5G base station deployment method and device

Abstract

The invention provides a 5G base station deployment method and apparatus. The method comprises the steps of obtaining 4G base station data of a target area; inputting the 4G base station data into a pre-trained base station traffic condition generation model to obtain a 5G base station traffic generation result; obtaining a base station deployment strategy according to the 5G base station flow generation result; wherein the base station traffic condition generation model comprises a traffic association extraction module, a traffic space association feature extraction module and a condition generation module; the traffic correlation extraction module comprises a decoder and an encoder based on an interactive attention mechanism, the traffic space correlation feature extraction module comprises a plurality of graph convolution layers, and the condition generation module comprises a pre-constructed condition diffusion model. According to the invention, 5G flow generation with low complexity, good controllability and good generalization is realized, so that city-level communication base station deployment is guided.

Description

Technical Field

[0001] The present invention relates to the field of mobile communication technologies, and in particular to a 5G base station deployment method and device. Background Art

[0002] The existing deployment of the fifth-generation mobile communication (5G) network has brought great convenience to production and life. Its high-speed data transmission, low latency, and massive connections provide reliable support for applications such as real-time communication, telemedicine, and intelligent manufacturing, and improve user experience. With the rapid development of 5G networks and the advancement of commercialization, telecom operators plan to deploy 5G base stations in larger areas, and the site selection of 5G base stations is a key issue. Traditional deployment strategies mainly refer to factors such as population distribution and 4G network traffic, while the demand for 5G base stations is not only related to the above factors, but in essence, it depends on the demand of regional users for 5G networks. Therefore, under traditional strategies, there will be a mismatch between network supply and demand, resulting in a decline in user experience and waste of resources.

[0003] In fact, the traffic data of 5G mobile networks can serve as an important reference for base station deployment, but the coverage of 5G networks is limited, and many areas lack historical traffic data of 5G base stations. A feasible solution is to estimate and predict 5G traffic based on real scenarios in a given area, thereby helping operators understand the spatiotemporal distribution of 5G network demand.

[0004] 5G traffic estimation is essentially a time series generation problem, which abstracts the traffic series into a time series to capture the temporal relationship. In recent years, machine learning has shown excellent performance in time series prediction tasks, and has also been widely studied in the field of traffic generation. Some studies use long-short term memory (LSTM) networks for 5G traffic prediction, and the long-term dependencies on time series can be effectively captured by designing memory units. Some studies use autoregressive (AR) models to adaptively obtain the cumulative density function of network traffic trajectories, but the requirements for data quality limit its robustness in complex environments.

[0005] On this basis, generative models such as generative adversarial networks (GANs) have shown outstanding performance in the field of data generation by strengthening the training between generators and discriminators. Some studies have proposed a GAN model HierTGAN combined with AR, as well as a knowledge-enhanced GAN. However, traffic generation based on GAN is unstable, prone to mode collapse, and the controllability of traffic generation is insufficient. In addition, machine learning methods are all based on rich 5G historical traffic data in a specific area, which may not be applicable in areas where 5G base stations have not yet been deployed.

[0006] 4G traffic data reflects the geographical location and user behavior characteristics of a region, and can serve as an important reference for generating 5G traffic. At present, major regions around the world have achieved high 4G coverage, but existing research has limited consideration of 4G traffic data information, resulting in less than ideal 5G traffic generation results. At the same time, such network model-based methods are highly complex, rely on specific network structures and protocols, are suitable for small devices, lack generalization, and are difficult to guide city-level communication base station deployment.

[0007] In summary, the existing technology has the problems of high complexity, insufficient controllability and insufficient generalization. Summary of the invention

[0008] The present invention provides a 5G base station deployment method and device to solve the defects of high complexity, insufficient controllability and insufficient generalization of 5G traffic generation in the prior art, and to achieve 5G traffic generation with low complexity, good controllability and good generalization, thereby guiding the deployment of communication base stations at the city level.

[0009] The present invention provides a 5G base station deployment method, comprising the following steps: Get 4G base station data in the target area; Inputting the 4G base station data into a pre-trained base station traffic condition generation model to obtain a 5G base station traffic generation result; Obtaining a base station deployment strategy according to the 5G base station traffic generation result; Among them, the base station traffic condition generation model includes a traffic association extraction module, a traffic space association feature extraction module and a condition generation module; the traffic association extraction module includes a decoder and an encoder based on an interactive attention mechanism, the traffic space association feature extraction module includes multiple graph convolution layers, and the condition generation module includes a pre-built conditional diffusion model.

[0010] According to a 5G base station deployment method provided by the present invention, obtaining 4G base station data in a target area specifically includes: Divide the target area into several sub-areas according to the coverage of the 5G base station; Get the spatial coordinates of the 4G base station and 4G historical traffic data for each sub-area; The 4G base station data of the corresponding sub-area is obtained according to the spatial coordinates and the 4G historical traffic data.

[0011] According to a 5G base station deployment method provided by the present invention, the 4G base station data is input into a pre-trained base station traffic condition generation model to obtain a 5G base station traffic generation result, which specifically includes: Inputting the 4G historical traffic data into the traffic association extraction module to extract components associated with 5G traffic in the 4G base station data to obtain traffic association; Inputting the traffic correlation and the spatial coordinates into the traffic spatial correlation feature extraction module to extract the spatial correlation of traffic between base stations, obtain traffic spatial correlation features, and further obtain generation condition information; The generation condition information and the pre-input traffic sample sequence are input into the condition generation module, and the 5G base station traffic generation result is obtained by sampling based on the denoising process of the conditional diffusion model.

[0012] According to a 5G base station deployment method provided by the present invention, the 4G historical traffic data is input into the traffic association extraction module to extract components associated with 5G traffic in the 4G base station data to obtain traffic association, specifically including: Input the 4G historical traffic data into an encoder, perform feature extraction on the 4G historical traffic data, and obtain 4G traffic features; Inputting the 4G traffic features and the 5G traffic in the denoising process into a decoder to obtain traffic correlation features output by the decoder; Among them, the 4G traffic features serve as the key and value of the decoder; and the 5G traffic in the denoising process serves as the query of the decoder.

[0013] According to a 5G base station deployment method provided by the present invention, the traffic association and the spatial coordinates are input into the traffic spatial association feature extraction module to extract the spatial association of traffic between base stations, obtain the traffic spatial association feature, and then obtain the generation condition information, which specifically includes: Taking the 4G base station as a node and the traffic association as a node feature in a graph, a degree matrix and an adjacency matrix are obtained; The edge weight is calculated according to the spatial distance information between the nodes; Based on the graph convolution layer, a graph convolution operation is performed according to the edge weight, the degree matrix and the adjacency matrix to obtain a node feature vector output by the last graph convolution layer as the node feature vector; The traffic space correlation characteristics are mapped to obtain generation condition information.

[0014] According to a 5G base station deployment method provided by the present invention, the conditional diffusion model includes:

[0015] in, Indicates 5G base stations are spreading in steps The flow sample sequence at time Represents the generation condition information added during the generation process. are the trainable parameters of the model, Represents the sample distribution of the diffusion process; Represents the sample distribution of the denoising process.

[0016] The present invention also provides a 5G base station deployment device, comprising the following modules: Data unit, used to obtain 4G base station data in the target area; A generating unit, configured to input the 4G base station data into a pre-trained base station traffic condition generating model to obtain a 5G base station traffic generating result; A strategy unit, configured to obtain a base station deployment strategy according to the 5G base station traffic generation result; Among them, the base station traffic condition generation model includes a traffic association extraction module, a traffic space association feature extraction module and a condition generation module; the traffic association extraction module includes a decoder and an encoder based on an interactive attention mechanism, the traffic space association feature extraction module includes multiple graph convolution layers, and the condition generation module includes a pre-built conditional diffusion model.

[0017] The present invention also provides an electronic device, 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, it implements any of the 5G base station deployment methods described above.

[0018] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the 5G base station deployment methods described above.

[0019] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements any of the 5G base station deployment methods described above.

[0020] The 5G base station deployment method and device provided by the present invention obtain 4G base station data in the target area; input the 4G base station data into the pre-trained base station traffic condition generation model to obtain the 5G base station traffic generation result; obtain the base station deployment strategy according to the 5G base station traffic generation result; wherein the base station traffic condition generation model includes a traffic association extraction module, a traffic space association feature extraction module and a condition generation module; the traffic association extraction module includes a decoder and an encoder based on an interactive attention mechanism, the traffic space association feature extraction module includes multiple graph convolution layers, and the condition generation module includes a pre-built conditional diffusion model. The present invention uses rich 4G historical traffic as the condition for generating the model, solves the complex network modeling problem in the traditional method, improves the generalization of the model, and is suitable for scenarios lacking historical 5G traffic data; at the same time, the complex spatiotemporal correlation between traffic is extracted by using the interactive attention mechanism and the graph convolution network, thereby obtaining the regional space and user behavior characteristics that affect the 5G traffic, and realizing the controllable generation of the traffic pattern; then the conditional diffusion model is used as the traffic generation architecture to generate 5G base station traffic that meets the actual situation, realizes stable traffic generation, and finally formulates a reasonable 5G base station deployment strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 It is one of the flow charts of the 5G base station deployment method provided by the present invention.

[0023] Figure 2 This is the second flow chart of the 5G base station deployment method provided by the present invention.

[0024] Figure 3 This is the third flow chart of the 5G base station deployment method provided by the present invention.

[0025] Figure 4 This is the fourth flow chart of the 5G base station deployment method provided by the present invention.

[0026] Figure 5 This is the fifth flow chart of the 5G base station deployment method provided by the present invention.

[0027] Figure 6 It is a schematic diagram of the traffic generation effect of the 5G base station deployment method provided by the present invention.

[0028] Figure 7 This is the regional grid division (left) and base station deployment effect (right) of an embodiment of the 5G base station deployment method provided by the present invention.

[0029] Figure 8 It is a structural schematic diagram of the 5G base station deployment device provided by the present invention.

[0030] Fig. 9 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0031] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] Combine the following Figure 1-Figure 7 Describe the 5G base station deployment method of the present invention, Figure 1 It is one of the flow charts of the 5G base station deployment method provided by the present invention, such as Figure 1 As shown, the method comprises the following steps: Step 110: Obtain 4G base station data in the target area.

[0033] It should be noted that the goal of the present invention is to use the conditional diffusion model to extract the regional space and user behavior characteristics that have an impact on 5G traffic through the rich historical 4G traffic data in the region, so as to generate 5G base station traffic that conforms to reality and formulate a reasonable 5G base station deployment strategy.

[0034] The target area refers to the area where 5G base stations need to be deployed. The target area includes the coverage of at least one 5G base station. The 4G base station data includes the relevant data of the 4G base stations in the target area, including the 4G historical traffic data of the 4G base stations and the spatial coordinates of the base stations. It should be emphasized that each pair of 4G base station data corresponds to the coverage of a 5G base station.

[0035] Furthermore, during the implementation process, when the target area includes the coverage areas of multiple 5G base stations, 4G base station data is extracted for the coverage areas of each 5G base station respectively.

[0036] Step 120: Input the 4G base station data into a pre-trained base station traffic condition generation model to obtain a 5G base station traffic generation result.

[0037] The 4G base station data obtained in step 110 is input into the base station traffic condition generation model. According to the spatiotemporal distribution of the existing 4G traffic in the area, the association between it and the 5G traffic is extracted, and the spatial position distribution characteristics between different base stations are captured, so as to obtain condition information reflecting the spatial and user behavior characteristics, and based on this, the 5G network traffic of the area is generated, such as Figure 2 shown.

[0038] Among them, the base station traffic condition generation model includes a traffic association extraction module, a traffic space association feature extraction module and a condition generation module; the traffic association extraction module includes a decoder and an encoder based on an interactive attention mechanism, the traffic space association feature extraction module includes multiple graph convolution layers, and the condition generation module includes a pre-built conditional diffusion model.

[0039] Specifically, during the application process, the traffic correlation extraction module is used to extract the components associated with 5G traffic in 4G traffic. ,Then, according to the spatial coordinates, the spatial correlation of traffic between base stations is extracted through the traffic spatial correlation feature extraction module ,Finally, the generation condition information representing the regional space and user behavior characteristics is obtained through mapping Afterwards, based on the generated condition information, the 5G base station traffic generation result is obtained through sampling of the conditional diffusion model.

[0040] Step 130: Obtain a base station deployment strategy based on the 5G base station traffic generation result.

[0041] In step 130, the 5G base station traffic generation result is used to guide the deployment of the 5G base station and obtain the base station deployment strategy.

[0042] It should be noted that the number and distribution of 5G base stations in a region should match the 5G traffic demand to meet the peak usage during high-demand periods. Traditional site deployment strategies mainly use population distribution as a proxy for 5G traffic demand or 4G traffic as a reference. However, user behavior characteristics vary in different regions, and population distribution may not accurately reflect traffic demand. In addition, 4G base stations and 5G base stations differ in coverage, power, bandwidth, etc., resulting in differences in detailed traffic fluctuation patterns. Therefore, using traffic generation methods to guide 5G base station deployment can be a better solution to better align 5G base station deployment with 5G traffic demand.

[0043] The present invention uses rich 4G historical traffic as the condition for generating the model, which solves the complex network modeling problem in the traditional method, improves the generalization of the model, and is suitable for scenarios where historical 5G traffic data is lacking. At the same time, the complex spatiotemporal correlations between traffic are extracted using the interactive attention mechanism and graph convolutional network. The correlation features can fully characterize the spatial and user behavior characteristics in the region, thereby realizing the controllable generation of traffic patterns. Finally, the conditional diffusion model is used as the traffic generation architecture to achieve stable traffic generation.

[0044] The following further describes step 110 in the case where the target area includes coverage of multiple 5G base stations. In some embodiments, obtaining 4G base station data of the target area specifically includes: Divide the target area into several sub-areas according to the coverage of the 5G base station; Get the spatial coordinates of the 4G base station and 4G historical traffic data for each sub-area; The 4G base station data of the corresponding sub-area is obtained according to the spatial coordinates and the 4G historical traffic data.

[0045] Specifically, the target area is divided into several sub-areas according to the coverage of 5G base stations. . One of the sub-regions contains 4G base stations, whose 4G historical traffic data is represented as a set , each 4G base station has spatial coordinates .

[0046] The following further describes step 120. In some embodiments, the 4G base station data is input into a pre-trained base station traffic condition generation model to obtain a 5G base station traffic generation result, specifically including: Inputting the 4G historical traffic data into the traffic association extraction module to extract components associated with 5G traffic in the 4G base station data to obtain traffic association; Inputting the traffic correlation and the spatial coordinates into the traffic spatial correlation feature extraction module to extract the spatial correlation of traffic between base stations, obtain traffic spatial correlation features, and further obtain generation condition information; The generation condition information and the pre-input traffic sample sequence are input into the condition generation module, and the 5G base station traffic generation result is obtained by sampling based on the denoising process of the conditional diffusion model.

[0047] Specifically, Figure 3 As shown in FIG. 1 , the present invention captures the correlation components between 4G and 5G traffic through a cross attention mechanism (Co-Atten), such as Figure 3(a); Use Graph Convolutional Network (GCN) to process the relationship between base station traffic and spatial distribution, such as Figure 3 (b) Through the above extraction, we can obtain the representation of space and user behavior characteristics (generate condition information) ,in and They represent the 4G historical traffic data and the spatial coordinates of the 4G base station in the corresponding sub-area respectively; The representative feature extraction algorithm includes two parts (a) and (b). Then, the model is generated by designing conditions. (correspond Figure 3 (c) in the above), generated under the condition 5G traffic in the following areas: .

[0048] The following further describes capturing the correlation components between 4G and 5G traffic through the interactive attention mechanism. In some embodiments, the 4G historical traffic data is input into the traffic correlation extraction module to extract the components associated with 5G traffic in the 4G base station data to obtain traffic correlation, specifically including: Inputting the 4G historical traffic data into an encoder, performing feature extraction on the 4G historical traffic data, and obtaining traffic correlation features; Inputting the traffic correlation feature and the 5G traffic in the denoising process into a decoder to obtain the traffic correlation output by the decoder; Among them, the traffic association features serve as the keys and values ​​of the decoder; and the 5G traffic in the denoising process serves as the query of the decoder.

[0049] Specifically, Figure 4 As shown in the figure, in the process of using the interactive attention mechanism to extract the correlation patterns between 4G and 5G traffic, the Co-Atten module (traffic correlation extraction module) has three inputs: query, key, and value, where the key and value usually have the same amount of data.

[0050] It should be noted that the principle of Co-Atten is to first calculate the similarity score between the query and the key, that is, the attention weight vector 𝑅, and then multiply the vector of values ​​by 𝑅. Through weighting, the result can reflect the correlation between the query and the key.

[0051] In the actual extraction process, for any area in the city, for each step in the denoising process, the 5G traffic in the denoising process, that is, the 5G traffic sample sequence restored at the current step, is used as the query value, and the traffic of all 4G base stations is used as the key value. More specifically, the Transformer encoder and decoder are used to implement it. The encoder uses the self-attention mechanism to extract features from the input 4G traffic, and the output is used as the Key and Value input of the decoder. The 5G traffic in the denoising process is input to the Query end of the decoder. The output of the decoder can be interpreted as the component of the 4G traffic features associated with the 5G traffic. Figure 4 The structure of the traffic correlation extraction module is shown. By calculating the similarity score and weighting the value, the components of 4G traffic related to 5G traffic can be captured. For area number i, 4G base station number j, attention weight The calculation is as follows:

[0052] It can be understood that when 4G and 5G traffic have higher similarity, the corresponding attention weights are also correspondingly higher, allowing the model to focus more on the relevant information between the two.

[0053] The following further describes the use of graph convolutional networks to process the association between base station traffic and spatial distribution. In some embodiments, the traffic association and the spatial coordinates are input into the traffic spatial association feature extraction module to extract the spatial association of traffic between base stations, obtain traffic spatial association features, and then obtain generation condition information, specifically including: Taking the 4G base station as a node and the traffic association as a node feature in a graph, a degree matrix and an adjacency matrix are obtained; The edge weight is calculated according to the spatial distance information between the nodes; Based on the graph convolution layer, a graph convolution operation is performed according to the edge weight, the degree matrix and the adjacency matrix to obtain a node feature vector output by the last graph convolution layer as the node feature vector; The traffic space correlation characteristics are mapped to obtain generation condition information.

[0054] Specifically, it can be understood that the graph convolutional network (GCN) is a type of deep learning model for processing graph-structured data. Its design is inspired by the convolutional neural network (CNN) and is improved to adapt to graph-structured data. GCN uses a graph to represent data, where nodes represent base stations, edges represent the relationship between nodes, and the entire graph can be mathematically represented by a degree matrix and an adjacency matrix. The core of GCN is the graph convolution operation, which allows the model to update node features while processing node neighbor information.

[0055] Assume that the degree matrix of a graph is D, the adjacency matrix is ​​A, the edge weight is W, and the number of layers is L. Any graph convolutional layer can be written as a nonlinear function:

[0056] is the input of the first layer, , N is the number of nodes in the graph, D is the dimension of each node feature vector, and A is the adjacency matrix. This embodiment first uses the spatial distance information between nodes to calculate the weight of the edge, which is defined as:

[0057] The traffic correlation output obtained by the traffic correlation extraction module contains the part of each 4G base station traffic that is associated with the 5G traffic pattern, and this output is used as the node feature in the graph.

[0058] On this basis, the nonlinear function of the convolutional layer GCNconv can be specifically expressed as:

[0059] in is the activation function. Through the graph convolution operation, the adjacency information of the nodes in the graph can be aggregated, thereby obtaining the node feature output containing the node spatial association. The traffic spatial association feature extraction module consists of multiple layers of graph convolution layers. Figure 5 The structure of the traffic spatial correlation feature extraction module in one embodiment is shown, which includes two graph convolution layers connected by activation function ReLu layer and random dropout layer. The traffic spatial correlation feature extraction module extracts the spatial correlation between 4G base stations and outputs the node feature vector , including the spatial relationship of 4G base stations. After mapping, the spatial latent features containing the spatial association of 4G base stations in the cell are obtained. As the generation condition information, where r is the mapping function.

[0060] Based on the above embodiment, the following Markov chain describes the process of generating traffic using the conditional diffusion model in the conditional generation module:

[0061] in, Indicates 5G base stations are spreading in steps The flow sample sequence at time Represents the conditional information added during the generation process, that is, the generation condition information. are trainable parameters of the model. represents the sample distribution of the diffusion process; Represents the sample distribution of the denoising process. Through the reverse process of the conditional diffusion model, the relative 5G traffic can be obtained. The conditional distribution of 5G traffic can be sampled and the traffic generation effect diagram is as follows: Figure 6 shown.

[0062] Furthermore, in order to further illustrate the 5G base station deployment method provided by the present invention, a specific embodiment of 5G base station deployment is given. Figure 7 As shown in Figure 1, this embodiment conducts 5G base station deployment experiments in Beijing and Nanchang. The goal is to deploy a certain number (100 in the experiment) of 5G base stations in the city to maximize the coverage of the demand. First, based on the typical coverage range of 5G base stations, the target area of ​​the city is selected for sub-area division, that is, gridding. The grid is divided into Figure 7 (Left), where the grid spacing , to achieve full coverage of urban areas. Then, three methods are used as references: population distribution, 4G traffic, and generated 5G traffic to determine how many 5G base stations need to be built in each grid. The following is a brief introduction to this method.

[0063] Table 1 shows the comparison of different deployment strategies with the actual site distribution of 5G base stations in Beijing and Nanchang datasets. MSE stands for mean square error, and COS stands for cosine similarity. ↓ means the lower the better, ↑ means the larger the better. Bold numbers represent the best results.

[0064] Table 1 Comparison of different deployment strategies and the actual site distribution of 5G base stations in Beijing and Nanchang datasets

[0065] The method provided by the present invention is used to generate 5G traffic in each grid. The circles in the right figure correspond to the grids in the left figure. The darker the red, the more 5G base stations need to be deployed.

[0066] Due to the diffusion process and neural network training requirements, the traffic should be normalized. Therefore, when obtaining the generated samples for downstream applications (such as base station deployment), the traffic is first denormalized using the mean and standard deviation of the 4G traffic. Then, in each grid, the number of base stations that need to be deployed is determined based on the maximum hourly traffic volume (a total of 100 in this embodiment). The distribution of the number of deployments in Beijing is as follows: Figure 7 (right) As shown. The base station deployment based on 4G traffic and population distribution is performed at the same time, and the comparison with the actual 5G base station distribution is shown in Table 1. The experiment shows that the deployment of 5G base stations based on the 5G traffic generated by this model can best approximate the actual site distribution. This model has broad application prospects in the construction of 5G networks in new fields.

[0067] The 5G base station deployment method provided by the present invention obtains 4G base station data in the target area; inputs the 4G base station data into a pre-trained base station traffic condition generation model to obtain a 5G base station traffic generation result; obtains a base station deployment strategy according to the 5G base station traffic generation result; wherein the base station traffic condition generation model includes a traffic association extraction module, a traffic space association feature extraction module and a condition generation module; the traffic association extraction module includes a decoder and an encoder based on an interactive attention mechanism, the traffic space association feature extraction module includes multiple graph convolution layers, and the condition generation module includes a pre-built conditional diffusion model. The present invention uses rich 4G historical traffic as a condition for generating a model, solves the complex network modeling problem in traditional methods, improves the generalization of the model, and is suitable for scenarios lacking historical 5G traffic data; at the same time, the complex spatiotemporal correlation between traffic is extracted using the interactive attention mechanism and the graph convolution network, thereby obtaining the regional space and user behavior characteristics that affect 5G traffic, and realizing the controllable generation of traffic patterns; then the conditional diffusion model is used as a traffic generation architecture to generate 5G base station traffic that meets the actual situation, achieving stable traffic generation, and finally formulating a reasonable 5G base station deployment strategy.

[0068] The following describes the 5G base station deployment device provided by the present invention. The 5G base station deployment device described below and the 5G base station deployment method described above can be referred to in correspondence with each other. Figure 8 As shown, the device includes the following modules: Data unit 810, used to obtain 4G base station data of the target area; A generating unit 820 is used to input the 4G base station data into a pre-trained base station traffic condition generation model to obtain a 5G base station traffic generation result; A strategy unit 830 is configured to obtain a base station deployment strategy according to the 5G base station traffic generation result; Among them, the base station traffic condition generation model includes a traffic association extraction module, a traffic space association feature extraction module and a condition generation module; the traffic association extraction module includes a decoder and an encoder based on an interactive attention mechanism, the traffic space association feature extraction module includes multiple graph convolution layers, and the condition generation module includes a pre-built conditional diffusion model.

[0069] According to a 5G base station deployment device provided by the present invention, obtaining 4G base station data in a target area specifically includes: Divide the target area into several sub-areas according to the coverage of the 5G base station; Get the spatial coordinates of the 4G base station and 4G historical traffic data for each sub-area; The 4G base station data of the corresponding sub-area is obtained according to the spatial coordinates and the 4G historical traffic data.

[0070] According to a 5G base station deployment device provided by the present invention, the 4G base station data is input into a pre-trained base station traffic condition generation model to obtain a 5G base station traffic generation result, which specifically includes: Inputting the 4G historical traffic data into the traffic association extraction module to extract components associated with 5G traffic in the 4G base station data to obtain traffic association; Inputting the traffic correlation and the spatial coordinates into the traffic spatial correlation feature extraction module to extract the spatial correlation of traffic between base stations, obtain traffic spatial correlation features, and further obtain generation condition information; The generation condition information and the pre-input traffic sample sequence are input into the condition generation module, and the 5G base station traffic generation result is obtained by sampling based on the denoising process of the conditional diffusion model.

[0071] According to a 5G base station deployment device provided by the present invention, the 4G historical traffic data is input into the traffic association extraction module to extract components associated with 5G traffic in the 4G base station data to obtain traffic association, specifically including: Input the 4G historical traffic data into an encoder, perform feature extraction on the 4G historical traffic data, and obtain 4G traffic features; Inputting the 4G traffic features and the 5G traffic in the denoising process into a decoder to obtain traffic correlation features output by the decoder; Among them, the 4G traffic features serve as the key and value of the decoder; and the 5G traffic in the denoising process serves as the query of the decoder.

[0072] According to a 5G base station deployment device provided by the present invention, the traffic association and the spatial coordinates are input into the traffic spatial association feature extraction module to extract the spatial association of traffic between base stations, obtain traffic spatial association features, and then obtain generation condition information, specifically including: Taking the 4G base station as a node and the traffic association as a node feature in a graph, a degree matrix and an adjacency matrix are obtained; The edge weight is calculated according to the spatial distance information between the nodes; Based on the graph convolution layer, a graph convolution operation is performed according to the edge weight, the degree matrix and the adjacency matrix to obtain a node feature vector output by the last graph convolution layer as the node feature vector; The traffic space correlation characteristics are mapped to obtain generation condition information.

[0073] According to a 5G base station deployment device provided by the present invention, the conditional diffusion model includes:

[0074] in, Indicates 5G base stations are spreading in steps The flow sample sequence at time Represents the generation condition information added during the generation process. are the trainable parameters of the model, represents the sample distribution of the diffusion process; Represents the sample distribution of the denoising process.

[0075] The 5G base station deployment device provided by the present invention obtains 4G base station data in the target area; inputs the 4G base station data into a pre-trained base station traffic condition generation model to obtain a 5G base station traffic generation result; obtains a base station deployment strategy according to the 5G base station traffic generation result; wherein the base station traffic condition generation model includes a traffic association extraction module, a traffic space association feature extraction module and a condition generation module; the traffic association extraction module includes a decoder and an encoder based on an interactive attention mechanism, the traffic space association feature extraction module includes multiple graph convolution layers, and the condition generation module includes a pre-built conditional diffusion model. The present invention uses rich 4G historical traffic as a condition for generating a model, solves the complex network modeling problem in traditional methods, improves the generalization of the model, and is suitable for scenarios lacking historical 5G traffic data; at the same time, the complex spatiotemporal correlation between traffic is extracted by using the interactive attention mechanism and the graph convolution network, thereby obtaining the regional space and user behavior characteristics that affect 5G traffic, and realizing the controllable generation of traffic patterns; then the conditional diffusion model is used as a traffic generation architecture to generate 5G base station traffic that meets the actual situation, realizing stable traffic generation, and finally formulating a reasonable 5G base station deployment strategy.

[0076] Fig. 9 An example of a physical structure diagram of an electronic device is shown in FIG. Fig. 9As shown, the electronic device may include: a processor 910, a communication interface 920, a memory 930 and a communication bus 940, wherein the processor 910, the communication interface 920 and the memory 930 communicate with each other through the communication bus 940. The processor 910 may call the logic instructions in the memory 930 to execute the 5G base station deployment method, which includes: obtaining 4G base station data of the target area; inputting the 4G base station data into a pre-trained base station traffic condition generation model to obtain a 5G base station traffic generation result; obtaining a base station deployment strategy according to the 5G base station traffic generation result; wherein the base station traffic condition generation model includes a traffic association extraction module, a traffic space association feature extraction module and a condition generation module; the traffic association extraction module includes a decoder and an encoder based on an interactive attention mechanism, the traffic space association feature extraction module includes multiple graph convolution layers, and the condition generation module includes a pre-built conditional diffusion model.

[0077] In addition, the logic instructions in the above-mentioned memory 930 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0078] On the other hand, the present invention also provides a computer program product, which includes a computer program, and the computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the 5G base station deployment method provided by the above methods, and the method includes: obtaining 4G base station data of the target area; inputting the 4G base station data into a pre-trained base station traffic condition generation model to obtain a 5G base station traffic generation result; obtaining a base station deployment strategy based on the 5G base station traffic generation result; wherein the base station traffic condition generation model includes a traffic association extraction module, a traffic space association feature extraction module and a condition generation module; the traffic association extraction module includes a decoder and an encoder based on an interactive attention mechanism, the traffic space association feature extraction module includes multiple graph convolution layers, and the condition generation module includes a pre-built conditional diffusion model.

[0079] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the 5G base station deployment method provided by the above methods, the method comprising: obtaining 4G base station data of the target area; inputting the 4G base station data into a pre-trained base station traffic condition generation model to obtain a 5G base station traffic generation result; obtaining a base station deployment strategy based on the 5G base station traffic generation result; wherein the base station traffic condition generation model comprises a traffic association extraction module, a traffic space association feature extraction module and a condition generation module; the traffic association extraction module comprises a decoder and an encoder based on an interactive attention mechanism, the traffic space association feature extraction module comprises multiple graph convolution layers, and the condition generation module comprises a pre-built conditional diffusion model.

[0080] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0081] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0082] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention 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 invention.

Claims

1. A 5G base station deployment method, characterized in that: include: Get 4G base station data in the target area; Inputting the 4G base station data into a pre-trained base station traffic condition generation model to obtain a 5G base station traffic generation result; Obtaining a base station deployment strategy according to the 5G base station traffic generation result; The base station traffic condition generation model includes a traffic correlation extraction module, a traffic space correlation feature extraction module and a condition generation module; The traffic association extraction module includes a decoder and an encoder based on an interactive attention mechanism, the traffic space association feature extraction module includes multiple graph convolution layers, and the conditional generation module includes a pre-built conditional diffusion model.

2. The 5G base station deployment method according to claim 1, characterized in that: Get 4G base station data in the target area, including: Divide the target area into several sub-areas according to the coverage of the 5G base station; Get the spatial coordinates of the 4G base station and 4G historical traffic data for each sub-area; The 4G base station data of the corresponding sub-area is obtained according to the spatial coordinates and the 4G historical traffic data.

3. The 5G base station deployment method according to claim 1 or 2, characterized in that: The 4G base station data is input into a pre-trained base station traffic condition generation model to obtain a 5G base station traffic generation result, specifically including: Inputting the 4G historical traffic data into the traffic association extraction module to extract components associated with 5G traffic in the 4G base station data to obtain traffic association; Inputting the traffic correlation and the spatial coordinates into the traffic spatial correlation feature extraction module to extract the spatial correlation of traffic between base stations, obtain traffic spatial correlation features, and further obtain generation condition information; The generation condition information and the pre-input traffic sample sequence are input into the condition generation module, and the 5G base station traffic generation result is obtained by sampling based on the denoising process of the conditional diffusion model.

4. The 5G base station deployment method according to claim 3 is characterized in that: Inputting the 4G historical traffic data into the traffic association extraction module to extract the components associated with the 5G traffic in the 4G base station data to obtain traffic association, specifically including: Inputting the 4G historical traffic data into an encoder, performing feature extraction on the 4G historical traffic data, and obtaining traffic correlation features; Inputting the traffic correlation feature and the 5G traffic in the denoising process into a decoder to obtain the traffic correlation output by the decoder; The traffic-related features serve as the key and value of the decoder; and the 5G traffic in the denoising process serves as the query of the decoder.

5. The 5G base station deployment method according to claim 3 is characterized in that: The traffic correlation and the spatial coordinates are input into the traffic spatial correlation feature extraction module to extract the spatial correlation of traffic between base stations, obtain traffic spatial correlation features, and then obtain generation condition information, specifically including: Taking the 4G base station as a node and the traffic association as a node feature in a graph, a degree matrix and an adjacency matrix are obtained; The edge weight is calculated according to the spatial distance information between the nodes; Based on the graph convolution layer, a graph convolution operation is performed according to the edge weight, the degree matrix and the adjacency matrix to obtain a node feature vector output by the last graph convolution layer as the node feature vector; The traffic space correlation characteristics are mapped to obtain generation condition information.

6. The 5G base station deployment method according to claim 1, characterized in that: The conditional diffusion model includes: ; in, Indicates 5G base stations are spreading in steps The flow sample sequence at time Represents the generation condition information added during the generation process. are the trainable parameters of the model, Represents the sample distribution of the diffusion process; Represents the sample distribution of the denoising process.

7. A 5G base station deployment device, characterized in that: include: Data unit, used to obtain 4G base station data in the target area; A generating unit, configured to input the 4G base station data into a pre-trained base station traffic condition generating model to obtain a 5G base station traffic generating result; A strategy unit, configured to obtain a base station deployment strategy according to the 5G base station traffic generation result; The base station traffic condition generation model includes a traffic correlation extraction module, a traffic space correlation feature extraction module and a condition generation module; The traffic association extraction module includes a decoder and an encoder based on an interactive attention mechanism, the traffic space association feature extraction module includes multiple graph convolution layers, and the conditional generation module includes a pre-built conditional diffusion model.

8. An electronic device 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, it implements the 5G base station deployment method as described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it implements the 5G base station deployment method as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, it implements the 5G base station deployment method as described in any one of claims 1 to 6.

Images