Network traffic prediction method and device, electronic equipment and storage medium

By generating graph structure data and determining the prediction results of cellular traffic based on path loss and signal quality, the problem of data quality impact in cellular traffic prediction is solved, and prediction accuracy and model adaptability are improved.

CN118921689APending Publication Date: 2024-11-08CHINA TELECOM CORP LTD TECHNOLOGY INNOVATION CENTER +1
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202411179257.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-26
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

In the prior art In cellular traffic prediction, the prediction accuracy of graph neural networks is greatly affected by the quality of cellular traffic data, and the random weight parameters lead to inaccurate feature aggregation, resulting in low prediction accuracy.

Method used

By generating graph structure data, the adjacency matrix element values ​​are determined based on the path loss and signal quality between base stations, and feature aggregation is performed based on the information weight and adjacency element values ​​to obtain accurate traffic prediction results, considering the path loss and signal quality of the signal to improve Practicality and accuracy of the model.

Benefits of technology

It improves the accuracy of network traffic prediction and the adaptability of models, enhances the prediction ability and stability in complex environments, especially in cities or complex terrain to quickly respond to environmental changes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118921689A_ABST
    Figure CN118921689A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of communication, in particular to a network traffic prediction method and device, electronic equipment and a storage medium, which are used for improving the accuracy of network traffic prediction. The method comprises the following steps: generating graph structure data by taking each base station in a network to be predicted as a node and a communication relationship between the base stations as an edge; determining each adjacent element value in the adjacent matrix based on path loss and signal quality of each base station for transmitting the signal to other base stations; based on the information weight and the adjacent element value corresponding to each neighbor node of each node in the graph structure data, at least one time of feature aggregation operation is carried out on each node, a flow prediction result of the network to be predicted is obtained, and the information weight represents the importance of the information of the neighbor node of the corresponding node to the corresponding node. Thus, the adjacency matrix is close to the actual communication environment, the model can more accurately capture the dependency relationship between the nodes, meanwhile, the aggregation mechanism is optimized, and the prediction accuracy of the network flow is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of communication technology, and in particular to a network traffic prediction method, device, electronic device and storage medium. Background Art

[0002] With the development of the fifth generation mobile communication technology (5G), the cellular traffic from various devices (especially mobile devices) continues to increase. As an indispensable part of the management of cellular network resources, the prediction of cellular traffic is crucial in the configuration of cellular network resources and the improvement of cellular network service quality.

[0003] Currently, cellular traffic is mainly predicted through Graph Neural Network (GNN). When predicting cellular traffic, GNN is first constructed, and then cellular traffic is predicted based on the constructed GNN.

[0004] In the prior art, when constructing a GNN, the adjacency matrix value is usually calculated based on the correlation of the node time series to construct the adjacency matrix; and the features of the neighboring nodes of each node are aggregated based on random weight parameters to obtain new features of each node and realize node aggregation.

[0005] However, when constructing the adjacency matrix, the correlation of the time series depends on the cellular traffic, making the adjacency matrix value affected by the data quality of the cellular traffic itself. When the data quality is average, the adjacency matrix value will be biased, resulting in low prediction accuracy when predicting cellular traffic through GNN. At the same time, when nodes are aggregated, the random weight parameters will lead to inaccurate features after aggregation, resulting in low prediction accuracy when predicting cellular traffic through GNN. Summary of the invention

[0006] The embodiments of the present application provide a network traffic prediction method, device, electronic device and storage medium to improve the accuracy of network traffic prediction.

[0007] The specific technical solutions provided by the embodiments of this application are as follows:

[0008] In a first aspect, a network traffic prediction method is provided, comprising:

[0009] Each base station in the network to be predicted is used as a node, and the communication relationship between base stations is used as an edge to generate graph structure data;

[0010] Determine the value of each adjacent element in the adjacency matrix based on the path loss and signal quality of each base station's transmission signal to other base stations;

[0011] Based on the information weights and adjacent element values ​​corresponding to each neighboring node of each node in the graph structure data, at least one feature aggregation operation is performed on each node to obtain the traffic prediction result of the network to be predicted, wherein the information weight represents: the importance of the information of the neighboring nodes of the corresponding node to the corresponding node.

[0012] In a possible embodiment, before determining the value of each adjacent element in the adjacency matrix based on the path loss and signal quality of each base station transmitting a signal to other base stations, the method further includes:

[0013] Obtaining a free space path loss, a reflection loss, a projection loss, and a diffraction loss of a signal transmitted from a first base station to a second base station among the base stations, wherein the first base station and the second base station are two different base stations among the base stations;

[0014] Based on the free space path loss, reflection loss, projection loss, and diffraction loss, the path loss of the first base station transmitting the signal to the second base station is determined.

[0015] Through the above method, in addition to considering the free path loss, the refraction and reflection information of the wireless signal when encountering walls and certain materials is also considered to obtain accurate path loss, making the element values ​​of the adjacency matrix of the graph neural network more accurate, making the adjacency matrix closer to the actual communication environment, and enhancing the practicality and accuracy of the model.

[0016] In a possible embodiment, obtaining a free space path loss, a reflection loss, a projection loss, and a diffraction loss of a signal transmitted from a first base station to a second base station in each base station includes:

[0017] Determine a free space path loss of a signal transmitted from the first base station to the second base station based on a distance and a signal frequency of the signal transmitted from the first base station to the second base station;

[0018] Determine the reflection loss of the signal transmitted from the first base station to the second base station based on the vertical reflection loss corresponding to the vertical polarization and the parallel reflection loss corresponding to the parallel polarization of the signal transmitted from the first base station to the second base station;

[0019] Determine the projection loss of the signal transmitted from the first base station to the second base station based on the vertical projection loss corresponding to the vertical polarization and the parallel projection loss corresponding to the parallel polarization of the signal transmitted from the first base station to the second base station;

[0020] The diffraction loss of the signal transmitted from the first base station to the second base station is determined based on the wave number, the diffraction angle and the distance from the diffraction edge to the second base station of the signal transmitted from the first base station to the second base station.

[0021] Through the above method, accurate free space path loss, reflection loss, projection loss, and diffraction loss can be obtained, thereby obtaining an accurate path loss.

[0022] In a possible embodiment, based on the information weights and adjacent element values ​​corresponding to each neighboring node of each node in the graph structure data, at least one feature aggregation operation is performed on each node to obtain the traffic prediction result of the network to be predicted, including:

[0023] For each node, the following operations are performed respectively: based on the path loss of a signal received by a node from each of its neighboring nodes, the information weight corresponding to each of the neighboring nodes of a node is calculated.

[0024] Through the above method, the importance of each neighbor node feature is adjusted according to the path loss of the signal, which is close to the actual communication environment and can obtain accurate information weight.

[0025] In a possible embodiment, based on the information weights and adjacent element values ​​corresponding to each neighboring node of each node in the graph structure data, at least one feature aggregation operation is performed on each node to obtain a traffic prediction result of the network to be predicted, wherein the following operations are performed during one feature aggregation operation:

[0026] Determine the weighted weights corresponding to each neighboring node of each node based on the information weights and adjacent element values ​​corresponding to each neighboring node of each node;

[0027] Based on the weighted weights corresponding to each neighboring node of each node, the features of each neighboring node of each node are aggregated to obtain new features of each node.

[0028] Through the above method, the aggregation process not only considers the signal transmission loss between nodes, that is, the information weight, but also combines the optimization effect of the graph structure, that is, the adjacent element value, which improves the accuracy of model prediction and also improves the performance of the model in complex traffic prediction tasks.

[0029] In a second aspect, the present application provides a network traffic prediction device, the device comprising:

[0030] A generation module, used to generate graph structure data by taking each base station in the network to be predicted as a node and the communication relationship between the base stations as an edge;

[0031] A determination module, configured to determine the value of each adjacent element in the adjacency matrix based on the path loss and signal quality of each base station transmitting a signal to other base stations;

[0032] The aggregation module is used to perform at least one feature aggregation operation on each node based on the information weight and adjacent element value corresponding to each neighbor node of each node in the graph structure data, so as to obtain the traffic prediction result of the network to be predicted, wherein the information weight represents: the importance of the information of the neighbor nodes of the corresponding node to the corresponding node.

[0033] In a possible embodiment, based on the path loss and signal quality of each base station transmitting a signal to other base stations, before determining the value of each adjacent element in the adjacency matrix, the determination module is further used to:

[0034] Obtaining a free space path loss, a reflection loss, a projection loss, and a diffraction loss of a signal transmitted from a first base station to a second base station among the base stations, wherein the first base station and the second base station are two different base stations among the base stations;

[0035] Based on the free space path loss, reflection loss, projection loss, and diffraction loss, the path loss of the first base station transmitting the signal to the second base station is determined.

[0036] In a possible embodiment, when obtaining the free space path loss, reflection loss, projection loss, and diffraction loss of a signal transmitted from a first base station to a second base station in each base station, the determination module is further used to:

[0037] Determine a free space path loss of a signal transmitted from the first base station to the second base station based on a distance and a signal frequency of the signal transmitted from the first base station to the second base station;

[0038] Determine the reflection loss of the signal transmitted from the first base station to the second base station based on the vertical reflection loss corresponding to the vertical polarization and the parallel reflection loss corresponding to the parallel polarization of the signal transmitted from the first base station to the second base station;

[0039] Determine the projection loss of the signal transmitted from the first base station to the second base station based on the vertical projection loss corresponding to the vertical polarization and the parallel projection loss corresponding to the parallel polarization of the signal transmitted from the first base station to the second base station;

[0040] The diffraction loss of the signal transmitted from the first base station to the second base station is determined based on the wave number, the diffraction angle and the distance from the diffraction edge to the second base station of the signal transmitted from the first base station to the second base station.

[0041] In a possible embodiment, based on the information weights and adjacent element values ​​corresponding to each neighboring node of each node in the graph structure data, at least one feature aggregation operation is performed on each node, and before obtaining the traffic prediction result of the network to be predicted, the aggregation module is further used to:

[0042] For each node, the following operations are performed respectively: based on the path loss of a signal received by a node from each of its neighboring nodes, the information weight corresponding to each of the neighboring nodes of a node is calculated.

[0043] In a possible embodiment, based on the information weights and adjacent element values ​​corresponding to each neighboring node of each node in the graph structure data, at least one feature aggregation operation is performed on each node to obtain a traffic prediction result of the network to be predicted, wherein, during one feature aggregation operation, the aggregation module is further used to:

[0044] Determine the weighted weights corresponding to each neighboring node of each node based on the information weights and adjacent element values ​​corresponding to each neighboring node of each node;

[0045] Based on the weighted weights corresponding to each neighboring node of each node, the features of each neighboring node of each node are aggregated to obtain new features of each node.

[0046] According to a third aspect, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of any one of the methods described in the first aspect are implemented.

[0047] According to a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method described in any one of the first aspects are implemented.

[0048] In a fifth aspect, the present application provides a computer program product, comprising: a computer program code, when the computer program code is run on a computer, the computer executes any one of the methods described in the first aspect.

[0049] In the embodiment of the present application, each base station in the network to be predicted is used as a node, and the communication relationship between base stations is used as an edge to generate graph structure data. Based on the path loss and signal quality of each base station's transmission signal to other base stations, the values ​​of each adjacent element in the adjacency matrix are determined. Then, based on the information weight and adjacent element value corresponding to each neighbor node of each node in the graph structure data, at least one feature aggregation operation is performed on each node to obtain the flow prediction result of the network to be predicted. In this way, the path loss and signal quality information derived from the physical model are used to determine the adjacency matrix, which is closer to the actual communication environment, enhancing the practicality and accuracy of the model. The model can more accurately capture the dependency between nodes, thereby improving the prediction accuracy of network traffic. At the same time, an aggregation mechanism based on information weight and adjacent element value is introduced, so that the model can more finely adjust the importance of information during the aggregation process, making the aggregated features more accurate and optimizing the learning process. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 A schematic diagram of graph structure data provided in an embodiment of the present application;

[0051] Figure 2 A schematic diagram of an adjacency matrix provided in an embodiment of the present application;

[0052] Figure 3 A schematic diagram of the implementation process of the network traffic prediction method provided in the embodiment of the present application;

[0053] Figure 4 A schematic diagram of determining an adjacency matrix in an embodiment of the present application;

[0054] Figure 5 Schematic diagram of the polymerization process in the embodiment of the present application;

[0055] Figure 6 This is a schematic diagram of the structure of a network traffic prediction device in an embodiment of the present application;

[0056] Figure 7 Schematic diagram of the structure of an electronic device in an embodiment of the present application. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0058] For ease of understanding, the following terms related to the embodiments of the present application are explained:

[0059] Graph Neural Network (GNN): A neural network used to process graph structured data. In GNN, graph structured data consists of nodes (representing entities) and edges (representing relationships between entities). It can capture complex dependencies between nodes and update the representation of each node by aggregating information from neighboring nodes. It is widely used in social network analysis, recommendation systems, bioinformatics and other fields.

[0060] The following is a brief introduction to the design concept of the embodiment of the present application:

[0061] With the development of the fifth generation mobile communication technology (5G), the cellular traffic from various devices (especially mobile devices) continues to increase. As an indispensable part of the management of cellular network resources, the prediction of cellular traffic is crucial in the configuration of cellular network resources and the improvement of cellular network service quality.

[0062] In the prediction of cellular traffic, artificial intelligence has shown great potential, especially deep learning methods such as long short-term memory networks (LSTM). The LSTM has been proven to be advantageous in time series prediction. However, LSTM processes cellular traffic in the time dimension and ignores the information of cellular traffic in the spatial dimension. Based on this, GNN, which can intuitively process and analyze spatial dimension information, has entered the public eye and become the key to improving the prediction accuracy of cellular traffic.

[0063] Currently, cellular traffic is predicted by GNN. When predicting cellular traffic, GNN is first constructed, and then the cellular traffic is predicted based on the constructed GNN.

[0064] In the prior art, when constructing a GNN, each base station constituting a mobile communication network is firstly regarded as a node, and the communication between base stations is regarded as an edge, so as to convert each base station into a graph structure (such as Figure 1 Then, based on the transformed graph structure, an adjacency matrix is ​​constructed (as shown in Figure 2 shown), in Figure 2 In the adjacency matrix shown, an element value of 1 indicates that there is a connection between nodes, and an element value of 0 indicates that there is no connection between nodes. When GNN processes graph structure data, it needs to extract spatial information, so all nodes need to be aggregated. For example, for node v, in the first aggregation, the new node hv will be obtained by weighted average of the features of neighboring nodes, which can be expressed as: Among them, h' v is the new feature of node v after aggregation, N(v) is the set of neighbor nodes of v. hu is the feature of neighbor node u, for example, Figure 1 For node 2 in , its new feature h'2 is calculated by weighted average (adjusted by weight matrix W and activated by ReLU function) of the features of nodes 1, 3, and 4. This aggregation method allows each node to capture the information of its neighbors, while at the same time obtaining information of more distant nodes after multiple aggregations.

[0065] However, when constructing the adjacency matrix, the correlation of the time series depends on the cellular traffic, making the adjacency matrix value affected by the data quality of the cellular traffic itself. When the data quality is average, the adjacency matrix value will be biased, resulting in low prediction accuracy when predicting cellular traffic through GNN. At the same time, when nodes are aggregated, the random weight parameters will lead to inaccurate features after aggregation, resulting in low prediction accuracy when predicting cellular traffic through GNN.

[0066] In view of this, in the embodiment of the present application, a network traffic prediction method, device, electronic device and storage medium are proposed, each base station in the network to be predicted is used as a node, and the communication relationship between base stations is used as an edge to generate graph structure data, and the path loss and signal quality of each base station's transmission signal to other base stations are used to determine the value of each adjacent element in the adjacency matrix, and then based on the information weight and adjacent element value corresponding to each neighbor node of each node in the graph structure data, at least one feature aggregation operation is performed on each node to obtain the traffic prediction result of the network to be predicted, wherein the information weight represents: the importance of the information of the neighbor node of the corresponding node to the corresponding node. In this way, the adjacency matrix is ​​determined by the path loss and signal quality information derived from the physical model, and the adjacency matrix is ​​closer to the actual communication environment, which enhances the practicality and accuracy of the model. The model can more accurately capture the dependency relationship between nodes, thereby improving the prediction accuracy of network traffic. At the same time, an aggregation mechanism based on information weight and adjacent element value is introduced, so that the model can more finely adjust the importance of information during the aggregation process, and optimize the learning process. In addition, the model can adapt to various complex communication environments, especially in cities or complex terrains, and the adaptability and generalization ability of the model are greatly enhanced. The model can adjust the prediction strategy based on the current network status in real time, especially in the event of emergencies, and can quickly respond to environmental changes and maintain the stability and accuracy of the prediction.

[0067] like Figure 3 As shown, it is a schematic diagram of the implementation process of the network traffic prediction method provided in the embodiment of the present application, which may specifically include the following steps:

[0068] Step 30: Generate graph structure data by taking each base station in the network to be predicted as a node and the communication relationship between the base stations as an edge.

[0069] In the embodiment of the present application, each base station in the network to be predicted is used as a node, the traffic data of each base station is used as a feature vector of the corresponding node, and the communication relationship between the base stations is used as an edge to generate graph structure data.

[0070] Among them, when the communication relationship between base stations is taken as the edge, it can be considered that there is a communication relationship between the two base stations, that is, it can be considered that there is an edge between their corresponding nodes. If there is no communication relationship between the two base stations, it can be considered that there is no edge between their corresponding nodes.

[0071] At the same time, in the embodiment of the present application, there is no restriction on the specific parameters of the traffic data of each base station, which can be set based on demand, for example, the traffic data per hour in a day.

[0072] Step 31: Determine the value of each adjacent element in the adjacency matrix based on the path loss and signal quality of each base station's transmission signal to other base stations.

[0073] Among them, each adjacent element value is used to represent the path loss and signal quality of a base station transmitting a signal to another base station. The signal quality can be determined based on factors such as the height, directivity, gain, and power of the base station antenna.

[0074] The adjacent element value in the embodiment of the present application can be expressed as:

[0075]

[0076] Among them, PL ij is the path loss of the signal transmitted from node i to node j, Q ij is the signal quality, α and β are adjustment parameters.

[0077] Optionally, in an embodiment of the present application, in order to obtain the path loss of each base station transmitting a signal to other base stations, a possible embodiment is provided, and the specific steps are as follows:

[0078] Step 310: Obtain free space path loss, reflection loss, projection loss, and diffraction loss of a signal transmitted from a first base station to a second base station in each base station.

[0079] The first base station and the second base station are two different base stations among the base stations.

[0080] Optionally, in an embodiment of the present application, a free space path loss of a signal transmitted from the first base station to the second base station is determined based on a distance and a signal frequency from the first base station to the second base station.

[0081] The free space path loss of the signal transmitted from the first base station to the second base station can be expressed as:

[0082] FSPL(dB)=20log 10 (d)+20log 10 (f)-147.55

[0083] Where d is the distance from the first base station to the second base station, f is the signal frequency, and the constant -147.55 is derived from the free space propagation model, which includes the conversion relationship between the speed of light C and frequency.

[0084] Since the relationship between the attenuation of electromagnetic waves and distance is exponential, a logarithmic function is used to measure the relationship between distance and attenuation. The logarithmic transformation linearizes this exponential relationship, making it easier to calculate and understand. The logarithmic function also helps convert multiplication operations into addition, simplifying the calculation process.

[0085] The coefficient of 20 comes from the relationship between power loss and distance, where the quadratic effect of distance leads to the coefficient of 20 (10×2), which reflects the physical property that the power of electromagnetic waves decays as the square of the distance increases.

[0086] Optionally, in an embodiment of the present application, the reflection loss of the signal transmitted by the first base station to the second base station is determined based on the vertical reflection loss corresponding to the vertical polarization of the signal transmitted by the first base station to the second base station and the parallel reflection loss corresponding to the parallel polarization.

[0087] The reflection loss of the signal transmitted from the first base station to the second base station can be expressed as:

[0088]

[0089] Among them, L reflection,∥ is the parallel reflection loss corresponding to parallel polarization, L reflection,⊥ is the vertical reflection loss corresponding to vertical polarization.

[0090] In the embodiment of the present application, the parallel reflection loss corresponding to the parallel polarization can be calculated by the parallel reflection coefficient corresponding to the parallel polarization, and the vertical reflection loss corresponding to the vertical polarization can be calculated by the vertical reflection coefficient corresponding to the vertical polarization.

[0091] The parallel reflection loss corresponding to parallel polarization can be expressed as:

[0092] L reflection,∥ =10·log 10 (|Γ ∥ | 2 )

[0093] The vertical reflection loss corresponding to vertical polarization can be expressed as:

[0094] L reflection,⊥ =10·log 10 (|Γ ⊥ |2)

[0095] Among them, Γ ∥ is the parallel reflection coefficient corresponding to parallel polarization, Γ ⊥ is the vertical reflection coefficient corresponding to vertical polarization.

[0096] In the embodiment of the present application, the parallel reflection coefficient corresponding to the parallel polarization and the vertical reflection coefficient corresponding to the vertical polarization can be calculated through the incident angle and the relative dielectric constant.

[0097] The parallel reflection coefficient corresponding to parallel polarization can be expressed as:

[0098]

[0099] The vertical reflection coefficient corresponding to vertical polarization can be expressed as:

[0100]

[0101] Among them, θ i is the incident angle, ε r is the relative dielectric constant.

[0102] Optionally, in an embodiment of the present application, the projection loss of the signal transmitted by the first base station to the second base station is determined based on the vertical projection loss corresponding to the vertical polarization of the signal transmitted by the first base station to the second base station and the parallel projection loss corresponding to the parallel polarization.

[0103] The projection loss of the signal transmitted from the first base station to the second base station can be expressed as:

[0104]

[0105] Among them, L transmission,∥ is the parallel projection loss corresponding to parallel polarization, L transmission,⊥ is the vertical projection loss corresponding to vertical polarization.

[0106] In the embodiment of the present application, the parallel projection loss corresponding to parallel polarization can be calculated by the parallel projection coefficient corresponding to parallel polarization, and the vertical projection loss corresponding to vertical polarization can be calculated by the vertical projection coefficient corresponding to vertical polarization.

[0107] The parallel projection loss corresponding to parallel polarization can be expressed as:

[0108] L transmission,∥ =10·log 10 (|T ∥ | 2 )

[0109] The vertical projection loss corresponding to vertical polarization can be expressed as:

[0110] L transmission,⊥ =10·log 10 (|T ⊥ | 2)

[0111] Among them, T ∥ is the parallel projection coefficient corresponding to parallel polarization, T ⊥ is the vertical projection coefficient corresponding to vertical polarization.

[0112] In the embodiment of the present application, the parallel projection coefficient corresponding to the parallel polarization is calculated by the parallel reflection coefficient corresponding to the parallel polarization, and the vertical projection coefficient corresponding to the vertical polarization is calculated by the vertical reflection coefficient corresponding to the vertical polarization.

[0113] The parallel projection coefficient corresponding to parallel polarization can be expressed as:

[0114] T ∥ =1-|Γ ∥ | 2

[0115] The vertical projection coefficient corresponding to vertical polarization can be expressed as:

[0116] T ⊥ =1-|Γ ⊥ | 2

[0117] Among them, Γ ∥ is the parallel reflection coefficient corresponding to parallel polarization, Γ ⊥ is the vertical reflection coefficient corresponding to vertical polarization.

[0118] Optionally, in an embodiment of the present application, the diffraction loss is determined using the Geometrical Theory of Diffraction (GTD), and the diffraction loss of the signal transmitted from the first base station to the second base station is determined based on the wave number, diffraction angle, and distance from the diffraction edge to the second base station of the signal transmitted from the first base station to the second base station. This is not limited in the embodiment of the present application. For example, the equivalent edge current method (Method of Equivalent Currents, MEC) may also be used.

[0119] The diffraction loss of the signal transmitted from the first base station to the second base station can be expressed as:

[0120]

[0121] Where d is the distance from the diffraction edge to the second base station, θ is the diffraction angle, and k is the wave number. The wave number describes the rate of change of the wave in space, that is, the number of cycles of the wave within a unit distance. For electromagnetic waves, the wave number is related to the frequency and propagation speed of the wave. The wave number is defined as the inverse of the wavelength multiplied by 2π.

[0122] In this way, through the above method, accurate free space path loss, reflection loss, projection loss, and diffraction loss can be obtained, thereby obtaining accurate path loss.

[0123] Step 311: Determine the path loss of a signal transmitted from the first base station to the second base station based on the free space path loss, reflection loss, projection loss, and diffraction loss.

[0124] In the embodiment of the present application, the free space path loss, reflection loss, projection loss, and diffraction loss are summed to obtain the path loss of the first base station transmitting the signal to the second base station.

[0125] The path loss of the signal transmitted from the first base station to the second base station can be expressed as:

[0126] PLij=FSPL(dB)+L reflection +L transmission +L diff

[0127] Wherein, FSPL (dB) is the free space path loss of the signal transmitted from the first base station to the second base station, L reflection is the reflection loss of the signal transmitted from the first base station to the second base station, L transmission is the projection loss of the signal transmitted from the first base station to the second base station, L diff It is the diffraction loss of the signal transmitted from the first base station to the second base station.

[0128] In this way, due to the free space path loss, it is difficult to fit the actual situation well. In actual applications, mobile communications will inevitably encounter factors such as refraction, reflection, and occlusion. The path loss is determined by multiple factors. These factors reflect the various physical obstacles and environmental characteristics encountered by electromagnetic waves during propagation. Therefore, in addition to considering the free path loss, the refraction and reflection information of wireless signals when encountering walls and certain materials are also considered to obtain accurate path loss, making the element values ​​of the adjacency matrix of the graph neural network more accurate, making the adjacency matrix closer to the actual communication environment, and enhancing the practicality and accuracy of the model.

[0129] See also Figure 4 As shown, it is a schematic diagram of determining the adjacency matrix in an embodiment of the present application. The free space path loss, reflection loss, projection loss, and diffraction loss are determined by reflection data, refraction data, base station data, electromagnetic wave data, distance, environmental parameters, multipath effect, etc., and the adjacency matrix is ​​optimized by the free space path loss, reflection loss, projection loss, and diffraction loss.

[0130] Step 32: Based on the information weights and adjacent element values ​​corresponding to each neighboring node of each node in the graph structure data, perform at least one feature aggregation operation on each node to obtain a traffic prediction result of the network to be predicted.

[0131] The information weight represents the importance of the information of the neighboring nodes of the corresponding node to the corresponding node.

[0132] Optionally, in the embodiment of the present application, during a feature aggregation operation, the following operations are performed:

[0133] Step 320: Determine the weighted weights corresponding to each neighboring node of each node based on the information weights and adjacent element values ​​corresponding to each neighboring node of each node.

[0134] In an embodiment of the present application, for each node, the following operations are performed respectively: based on the information weights and adjacent element values ​​corresponding to each neighbor node of a node, the weighted weights corresponding to each neighbor node of the node are determined, wherein the information weight corresponding to a neighbor node of a node represents: the importance of the information of the neighbor node of the node to the node.

[0135] Optionally, in an embodiment of the present application, a possible embodiment is provided to obtain the information weight corresponding to each neighboring node of each node, and the specific steps are as follows:

[0136] For each node, the following operations are performed respectively: based on the path loss of a signal received by a node from each of its neighboring nodes, the information weight corresponding to each of the neighboring nodes of a node is calculated.

[0137] The information weight corresponding to a neighbor node j of a node i can be expressed as:

[0138] ωj=exp(-α·PL ji )

[0139] Among them, PL ji is the path loss of the signal received by node i from neighbor node j, that is, the path loss of the signal transmitted by neighbor node j to node i, and α is an adjustment parameter.

[0140] In this way, the importance of each neighbor node feature is adjusted according to the path loss of the signal, which is close to the actual communication environment and can obtain accurate information weight.

[0141] In the embodiment of the present application, the weighted weight corresponding to a neighbor node j of a node i can be expressed as:

[0142]

[0143] Among them, A ij is the adjacent element value corresponding to node i to neighbor node j, ωj is the information weight corresponding to neighbor node j of node i, and λ is the adjustment parameter used to balance the adjustment of the adjacency matrix and the adjustment of the information weight, which is between 0 and 1.

[0144] In this way, the adjacent element value can characterize the connection strength between the node and the neighboring node. The weighted weight is determined by the information weight and the adjacent element value, which is close to the actual communication environment and can effectively improve the prediction accuracy and model robustness in tasks such as network traffic prediction.

[0145] Step 321: Based on the weighted weights corresponding to the neighboring nodes of each node, the features of the neighboring nodes of each node are aggregated to obtain new features of each node.

[0146] In the embodiment of the present application, the following operations are performed for each node: based on the weighted weights corresponding to each neighboring node of a node, the features of each neighboring node of the node are aggregated to obtain new features of the node.

[0147] In the embodiment of the present application, the characteristics of a node i at layer l+1 can be expressed as:

[0148]

[0149] in, is the weighted weight corresponding to neighbor node j, is the feature of the neighbor node j in the lth layer, W and b are learning parameters, N(i) is the set of neighbor nodes of node i, and σ is the activation function, which can usually use sigmoid.

[0150] In this way, the aggregation process not only considers the signal transmission loss between nodes, that is, the information weight, but also combines the optimization effect of the graph structure, that is, the adjacent element value, which improves the accuracy of the model prediction and also improves the performance of the model in complex traffic prediction tasks.

[0151] See also Figure 5 As shown, it is a schematic diagram of the aggregation process in the embodiment of the present application. The information weight and the adjacent element value are determined by reflection data, refraction data, base station data, electromagnetic wave data, distance, environmental parameters, multipath effect, etc., and the aggregation process is changed by the information weight and the adjacent element value.

[0152] Based on the same inventive concept, a network traffic prediction device is also provided in the embodiment of the present application. Figure 6 FIG. 1 is a schematic diagram of the structure of a network traffic prediction device in an embodiment of the present application, which specifically includes:

[0153] A generation module 601 is used to generate graph structure data by taking each base station in the network to be predicted as a node and the communication relationship between the base stations as an edge;

[0154] A determination module 602 is used to determine the value of each adjacent element in the adjacency matrix based on the path loss and signal quality of each base station transmitting a signal to other base stations;

[0155] Aggregation module 603 is used to perform at least one feature aggregation operation on each node based on the information weight and adjacent element value corresponding to each neighbor node of each node in the graph structure data, so as to obtain the traffic prediction result of the network to be predicted, wherein the information weight represents: the importance of the information of the neighbor nodes of the corresponding node to the corresponding node.

[0156] In a possible embodiment, based on the path loss and signal quality of each base station transmitting a signal to other base stations, before determining the value of each adjacent element in the adjacency matrix, the determination module 602 is further configured to:

[0157] Obtaining a free space path loss, a reflection loss, a projection loss, and a diffraction loss of a signal transmitted from a first base station to a second base station among the base stations, wherein the first base station and the second base station are two different base stations among the base stations;

[0158] Based on the free space path loss, reflection loss, projection loss, and diffraction loss, the path loss of the first base station transmitting the signal to the second base station is determined.

[0159] In a possible embodiment, when obtaining the free space path loss, reflection loss, projection loss, and diffraction loss of a signal transmitted from a first base station to a second base station in each base station, the determination module 602 is further used to:

[0160] Determine a free space path loss of a signal transmitted from the first base station to the second base station based on a distance and a signal frequency of the signal transmitted from the first base station to the second base station;

[0161] Determine the reflection loss of the signal transmitted from the first base station to the second base station based on the vertical reflection loss corresponding to the vertical polarization and the parallel reflection loss corresponding to the parallel polarization of the signal transmitted from the first base station to the second base station;

[0162] Determine the projection loss of the signal transmitted from the first base station to the second base station based on the vertical projection loss corresponding to the vertical polarization and the parallel projection loss corresponding to the parallel polarization of the signal transmitted from the first base station to the second base station;

[0163] The diffraction loss of the signal transmitted from the first base station to the second base station is determined based on the wave number, the diffraction angle and the distance from the diffraction edge to the second base station of the signal transmitted from the first base station to the second base station.

[0164] In a possible embodiment, based on the information weights and adjacent element values ​​corresponding to each neighboring node of each node in the graph structure data, at least one feature aggregation operation is performed on each node, and before obtaining the traffic prediction result of the network to be predicted, the aggregation module 603 is further used to:

[0165] For each node, the following operations are performed respectively: based on the path loss of a signal received by a node from each of its neighboring nodes, the information weight corresponding to each of the neighboring nodes of a node is calculated.

[0166] In a possible embodiment, based on the information weights and adjacent element values ​​corresponding to each neighboring node of each node in the graph structure data, at least one feature aggregation operation is performed on each node to obtain a traffic prediction result of the network to be predicted, wherein, during one feature aggregation operation, the aggregation module 603 is further used to:

[0167] Determine the weighted weights corresponding to each neighboring node of each node based on the information weights and adjacent element values ​​corresponding to each neighboring node of each node;

[0168] Based on the weighted weights corresponding to each neighboring node of each node, the features of each neighboring node of each node are aggregated to obtain new features of each node.

[0169] Based on the above embodiments, see Figure 7 Shown is a schematic diagram of the structure of an electronic device in an embodiment of the present application.

[0170] An embodiment of the present application provides an electronic device, which may include a processor 710 (Center Processing Unit, CPU), a memory 720, an input device 730 and an output device 740, etc. The input device 730 may include a keyboard, a mouse, a touch screen, etc., and the output device 740 may include a display device, such as a liquid crystal display (Liquid Crystal Display, LCD), a cathode ray tube (Cathode Ray Tube, CRT), etc.

[0171] The memory 720 may include a read-only memory (ROM) and a random access memory (RAM), and provides the processor 710 with program instructions and data stored in the memory 720. In the embodiment of the present application, the memory 720 may be used to store the program of any network traffic prediction method in the embodiment of the present application.

[0172] The processor 710 calls the program instructions stored in the memory 720, and the processor 710 is used to execute any network traffic prediction method in the embodiments of the present application according to the obtained program instructions.

[0173] Based on the above embodiments, in an embodiment of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the network traffic prediction method in any of the above method embodiments is implemented.

[0174] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0175] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0176] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0177] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0178] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A network traffic prediction method, characterized in that: include: Each base station in the network to be predicted is used as a node, and the communication relationship between base stations is used as an edge to generate graph structure data; Determine the value of each adjacent element in the adjacency matrix based on the path loss and signal quality of each base station transmitting a signal to other base stations; Based on the information weights corresponding to each neighboring node of each node in the graph structure data and the adjacent element values, at least one feature aggregation operation is performed on each node to obtain a traffic prediction result of the network to be predicted, wherein the information weight represents: the importance of the information of the neighboring nodes of the corresponding node to the corresponding node.

2. The method according to claim 1, characterized in that Before determining the value of each adjacent element in the adjacency matrix based on the path loss and signal quality of each base station transmitting a signal to other base stations, the method further includes: Obtaining a free space path loss, a reflection loss, a projection loss, and a diffraction loss of a signal transmitted from a first base station to a second base station among the base stations, wherein the first base station and the second base station are two different base stations among the base stations; Based on the free space path loss, the reflection loss, the projection loss, and the diffraction loss, a path loss of a signal transmitted from the first base station to the second base station is determined.

3. The method according to claim 2, characterized in that The obtaining of free space path loss, reflection loss, projection loss, and diffraction loss of a signal transmitted from a first base station to a second base station among the base stations includes: Determine a free space path loss of a signal transmitted from the first base station to the second base station based on a distance and a signal frequency from the first base station to the second base station; Determine the reflection loss of the signal transmitted from the first base station to the second base station based on the vertical reflection loss corresponding to the vertical polarization and the parallel reflection loss corresponding to the parallel polarization of the signal transmitted from the first base station to the second base station; Determine a projection loss of a signal transmitted from the first base station to the second base station based on a vertical projection loss corresponding to a vertical polarization and a parallel projection loss corresponding to a parallel polarization of a signal transmitted from the first base station to the second base station; The diffraction loss of the signal transmitted from the first base station to the second base station is determined based on the wave number, diffraction angle and distance from the diffraction edge to the second base station of the signal transmitted from the first base station to the second base station.

4. The method according to claim 1, characterized in that: Before performing at least one feature aggregation operation on each node based on the information weights and the adjacent element values ​​corresponding to each neighboring node of each node in the graph structure data to obtain the traffic prediction result of the network to be predicted, the method includes: For each of the nodes, the following operations are performed respectively: based on the path loss of signals received by a node from each of its neighboring nodes, the information weight corresponding to each of the neighboring nodes of the node is calculated.

5. The method according to claim 1, characterized in that Based on the information weights and the adjacent element values ​​corresponding to the neighboring nodes of each node in the graph structure data, at least one feature aggregation operation is performed on each node to obtain the traffic prediction result of the network to be predicted, wherein during one feature aggregation operation, the following operations are performed: Determine the weighted weights corresponding to the neighboring nodes of each node based on the information weights corresponding to the neighboring nodes of each node and the adjacent element values; Based on the weighted weights corresponding to the neighboring nodes of each node, the features of the neighboring nodes of each node are aggregated to obtain new features of each node.

6. A network traffic prediction device, characterized in that: include: A generation module, used to generate graph structure data by taking each base station in the network to be predicted as a node and the communication relationship between the base stations as an edge; A determination module, configured to determine the value of each adjacent element in the adjacency matrix based on the path loss and signal quality of each base station transmitting a signal to other base stations; An aggregation module is used to perform at least one feature aggregation operation on each node based on the information weight corresponding to each neighbor node of each node in the graph structure data and the adjacent element value, so as to obtain the traffic prediction result of the network to be predicted, wherein the information weight represents: the importance of the information of the neighbor nodes of the corresponding node to the corresponding node.

7. The device according to claim 6, characterized in that Before determining the value of each adjacent element in the adjacency matrix based on the path loss and signal quality of each base station transmitting a signal to other base stations, the determining module is further used to: Obtaining a free space path loss, a reflection loss, a projection loss, and a diffraction loss of a signal transmitted from a first base station to a second base station among the base stations, wherein the first base station and the second base station are two different base stations among the base stations; Based on the free space path loss, the reflection loss, the projection loss, and the diffraction loss, a path loss of a signal transmitted from the first base station to the second base station is determined.

8. The device according to claim 7, characterized in that When obtaining the free space path loss, reflection loss, projection loss, and diffraction loss of a signal transmitted from a first base station to a second base station among the base stations, the determination module is further used to: Determine a free space path loss of a signal transmitted from the first base station to the second base station based on a distance and a signal frequency from the first base station to the second base station; Determine the reflection loss of the signal transmitted from the first base station to the second base station based on the vertical reflection loss corresponding to the vertical polarization and the parallel reflection loss corresponding to the parallel polarization of the signal transmitted from the first base station to the second base station; Determine a projection loss of a signal transmitted from the first base station to the second base station based on a vertical projection loss corresponding to a vertical polarization and a parallel projection loss corresponding to a parallel polarization of a signal transmitted from the first base station to the second base station; The diffraction loss of the signal transmitted from the first base station to the second base station is determined based on the wave number, diffraction angle and distance from the diffraction edge to the second base station of the signal transmitted from the first base station to the second base station.

9. The device according to claim 6, characterized in that Before performing at least one feature aggregation operation on each node based on the information weights and the adjacent element values ​​corresponding to each neighboring node of each node in the graph structure data to obtain the traffic prediction result of the network to be predicted, the aggregation module is further used to: For each of the nodes, the following operations are performed respectively: based on the path loss of signals received by a node from each of its neighboring nodes, the information weight corresponding to each of the neighboring nodes of the node is calculated.

10. The device according to claim 6, characterized in that Based on the information weights and the adjacent element values ​​corresponding to the neighboring nodes of each node in the graph structure data, at least one feature aggregation operation is performed on each node to obtain the traffic prediction result of the network to be predicted, wherein during one feature aggregation operation, the aggregation module is further used to: Determine the weighted weights corresponding to the neighboring nodes of each node based on the information weights corresponding to the neighboring nodes of each node and the adjacent element values; Based on the weighted weights corresponding to the neighboring nodes of each node, the features of the neighboring nodes of each node are aggregated to obtain new features of each node.

11. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the method according to any one of claims 1 to 5 are implemented.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Citation Information

Cited By

  • Wireless charging power adaptive adjustment method and system based on machine learning

    CN120824874A