CN110267292B - Cellular network flow prediction method based on three-dimensional convolutional neural network - Google Patents

Cellular network flow prediction method based on three-dimensional convolutional neural network Download PDF

Info

Publication number
CN110267292B
CN110267292B CN201910408711.5A CN201910408711A CN110267292B CN 110267292 B CN110267292 B CN 110267292B CN 201910408711 A CN201910408711 A CN 201910408711A CN 110267292 B CN110267292 B CN 110267292B
Authority
CN
China
Prior art keywords
data
training
network
short
long
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201910408711.5A
Other languages
Chinese (zh)
Other versions
CN110267292A (en
Inventor
陈岑
符潇
李肯立
李克勤
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hunan University
Original Assignee
Hunan University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hunan University filed Critical Hunan University
Priority to CN201910408711.5A priority Critical patent/CN110267292B/en
Publication of CN110267292A publication Critical patent/CN110267292A/en
Application granted granted Critical
Publication of CN110267292B publication Critical patent/CN110267292B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/14—Network analysis or design
    • H04L41/145—Network analysis or design involving simulating, designing, planning or modelling of a network
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/14—Network analysis or design
    • H04L41/147—Network analysis or design for predicting network behaviour
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/06—Testing, supervising or monitoring using simulated traffic
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D30/00—Reducing energy consumption in communication networks
    • Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Data Mining & Analysis (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Traffic Control Systems (AREA)

Abstract

The invention provides a cellular network flow prediction method based on a three-dimensional convolutional neural network, which comprises the following steps: modeling the network traffic data into a three-dimensional tensor input form to obtain a three-dimensional network traffic data model; acquiring training set data and test set data according to the three-dimensional network flow data; constructing a basic three-dimensional convolution neural network; training the three-dimensional convolution neural network on the short-time dependent data to obtain short-time characteristics, and training the three-dimensional convolution neural network on the long-time dependent data to obtain long-time characteristics; performing fusion training on the short-term features and the long-term features to obtain a feature matrix, and forming a training model by using the feature matrix as the output of the basic three-dimensional convolution neural network; and predicting the network traffic data to be predicted by using the training model to obtain a network traffic prediction result. The prediction method provided by the invention simultaneously considers the short-term correlation and the long-term trend of the network traffic data and captures the characteristic correlation of the network traffic data in time sequence.

Description

Cellular network flow prediction method based on three-dimensional convolutional neural network
[ technical field ] A method for producing a semiconductor device
The invention relates to the field of computer time sequence prediction application, in particular to a cellular network flow prediction method based on a three-dimensional convolutional neural network.
[ background of the invention ]
In recent years, with the popularization of mobile devices and mobile applications, wireless network technology has played a key role in daily life of people worldwide, and more people use mobile devices to access cellular networks, and the demands for cellular network traffic and network traffic are rapidly increasing. Recent industry forecasts show that by 2021, cellular network traffic for global mobile devices is expected to exceed 48.3EB per person, 7 times the amount currently used, and smartphone traffic will exceed PC traffic in the same year. In order for cellular network service providers and infrastructure providers to cope with the increasing demand, to provide users with stable cellular network services and guaranteed quality of service (Qos), it is crucial to accurately predict mobile communication demands. For example, by accurately predicting the traffic demand of the cellular network, timely traffic scheduling can be realized, and part of the demand is dispersed from a busy transmitting tower to an idle transmitting tower, so that network congestion is avoided and user experience is not influenced. Obviously, the flow prediction can optimize the resource allocation, improve the energy efficiency and lay a good foundation for realizing the intelligent cellular network.
In the related art, cellular network traffic prediction is usually modeled as a general time series analysis problem. The most widely used linear statistical models generated by studying such problems are the autoregressive integrated moving average (ARIMA) and Support Vector Regression (SVR). The ARIMA method, however, tends to focus only on the average of historical sequence data, and thus cannot capture the rapid variation process of the underlying traffic load, and cannot model the nonlinear relationship in a real system; although the SVR method can process the nonlinear relationship, it needs to adjust the key parameters to obtain an accurate prediction result. Meanwhile, considering the influence of factors such as cellular network user mobility, arrival mode and user demand diversity, the potential correlation between traffic sequences in the cellular network is mostly ignored by the method. For example, due to spatial dependencies in a cellular network, it is obvious that the movement of a user may drive traffic demand to shift, resulting in significant spatial dependencies between traffic between different base stations, and the basic traffic demand in each area may be influenced by the surrounding environment, and the traffic demand in a busy area is significantly greater than that in a remote area, which dependencies cannot be captured by the conventional method. In recent years, the recent progress of deep learning models in various fields also provides a new idea for problems such as flow prediction.
Therefore, it is necessary to provide a cellular network traffic prediction method based on a three-dimensional convolutional neural network to solve the above problems.
[ summary of the invention ]
The invention aims to provide a cellular network flow prediction method based on a three-dimensional convolutional neural network, which simultaneously considers the short-term correlation and the long-term trend of network flow data and captures the characteristic correlation of the network flow data in time sequence.
In order to solve the technical problem, the invention provides a cellular network flow prediction method based on a three-dimensional convolutional neural network, which comprises the following steps:
s1: modeling network traffic data into a three-dimensional tensor input form to obtain a three-dimensional network traffic data model, wherein the three-dimensional network traffic data model comprises long-time dependency data and short-time dependency data;
s2: acquiring training set data and test set data according to the three-dimensional network traffic data;
s3: constructing a basic three-dimensional convolution neural network;
s4: training the short-time dependent data by a three-dimensional convolutional neural network to obtain short-time characteristics, and training the long-time dependent data by the three-dimensional convolutional neural network to obtain long-time characteristics;
s5: performing fusion training on the short-term features and the long-term features to obtain a feature matrix, and taking the feature matrix as the output of the basic three-dimensional convolution neural network to form a training model;
s6: and predicting the network traffic data to be predicted by using the training model to obtain a network traffic prediction result.
Preferably, the step S1 includes the following steps:
s11: dividing a city into an H multiplied by W grid graph, recording network flow data of all areas in the grid graph at an interval of 15 minutes, and combining the network flow data into 1-hour network flow data, wherein the grid area without the network flow data is filled with a numerical value of 0;
s12: let tensor Xt∈RH×WRepresenting the total network flow value transmitted in all grids of the whole city in the t time slot; make tensor
Figure BDA0002062124230000031
Representing network traffic generated within a grid area of coordinates (i, j), wherein a time slot represents an interval of one hour;
s13: modeling the time correlation of the network flow from short-time dependence and long-time dependence, wherein the short-time dependence refers to the time correlation reflected by the network flow in a time slot interval; long-term dependencies refer to the time dependence embodied by network traffic over twenty-four time slot intervals.
Preferably, the step S2 includes the following steps:
s21: defining the data length of a test set as n, extracting m samples from the three-dimensional network traffic data model as a test sample set, and taking the remaining n-m samples as a training sample set;
s22: and respectively carrying out minimum-maximum normalization on the data in the training sample set and the data in the testing sample set, so that the finally input data vector values of the training sample set and the testing sample set are mapped in a range of [0,1 ].
Preferably, the conversion process of the data vector in step S22 is as follows:
Figure BDA0002062124230000032
wherein min is the minimum of the data in the training sample set or the test sample set, and max is the maximum of the data in the training sample set or the test sample set.
Preferably, the step S4 includes the following steps:
s41: based on step S3, two basic three-dimensional convolutional neural networks c _3DCNN and p _3DCNN with the same structure are constructed, and the short-time dependent data stream and the long-time dependent data stream are respectively trained.
S42: initializing parameters of c _3DCNN and p _3DCNN networks;
s43: setting the training iteration times of c _3DCNN and p _3DCNN to be epochs, taking root mean square error val _ RMSE on the test set as monitoring data, and setting pt numerical values;
s44: respectively taking the short-term dependence tensor and the long-term dependence tensor as input data of c _3DCNN and p _3DCNN, and respectively extracting short-term features VcAnd long term characteristic Vp。
Preferably, epochs is 50 and pt is 10.
Preferably, the fusion training in step S5 specifically includes:
Figure BDA0002062124230000041
wherein, VfusionRepresenting the features obtained after fusion, WcAnd WpRespectively representing the weight matrix to be learned to fit the influence of the short-term dependence and the long-term dependence, VcAnd VpIndicating the short-term features and the long-term features extracted in step S4,
Figure BDA0002062124230000042
representing a dot product operation between vectors.
Compared with the related technology, the cellular network flow prediction method based on the three-dimensional convolutional neural network has the beneficial effects that:
1) different from the traditional prediction method which can only capture the short-term correlation of the traffic data, the method not only considers the short-term correlation of the network traffic data, but also considers the long-term trend, and can more perfectly capture the characteristic correlation of the network traffic data in time sequence;
2) the advantages of the deep neural network model are fully utilized, and the spatial correlation among the urban grids is fully excavated. Modeling a network flow data sequence into a three-dimensional tensor model, and regarding flow data collected from a city as a panoramic picture, so that spatial correlation among city grids can be mined by using a convolutional neural network, a time sequence prediction problem is converted into an image recognition problem, and the advantages of the convolutional neural network are fully exerted;
3) the basic three-dimensional convolutional neural network is used for replacing a traditional time sequence prediction network LSTM (long and short time memory neural network), network parameters are greatly reduced, and network training time and prediction precision are effectively improved.
[ description of the drawings ]
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without inventive efforts, wherein:
FIG. 1 is a flow chart illustrating the steps of a method for predicting the traffic of a cellular network based on a three-dimensional convolutional neural network according to the present invention;
FIG. 2 is a flowchart illustrating the steps of step S1 shown in FIG. 1;
FIG. 3 is a flowchart illustrating the steps of step S2 shown in FIG. 1;
fig. 4 is a flowchart of the step S4 shown in fig. 1.
[ detailed description ] A
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1 to 4, the present invention provides a cellular network traffic prediction method based on a three-dimensional convolutional neural network, including the following steps:
s1: modeling the network traffic data into a three-dimensional tensor input form to obtain a three-dimensional network traffic data model, wherein the three-dimensional network traffic data model comprises long-time dependency data and short-time dependency data.
In this embodiment, one city M is selected, and network traffic data in the city M is modeled as a three-dimensional tensor input form.
Specifically, the step S1 includes the following steps:
s11: dividing a city into an H multiplied by W grid graph, recording network traffic data of all areas in the grid graph at an interval of 15 minutes, and combining the network traffic data into 1-hour network traffic data, wherein the grid area without the network traffic data is filled with a value of 0.
Specifically, in this embodiment, H is 100, and in other embodiments, the values of H and W may be adjusted according to actual needs, which is not limited by the present invention.
S12: let tensor Xt∈RH×WRepresenting the total network flow value transmitted in all grids of the whole city in the t time slot; make tensor
Figure BDA0002062124230000051
Representing network traffic generated within a grid area of coordinates (i, j), where a time slot represents an interval of one hour.
Thus each tensor X can be dividedtThe image with the channel 1 is regarded as one image, and therefore the spatial dependency of network traffic is achieved.
S13: modeling the time correlation of the network flow from short-time dependence and long-time dependence, wherein the short-time dependence refers to the time correlation reflected by the network flow in a time slot interval; long-term dependencies refer to the time dependence embodied by network traffic over twenty-four time slot intervals.
Specifically, c is defined as the length of the short-time dependent sequence, and the short-time dependent three-dimensional tensor stream is constructed and expressed as [ X [ ]t-c，Xt-(c-1)，…，Xt-1]. Connecting the network traffic of each time slot along the first axis, similarly defining p as the length of the long-term dependence sequence, and constructing a long-term dependence three-dimensional tensor flow expressed as [ X ]t-p*24，Xt-(p-1)*24，…，Xt-24]. Each time slotThe network flow is connected along the first shaft, and then the data model which is used as the urban cellular network flow data and respectively expressed as the short-time dependence tensor X is obtainedc∈Rc×H×WAnd the long-term dependence tensor Xp∈Rp×H×W。
S2: and acquiring training set data and test set data according to the three-dimensional network flow data.
Specifically, the step S2 includes the following steps:
s21: defining the data length of a test set as n, extracting m samples from the three-dimensional network flow data model to serve as a test sample set, and taking the remaining n-m samples as a training sample set.
S22: and respectively carrying out minimum-maximum normalization on the data in the training sample set and the data in the testing sample set, so that the finally input data vector values of the training sample set and the testing sample set are mapped in a range of [0,1 ].
Specifically, the conversion process of the data vector is as follows:
Figure BDA0002062124230000061
wherein min is the minimum value of the data in the training sample set or the test sample set, and max is the maximum value of the data in the training sample set or the test sample set.
Step S3: and constructing a basic three-dimensional convolution neural network.
Constructing a three-dimensional convolution neural network, sequentially connecting three convolution layers with convolution kernel sizes of (3,3,3), using a ReLU function f (X) ═ max {0, X } as an activation function, using a convolved output feature map as an input of a maximum pooling layer, and using Dropout to randomly suppress a part of network neurons to prevent overfitting.
S4: and training the three-dimensional convolution neural network on the short-time dependent data to obtain short-time characteristics, and training the three-dimensional convolution neural network on the long-time dependent data to obtain long-time characteristics.
Specifically, the step S4 includes the following steps:
s41: based on step S3, two basic three-dimensional convolutional neural networks c _3DCNN and p _3DCNN with the same structure are constructed, and the short-time dependent data stream and the long-time dependent data stream are respectively trained.
S42: parameters of c _3DCNN and p _3DCNN networks are initialized.
The weight matrixes Wi and bi in the two three-dimensional convolution neural network input layers and the hidden layer are initialized to random numbers in the range of [0,1], so that parameter adjustment in the training process is facilitated.
S43: setting the training iteration times of c _3DCNN and p _3DCNN to be epochs, taking root mean square error val _ RMSE on the test set as monitoring data, and setting pt values.
If the val _ RMSE is not promoted within pt iteration times, even if the set iteration times epochs are not reached, the network training is immediately stopped, and overfitting is prevented.
Specifically, in the present embodiment, epochs is 50 and pt is 10, that is, val _ RMSE is not raised within 10 iterations, and even if 10 iterations are not reached, the network training is immediately stopped to prevent the occurrence of overfitting.
S44: respectively taking the short-term dependence tensor and the long-term dependence tensor as input data of c _3DCNN and p _3DCNN, and respectively extracting short-term features VcAnd long term characteristic Vp。
S5: and performing fusion training on the short-term features and the long-term features to obtain a feature matrix, and taking the feature matrix as the output of the basic three-dimensional convolution neural network to form a training model.
Since each grid in the city is affected by short-term dependence and long-term dependence, and the degrees of influence are different, the short-term feature V obtained in step S4 is usedcAnd long term characteristic VpFusion retraining was performed as follows:
Figure BDA0002062124230000071
wherein, VfusionRepresentative of the fusion after obtainingIs characterized by WcAnd WpRespectively representing weight matrixes to be learned to fit influences generated by short-term dependence and long-term dependence, VcAnd VpIndicating the short-term features and the long-term features extracted in step S4,
Figure BDA0002062124230000072
representing a dot product operation between vectors.
Feature matrix V obtained after fusionfusionIs flattened into a characteristic vector VoutAs the output of the entire network. And then minimizing the cross entropy of the whole network through a back propagation algorithm, optimizing the network by taking an optimization function as an Adam optimizer, and setting the learning rate of the optimizer as lr.
Step 6: and predicting the network traffic data to be predicted by using the training model to obtain a network traffic prediction result.
Inputting the samples in the test set into the trained network to obtain a normalized network flow prediction result YpreAnd performing inverse normalization operation on the network flow to obtain a network flow prediction test result.
Compared with the related technology, the cellular network flow prediction method based on the three-dimensional convolutional neural network has the beneficial effects that:
1) different from the traditional prediction method which can only capture the short-term correlation of the traffic data, the method not only considers the short-term correlation of the network traffic data, but also considers the long-term trend, and can more perfectly capture the characteristic correlation of the network traffic data in time sequence;
2) and the advantages of the deep neural network model are fully utilized, and the spatial correlation among the urban grids is fully excavated. Modeling a network flow data sequence into a three-dimensional tensor model, and regarding flow data collected from a city as a panoramic picture, so that spatial correlation among city grids can be mined by using a convolutional neural network, a time sequence prediction problem is converted into an image recognition problem, and the advantages of the convolutional neural network are fully exerted;
3) the basic three-dimensional convolutional neural network is used for replacing a traditional time sequence prediction network LSTM (long and short time memory neural network), network parameters are greatly reduced, and network training time and prediction precision are effectively improved.
While the foregoing is directed to embodiments of the present invention, it will be understood by those skilled in the art that various changes may be made without departing from the spirit and scope of the invention.

Claims (4)

1. A cellular network flow prediction method based on a three-dimensional convolutional neural network comprises the following steps:
s1: modeling network traffic data into a three-dimensional tensor input form to obtain a three-dimensional network traffic data model, wherein the three-dimensional network traffic data model comprises long-time dependency data and short-time dependency data;
s2: acquiring training set data and test set data according to the three-dimensional network traffic data;
s3: constructing a basic three-dimensional convolution neural network;
s4: training the short-time dependent data by a three-dimensional convolutional neural network to obtain short-time characteristics, and training the long-time dependent data by the three-dimensional convolutional neural network to obtain long-time characteristics;
s5: performing fusion training on the short-term features and the long-term features to obtain a feature matrix, and taking the feature matrix as the output of the basic three-dimensional convolution neural network to form a training model;
s6: predicting the network traffic data to be predicted by using the training model to obtain a network traffic prediction result;
the step S1 includes the steps of:
s11: dividing a city into an H multiplied by W grid graph, recording network traffic data of all areas in the grid graph at an interval of 15 minutes, and combining the network traffic data into 1-hour network traffic data, wherein the grid area without the network traffic data is filled with a numerical value of 0;
s12: let tensor Xt∈RH×WRepresenting the whole in the t time slotThe total network flow value transmitted in all grids of a city; order tensor
Figure FDA0003539117070000011
Representing network traffic generated within a grid area of coordinates (i, j), wherein a time slot represents an interval of one hour;
s13: modeling the time correlation of the network flow from short-time dependence and long-time dependence, wherein the short-time dependence refers to the time correlation reflected by the network flow in a time slot interval; the long-term dependence refers to the time correlation embodied by network traffic in twenty-four time slot intervals;
the step S4 includes the following steps:
s41: constructing two basic three-dimensional convolutional neural networks c _3DCNN and p _3DCNN with the same structure based on the step S3, and respectively training short-time dependent data streams and long-time dependent data streams;
s42: initializing parameters of c _3DCNN and p _3DCNN networks;
s43: setting the training iteration times of c _3DCNN and p _3DCNN to be epochs, taking root mean square error val _ RMSE on the test set as monitoring data, and setting pt numerical values;
s44: respectively taking the short-term dependence tensor and the long-term dependence tensor as input data of c _3DCNN and p _3DCNN, and respectively extracting short-term features Vc and long-term features Vp;
the fusion training in step S5 specifically includes:
Figure FDA0003539117070000021
wherein, VfusionRepresenting features obtained after fusion, WcAnd WpRespectively representing the weight matrix to be learned to fit the influence of the short-term dependence and the long-term dependence, VcAnd VpIndicating the short-term features and the long-term features extracted in step S4,
Figure FDA0003539117070000022
representing a dot product operation between vectors.
2. The method according to claim 1, wherein the step S2 comprises the steps of:
s21: defining the data length of a test set as n, extracting m samples from the three-dimensional network flow data model as a test sample set, and taking the rest n-m samples as a training sample set;
s22: and respectively carrying out minimum-maximum normalization on the data in the training sample set and the data in the testing sample set, so that the finally input data vector values of the training sample set and the testing sample set are mapped in a range of [0,1 ].
3. The method according to claim 2, wherein the conversion process of the data vector in step S22 is:
Figure FDA0003539117070000023
wherein min is the minimum of the data in the training sample set or the test sample set, and max is the maximum of the data in the training sample set or the test sample set.
4. The method of claim 1 wherein epochs is 50 and pt is 10.
CN201910408711.5A 2019-05-16 2019-05-16 Cellular network flow prediction method based on three-dimensional convolutional neural network Active CN110267292B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910408711.5A CN110267292B (en) 2019-05-16 2019-05-16 Cellular network flow prediction method based on three-dimensional convolutional neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910408711.5A CN110267292B (en) 2019-05-16 2019-05-16 Cellular network flow prediction method based on three-dimensional convolutional neural network

Publications (2)

Publication Number Publication Date
CN110267292A CN110267292A (en) 2019-09-20
CN110267292B true CN110267292B (en) 2022-07-08

Family

ID=67914727

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910408711.5A Active CN110267292B (en) 2019-05-16 2019-05-16 Cellular network flow prediction method based on three-dimensional convolutional neural network

Country Status (1)

Country Link
CN (1) CN110267292B (en)

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111159149B (en) * 2019-12-13 2023-06-30 国网浙江省电力有限公司紧水滩水力发电厂 River flow prediction method based on three-dimensional convolutional neural network
CN111327453B (en) * 2020-01-19 2023-04-07 国网福建省电力有限公司经济技术研究院 Communication bandwidth estimation method considering gridding dynamic and static components
CN111404942A (en) * 2020-03-18 2020-07-10 广东技术师范大学 Vertical malicious crawler flow identification method based on deep learning
CN111949704A (en) * 2020-07-17 2020-11-17 网络通信与安全紫金山实验室 Interpretable multidimensional time sequence data analysis method
CN111866024B (en) * 2020-08-05 2022-10-14 国家计算机网络与信息安全管理中心 Network encryption traffic identification method and device
CN112291807B (en) * 2020-10-15 2022-09-23 山东科技大学 Wireless cellular network traffic prediction method based on deep migration learning and cross-domain data fusion
CN114726745B (en) * 2021-01-05 2024-05-17 中国移动通信有限公司研究院 Network traffic prediction method, device and computer readable storage medium
CN116566841B (en) * 2023-05-09 2023-12-01 北京有元科技有限公司 Flow trend prediction method based on network flow query

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107451552A (en) * 2017-07-25 2017-12-08 北京联合大学 A kind of gesture identification method based on 3D CNN and convolution LSTM
CN109508375A (en) * 2018-11-19 2019-03-22 重庆邮电大学 A kind of social affective classification method based on multi-modal fusion
CN109697852A (en) * 2019-01-23 2019-04-30 吉林大学 Urban road congestion degree prediction technique based on timing traffic events

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10417788B2 (en) * 2016-09-21 2019-09-17 Realize, Inc. Anomaly detection in volumetric medical images using sequential convolutional and recurrent neural networks

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107451552A (en) * 2017-07-25 2017-12-08 北京联合大学 A kind of gesture identification method based on 3D CNN and convolution LSTM
CN109508375A (en) * 2018-11-19 2019-03-22 重庆邮电大学 A kind of social affective classification method based on multi-modal fusion
CN109697852A (en) * 2019-01-23 2019-04-30 吉林大学 Urban road congestion degree prediction technique based on timing traffic events

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
Cen Chen ; Kenli Li ; Sin G. Teo ; Guizi Chen ; Xiaofeng Zou ; Xulei.Exploiting Spatio-Temporal Correlations with Multiple 3D Convolutional Neural Networks for Citywide Vehicle Flow Prediction.《 2018 IEEE International Conference on Data Mining (ICDM)》.2018, *
Xi Ouyang ; Shuangjie Xu ; Chaoyun Zhang ; Pan Zhou ; Yang Yang ; Gua.A 3D-CNN and LSTM Based Multi-Task Learning Architecture for Action Recognition.《IEEE Access》.2019, *

Also Published As

Publication number Publication date
CN110267292A (en) 2019-09-20

Similar Documents

Publication Publication Date Title
CN110267292B (en) Cellular network flow prediction method based on three-dimensional convolutional neural network
CN112532439B (en) Network flow prediction method based on attention multi-component space-time cross-domain neural network model
JP5888640B2 (en) Photovoltaic power generation prediction apparatus, solar power generation prediction method, and solar power generation prediction program
CN113316163B (en) Long-term network traffic prediction method based on deep learning
CN111210082A (en) Optimized BP neural network algorithm-based precipitation prediction method
CN113064700B (en) Container deployment method and device
CN112910711B (en) Wireless service flow prediction method, device and medium based on self-attention convolutional network
CN112532746B (en) Cloud edge cooperative sensing method and system
Zhang et al. LNTP: An end-to-end online prediction model for network traffic
CN108829766B (en) Interest point recommendation method, system, equipment and computer readable storage medium
CN112328914A (en) Task allocation method based on space-time crowdsourcing worker behavior prediction
CN114330935B (en) New energy power prediction method and system based on multiple combination strategies integrated learning
CN112287990A (en) Model optimization method of edge cloud collaborative support vector machine based on online learning
Xiao et al. Collaborative cloud-edge service cognition framework for DNN configuration toward smart IIoT
Bejarano-Luque et al. A deep-learning model for estimating the impact of social events on traffic demand on a cell basis
CN116106988A (en) Weather prediction method and device, electronic equipment and storage medium
He et al. Network traffic prediction method based on multi-channel spatial-temporal graph convolutional networks
Nabi et al. Deep learning based fusion model for multivariate LTE traffic forecasting and optimized radio parameter estimation
Ajagunsegun et al. Machine learning-based system for managing energy efficiency of public buildings: An approach towards smart cities
Phung et al. A prediction based autoscaling in serverless computing
CN115604131B (en) Link flow prediction method, system, electronic device and medium
Gungor et al. Lenard: Lightweight ensemble learner for medium-term electricity consumption prediction
CN111935766B (en) Wireless network flow prediction method based on global spatial dependency
CN117528552B (en) Mobile phone data transmission method and system based on 5G
CN117539640B (en) Heterogeneous reasoning task-oriented side-end cooperative system and resource allocation method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant