Link-level traffic prediction method and device for 6G heterogeneous aggregation networks

Abstract

The present application discloses a link-level traffic prediction method and device for a 6G heterogeneous aggregation network, the method comprising: based on a preset test environment, using a preset 6G traffic generation model to generate a 6G link-level user traffic data set by link allocation; using the 6G link-level user traffic data set to perform traffic prediction training on the initial traffic prediction model to obtain a 6G user traffic prediction model; inputting the historical link-level user traffic data of the target user into the 6G user traffic prediction model to perform future traffic prediction to obtain a target traffic prediction result. Generating user traffic data by using a preset 6G traffic generation model can provide more reliable analysis data; using a traffic prediction model for prediction can ensure the accuracy of the prediction results. Therefore, the present application can solve the technical problem that the prior art cannot simultaneously capture the spatiotemporal characteristics of user traffic, and the data construction method is not suitable for 6G heterogeneous aggregation networks, resulting in a lack of accuracy in the prediction results.

Description

Technical Field

[0001] The present application relates to the technical field of network traffic analysis, and in particular to a link-level traffic prediction method and device for 6G heterogeneous aggregation networks. Background Art

[0002] Based on the widespread deployment of 5G wireless communication networks, 6G technology has also begun to be studied and will achieve breakthroughs, including the trend of "Internet of Everything" and the integration of "air-space-ground" communications. Therefore, more types of end users will access the 6G network. As a key technology for realizing 6G communication, heterogeneous link aggregation technology can allocate user traffic requirements to different links. Predicting the traffic requirements of dynamic users on heterogeneous links in advance can manage link resources in advance, increase network capacity, reduce network congestion, and improve user experience.

[0003] However, traditional time series prediction methods all assume that there is a linear relationship between time series data, and the linear model based on this cannot capture the complex nonlinear spatiotemporal characteristics of mobile traffic. The deep learning-based methods only focus on the time dimension of the underlying data and lack modeling of the spatial dimension. In addition, the goal of most existing studies is to predict the traffic generated by users in a specific cell. Unlike the user traffic in conventional cellular-based wireless networks, the user traffic within the 6G heterogeneous link aggregation network varies greatly; therefore, the traffic dataset constructed based on the cellular-based wireless network cannot reflect the traffic characteristics of users within the 6G heterogeneous link aggregation network, resulting in inaccurate prediction results. Summary of the invention

[0004] The present application provides a link-level traffic prediction method and device for 6G heterogeneous aggregation networks, which are used to solve the technical problems that the existing technology cannot simultaneously capture the temporal and spatial characteristics of user traffic, and the data construction method is not suitable for 6G heterogeneous aggregation networks, resulting in a lack of accuracy in the prediction results.

[0005] In view of this, the first aspect of the present application provides a link-level traffic prediction method for a 6G heterogeneous aggregation network, including:

[0006] Based on the preset test environment, a preset 6G traffic generation model is used to generate a 6G link-level user traffic dataset through link allocation;

[0007] The 6G link-level user traffic data set is used to perform traffic prediction training on the initial traffic prediction model to obtain a 6G user traffic prediction model, wherein the initial traffic prediction model includes an LSTM time prediction module and a GAT space prediction module;

[0008] The historical link-level user traffic data of the target user is input into the 6G user traffic prediction model to perform future traffic prediction to obtain the target traffic prediction result.

[0009] Preferably, based on a preset test environment, generating a 6G link-level user traffic data set by link allocation using a preset 6G traffic generation model includes:

[0010] Generate link request information of multiple users using a preset 6G traffic generation model, where the link request information includes a request start time, a request rate, a request duration, and a request link;

[0011] Based on the preset test environment, an intelligent scheduling algorithm is used to perform link allocation operations according to the link request information to generate link-level user traffic information;

[0012] Collect user traffic data from the link-level user traffic information according to a preset time period to generate a 6G link-level user traffic data set.

[0013] Preferably, the method of generating a 6G link-level user traffic data set by link allocation based on a preset test environment using a preset 6G traffic generation model also includes:

[0014] Visualize the temporal and spatial characteristics of 5G user traffic data to obtain a 5G spatiotemporal traffic data set;

[0015] The 5G spatiotemporal traffic data set is used to train an initial 5G traffic generation model to obtain an optimized 5G traffic generation model;

[0016] The optimized 5G traffic generation model is differentiatedly adjusted by presetting 6G user characteristic parameters to obtain a preset 6G traffic generation model.

[0017] Preferably, the method of using the 6G link-level user traffic data set to perform traffic prediction training on the initial traffic prediction model to obtain the 6G user traffic prediction model further includes:

[0018] Construct a time prediction network based on the LSTM network framework to obtain the LSTM time prediction module;

[0019] Based on the GAT framework and attention mechanism, a spatial prediction network is constructed to obtain the GAT spatial prediction module.

[0020] The LSTM time prediction module and the GAT space prediction module are combined to generate an initial traffic prediction model.

[0021] Preferably, the method of using the 6G link-level user traffic data set to perform traffic prediction training on the initial traffic prediction model to obtain the 6G user traffic prediction model further includes:

[0022] Converting the 6G link-level user traffic data in the 6G link-level user traffic data set into graph structure data to obtain a 6G graph structure data set;

[0023] The using the 6G link-level user traffic data set to perform traffic prediction training on the initial traffic prediction model to obtain the 6G user traffic prediction model includes:

[0024] The 6G graph structure data set is used to perform traffic prediction training on the initial traffic prediction model to obtain a 6G user traffic prediction model.

[0025] The second aspect of the present application provides a link-level traffic prediction device for a 6G heterogeneous aggregation network, including:

[0026] A data generation unit, configured to generate a 6G link-level user traffic data set by link allocation based on a preset test environment and using a preset 6G traffic generation model;

[0027] A model training unit, used to perform traffic prediction training on an initial traffic prediction model using the 6G link-level user traffic data set to obtain a 6G user traffic prediction model, wherein the initial traffic prediction model includes an LSTM time prediction module and a GAT space prediction module;

[0028] The traffic prediction unit is used to input the historical link-level user traffic data of the target user into the 6G user traffic prediction model to perform future traffic prediction and obtain a target traffic prediction result.

[0029] Preferably, the data generating unit is specifically used for:

[0030] Generate link request information of multiple users using a preset 6G traffic generation model, where the link request information includes a request start time, a request rate, a request duration, and a request link;

[0031] Based on the preset test environment, an intelligent scheduling algorithm is used to perform link allocation operations according to the link request information to generate link-level user traffic information;

[0032] Collect user traffic data from the link-level user traffic information according to a preset time period to generate a 6G link-level user traffic data set.

[0033] Preferably, it also includes:

[0034] The 5G analysis unit is used to visualize the temporal and spatial characteristics of 5G user traffic data to obtain a 5G spatiotemporal traffic data set;

[0035] A 5G training unit, configured to train an initial 5G traffic generation model using the 5G spatiotemporal traffic data set to obtain an optimized 5G traffic generation model;

[0036] A difference adjustment unit is used to perform differentiated adjustment on the optimized 5G traffic generation model by presetting 6G user characteristic parameters to obtain a preset 6G traffic generation model.

[0037] Preferably, it also includes:

[0038] The time prediction unit is used to construct a time prediction network based on the LSTM network framework to obtain an LSTM time prediction module;

[0039] The spatial prediction unit is used to build a spatial prediction network based on the GAT framework and attention mechanism to obtain the GAT spatial prediction module;

[0040] The model generation unit is used to generate an initial traffic prediction model by combining the LSTM time prediction module and the GAT space prediction module.

[0041] Preferably, it also includes:

[0042] A data conversion unit, used to convert the 6G link-level user traffic data in the 6G link-level user traffic data set into graph structure data to obtain a 6G graph structure data set;

[0043] The model training unit is specifically used for:

[0044] The 6G graph structure data set is used to perform traffic prediction training on the initial traffic prediction model to obtain a 6G user traffic prediction model.

[0045] It can be seen from the above technical solutions that the embodiments of the present application have the following advantages:

[0046] In the present application, a link-level traffic prediction method for 6G heterogeneous aggregation networks is provided, including: based on a preset test environment, using a preset 6G traffic generation model to generate a 6G link-level user traffic data set by link allocation; using the 6G link-level user traffic data set to perform traffic prediction training on an initial traffic prediction model to obtain a 6G user traffic prediction model, wherein the initial traffic prediction model includes an LSTM time prediction module and a GAT space prediction module; inputting the historical link-level user traffic data of the target user into the 6G user traffic prediction model to perform future traffic prediction to obtain a target traffic prediction result.

[0047] The link-level traffic prediction method for 6G heterogeneous aggregation networks provided by this application generates user traffic data by presetting a 6G traffic generation model, which is more in line with the actual situation, so it can provide more reliable analysis data; in addition, a traffic prediction model including an LSTM time prediction module and a GAT space prediction module is used to predict 6G link-level user traffic data, which can capture the temporal and spatial characteristics of user traffic at the same time, so the accuracy of the prediction results can be ensured. Therefore, this application can solve the technical problem that the prior art cannot capture the temporal and spatial characteristics of user traffic at the same time, and the data construction method is not suitable for 6G heterogeneous aggregation networks, resulting in a lack of accuracy in the prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A schematic diagram of a flow chart of a link-level traffic prediction method for a 6G heterogeneous aggregation network provided in an embodiment of the present application;

[0049] Figure 2 A schematic diagram of the structure of a link-level traffic prediction device for a 6G heterogeneous aggregation network provided in an embodiment of the present application;

[0050] Figure 3 A visualization graph of a 6G link-level user traffic data set provided in an embodiment of the present application;

[0051] Figure 4 A schematic diagram of the training optimization process of the initial traffic prediction model provided in an embodiment of the present application;

[0052] Figure 5 A schematic diagram of the network structure of the initial traffic prediction model provided in an embodiment of the present application;

[0053] Figure 6 A curve chart of prediction results when the prediction step length is 30 minutes provided in the embodiment of the present application. DETAILED DESCRIPTION

[0054] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings 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.

[0055] For easier understanding, see Figure 1 , an embodiment of a link-level traffic prediction method for a 6G heterogeneous aggregation network provided by the present application includes:

[0056] Step 101: Based on a preset test environment, a preset 6G traffic generation model is used to generate a 6G link-level user traffic data set by link allocation.

[0057] Furthermore, step 101 includes:

[0058] A preset 6G traffic generation model is used to generate link request information of multiple users, where the link request information includes request start time, request rate, request duration and request link;

[0059] Based on the preset test environment, an intelligent scheduling algorithm is used to perform link allocation operations according to link request information to generate link-level user traffic information;

[0060] Collect user traffic data from link-level user traffic information according to a preset time period to generate a 6G link-level user traffic data set.

[0061] It should be noted that the preset 6G traffic generation model is a model dedicated to generating user request data for 6G heterogeneous aggregation networks. It can be pre-built based on relevant characteristic parameters set for users of the 6G heterogeneous aggregation network.

[0062] The preset 6G traffic generation model does not directly generate user data, but generates user request-related data before link allocation, which may include but is not limited to request start time, request rate, request duration and request link; then, under the test environment, link allocation is performed based on the generated user link request information, and link-level user traffic information can be generated according to the link allocation situation; link-level user traffic information is the total link traffic information, and each link can contain multiple users, each user can generate multiple traffic information, and link-level user traffic information is sent in the form of data packets. Therefore, this embodiment collects user traffic data in units of users in the link-level user traffic information according to the preset time period to generate a 6G link-level user traffic data set.

[0063] It is understandable that the preset test environment can be simulated and set according to the relevant characteristic parameters of the 6G heterogeneous aggregation network. The preset time period can also be configured according to the actual situation, and the specific time period is not limited here, such as 5 minutes, 15 minutes, 30 minutes, etc., as long as it is reasonable; data of multiple time periods can be collected according to the preset time period, for example, data collection can last for 30 days, and data is collected every day at a preset time period. The number of user traffic data collected can also be selected according to the actual situation. This embodiment collects data for 30 days and obtains 8640 lines of user traffic data by collecting it once every 5 minutes every day. The intelligent scheduling algorithm can reasonably allocate users to different links of the 6G heterogeneous aggregation network according to user request information, ensuring that user needs can be met and link congestion can be avoided, thereby improving the stability of system operation.

[0064] In addition, in order to improve the reliability of data and the accuracy of subsequent predictions, this embodiment can also perform data preprocessing operations on the collected 6G link-level user traffic data set, such as deleting invalid information, supplementing blank information according to traffic data characteristics, and drawing line graphs to visualize the data. Please refer to Figure 3 ; Then analyze whether it conforms to the characteristics of time periodicity and spatial similarity of user traffic.

[0065] Furthermore, step 101, before that, also includes:

[0066] Visualize the temporal and spatial characteristics of 5G user traffic data to obtain a 5G spatiotemporal traffic data set;

[0067] The 5G spatiotemporal traffic dataset is used to train the initial 5G traffic generation model to obtain the optimized 5G traffic generation model;

[0068] By presetting 6G user characteristic parameters, the optimized 5G traffic generation model is differentiated and adjusted to obtain a preset 6G traffic generation model.

[0069] It should be noted that 5G user traffic data can be obtained from operators and is the 5G user traffic data that has been generated. Time characteristic analysis of 5G user traffic data refers to time series analysis of the historical 5G user traffic data that can be obtained, observing the daily, weekly, and monthly user traffic change trends; then analyzing the peaks and troughs of user traffic data in different time periods to find out the peak and trough periods of user traffic; the analysis process can display the change trends in the form of graphs for easy observation and analysis.

[0070] Spatial characteristic analysis of 5G user traffic data refers to using a geographic information system (GIS) to perform spatial analysis based on the geographical location information of user traffic data, and then observing the distribution of user traffic in different regions; analyzing the spatial correlation of user traffic data between different regions through spatial statistical methods, and spatial statistical methods include but are not limited to spatial autocorrelation analysis, geographically weighted regression analysis, etc.

[0071] By using the 5G spatiotemporal traffic data set to train the initial 5G traffic generation model, an optimized 5G traffic generation model can be obtained. In order to obtain a preset 6G traffic generation model without a training data basis, task migration can be performed based on the optimized 5G traffic generation model, that is, the preset 6G user characteristic parameters are added to the optimized 5G traffic generation model for model adjustment, and a traffic generation model suitable for 6G heterogeneous aggregation networks, namely, the preset 6G traffic generation model, can be obtained.

[0072] It is understandable that before building a traffic generation model, it is necessary to clarify the characteristics of the arrival of user traffic data and establish a model that conforms to the characteristics of user traffic arrival and duration. Therefore, this embodiment uses the Pareto distribution model framework to build a traffic generation model, which is an ON / OFF cycle model that conforms to the periodic change characteristics of user traffic data.

[0073] Based on the above analysis of spatiotemporal characteristics, we can draw line graphs for 5G user traffic data in the same area according to 24 hours, 7×24 hours, and 30×24 hours, and count the changes in 5G user traffic data over time to find peak points and trough points. Draw histograms for user traffic data in different regions at the same time, count the relationship between user traffic data and geographical location, and identify peak areas and underestimated areas in geographical locations. The same time can be 8:00, 12:00, 14:00, 18:00, 22:00, 24:00, etc., without specific restrictions.

[0074] The preset 6G user characteristic parameters are a set of parameters obtained based on actual application experience, including but not limited to the peak and trough values ​​of 6G user traffic, model parameter values, mean traffic requests of different user types, and duration of traffic requests of different user types. Since user traffic in future 6G heterogeneous aggregation networks has the characteristics of high peak value and large differences, these two points need to be set as 6G user characteristic parameters to adjust the 5G traffic generation model, and set different mean traffic request values, model parameter values, and duration of traffic request for users with different peak demands. In addition, the preset 6G traffic generation model is used to generate request-related information of users in 6G heterogeneous aggregation networks, rather than directly generating link-level user traffic information.

[0075] Step 102: Use the 6G link-level user traffic data set to perform traffic prediction training on the initial traffic prediction model to obtain a 6G user traffic prediction model. The initial traffic prediction model includes an LSTM time prediction module and a GAT space prediction module.

[0076] Furthermore, step 102, before that, also includes:

[0077] Construct a time prediction network based on the LSTM network framework to obtain the LSTM time prediction module;

[0078] Based on the GAT framework and attention mechanism, a spatial prediction network is constructed to obtain the GAT spatial prediction module.

[0079] The initial traffic prediction model is generated by combining the LSTM time prediction module and the GAT space prediction module.

[0080] Furthermore, step 102, before that, also includes:

[0081] Convert the 6G link-level user traffic data in the 6G link-level user traffic data set into graph structure data to obtain a 6G graph structure data set;

[0082] Step 102 includes:

[0083] The 6G graph structure dataset is used to train the initial traffic prediction model and obtain the 6G user traffic prediction model.

[0084] It should be noted that the 6G user traffic prediction model is a future traffic prediction model that is pre-built and trained and optimized, and the core of the model mainly includes the LSTM time prediction module and the GAT space prediction module. Figure 4 and Figure 5 , the LSTM time prediction module is a time prediction network built based on the LSTM network framework, which can be used to analyze the time characteristics of 6G link-level user traffic data; the GAT space prediction module is a space prediction network built based on the GAT framework and the attention mechanism, which can analyze the space characteristics of 6G link-level user traffic data. Therefore, the use of the 6G user traffic prediction model to predict the link-level user traffic of the 6G heterogeneous aggregation network can capture the temporal and spatial characteristics of the traffic data at the same time, and can ensure the accuracy and reliability of the prediction results. The temporal and spatial characteristics analysis here are essentially the same as the temporal and spatial characteristics analysis of the 5G user traffic data mentioned above, and the specific analysis content will not be repeated. Among them, a unit of LSTM will determine the retention and modification of information through the three parts of the forget gate, the input gate, and the output gate. The prediction accuracy of the traffic data in time is improved through the LSTM module.

[0085] In addition, the 6G link-level user traffic data in the 6G link-level user traffic data set is converted into graph structure data to obtain a 6G graph structure data set. For users in the coverage area of ​​an aggregated base station containing multiple heterogeneous communication links, each link of the heterogeneous aggregated base station is regarded as a point. Each link carries multiple users, and the information of each point changes over time; if two links are adjacent, there is an edge between the two points, and the weight of the edge is determined by the distance between the two links.

[0086] The state of the 6G heterogeneous link aggregation network at each moment is represented as a graph Gt. = (Vt., E, W), where Vt. represents the set of points, E represents the set of edges, and W represents the weights between adjacent points. As time changes, a series of graphs can be obtained, which can be expressed as G = {Gtn.,......,Gt-2.,Gt-1.,Gt.}.

[0087] The GAT spatial prediction module is a spatial prediction network built based on the GAT framework and attention mechanism. The input of the GAT framework is the features of each point, that is, the user traffic information carried on each link, which can be expressed as in, N is the total number of links, F is the information dimension on each link; the output is Although the total number of links N remains unchanged, the information on the links will change as user traffic changes.

[0088] In the GAT spatial prediction module, the degree of association between each point and its surrounding points is calculated through the attention mechanism. The degree of association is the similarity between any two feature points. The specific calculation method of the similarity can be designed according to the actual situation, and this embodiment is not limited to it, such as Euclidean distance, etc. By normalizing the calculated degree of association, the attention coefficient can be obtained to measure the importance of each point to other points. Based on the weighted aggregation calculation of the attention coefficient and the features of each feature point, the aggregated features of each point can be obtained, which is the spatial feature of each point.

[0089] Integrating the LSTM time prediction module and the GAT space prediction module can obtain the initial traffic prediction model; then the 6G link-level user traffic dataset generated by the traffic generation model is used for model training. The actual process uses the 6G graph structure dataset, which is the 6G link-level user traffic dataset after graph structure conversion. Figure 4 It is understandable that the data set can be divided into a training data set, a test data set and a validation data set. The division ratio of this embodiment is 6:2:2. During the training process, the model can output the number of user traffic on the link at m moments. m can be 15 minutes, 20 minutes, 30 minutes, etc. The prediction accuracy of user traffic data of different time lengths will have a certain difference. For example, when m = 30 minutes, the prediction result is Figure 6 shown.

[0090] Step 103: Input the historical link-level user traffic data of the target user into the 6G user traffic prediction model to perform future traffic prediction and obtain the target traffic prediction result.

[0091] The 6G user traffic prediction model obtained through training and optimization can be applied to the link-level user traffic prediction task of the actual 6G heterogeneous aggregation network. It can be used to predict the traffic of users in the 6G heterogeneous aggregation network at a certain moment or certain moments in the future. It can provide theoretical support for the planning and management of the 6G heterogeneous aggregation network and optimize decision-making.

[0092] The link-level traffic prediction method for 6G heterogeneous aggregation networks provided in the embodiment of the present application generates user traffic data by presetting a 6G traffic generation model, which is more in line with the actual situation, so it can provide more reliable analysis data; in addition, the traffic prediction model including the LSTM time prediction module and the GAT space prediction module is used to predict 6G link-level user traffic data, which can capture the temporal and spatial characteristics of user traffic at the same time, so the accuracy of the prediction results can be ensured. Therefore, the embodiment of the present application can solve the technical problem that the prior art cannot capture the temporal and spatial characteristics of user traffic at the same time, and the data construction method is not suitable for 6G heterogeneous aggregation networks, resulting in the lack of accuracy of the prediction results.

[0093] For easier understanding, see Figure 2 The present application provides an embodiment of a link-level traffic prediction device for a 6G heterogeneous aggregation network, including:

[0094] The data generating unit 201 is used to generate a 6G link-level user traffic data set by link allocation based on a preset test environment and using a preset 6G traffic generation model;

[0095] A model training unit 202 is used to perform traffic prediction training on an initial traffic prediction model using a 6G link-level user traffic data set to obtain a 6G user traffic prediction model, wherein the initial traffic prediction model includes an LSTM time prediction module and a GAT space prediction module;

[0096] The traffic prediction unit 203 is used to input the historical link-level user traffic data of the target user into the 6G user traffic prediction model to perform future traffic prediction and obtain a target traffic prediction result.

[0097] Furthermore, the data generating unit 201 is specifically configured to:

[0098] A preset 6G traffic generation model is used to generate link request information of multiple users, where the link request information includes request start time, request rate, request duration and request link;

[0099] Based on the preset test environment, an intelligent scheduling algorithm is used to perform link allocation operations according to link request information to generate link-level user traffic information;

[0100] Collect user traffic data from link-level user traffic information according to a preset time period to generate a 6G link-level user traffic data set.

[0101] Furthermore, it also includes:

[0102] The 5G analysis unit 204 is used to perform a visual analysis of the time characteristics and spatial characteristics of the 5G user traffic data to obtain a 5G spatiotemporal traffic data set;

[0103] A 5G training unit 205 is used to train an initial 5G traffic generation model using a 5G spatiotemporal traffic data set to obtain an optimized 5G traffic generation model;

[0104] The difference adjustment unit 206 is used to perform differentiated adjustment on the optimized 5G traffic generation model by presetting 6G user characteristic parameters to obtain a preset 6G traffic generation model.

[0105] Furthermore, it also includes:

[0106] The time prediction unit 207 is used to construct a time prediction network according to the LSTM network framework to obtain an LSTM time prediction module;

[0107] A spatial prediction unit 208 is used to construct a spatial prediction network based on the GAT framework and the attention mechanism to obtain a GAT spatial prediction module;

[0108] The model generation unit 209 is used to generate an initial traffic prediction model by combining the LSTM time prediction module and the GAT space prediction module.

[0109] Furthermore, it also includes:

[0110] A data conversion unit 210 is used to convert the 6G link-level user traffic data in the 6G link-level user traffic data set into graph structure data to obtain a 6G graph structure data set;

[0111] The model training unit 202 is specifically used for:

[0112] The 6G graph structure dataset is used to train the initial traffic prediction model and obtain the 6G user traffic prediction model.

[0113] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0114] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0115] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0116] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for executing all or part of the steps of the method described in each embodiment of the present application through a computer device (which can be a personal computer, a server, or a network device, etc.). The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (full name in English: Read-Only Memory, English abbreviation: ROM), random access memory (full name in English: Random Access Memory, English abbreviation: RAM), disk or optical disk and other media that can store program codes.

[0117] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A link-level traffic prediction method for 6G heterogeneous aggregation networks, characterized in that: include: Visualize the temporal and spatial characteristics of 5G user traffic data to obtain a 5G spatiotemporal traffic data set; The 5G spatiotemporal traffic data set is used to train an initial 5G traffic generation model to obtain an optimized 5G traffic generation model; The optimized 5G traffic generation model is differentiatedly adjusted by presetting 6G user characteristic parameters to obtain a preset 6G traffic generation model; Based on the preset test environment, the preset 6G traffic generation model is used to generate a 6G link-level user traffic data set by link allocation; The 6G link-level user traffic data set is used to perform traffic prediction training on the initial traffic prediction model to obtain a 6G user traffic prediction model, wherein the initial traffic prediction model includes an LSTM time prediction module and a GAT space prediction module; The historical link-level user traffic data of the target user is input into the 6G user traffic prediction model to perform future traffic prediction to obtain the target traffic prediction result.

2. The link-level traffic prediction method for 6G heterogeneous aggregation networks according to claim 1 is characterized in that: Based on the preset test environment, the preset 6G traffic generation model is used to generate a 6G link-level user traffic data set by link allocation, including: Generate link request information of multiple users using a preset 6G traffic generation model, where the link request information includes a request start time, a request rate, a request duration, and a request link; Based on the preset test environment, an intelligent scheduling algorithm is used to perform link allocation operations according to the link request information to generate link-level user traffic information; Collect user traffic data from the link-level user traffic information according to a preset time period to generate a 6G link-level user traffic data set.

3. The link-level traffic prediction method for 6G heterogeneous aggregation networks according to claim 1 is characterized in that: The 6G link-level user traffic data set is used to perform traffic prediction training on the initial traffic prediction model to obtain a 6G user traffic prediction model, and the method also includes: Construct a time prediction network based on the LSTM network framework to obtain the LSTM time prediction module; Based on the GAT framework and attention mechanism, a spatial prediction network is constructed to obtain the GAT spatial prediction module. The LSTM time prediction module and the GAT space prediction module are combined to generate an initial traffic prediction model.

4. The link-level traffic prediction method for 6G heterogeneous aggregation networks according to claim 1 is characterized in that: The 6G link-level user traffic data set is used to perform traffic prediction training on the initial traffic prediction model to obtain a 6G user traffic prediction model, and the method also includes: Converting the 6G link-level user traffic data in the 6G link-level user traffic data set into graph structure data to obtain a 6G graph structure data set; The using the 6G link-level user traffic data set to perform traffic prediction training on the initial traffic prediction model to obtain the 6G user traffic prediction model includes: The 6G graph structure data set is used to perform traffic prediction training on the initial traffic prediction model to obtain a 6G user traffic prediction model.

5. A link-level traffic prediction device for 6G heterogeneous aggregation networks, characterized in that: include: The 5G analysis unit is used to visualize the temporal and spatial characteristics of 5G user traffic data to obtain a 5G spatiotemporal traffic data set; A 5G training unit, configured to train an initial 5G traffic generation model using the 5G spatiotemporal traffic data set to obtain an optimized 5G traffic generation model; A difference adjustment unit, used to perform differential adjustment on the optimized 5G traffic generation model by presetting 6G user characteristic parameters to obtain a preset 6G traffic generation model; A data generation unit, configured to generate a 6G link-level user traffic data set by link allocation based on a preset test environment and using the preset 6G traffic generation model; A model training unit, used to perform traffic prediction training on an initial traffic prediction model using the 6G link-level user traffic data set to obtain a 6G user traffic prediction model, wherein the initial traffic prediction model includes an LSTM time prediction module and a GAT space prediction module; The traffic prediction unit is used to input the historical link-level user traffic data of the target user into the 6G user traffic prediction model to perform future traffic prediction and obtain a target traffic prediction result.

6. The link-level traffic prediction device for 6G heterogeneous aggregation networks according to claim 5 is characterized in that: The data generating unit is specifically used for: Generate link request information of multiple users using a preset 6G traffic generation model, where the link request information includes a request start time, a request rate, a request duration, and a request link; Based on the preset test environment, an intelligent scheduling algorithm is used to perform link allocation operations according to the link request information to generate link-level user traffic information; Collect user traffic data from the link-level user traffic information according to a preset time period to generate a 6G link-level user traffic data set.

7. The link-level traffic prediction device for 6G heterogeneous aggregation networks according to claim 5 is characterized in that: Also includes: The time prediction unit is used to construct a time prediction network based on the LSTM network framework to obtain an LSTM time prediction module; The spatial prediction unit is used to build a spatial prediction network based on the GAT framework and attention mechanism to obtain the GAT spatial prediction module; The model generation unit is used to generate an initial traffic prediction model by combining the LSTM time prediction module and the GAT space prediction module.

8. The link-level traffic prediction device for 6G heterogeneous aggregation networks according to claim 5 is characterized in that: Also includes: A data conversion unit, used to convert the 6G link-level user traffic data in the 6G link-level user traffic data set into graph structure data to obtain a 6G graph structure data set; The model training unit is specifically used for: The 6G graph structure data set is used to perform traffic prediction training on the initial traffic prediction model to obtain a 6G user traffic prediction model.

Images