CN115987816A - Network flow prediction method and device, electronic equipment and readable storage medium - Google Patents
Network flow prediction method and device, electronic equipment and readable storage medium Download PDFInfo
- Publication number
- CN115987816A CN115987816A CN202211620622.5A CN202211620622A CN115987816A CN 115987816 A CN115987816 A CN 115987816A CN 202211620622 A CN202211620622 A CN 202211620622A CN 115987816 A CN115987816 A CN 115987816A
- Authority
- CN
- China
- Prior art keywords
- network
- network traffic
- data
- prediction
- prediction model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 53
- 238000012549 training Methods 0.000 claims abstract description 17
- 239000011159 matrix material Substances 0.000 claims description 15
- 238000012545 processing Methods 0.000 claims description 14
- 230000006870 function Effects 0.000 description 10
- 238000010586 diagram Methods 0.000 description 4
- 238000004590 computer program Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 235000011888 snacks Nutrition 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 235000012054 meals Nutrition 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Images
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The application provides a network traffic prediction method, a network traffic prediction device, an electronic device and a readable storage medium. The method comprises the following steps: acquiring position data of a plurality of network devices in a target area and network flow use data of the plurality of network devices at a plurality of historical moments; dividing network traffic usage data into a first data set and a second data set; inputting the position data of the plurality of network devices and the first data set into a convolution network model, and constructing a network flow prediction model to obtain prediction data at a moment corresponding to the second data set; comparing the prediction data with the second data set to obtain a prediction error of the network flow prediction model; training a network traffic prediction model according to the prediction error to obtain a trained network traffic prediction model; and obtaining a network flow prediction result at a future moment based on the trained network flow prediction model. The network flow prediction method has high accuracy.
Description
Technical Field
The present application relates to the field of network technologies, and in particular, to a method and an apparatus for predicting network traffic, an electronic device, and a readable storage medium.
Background
The prediction of network traffic can provide a basis for an energy-saving scheme for the network device. The network traffic prediction is a method for predicting network traffic use data of a to-be-detected area at a future moment according to the network traffic use data of a plurality of network devices in the to-be-detected area at a historical moment.
The current network traffic prediction method adopts a Recurrent Neural Network (RNN) -Graph Neural Network (GNN) model to perform prediction. The GNN is capable of learning correlation information between the plurality of network devices based on network traffic usage data of the plurality of network devices at a specific time, where the correlation information may be referred to as geographical features or node relationship information; the RNN is capable of learning a change law of network traffic usage data of the plurality of network devices over time, which may be referred to as a timing characteristic, based on the network traffic usage data of the plurality of network devices at a plurality of times. In this way, the RNN-GNN model trained based on the network traffic usage data of the plurality of network devices at historical times can predict the network traffic usage data of one or more of the plurality of network devices at a future time.
However, the accuracy of the current network traffic prediction method is low.
Disclosure of Invention
The application provides a network traffic prediction method, a network traffic prediction device, an electronic device and a readable storage medium, which can improve the accuracy of network traffic prediction.
In a first aspect, the present application provides a method for predicting network traffic, including:
acquiring position data of a plurality of network devices in a target area and network traffic use data of the plurality of network devices at a plurality of historical moments; dividing the network traffic usage data into a first data set and a second data set; inputting the position data of the plurality of network devices and the first data set into a convolutional network model, and constructing a network flow prediction model to obtain prediction data at a moment corresponding to the second data set; comparing the predicted data with the second data set to obtain a prediction error of the network flow prediction model; training the network traffic prediction model according to the prediction error to obtain a trained network traffic prediction model; and obtaining a network flow prediction result at a future moment based on the trained network flow prediction model.
In some implementation manners of the first aspect, the training the network traffic prediction model according to the prediction error to obtain a trained network traffic prediction model includes: under the condition that the prediction error is larger than or equal to an error threshold value, adjusting parameters in the network traffic prediction model to obtain an adjusted network traffic prediction model; inputting the time corresponding to the second data set into the adjusted network traffic prediction model to obtain the second prediction data of the time corresponding to the second data set; comparing the second data set with the second prediction data to obtain a second prediction error of the network flow prediction model; and under the condition that the re-prediction error is smaller than the error threshold value, determining the adjusted network traffic prediction model as the trained network traffic prediction model.
In certain implementations of the first aspect, the convolutional network model includes a graph structure learning module and a graph convolutional network module; inputting the position data of the plurality of network devices and the first data set into a convolutional network model to construct a network traffic prediction model, including: inputting the position data of the plurality of network devices into the graph structure learning module to obtain an adjacency matrix of the plurality of network devices; inputting the adjacency matrix and the first data set into the graph convolution network module to construct the network traffic prediction model.
In certain implementations of the first aspect, prior to dividing the network traffic usage data into the first data set and the second data set, the method further comprises: judging whether null values exist in the network traffic usage data or not; and if the null value exists in the network traffic usage data, filling up the null value.
In certain implementations of the first aspect, the network traffic usage data of the first network device at the first time instance is null; the filling up the null value comprises: and determining the network traffic usage data of the first network device at the first moment according to the network traffic usage data of the first network device at the moment before the first moment and the network traffic usage data of the first network device at the moment after the first moment.
In certain implementations of the first aspect, the second set of data includes network traffic usage data for a plurality of time instants, the prediction data including prediction data for each of the plurality of time instants; the comparing the prediction data with the second data set to obtain the prediction error of the network traffic prediction model includes: calculating a difference between the predicted data at each of the plurality of time instants and the network traffic usage data at each of the plurality of time instants; and calculating the average value of the difference values corresponding to the plurality of moments to obtain the prediction error of the network flow prediction model.
In a second aspect, the present application provides a network traffic prediction apparatus, including:
the system comprises an acquisition module, a storage module and a processing module, wherein the acquisition module is used for acquiring the position data of a plurality of network devices in a target area and the network traffic use data of the plurality of network devices at a plurality of historical moments;
a processing module for dividing the network traffic usage data into a first data set and a second data set; inputting the position data of the plurality of network devices and the first data set into a convolutional network model to construct the network traffic prediction model; inputting the time corresponding to the second data set into the network traffic prediction model to obtain prediction data of the time corresponding to the second data set; comparing the predicted data with the second data set to obtain a prediction error of the network flow prediction model; training the network traffic prediction model according to the prediction error to obtain the trained network traffic prediction model; and inputting the time to be predicted into the trained network traffic prediction model to obtain a network traffic prediction result.
In a third aspect, the present application provides an electronic device, comprising:
at least one processor and memory;
the memory stores computer execution instructions;
the at least one processor executing the computer-executable instructions stored by the memory causes the at least one processor to perform the method of network traffic prediction as described above in the first aspect.
In a fourth aspect, the present application provides a computer-readable storage medium, in which computer-executable instructions are stored, and when a processor executes the computer-executable instructions, the method for predicting network traffic according to the first aspect is implemented.
According to the network traffic prediction method, the network traffic prediction device, the electronic equipment and the readable storage medium, the network traffic prediction model is constructed through the position data of the plurality of network equipment and the network traffic use data, and the parameters of the network traffic prediction model are adjusted according to the prediction error of the network traffic prediction model until the prediction error of the network traffic prediction model is smaller than the error threshold value, so that the network traffic prediction model can more accurately learn the time sequence characteristics of the network traffic use data and the node relation information among the plurality of network equipment, and thus, the accuracy of network traffic prediction can be improved.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and together with the description, serve to explain the principles of the application.
Fig. 1 is a schematic flowchart of a network traffic prediction method according to an embodiment of the present disclosure;
fig. 2 is a schematic diagram of a node relationship provided in an embodiment of the present application;
fig. 3 is a schematic structural diagram of a network traffic prediction apparatus according to an embodiment of the present disclosure;
fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present application.
With the above figures, there are shown specific embodiments of the present application, which will be described in more detail below. These drawings and written description are not intended to limit the scope of the inventive concepts in any manner, but rather to illustrate the inventive concepts to those skilled in the art by reference to specific embodiments.
Detailed Description
Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. The following description refers to the accompanying drawings in which the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.
In the embodiments of the present application, the terms "first", "second", and the like are used to distinguish the same or similar items having substantially the same function and effect. For example, the first chip and the second chip are only used for distinguishing different chips, and the sequence order thereof is not limited. Those skilled in the art will appreciate that the terms "first," "second," etc. do not denote any order or quantity, nor do the terms "first," "second," etc. denote any order or importance.
It should be noted that in the embodiments of the present application, words such as "exemplary" or "for example" are used to mean serving as examples, illustrations or descriptions. Any embodiment or design described herein as "exemplary" or "e.g.," is not necessarily to be construed as preferred or advantageous over other embodiments or designs. Rather, use of the word "exemplary" or "such as" is intended to present relevant concepts in a concrete fashion.
In the embodiments of the present application, "at least one" means one or more, "a plurality" means two or more. "and/or" describes the association relationship of the associated objects, meaning that there may be three relationships, e.g., a and/or B, which may mean: a exists alone, A and B exist simultaneously, and B exists alone, wherein A and B can be singular or plural. The character "/" generally indicates that the former and latter associated objects are in an "or" relationship. "at least one of the following" or similar expressions refer to any combination of these items, including any combination of the singular or plural items. For example, at least one (one) of a, b, or c, may represent: a, b, c, a-b, a-c, b-c, or a-b-c, wherein a, b, c can be single or multiple.
With the development of society, network traffic is continuously increasing, so that the investment of network equipment is larger and larger. By predicting the network traffic of the network device, comprehensive utilization of network device resources is facilitated, and energy conservation of the network device is facilitated.
RNN-GNN models are currently commonly employed to predict network traffic usage data for one or more network devices at a future time. Wherein the GNN is capable of learning node relationship information between the plurality of network devices; the RNN is capable of learning timing characteristics of network traffic usage data for a plurality of network devices.
However, GNNs generally correlate neighboring network devices based on the location relationship between the network devices, so that the accuracy of the obtained node relationship between the network devices is generally low, and further the accuracy of the network traffic usage data predicted by the RNN-GNN model is low.
In order to improve the accuracy of network traffic prediction, the application provides a network traffic prediction method, and the network traffic prediction method carries out prediction through a network traffic prediction model. The method comprises the steps of firstly establishing a network flow prediction model through position information of a plurality of network devices and network flow use data, and adjusting parameters of the network flow prediction model according to errors obtained in the training process of the network flow prediction model, so that the accuracy of node relationships among the plurality of network devices learned by the network flow prediction model can be improved, and the accuracy of predicted network flow use data is further improved.
The network traffic prediction method of the present application is described in detail below with reference to fig. 1 to 2. The network traffic prediction method in the embodiment of the present application may be executed by one or more servers, or may be executed by a terminal device or a chip, a chip system, or a processor that supports the terminal device to implement the network traffic prediction method, or may be executed by a logic module or software that can implement all or part of the terminal device, which is not limited in this application. The network traffic prediction method according to the embodiment of the present application will be described in detail below with a server as an execution subject.
The following describes the technical solutions of the present application and how to solve the above technical problems with specific embodiments. The following several specific embodiments may be combined with each other, and details of the same or similar concepts or processes may not be repeated in some embodiments.
Fig. 1 is a flowchart illustrating a network traffic prediction method 100 according to an embodiment of the present disclosure. The method 100 is performed by a server, the method 100 comprising:
s101, obtaining position data of a plurality of network devices in a target area and network traffic using data of the plurality of network devices at a plurality of historical moments.
It should be understood that the target area refers to an area where traffic prediction is required, and the area includes at least one network device, which may be a base station. The location data may refer to coordinates of a location where the network device is located, for example, if the location data of the base station a is (40 ° N,116 ° E), the location data is coordinates of the location where the base station a is located. The position data indicates that the base station a is located at 40 degrees north latitude and 116 degrees east longitude. The historical time refers to the past time, for example, the current time is 15/2/1/2022, and the historical time may refer to 12/1/2/2022. The network traffic usage data may include data of a network traffic usage amount of each of the plurality of network devices, a number of users corresponding to each of the network devices, a time corresponding to the network traffic usage amount of each of the network devices, and the like.
In a possible implementation, the network device is a base station, and the network traffic usage data includes a network traffic usage amount of a cell corresponding to each base station in the plurality of base stations, a number of users corresponding to each cell, a time corresponding to each network traffic usage data, and the like. The network traffic usage of the cell corresponding to each base station may refer to uplink traffic and downlink traffic of the cell corresponding to each base station.
In another possible implementation, the time intervals between every two adjacent historical time instants in the plurality of historical time instants are all equal. When the server obtains the network flow use data at the historical moment, the network flow use data of the plurality of network devices are obtained once every preset time, and the network flow use data are stored in the server. The preset time period may be any time period, such as 1 hour, 2 hours, 1 day, etc.
S102, dividing the network traffic use data into a first data set and a second data set.
It should be understood that the above-mentioned network traffic usage data is data of network traffic used by a plurality of network devices at a plurality of historical times. If the network traffic usage data includes network traffic usage data of a plurality of network devices at N times, sorting the N times according to a time sequence, and dividing the sorted network traffic usage data into a first data set and a second data set. The part of data in the front time can be a first data set, and the first data set comprises network traffic use data of a plurality of network devices at N1 moments; the later part of data may be a second data set, where the second data set includes network traffic usage data of the plurality of network devices at N2 times, and a sum of N1 and N2 is equal to or less than N.
In a specific example, the server acquires the network traffic usage data of five base stations in the target area once every hour, the server acquires 00 for the first time at 2 month 1 day 2022, and the server acquires and stores 36 network traffic usage data of 00. According to the sequence of time, wherein the network traffic use data at 24 moments from 00 on 2 months and 1 days in 2022 to 00 on 2 months and 2 days in 2022 are divided into a first data set; network traffic usage data for 12 time instants 00 on 2/2022 to 12 on 2/2022 at 00.
In one possible implementation, the network traffic usage data is divided proportionally into a first data set and a second data set. Illustratively, the ratio may be 2:1, dividing the data of the first 2/3 historical moments in the network traffic use data into a first data set and dividing the data of the last 1/3 historical moments into a second data set according to the time sequence.
S103, inputting the position data of the plurality of network devices and the first data set into a convolutional network model, and constructing a network flow prediction model to obtain prediction data at a moment corresponding to the second data set.
The convolutional network model is a known model, and the network traffic prediction model can be constructed by inputting the first data set and the position data of the plurality of network devices into the convolutional network model.
It should be understood that the second data set may correspond to one or more times. The server inputs the first data set and the position data of the plurality of network devices into the convolutional network model, can establish a network traffic prediction model, and outputs the network traffic use data of the plurality of network devices at the time to be predicted. And predicting data which is network flow use data obtained through the prediction of the network flow prediction model. Illustratively, assume that the first data set includes 24 time points of network traffic usage data of 00 on 1/2/2022 to 00 on 2/2022; the second data set includes 12 time points of network traffic usage data of 00 on 2/2022 to 12 on 2/2022/00. By inputting network traffic usage data at 24 times 00 on 2022 year 2 month 1 day 00 to 00 on 2022 year 2 month 2 day 00 and location data of a plurality of networks into the convolutional network model, it is possible to construct a network traffic prediction model and output prediction data at 12 times 00 on 2022 year 2 month 2 day 00.
And S104, comparing the prediction data with the second data set to obtain the prediction error of the network flow prediction model.
It is to be understood that the number of prediction data may be one or more and that the time instants to which the prediction data corresponds are the same as the time instants to which the second data set corresponds. Therefore, for each time corresponding to the prediction data, the second data set includes network traffic usage data corresponding to the time, the network traffic usage data corresponding to the time is real data of network traffic used by the plurality of network devices at the time, and a prediction error of the network traffic prediction model can be obtained from the real data and the prediction data.
In one possible embodiment, the second data set includes network traffic usage data for a plurality of time instances, and the forecast data includes forecast data for each of the plurality of time instances; comparing the predicted data with the second data set to obtain a prediction error of the network flow prediction model, wherein the method comprises the following steps: calculating a difference between the predicted data at each of the plurality of times and the network traffic usage data at each of the plurality of times; and calculating the average value of the difference values corresponding to a plurality of moments to obtain the prediction error of the network flow prediction model.
The above formula for calculating the prediction error is:
wherein E is a prediction error; n is the number of the time corresponding to the network flow use data; a. The t Using data for the network flow at the time t; f t Is the predicted data at time t. The prediction error of the network flow prediction model can be calculated through the formula.
The average value of the difference values corresponding to a plurality of moments is used as the prediction error of the network traffic prediction model, the calculation process of the prediction error is simple, the accuracy of the network traffic prediction model in predicting traffic can be accurately reflected, and the larger the prediction error is, the lower the accuracy of the network traffic prediction model in predicting traffic is.
In another possible embodiment, the prediction error is the difference between the predicted data at each of the plurality of time instances and the network traffic usage data at each of the plurality of time instancesThe root error, the prediction error is calculated by the following formula:
A t using data F for network traffic at time t t Is the predicted data at time t.
In yet another possible embodiment, the prediction error is a weighted average absolute error percentage between the predicted data at each of the plurality of time instants and the network traffic usage data at each of the plurality of time instants, and the prediction error is calculated by the following formula:
A t using data for the network flow at the time t; f t Is the predicted data at time t.
And S105, training the network traffic prediction model according to the prediction error to obtain the trained network traffic prediction model.
And (3) training a network traffic prediction model according to the prediction error, namely, adjusting parameters in the network traffic prediction model so as to reduce the prediction error. And when the prediction error is not reduced or is smaller than a preset threshold value, finishing the training of the network flow prediction model.
And S106, obtaining a network flow prediction result at the future moment based on the trained network flow prediction model.
It should be understood that the future time may be one or more times in the future. The network flow prediction result is data of network flow used by a plurality of network devices corresponding to the time to be predicted, which is obtained by the prediction of the trained network flow prediction model. The trained network traffic prediction model needs to predict network traffic prediction data at a future moment based on a plurality of historical moments, and the closer the historical moments are to the future moment, the more accurate the predicted network traffic prediction result is. For example, assuming that 12% of 2, 3 and 3 days in 2022 at the current time, staff need to predict the network traffic usage data at each hour of 13% of 2, 3 and 3 days in 2022 and 3. And obtaining a network flow prediction result at the future moment based on the trained network flow prediction model. The network traffic use data at a plurality of historical moments needs to be input into the trained network traffic prediction model, so that the trained network traffic prediction model can output the network traffic use data at the future moments.
According to the network traffic prediction method, the network traffic prediction model is constructed through the position data of the plurality of network devices and the network traffic use data, and the network traffic prediction model is trained according to the prediction error, so that the network traffic prediction model can more accurately learn the time sequence characteristics of the network traffic use data and the node relation information among the plurality of network devices, and therefore the accuracy of the network traffic prediction model can be improved.
In an alternative embodiment, the convolutional network model includes a graph structure learning module and a graph convolutional network module, and the above S103 is implemented by: inputting the position data of a plurality of network devices into a graph structure learning module to obtain an adjacency matrix of the plurality of network devices; and inputting the adjacency matrix and the first data set into a graph convolution network module to construct a network traffic prediction model.
It should be understood that the adjacency matrix is a two-dimensional array used to describe associations between multiple network devices. Based on the position data and the graph structure learning module, the adjacency matrixes of the network devices can be determined more accurately, and the connection relation among the network devices is obtained, so that a network traffic prediction model with higher accuracy can be constructed based on the adjacency matrixes and the first data set, and the accuracy of network traffic prediction is improved.
In one possible implementation, inputting location data of a plurality of network devices to a graph structure learning module to obtain an adjacency matrix of the plurality of network devices includes: converting 2-dimensional position data into 20-dimensional space data as characteristic data of each node through a Liner () function by linear operation; calculating Euclidean distance between every two network devices in the plurality of network devices according to the characteristic data; and processing the Euclidean distance by a normalized exponential (Softmax) function to obtain an adjacency matrix.
It should be understood that the formula for the Liner () function is: linear (x) = Wx + b; where x is the position data, W is the training weight, and b is the bias value. The data of the 20-dimensional space is a 20-dimensional vector. The Softmax function can convert data in 20-dimensional space to values between 0 and 1.
In another possible implementation, the graph convolution network module includes a graph convolution network and a one-dimensional convolution network, and inputting the adjacency matrix and the first data set to the graph convolution network module includes: inputting the adjacency matrix and the first data set into a graph convolution network, and outputting network traffic use data processed by the graph convolution network; and inputting the network traffic use data processed by the graph convolution network into the one-dimensional convolution network, and outputting the predicted traffic data.
Correspondingly, training a network traffic prediction model according to the prediction error comprises the following steps: and adjusting parameters in the graph structure learning module, parameters in the graph convolution network and parameters in the one-dimensional convolution network according to the prediction error.
It should be understood that, the graph convolution network obtains topology structure information between a plurality of network devices corresponding to each of a plurality of time instants according to the adjacency matrix and the first data set, and the topology structure information is used for describing an association relationship between the plurality of network devices.
Taking a base station as an example of a network device, the following describes a graph volume network in detail with reference to fig. 2.
As shown in fig. 2, it is assumed that the wireless signal of the base station a covers an office building of a commercial street, the wireless signal of the base station B covers a snack street, and the wireless signal of the base station C covers a residential area. The node relationship of the three base stations at time 13; the node relationships of the three base stations at time 19. 13, at the noon break moment of a user working in an office building, at which moment the user may go from the office building to a snack street for eating, and return to the office building after the eating is finished, so that a mutual relationship exists between the base station a and the base station B; in addition, the user may go from the office building to the residential area to have a meal and return to the office building after the meal is finished, so that the base station a and the base station C have a mutual relationship. However, at time 19. Further, at this time, after going to the snack street, the user returns home, i.e., goes to the residential area, so that there is also a correlation between the base station B and the base station C. It follows that GNNs may learn different node relationship information between multiple network devices at different times.
The one-dimensional convolution network learns the change rule of the network traffic usage data over time based on the network traffic usage data output and processed by the graph convolution network, and outputs predicted traffic data. The output predicted flow data is processed by a Liner () function or a Softmax function, and predicted network flow use data, namely a network flow prediction result, can be obtained.
In another alternative embodiment, the above S106 is implemented as follows: under the condition that the prediction error is larger than or equal to the error threshold value, adjusting parameters in the network traffic prediction model to obtain an adjusted network traffic prediction model; inputting the time corresponding to the second data set into the adjusted network traffic prediction model to obtain the second prediction data of the time corresponding to the second data set; comparing the re-prediction data with the second data set to obtain a re-prediction error of the network flow prediction model; and under the condition that the prediction error is smaller than the error threshold value again, determining the adjusted network traffic prediction model as a trained network traffic prediction model.
It should be understood that the error threshold may be any value that is preset. The server adjusts parameters in the network traffic prediction model, so that the obtained prediction error changes continuously, when the prediction error is smaller than an error threshold value, the prediction error of the network traffic prediction model is smaller, the accuracy is high, and the optimization of the network traffic prediction model is completed at the moment. For example, assuming that the error threshold is 2, the prediction error obtained by the network traffic prediction model is 3, and since 3 is greater than 2, the parameters of the network traffic prediction model will be adjusted to obtain an adjusted network traffic prediction model. And inputting the time corresponding to the second data set into the adjusted network traffic prediction model to obtain a secondary prediction error 1, and considering that the training of the network traffic prediction model is finished because 1 is less than 2.
And adjusting parameters in the network traffic prediction model based on the prediction error until the re-prediction error of the network traffic prediction model is smaller than the error threshold value, so as to obtain the trained network traffic prediction model. In this way, parameters in the network traffic prediction model are adjusted according to the prediction error, so that the accuracy of the adjacency matrix and the accuracy of the network traffic prediction model can be improved.
In yet another alternative embodiment, the above S106 is implemented as follows: repeatedly adjusting parameters in the network traffic prediction model to obtain an adjusted network traffic prediction model, and inputting the time corresponding to the second data set into the adjusted network traffic prediction model to obtain a re-prediction error of the network traffic prediction model until the re-prediction error is not reduced within the preset duration and/or the preset repetition times; and determining the adjusted network traffic prediction model as a trained network traffic prediction model.
It is to be understood that the preset time period is any time period preset, for example, 1 hour. The preset number of repetitions is any positive integer, e.g., 5, 7, etc. The process of training the network traffic prediction model is the process of adjusting the parameters of the network traffic prediction model, when the obtained prediction error is not reduced any more by adjusting the parameters of the network traffic prediction model, the network traffic prediction model is in an optimal state, and the training of the network traffic prediction model is completed at this moment.
In yet another alternative embodiment, prior to dividing the network traffic usage data into the first data set and the second data set, the method 100 further comprises: judging whether null values exist in the network traffic use data or not; when the network traffic usage data includes a null value, the null value is filled.
As will be understood by those skilled in the art, when a network device is disconnected, the network traffic usage data corresponding to a part of the time of the network device stored by the server may be null, and at this time, the network traffic usage data corresponding to the null needs to be filled. Therefore, the reliability of the data of the first data set and the second data set can be improved, and the accuracy of the network traffic prediction model for predicting traffic is improved.
In one possible embodiment, the network traffic usage data of a first network device of the plurality of network devices at a first time is null; filling up the null value, comprising: and determining the network traffic usage data of the first network device at the first moment according to the network traffic usage data of the first network device at the moment before the first moment and the network traffic usage data of the first network device at the moment after the first moment.
For example, assuming that the server stores network traffic usage data of 12. According to the network traffic usage data corresponding to 13.
It is understood that the above-described method of filling the null value may be referred to as an interpolation method, and the null value can be efficiently filled by the interpolation method.
In an alternative embodiment, the network traffic usage data includes a third set of data in addition to the first set of data and the second set of data; the method 100 further comprises: after the trained network traffic prediction model is obtained, inputting the time corresponding to the third data set into the trained network traffic prediction model to obtain a verification error; and determining that the trained network traffic prediction model is trained and optimized under the condition that the difference value between the verification error and the prediction error is smaller than a preset threshold value.
It should be understood that the validation error is calculated in a similar manner to the prediction error and will not be described in detail herein. The third data set is used for verifying the reliability of the trained network traffic prediction model, and under the condition that the difference value between the verification error and the prediction error is small, the reliability of the network traffic prediction model is considered to be high, so that the accuracy is extremely high. Therefore, after the network traffic prediction model is trained, the performance of the network traffic prediction model is further verified through the third data set, and therefore the accuracy of the network traffic prediction model in predicting traffic can be more accurately determined.
It should be understood that the sequence numbers of the above steps do not mean the execution sequence, and the execution sequence of the steps should be determined by the function and the inherent logic.
The network traffic prediction method according to the embodiment of the present application is described in detail above with reference to fig. 1 and 2, and the network traffic prediction apparatus is described in detail below with reference to fig. 3 and 4.
Fig. 3 is a schematic structural diagram of a network traffic prediction apparatus 300 according to an embodiment of the present disclosure. As shown in fig. 3, the apparatus 300 includes: an acquisition module 301 and a processing module 302.
An obtaining module 301, configured to obtain location data of multiple network devices in a target area and network traffic usage data of the multiple network devices at multiple historical times; a processing module 302 for dividing the network traffic usage data into a first data set and a second data set; inputting the position data of the plurality of network devices and the first data set into a convolution network model, and constructing a network flow prediction model to obtain prediction data at a moment corresponding to the second data set; comparing the prediction data with the second data set to obtain a prediction error of the network flow prediction model; training a network traffic prediction model according to the prediction error to obtain a trained network traffic prediction model; and obtaining a network flow prediction result at the future moment based on the trained network flow prediction model.
Optionally, the processing module 302 is specifically configured to: under the condition that the prediction error is larger than or equal to the error threshold value, adjusting parameters in the network traffic prediction model to obtain an adjusted network traffic prediction model; inputting the time corresponding to the second data set into the adjusted network traffic prediction model to obtain the second prediction data of the time corresponding to the second data set; comparing the re-prediction data with the second data set to obtain a re-prediction error of the network flow prediction model; and under the condition that the prediction error is smaller than the error threshold value again, determining the adjusted network traffic prediction model as a trained network traffic prediction model.
Optionally, the convolutional network model includes a graph structure learning module and a graph convolutional network module; the processing module 302 is specifically configured to: inputting the position data of the plurality of network devices into a graph structure learning module to obtain an adjacency matrix of the plurality of network devices; and inputting the adjacency matrix and the first data set into a graph convolution network module to construct a network traffic prediction model.
Optionally, the processing module 302 is further configured to: judging whether null values exist in the network traffic use data or not; when the network traffic usage data includes a null value, the null value is filled.
Optionally, the network traffic usage data of the first network device in the plurality of network devices at the first time is null; the processing module 302 is specifically configured to: and determining the network traffic usage data of the first network device at the first moment according to the network traffic usage data of the first network device at the moment before the first moment and the network traffic usage data of the first network device at the moment after the first moment.
Optionally, the second data set comprises network traffic usage data at a plurality of time instants, and the forecast data comprises forecast data at each of the plurality of time instants; the processing module 302 is specifically configured to: calculating a difference between the predicted data at each of the plurality of times and the network traffic usage data at each of the plurality of times; and calculating the average value of the difference values corresponding to a plurality of moments to obtain the prediction error of the network flow prediction model.
The network traffic prediction apparatus provided in the embodiment of the present application is applicable to the method embodiments described above, and is not described herein again.
An embodiment of the present application further provides an electronic device 400, as shown in fig. 4, the electronic device shown in fig. 4 includes: a processor 401 and a memory 402. Wherein the processor 401 is coupled to the memory 402, such as via a bus 403. Optionally, the electronic device may further comprise a transceiver. It should be noted that the transceiver in practical application is not limited to one, and the structure of the electronic device does not constitute a limitation to the embodiments of the present application.
The Processor 401 may be a CPU (Central Processing Unit), a general purpose Processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other Programmable logic device, a transistor logic device, a hardware component, or any combination thereof. Which may implement or perform the various illustrative logical blocks, modules, and circuits described in connection with the disclosure. The processor 401 may also be a combination of computing functions, e.g., comprising one or more microprocessors, a combination of a DSP and a microprocessor, or the like.
The Memory 402 may be a ROM (Read Only Memory) or other type of static storage device that can store static information and instructions, a RAM (Random Access Memory) or other type of dynamic storage device that can store information and instructions, an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical Disc storage, optical Disc storage (including Compact Disc, laser Disc, optical Disc, digital versatile Disc, blu-ray Disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer, but is not limited to these.
The memory 402 is used for storing application program codes for executing the scheme of the application, and the execution is controlled by the processor 401. The processor 401 is configured to execute application program code stored in the memory 402 to implement the aspects illustrated in the foregoing method embodiments.
Among them, electronic devices include but are not limited to: mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and the like, and fixed terminals such as digital TVs, desktop computers, and the like. But also a server, etc. The electronic device shown in fig. 4 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present disclosure.
The present application provides a computer-readable storage medium having stored thereon a computer program which, when run on a computer, enables the computer to perform the respective content of the aforementioned method embodiments.
The present application also provides a computer program product comprising a computer program (which may also be referred to as code, or instructions) which, when run on a computer, can carry out the respective content of the above-described method embodiments.
Other embodiments of the present application will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the invention following, in general, the principles of the application and including such departures from the present disclosure as come within known or customary practice within the art to which the invention pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the application being indicated by the following claims.
It will be understood that the present application is not limited to the precise arrangements described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the application is limited only by the appended claims.
Claims (10)
1. A method for predicting network traffic, comprising:
acquiring position data of a plurality of network devices in a target area and network flow use data of the plurality of network devices at a plurality of historical moments;
dividing the network traffic usage data into a first data set and a second data set;
inputting the position data of the plurality of network devices and the first data set into a convolutional network model, and constructing a network traffic prediction model to obtain prediction data at a moment corresponding to the second data set;
comparing the predicted data with the second data set to obtain a prediction error of the network flow prediction model;
training the network traffic prediction model according to the prediction error to obtain a trained network traffic prediction model;
and obtaining a network flow prediction result at the future moment based on the trained network flow prediction model.
2. The method of claim 1, wherein the training the network traffic prediction model according to the prediction error to obtain a trained network traffic prediction model comprises:
under the condition that the prediction error is larger than or equal to an error threshold value, adjusting parameters in the network traffic prediction model to obtain an adjusted network traffic prediction model;
inputting the time corresponding to the second data set into the adjusted network traffic prediction model to obtain the second prediction data of the time corresponding to the second data set;
comparing the re-prediction data with the second data set to obtain a re-prediction error of the network flow prediction model;
and under the condition that the re-prediction error is smaller than the error threshold value, determining the adjusted network traffic prediction model as the trained network traffic prediction model.
3. The method of claim 1 or 2, wherein the convolutional network model comprises a graph structure learning module and a graph convolutional network module;
inputting the location data of the plurality of network devices and the first data set into a convolutional network model to construct a network traffic prediction model, including:
inputting the position data of the plurality of network devices into the graph structure learning module to obtain an adjacency matrix of the plurality of network devices;
inputting the adjacency matrix and the first data set into the graph convolution network module to construct the network traffic prediction model.
4. The method of claim 1 or 2, wherein prior to dividing the network traffic usage data into a first data set and a second data set, the method further comprises:
judging whether null values exist in the network flow use data or not;
and if the null value exists in the network traffic usage data, filling up the null value.
5. The method of claim 4, wherein the network traffic usage data of a first network device of the plurality of network devices at a first time is null;
the filling up the null value comprises:
and determining the network traffic usage data of the first network device at the first moment according to the network traffic usage data of the first network device at the moment before the first moment and the network traffic usage data of the first network device at the moment after the first moment.
6. The method of claim 1 or 2, wherein the second data set comprises network traffic usage data for a plurality of time instances, and the prediction data comprises prediction data for each of the plurality of time instances;
the comparing the prediction data with the second data set to obtain the prediction error of the network traffic prediction model includes:
calculating a difference between the predicted data at each of the plurality of time instants and the network traffic usage data at each of the plurality of time instants;
and calculating the average value of the difference values corresponding to the moments to obtain the prediction error of the network flow prediction model.
7. A network traffic prediction apparatus, comprising:
the system comprises an acquisition module, a storage module and a processing module, wherein the acquisition module is used for acquiring the position data of a plurality of network devices in a target area and the network traffic use data of the plurality of network devices at a plurality of historical moments;
a processing module for dividing the network traffic usage data into a first data set and a second data set; inputting the position data of the plurality of network devices and the first data set into a convolutional network model to construct the network traffic prediction model; inputting the time corresponding to the second data set into the network traffic prediction model to obtain prediction data of the time corresponding to the second data set; comparing the predicted data with the second data set to obtain a prediction error of the network flow prediction model; training the network traffic prediction model according to the prediction error to obtain the trained network traffic prediction model; and inputting the time to be predicted into the trained network traffic prediction model to obtain a network traffic prediction result.
8. The apparatus of claim 7, wherein the processing module is specifically configured to:
under the condition that the prediction error is larger than or equal to an error threshold value, adjusting parameters in the network traffic prediction model to obtain an adjusted network traffic prediction model;
inputting the time corresponding to the second data set into the adjusted network traffic prediction model to obtain the second prediction data of the time corresponding to the second data set;
comparing the re-prediction data with the second data set to obtain a re-prediction error of the network flow prediction model;
and under the condition that the re-prediction error is smaller than the error threshold value, determining the adjusted network traffic prediction model as the trained network traffic prediction model.
9. An electronic device, comprising: at least one processor and memory;
the memory stores computer execution instructions;
the at least one processor executing computer-executable instructions stored by the memory cause the at least one processor to perform the network traffic prediction method of any of claims 1-6.
10. A computer-readable storage medium having computer-executable instructions stored thereon, which when executed by a processor, are configured to implement the network traffic prediction method of any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211620622.5A CN115987816A (en) | 2022-12-15 | 2022-12-15 | Network flow prediction method and device, electronic equipment and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211620622.5A CN115987816A (en) | 2022-12-15 | 2022-12-15 | Network flow prediction method and device, electronic equipment and readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115987816A true CN115987816A (en) | 2023-04-18 |
Family
ID=85958936
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211620622.5A Pending CN115987816A (en) | 2022-12-15 | 2022-12-15 | Network flow prediction method and device, electronic equipment and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115987816A (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113497717A (en) * | 2020-03-19 | 2021-10-12 | 中国移动通信有限公司研究院 | Network flow prediction method, device, equipment and storage medium |
| CN113505536A (en) * | 2021-07-09 | 2021-10-15 | 兰州理工大学 | Optimized traffic flow prediction model based on space-time diagram convolution network |
| CN113743594A (en) * | 2021-09-30 | 2021-12-03 | 中国联合网络通信集团有限公司 | Network flow prediction model establishing method and device, electronic equipment and storage medium |
| CN114205251A (en) * | 2021-12-09 | 2022-03-18 | 西安电子科技大学 | Switch link resource prediction method based on space-time characteristics |
| US20220110021A1 (en) * | 2020-10-05 | 2022-04-07 | Continual Ltd. | Flow forecasting for mobile users in cellular networks |
| CN115314925A (en) * | 2022-08-04 | 2022-11-08 | 南京智联达科技有限公司 | Intelligent power grid fusion network flow prediction method based on GNN and LSTM |
| US20220368650A1 (en) * | 2021-05-11 | 2022-11-17 | Beijing University Of Posts And Telecommunications | Method and Device of Network Resource Allocation |
| CN115442253A (en) * | 2022-09-15 | 2022-12-06 | 西安电子科技大学 | Network flow prediction method using attention mechanism |
-
2022
- 2022-12-15 CN CN202211620622.5A patent/CN115987816A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113497717A (en) * | 2020-03-19 | 2021-10-12 | 中国移动通信有限公司研究院 | Network flow prediction method, device, equipment and storage medium |
| US20220110021A1 (en) * | 2020-10-05 | 2022-04-07 | Continual Ltd. | Flow forecasting for mobile users in cellular networks |
| US20220368650A1 (en) * | 2021-05-11 | 2022-11-17 | Beijing University Of Posts And Telecommunications | Method and Device of Network Resource Allocation |
| CN113505536A (en) * | 2021-07-09 | 2021-10-15 | 兰州理工大学 | Optimized traffic flow prediction model based on space-time diagram convolution network |
| CN113743594A (en) * | 2021-09-30 | 2021-12-03 | 中国联合网络通信集团有限公司 | Network flow prediction model establishing method and device, electronic equipment and storage medium |
| CN114205251A (en) * | 2021-12-09 | 2022-03-18 | 西安电子科技大学 | Switch link resource prediction method based on space-time characteristics |
| CN115314925A (en) * | 2022-08-04 | 2022-11-08 | 南京智联达科技有限公司 | Intelligent power grid fusion network flow prediction method based on GNN and LSTM |
| CN115442253A (en) * | 2022-09-15 | 2022-12-06 | 西安电子科技大学 | Network flow prediction method using attention mechanism |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110166991B (en) | Method, device, apparatus and storage medium for locating electronic device | |
| CN112862176B (en) | Public service facility site selection method and device | |
| CN112418518A (en) | Passenger flow prediction method and device based on time characteristic weight and network topology | |
| CN111935820B (en) | Positioning implementation method based on wireless network and related equipment | |
| JP2018521311A (en) | Method and apparatus for improving the positioning performance of an artificial neural network | |
| CN113515837B (en) | Method and device for establishing simulation test platform and electronic equipment | |
| CN114614989A (en) | Feasibility verification method and device of network service based on digital twin technology | |
| Ghafourian et al. | Wireless localization with diffusion maps | |
| CN115759365A (en) | Photovoltaic power generation power prediction method and related equipment | |
| CN116528282B (en) | Coverage scene recognition method, device, electronic equipment and readable storage medium | |
| CN112469116B (en) | Positioning method, positioning device, electronic equipment and computer readable storage medium | |
| CN113269379B (en) | Method and device for determining attributes of resource objects, storage medium and computer equipment | |
| CN115987816A (en) | Network flow prediction method and device, electronic equipment and readable storage medium | |
| CN114501518B (en) | Flow prediction method, flow prediction device, flow prediction apparatus, flow prediction medium, and program product | |
| CN112243247A (en) | Method and device for determining optimization priority of base station and computing equipment | |
| CN107657474B (en) | Method for determining business circle boundary and server | |
| CN113015193B (en) | Method and device for generating base station configuration file, server and storage medium | |
| CN114390563A (en) | 5G cell residence capacity evaluation method and device and electronic equipment | |
| CN112235723A (en) | Positioning method, positioning device, electronic equipment and computer readable storage medium | |
| CN114726463A (en) | Method and device for predicting space-time distribution of mobile communication user based on neural network | |
| CN114867033B (en) | Cell position determining method and device | |
| CN117808270B (en) | Method, device, equipment and storage medium for determining facility layout | |
| CN115511253B (en) | Method, device, equipment and medium for evaluating land development capacity | |
| CN116484201B (en) | New energy power grid load prediction method and device and electronic equipment | |
| CN113935402B (en) | Training method and device of time difference positioning model and electronic equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |