通流量和所述上下路流量。 Live function, the α jt is the convolution kernel parameter of the j-th dimension traffic attribute of the topology node at the t-th moment, the L is the normalized graph Laplacian, and the x jt is The jth dimension traffic attribute of the topology node at the t th time, the k is the total dimension of the traffic attribute, the M is the total number at the time, and the dimension of the traffic attribute includes the total traffic, the through flow and the add and drop flow.
- 根据权利要求2所述的流量信号转换方法,其中,所述获取所述OTN的ODU电层拓扑若干个时刻的流量信号,具体为:The traffic signal conversion method according to claim 2, wherein the acquiring the traffic signals of the ODU electrical layer topology of the OTN at several times is specifically:获取拓扑节点所包含业务在若干个时刻的流量信号,所述拓扑节点为所述ODU电层拓扑所包含的节点。Acquire traffic signals of services included in a topology node at several times, where the topology node is a node included in the ODU electrical layer topology.
- 根据权利要求5所述的流量信号转换方法,其中,所述利用图卷积网络将所述流量信号转换为具有时间特征和空间特征的流量信号,包括:The traffic signal conversion method according to claim 5, wherein the converting the traffic signal into a traffic signal with temporal characteristics and spatial characteristics by using a graph convolutional network comprises:活函数,所述α jt为所述拓扑节点在第t个时刻的第j条业务的流量属性的卷积 核参数,所述L为归一化的图拉普拉斯算子,所述x jt为所述拓扑节点在第t个时刻的第j条业务的流量属性,所述r为所述拓扑节点所包含业务的总条数,所述M为所述时刻的总个数。 Live function, the α jt is the convolution kernel parameter of the traffic attribute of the jth service of the topology node at the tth moment, the L is the normalized graph Laplacian, and the x jt is the traffic attribute of the jth service of the topology node at the tth time, the r is the total number of services included in the topology node, and the M is the total number at the time.
- 根据权利要求5所述的流量信号转换方法,其中,所述利用图卷积网络将所述流量信号转换为具有时间特征和空间特征的流量信号,包括:The traffic signal conversion method according to claim 5, wherein the converting the traffic signal into a traffic signal with temporal characteristics and spatial characteristics by using a graph convolutional network comprises:利用图卷积网络根据 将所述流量信号转换为具有时间特征和空间特征的流量信号,其中,y'为转换后的流量信号,所述σ为激活函数,所述α t为所述拓扑节点在第t个时刻的上下路流量属性的卷积核参数,所述L为归一化的图拉普拉斯算子,所述x t为所述拓扑节点在第t个时刻的上下路流量属性,所述M为所述时刻的总个数。 Using a graph convolutional network according to Convert the flow signal into a flow signal with temporal and spatial characteristics, where y' is the converted flow signal, the σ is the activation function, and the α t is the topological node at the t-th moment. The convolution kernel parameter of the add/drop traffic attribute, the L is the normalized graph Laplacian operator, the x t is the add/drop traffic attribute of the topology node at the t-th moment, and the M is the total number of times at that time.
- 根据权利要求1-7任一项所述的流量信号转换方法,其中,在利用图卷积网络将所述流量信号转换为具有时间特征和空间特征的流量信号之后,还包括:The traffic signal conversion method according to any one of claims 1-7, wherein after using a graph convolution network to convert the traffic signal into a traffic signal with temporal characteristics and spatial characteristics, the method further comprises:根据所述具有时间特征和空间特征的流量信号预测所述OTN的流量。The traffic of the OTN is predicted according to the traffic signal with temporal and spatial characteristics.
- 根据权利要求8所述的流量信号转换方法,其中,所述根据所述具有时间特征和空间特征的流量信号预测所述OTN的流量,包括:The traffic signal conversion method according to claim 8, wherein the predicting the traffic of the OTN according to the traffic signal with temporal characteristics and spatial characteristics comprises:利用深度神经网络根据所述具有时间特征和空间特征的流量信号预测所述OTN的流量。A deep neural network is used to predict the traffic of the OTN according to the traffic signal with temporal and spatial characteristics.
- 一种流量信号转换装置,包括:A flow signal conversion device, comprising:获取模块,用于获取OTN若干个时刻的流量信号;The acquisition module is used to acquire OTN traffic signals at several moments;转换模块,用于利用图卷积网络将所述流量信号转换为具有时间特征和空间特征的流量信号。The conversion module is used to convert the traffic signal into a traffic signal with temporal characteristics and spatial characteristics by using a graph convolutional network.
- 一种电子设备,包括:An electronic device comprising:至少一个处理器;以及,at least one processor; and,与所述至少一个处理器通信连接的存储器;其中,a memory communicatively coupled to the at least one processor; wherein,所述存储器存储有可被所述至少一个处理器执行的指令,所述指令被所述至少一个处理器执行,以使所述至少一个处理器能够执行如权利要求1至9任一项所述的流量信号转换方法。The memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the execution of any one of claims 1 to 9 The flow signal conversion method.
- 一种计算机可读存储介质,存储有计算机程序,所述计算机程序被处理器执行时实现如权利要求1至9任一项所述的流量信号转换方法。A computer-readable storage medium storing a computer program, when the computer program is executed by a processor, the flow signal conversion method according to any one of claims 1 to 9 is implemented.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011127237.8 | 2020-10-20 | ||
CN202011127237.8A CN114448840A (en) | 2020-10-20 | 2020-10-20 | Flow signal conversion method and device, electronic equipment and storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022083549A1 true WO2022083549A1 (en) | 2022-04-28 |
Family
ID=81291556
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2021/124448 WO2022083549A1 (en) | 2020-10-20 | 2021-10-18 | Traffic signal conversion method and apparatus, electronic device, and storage medium |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN114448840A (en) |
WO (1) | WO2022083549A1 (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116994700A (en) * | 2023-03-31 | 2023-11-03 | 北京诺道认知医学科技有限公司 | Quetiapine dose individuation recommendation method and device based on deep learning |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116192669B (en) * | 2023-03-07 | 2024-06-07 | 西安电子科技大学 | Network flow prediction method based on dynamic space-time diagram convolution |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20190251480A1 (en) * | 2018-02-09 | 2019-08-15 | NEC Laboratories Europe GmbH | Method and system for learning of classifier-independent node representations which carry class label information |
CN110264709A (en) * | 2019-05-06 | 2019-09-20 | 北京交通大学 | The prediction technique of the magnitude of traffic flow of road based on figure convolutional network |
CN110995520A (en) * | 2020-02-28 | 2020-04-10 | 清华大学 | Network flow prediction method and device, computer equipment and readable storage medium |
CN111696345A (en) * | 2020-05-08 | 2020-09-22 | 东南大学 | Intelligent coupled large-scale data flow width learning rapid prediction algorithm based on network community detection and GCN |
-
2020
- 2020-10-20 CN CN202011127237.8A patent/CN114448840A/en active Pending
-
2021
- 2021-10-18 WO PCT/CN2021/124448 patent/WO2022083549A1/en active Application Filing
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20190251480A1 (en) * | 2018-02-09 | 2019-08-15 | NEC Laboratories Europe GmbH | Method and system for learning of classifier-independent node representations which carry class label information |
CN110264709A (en) * | 2019-05-06 | 2019-09-20 | 北京交通大学 | The prediction technique of the magnitude of traffic flow of road based on figure convolutional network |
CN110995520A (en) * | 2020-02-28 | 2020-04-10 | 清华大学 | Network flow prediction method and device, computer equipment and readable storage medium |
CN111696345A (en) * | 2020-05-08 | 2020-09-22 | 东南大学 | Intelligent coupled large-scale data flow width learning rapid prediction algorithm based on network community detection and GCN |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116994700A (en) * | 2023-03-31 | 2023-11-03 | 北京诺道认知医学科技有限公司 | Quetiapine dose individuation recommendation method and device based on deep learning |
CN116994700B (en) * | 2023-03-31 | 2024-06-04 | 北京诺道认知医学科技有限公司 | Quetiapine dose individuation recommendation method and device based on deep learning |
Also Published As
Publication number | Publication date |
---|---|
CN114448840A (en) | 2022-05-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2022083549A1 (en) | Traffic signal conversion method and apparatus, electronic device, and storage medium | |
CN110839184B (en) | Method and device for adjusting bandwidth of mobile fronthaul optical network based on flow prediction | |
CN114167760A (en) | Intention-driven network management system and method | |
CN104486194A (en) | Control system and control method for virtual network with multiple reliability levels | |
CN113992769B (en) | Industrial Internet information exchange method | |
CN109947574A (en) | A kind of vehicle big data calculating discharging method based on mist network | |
CN110213175A (en) | A kind of intelligent managing and control system and management-control method towards knowledge definition network | |
WO2022111068A1 (en) | Rru undervoltage risk prediction method, apparatus, and system, device, and medium | |
Kovalenko et al. | Horizontal scaling method for a hyperconverged network | |
CN112187891A (en) | Load optimization method and device of edge computing node set based on multiple services | |
Wang et al. | Edge intelligence for mission cognitive wireless emergency networks | |
Zheng et al. | Minimizing the latency of embedding dependence-aware SFCs into MEC network via graph theory | |
CN107659505A (en) | The route selecting method and SDN controllers of a kind of SDN | |
CN114691372A (en) | Group intelligent control method of multimedia end edge cloud system | |
CN117196014B (en) | Model training method and device based on federal learning, computer equipment and medium | |
WO2021004478A1 (en) | Distributed ai system | |
Mobasheri et al. | Toward developing fog decision making on the transmission rate of various IoT devices based on reinforcement learning | |
CN101552797B (en) | System for realizing network system structure of service-oriented provision | |
CN116760722A (en) | Storage auxiliary MEC task unloading system and resource scheduling method | |
WO2024011376A1 (en) | Task scheduling method and device for artificial intelligence (ai) network function service | |
CN111245878A (en) | Method for computing and offloading communication network based on hybrid cloud computing and fog computing | |
CN116109058A (en) | Substation inspection management method and device based on deep reinforcement learning | |
WO2022111356A1 (en) | Data migration method and system, and server and storage medium | |
CN112738225B (en) | Edge calculation method based on artificial intelligence | |
Zhang et al. | Accelerate deep learning in IoT: Human-interaction co-inference networking system for edge |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 04/09/2023) |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 21881965 Country of ref document: EP Kind code of ref document: A1 |