Traffic load balancing method and apparatus, electronic device, and storage medium

Abstract

The application provides a traffic load balancing method and device, electronic equipment and storage medium, comprising: obtaining traffic data of a base station in a preset period, performing feature analysis on the traffic data by using a pre-trained network traffic prediction model to obtain a space-time feature representation of the traffic data, obtaining predicted traffic data of the base station at a next moment according to the space-time feature representation, calculating pheromone between the base station and adjacent base stations by using actual traffic data, predicted traffic data and an adjacency matrix, calculating heuristic information between the base station and adjacent base stations by using residual traffic capacity and the adjacency matrix, calculating path selection probability of traffic transmission between the base station and adjacent base stations according to the pheromone and the heuristic information, determining a target transmission path, and performing traffic distribution on the base station on the target transmission path. The application ensures high performance under high traffic load by network traffic prediction, avoids network port overload by load balancing, and guarantees network stability and efficiency.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of communication, and particularly relates to a traffic load balancing method, a traffic load balancing device, an electronic device and a computer readable storage medium. BACKGROUND

[0002] Under the background of rapid development of today's communication technology, the growth of network scale and the massive access demand of terminal users lead to the exponential growth of communication network traffic. In this case, through the measurement and monitoring of network traffic, the network and server can maintain high performance under high traffic load, while avoiding network port overload, which becomes the key to improving network performance and user experience.

[0003] At present, communication network traffic prediction uses real-time analysis and deep learning model to predict future traffic trends, so as to optimize network load distribution. Since network traffic has the characteristics of self-similarity, periodicity and chaos, and is affected by the spatial node topology structure, the network traffic is essentially a nonlinear spatiotemporal sequence data. However, the traditional network traffic prediction is mainly based on time series analysis, ignoring the spatiotemporal characteristics of network traffic, which affects the accuracy of traffic prediction results, and it is difficult to cope with load fluctuations in complex network environment, and network load distribution is uneven, which further affects network performance. SUMMARY

[0004] Therefore, the present application aims to provide a traffic load balancing method, device, electronic device and storage medium, which solves the problem that the current network traffic prediction ignores the spatiotemporal characteristics of network traffic, the traffic prediction result is inaccurate, the network load distribution is uneven, and the network performance is affected.

[0005] According to a first aspect of the present application, a traffic load balancing method is provided, the method comprising:

[0006] obtaining traffic data of a base station in a preset period;

[0007] performing feature analysis on the traffic data by using a pre-trained network traffic prediction model to obtain a spatiotemporal feature representation of the traffic data, and obtaining predicted traffic data of the next moment of the base station according to the spatiotemporal feature representation;

[0008] obtaining an adjacency matrix of the base station, actual traffic data of the current moment and residual traffic capacity;

[0009] calculating pheromone between the base station and adjacent base stations by using the actual traffic data, the predicted traffic data and the adjacency matrix;

[0010] calculating heuristic information between the base station and adjacent base stations by using the residual traffic capacity and the adjacency matrix;

[0011] calculating a path selection probability of traffic transmission between the base station and the adjacent base station according to the pheromone and the heuristic information, determining a target transmission path;

[0012] allocating traffic to the base stations on the target transmission path.

[0013] Optionally, the feature analysis of the traffic data by using the pre-trained network traffic prediction model obtains a spatio-temporal feature representation of the traffic data, and the predicted traffic data of the base station at the next moment is obtained according to the spatio-temporal feature representation, including:

[0014] The feature analysis of the traffic data by using the pre-trained network traffic prediction model obtains a time sequence feature representation and a space feature representation of the traffic data;

[0015] The time sequence feature representation and the space feature representation are fused to obtain a spatio-temporal feature representation of the traffic data;

[0016] The spatio-temporal feature representation is mapped to obtain the predicted traffic data of the base station at the next moment.

[0017] Optionally, the feature analysis of the traffic data by using the pre-trained network traffic prediction model obtains a time sequence feature representation and a space feature representation of the traffic data, including:

[0018] The traffic data is convoluted with a preset time embedding matrix to obtain the time sequence feature representation of the traffic data;

[0019] The static adaptive graph learning and the dynamic adaptive graph learning are performed on the traffic data to obtain a static space relationship and a dynamic space relationship;

[0020] The static space relationship and the dynamic space relationship are fused to obtain the space feature representation of the traffic data.

[0021] Optionally, the calculation of the pheromone between the base station and the adjacent base station by using the actual traffic data, the predicted traffic data and the adjacency matrix includes:

[0022] The traffic difference is calculated according to the actual traffic data and the predicted traffic data, and the adjacent base stations of the base station are determined according to the adjacency matrix of the base station;

[0023] The pheromone increment of the adjacent base station is calculated by using the traffic difference and a preset increment parameter;

[0024] The current pheromone of the adjacent base station is adjusted by using the pheromone increment to obtain the pheromone between the base station and the adjacent base station.

[0025] Optionally, the remaining traffic capacity and the adjacency matrix are used to calculate heuristic information between the base station and the adjacent base station, including:

[0026] The remaining traffic capacity of the base station is calculated with the adjacency matrix, and the non-adjacent matrix in the adjacency matrix is adjusted to zero to obtain the heuristic information between the base station and the adjacent base station.

[0027] Optionally, the pheromone and the heuristic information are used to calculate the path selection probability of traffic transmission between the base station and the adjacent base station, and the target transmission path is determined, including:

[0028] The pheromone and the heuristic information are used to calculate the selection probability of each path connected to the base station to obtain the path selection probability of traffic transmission between the base station and the adjacent base station;

[0029] According to the path selection probability, the target transmission path is selected from each path connected to the base station.

[0030] Optionally, after the traffic is allocated to the base station on the target transmission path, the method further includes:

[0031] The target transmission path is stored in an initial solution space matrix;

[0032] The traffic allocation of the base station is monitored, and the remaining traffic capacity of the target transmission path is updated;

[0033] The resource utilization rate of the target transmission path and the variance of the resource utilization rate are calculated using the remaining traffic capacity;

[0034] The target transmission path with a variance less than a preset threshold is determined as an optimal transmission path.

[0035] According to a second aspect of the present application, a traffic load balancing device is provided, and the device includes:

[0036] A first acquisition module is configured to acquire traffic data of a base station in a preset period;

[0037] A traffic prediction module is configured to analyze the traffic data using a pre-trained network traffic prediction model to obtain a spatio-temporal feature representation of the traffic data, and to obtain predicted traffic data of the base station at a next time according to the spatio-temporal feature representation;

[0038] A second acquisition module is configured to acquire an adjacency matrix of the base station, actual traffic data at a current time, and remaining traffic capacity;

[0039] a pheromone calculation module configured to calculate pheromones between base stations and adjacent base stations by using the actual traffic data, the predicted traffic data and the adjacency matrix;

[0040] a heuristic information calculation module configured to calculate heuristic information between base stations and adjacent base stations by using the residual traffic capacity and the adjacency matrix;

[0041] a path determination module configured to determine a target transmission path according to a path selection probability of traffic transmission between base stations and adjacent base stations calculated by the pheromones and the heuristic information;

[0042] a traffic allocation module configured to allocate traffic to base stations on the target transmission path.

[0043] According to still another aspect of the present application, there is also provided an electronic device comprising:

[0044] a processor;

[0045] a memory for storing instructions executable by the processor;

[0046] wherein the processor is configured to execute the instructions to implement the traffic load balancing method as described above.

[0047] According to still another aspect of the present application, there is also provided a readable storage medium having stored thereon a computer program, the computer program being executable by a processor to implement the steps of the traffic load balancing method as described above.

[0048] The traffic load balancing method provided by the embodiment of the application comprises the following steps: obtaining traffic data of a base station in a preset period, performing feature analysis on the traffic data by using a pre-trained network traffic prediction model to obtain a space-time feature representation of the traffic data, obtaining predicted traffic data of the base station at a next moment according to the space-time feature representation, obtaining an adjacency matrix of the base station, actual traffic data at a current moment and a remaining traffic capacity, calculating pheromone between the base station and adjacent base stations by using the actual traffic data, the predicted traffic data and the adjacency matrix, calculating heuristic information between the base station and the adjacent base stations by using the remaining traffic capacity and the adjacency matrix, calculating path selection probability of traffic transmission between the base station and the adjacent base stations according to the pheromone and the heuristic information, determining a target transmission path, and performing traffic distribution on the base stations on the target transmission path. The embodiment of the application realizes space-time characteristic fusion of traffic data by using a pre-trained network traffic prediction model, realizes traffic prediction conforming to space-time characteristics of network traffic, improves the accuracy of service traffic prediction in a complex and changeable cellular network environment, provides reliable data support for load balancing, calculates pheromone and heuristic information based on a traffic prediction result, obtains a change trend of current and future loads, dynamically adjusts load distribution between base stations, more flexibly allocates network resources to base stations with expected high loads to cope with future traffic fluctuations, ensures high performance under high traffic load by predicting and monitoring network traffic, and avoids network port overload by using load balancing, thereby ensuring stability and efficiency of the network.

[0049] The above description is only a summary of the technical scheme of the application. In order to more clearly understand the technical means of the application, the application can be implemented according to the content of the specification, and in order to make the above and other purposes, characteristics and advantages of the application more obvious and easy to understand, the following specific embodiments of the application are described. BRIEF DESCRIPTION OF DRAWINGS

[0050] Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The accompanying drawings are intended to further assist in understanding the preferred embodiments, and are not considered limiting of the application. Moreover, like reference numerals denote like parts throughout the several views in the drawings. In the drawings:

[0051] Figure 1 is a step flowchart of a traffic load balancing method provided by an embodiment of the application;

[0052] Figure 2 is Figure 1 is a flowchart of step 102 in the traffic load balancing method provided by an embodiment of the application in

[0053] Figure 3 is Figure 1 is a flowchart of step 104 in the traffic load balancing method provided by an embodiment of the application in

[0054] Figure 4 is Figure 1 The flow chart of step 106 in the traffic load balancing method provided by the embodiment of the present application is shown in

[0055] Figure 5 The flow chart of step of another traffic load balancing method provided by the embodiment of the present application is shown in

[0056] Figure 6 The scenario schematic diagram of the traffic load balancing method provided by the embodiment of the present application is shown in

[0057] Figure 7 The structural schematic diagram of a traffic load balancing device provided by the embodiment of the present application is shown in

[0058] Figure 8 The structural schematic diagram of an electronic device provided by the embodiment of the present application is shown in DETAILED DESCRIPTION

[0059] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the embodiments of the present application will be described in detail below with reference to the drawings. However, those skilled in the art can understand that, in the embodiments of the present application, many technical details are proposed in order to make the readers better understand the present application. However, the technical solutions claimed by the present application can be realized even without these technical details and based on various changes and modifications of the following embodiments. The division of the following embodiments is for the convenience of description, and should not constitute any limitation on the specific implementation of the present application, and the embodiments can be combined and referenced with each other on the premise of no contradiction.

[0060] Referring to Figure 1 , the flow chart of step of the traffic load balancing method provided by the embodiment of the present application is shown, which can include:

[0061] Step 101, obtaining traffic data of a base station preset period.

[0062] In the field of network traffic management, traffic prediction is a key task in the embodiment of the present application, which aims to accurately predict the network traffic in the future time period. However, with the increase of network size and application complexity, the traditional network prediction model has been difficult to meet the demand of traffic prediction in modern network. The embodiment of the present application proposes a traffic prediction model based on adaptive hybrid spatio-temporal graph neural network for cellular network traffic prediction problem, which dynamically considers time information by inputting time embedding matrix, realizes spatio-temporal characteristics fusion, and improves the accuracy and real-time performance of traffic prediction. Secondly, based on the traffic prediction result, the network resource allocation is dynamically adjusted to cope with the future traffic peak and load imbalance, so as to improve the network performance and resource utilization efficiency.

[0063] It should be noted that the embodiment takes the cellular network as an example for illustration, the cellular network provides extensive coverage by dividing a geographical area into a plurality of small "cellular" areas (usually referred to as "cells"), each cell is managed by a base station or called "base transceiver station", the embodiment adopts an improved adaptive hybrid spatial-temporal graph neural network (AHSTGNN), which comprehensively considers the spatial correlation between cellular towers, nonlinear time dependence and heterogeneity through hybrid graph learning, time convolution and spatial-temporal adaptive module, and better network traffic prediction effect is achieved.

[0064] Specifically, the traffic data of the base station in a preset period is obtained, the traffic data includes three periodic spatio-temporal data, which are the recent, daily and weekly traffic of the cellular tower base station respectively, each periodic traffic data has a different length, which is T R , T D , T W , X∈R T×N×F , wherein N represents the number of cellular towers, and F represents the number of features.

[0065] In step 102, a pre-trained network traffic prediction model is used to analyze the features of the traffic data, to obtain the spatio-temporal feature representation of the traffic data, and to obtain the predicted traffic data of the next moment of the base station according to the spatio-temporal feature representation.

[0066] The embodiment of the application adopts a pre-trained network traffic prediction model to analyze the features of the traffic data, to obtain the spatio-temporal feature representation of the traffic data, wherein the pre-trained network traffic prediction model is an improved adaptive hybrid spatial-temporal graph neural network model, the pre-trained network traffic prediction model includes an input layer, an adaptive hybrid spatio-temporal learning module and an output layer, the obtained traffic data of the base station in a preset period is input to the input layer of the model, the adaptive hybrid spatio-temporal learning module is used to analyze the features of the traffic data, to obtain the spatio-temporal feature representation of the traffic data, the spatio-temporal feature representation obtained by the adaptive hybrid spatio-temporal learning module is connected to the output layer, and the predicted traffic data of the next moment of the base station is output.

[0067] It should be noted that the improved adaptive hybrid graph neural network model is used for traffic prediction in the embodiment, and the convolution operation of GCN is specifically used to capture the spatial relationship between base stations in combination with the time embedding matrix to fully consider the space-time interaction. Specifically, the time embedding matrix is placed in the traffic data, so that the model can capture the relationship in the time dimension through learning and updating. In this case, the weight of the time embedding matrix will be trained as part of the model, so that it can adapt to the time-related patterns in the task. The time information is dynamically considered through the input time embedding matrix, which can effectively integrate spatial and temporal information. Through the combination of the graph convolution network and the sequence model, the modeling of complex space-time dependence is realized, thereby improving the prediction accuracy and robustness, and providing a more reliable reference basis for network planning and resource allocation.

[0068] Specifically, referring to Figure 6 , the architecture of the adaptive hybrid spatio-temporal graph neural network model of the traffic load balancing method provided by the embodiment of the application is shown, and the adaptive hybrid spatio-temporal learning module includes three important sub-modules: a temporal convolution module (TCM), an adaptive hybrid graph learning module (AHGLM), and a spatio-temporal adaptive module (STAM). The temporal convolution module TCM is used to process traffic data, and a gated temporal convolution network can be used to capture the complex relationship and periodicity of time series data to obtain time series feature representation of the traffic data. The adaptive hybrid graph learning module AHGLM is used to capture the spatial relationship between the cell towers. The static adaptive graph learning (SAGL) therein captures the relatively stable spatial relationship, and the dynamic adaptive graph learning (DAGL) captures the dynamic influence between neighbors to process dynamic spatial features. The outputs of SAGL and DGL are fused to adaptively control the flow of static and dynamic spatial dependence to obtain spatial feature representation of the traffic data, wherein the data dimensions are all T x N x D. The spatio-temporal adaptive module STAM is used to capture the node-level spatio-temporal adaptive trend, and the time series feature representation and the spatial feature representation obtained by TCM and AHGLM are fused and processed to obtain the spatio-temporal feature representation of the traffic data that integrates the space-time information.

[0069] In this embodiment of the invention, after obtaining the spatiotemporal feature representation of the traffic data incorporating spatiotemporal information, the predicted traffic data for the base station at the next moment is obtained based on the spatiotemporal feature representation. Specifically, the output layer performs skip connections on the outputs of each adaptive hybrid spatiotemporal learning module, directly connecting them to the output layer. As multiple adaptive hybrid spatiotemporal learning modules are stacked, the temporal awareness of the modules also increases. The lower-level blocks focus more on temporally adjacent traffic features, while the higher-level blocks focus on long-term temporal information. Skip connections are used to solve the problem of spatial dependency modeling at different time levels. After the skip connections, the outputs of all adaptive hybrid spatiotemporal learning modules are fused. The output layer consists of two fully connected layers to generate the final multi-step prediction, i.e., the predicted traffic data for the base station at the next moment.

[0070]

[0071] It should be noted that this embodiment adds a temporal embedding matrix to the initialization part of the adaptive hybrid graph neural network model and updates it in each iteration to represent the relationship in the time dimension. The temporal embedding matrix is ​​generated by sine and cosine functions to dynamically represent time information. Specifically, during the forward propagation of the static adaptive graph learning (SAGL) module in the adaptive hybrid graph learning module, the temporal embedding matrix is ​​applied to the traffic data, enabling the model to dynamically consider time information.

[0072] Step 103: Obtain the adjacency matrix of the base station, the actual traffic data at the current moment, and the remaining traffic capacity.

[0073] In this embodiment of the invention, in cellular networks, traditional load balancing strategies allocate resources based solely on the current network load, neglecting the dynamic nature of the network load. This makes it difficult to cope with load fluctuations in complex network environments, resulting in uneven load distribution. This embodiment considers future traffic conditions and adjusts resource allocation strategies based on predicted traffic data to address potential future load changes. Specifically, it establishes models, calculates pheromone increments, updates pheromone levels, and redefines path selection strategies. The difference between actual and predicted traffic is used as the basis for calculating pheromone increments, and heuristic information is combined to redefine path selection strategies, thereby achieving more intelligent and efficient load balancing.

[0074] Specifically, first, the adjacency matrix of the base station, the actual traffic data at the current moment, and the remaining traffic capacity are obtained. The adjacency matrix of the base station is a square matrix, where each element indicates whether there is a direct adjacency relationship between two base stations. Specifically, if base station i and base station j are adjacent, then the element in the i-th row and j-th column of the adjacency matrix is ​​1, otherwise it is 0. The actual traffic data of the base station at the current moment refers to the actual data traffic that the base station is processing at a certain moment. The actual data traffic can include uplink traffic (from user equipment to base station) and downlink traffic (from base station to user equipment), which are not specifically limited here. The remaining traffic capacity of the base station refers to the additional number of users or traffic that the base station can support under the current load.

[0075] In this embodiment, Y t ={y t,1 ,y t,2 ,...,y t,N Let y be the actual traffic data of N base stations at time t, where y t,n Y represents the actual traffic situation of the nth base station at time t. t+1 ={y t+1,1 ,y t+1,2 ,...,y t+1,n} represents the predicted traffic data of each base station at time t+1, where y t+1,n This represents the predicted traffic situation of the nth base station at time t+1. A ij ∈R N×N , represents the adjacency matrix of the base stations in the cellular network, where N represents the number of base stations. S ij ∈R N×N This represents a load balancing scheduling scheme, indicating the traffic that base station i needs to transfer to base station j in this scheme. This embodiment requires a load balancing decision strategy D to implement the load balancing scheduling scheme, adjust the user traffic allocation scheme, determine which base stations to allocate user traffic to, minimize the load differences between base stations or maximize the overall network resource utilization, and improve network performance.

[0076] Step 104: Calculate the pheromone between the base station and its neighboring base stations using actual traffic data, predicted traffic data, and adjacency matrix.

[0077] In this embodiment of the invention, the pheromone between a base station and its neighboring base stations is calculated using actual traffic data, predicted traffic data, and an adjacency matrix. Specifically, firstly, the difference between actual and predicted traffic is calculated to determine the preferred direction of path selection. If the load increases in the next time step, the difference value is positive, and the pheromone increment of the transfer path of traffic flowing out of that base station is less than 0; conversely, the pheromone increment of the transfer path of traffic flowing out of that base station is greater than 0. Secondly, the traffic difference is multiplied by a set maximum increment parameter to calculate the pheromone increment. The maximum increment parameter is used to control the adjustment range of the pheromone. When the base station is expected to be a high-load base station, the corresponding maximum increment parameter will increase, significantly reducing its pheromone concentration. Finally, for each pair of neighboring base stations, the pheromone value of the corresponding path is adjusted according to the pheromone increment.

[0078] It's important to note that pheromones are a mechanism that mimics biological behavior and is commonly used in optimization algorithms, such as Ant Colony Optimization (ACO). In ACO, pheromones simulate the chemical substances left by ants on their paths to guide other ants in choosing their routes. The concentration of pheromones reflects the quality or detriment of the path. ACO is a heuristic optimization algorithm inspired by the path-finding behavior of ants. Its core idea is to find the optimal solution by simulating the process of ants releasing pheromones while searching for food. In ACO, ants search for solutions in the solution space and are guided to choose their routes through the deposition and volatilization of pheromones, thus achieving a balance between global and local search and finding the optimal solution. In cellular networks, load balancing among multiple base stations is necessary, meaning that the load conditions of multiple base stations need to be considered simultaneously. The distributed search capability of the ant colony algorithm enables it to effectively handle this multivariate, multi-objective problem. In this embodiment, pheromones are used as a mechanism to adjust and optimize path selection in the network.

[0079] Specifically, for each pair of adjacent base stations, the pheromone value of the corresponding path is adjusted according to the pheromone increment. For each base station i, its list of adjacent base stations is obtained and the list of adjacent base stations is traversed. For each adjacent base station j, the pheromone value on the path from base station i to base station j is updated. The update rule is to add the increment to the original pheromone value and ensure that the pheromone value is not less than 0. This will not be elaborated on here.

[0080] Step 105: Use the remaining traffic capacity and adjacency matrix to calculate the heuristic information between the base station and its neighboring base stations.

[0081] In this embodiment of the invention, heuristic information between a base station and its neighboring base stations is calculated using the remaining traffic capacity and the adjacency matrix. The heuristic information is calculated based on the remaining traffic capacity and the adjacency matrix of the current base station. For each base station, the remaining traffic capacity is multiplied by the adjacency matrix, and the remaining capacity of non-adjacent base stations is set to zero. This yields a heuristic information matrix corresponding to the adjacency matrix. The heuristic information is used to measure the merits of each path during the path selection process. It is mainly used to determine the probability of choosing a path, thereby influencing the base station's decision to transfer traffic to another base station.

[0082] Step 106: Calculate the path selection probability of traffic transmission between the base station and neighboring base stations based on pheromones and heuristic information, and determine the target transmission path.

[0083] In this embodiment of the invention, for each base station, the path selection probability of traffic transmission between the base station and neighboring base stations is calculated based on pheromone and heuristic information. That is, the selection probability of each path connected to the base station is calculated based on pheromone concentration and heuristic information. Specifically, the path selection probability is calculated based on the heuristic information value of the path combined with the pheromone value of the path. The heuristic information value can be the estimated cost, distance, delay, etc. of the path, and the pheromone value reflects the quality or inferiority of the path.

[0084] Step 107: Distribute traffic to base stations on the target transmission path.

[0085] In this embodiment of the invention, traffic allocation is performed on base stations along the target transmission path. For each base station, the selection probability of each path connected to that base station is calculated based on pheromone concentration and heuristic information, thus optimizing traffic allocation. During the load balancing process at each time step, the pheromone concentration is updated based on the traffic prediction data for the next time step to adapt to network changes, helping the network better cope with traffic fluctuations and improving the load balancing effect. The load balancing in this embodiment combines an ant colony optimization strategy, utilizing traffic prediction results to achieve intelligent scheduling and optimized allocation of cellular network resources. This enables more intelligent scheduling of network resources, improves cellular network resource utilization, and provides a solution for the sustainable development of mobile communications.

[0086] The traffic load balancing method provided in this invention obtains traffic data from base stations for a preset period, performs feature analysis on the traffic data using a pre-trained network traffic prediction model to obtain a spatiotemporal feature representation of the traffic data, obtains the predicted traffic data of the base station at the next moment based on the spatiotemporal feature representation, obtains the adjacency matrix of the base station, the actual traffic data at the current moment, and the remaining traffic capacity, calculates the pheromone between the base station and neighboring base stations using the actual traffic data, predicted traffic data, and adjacency matrix, calculates the heuristic information between the base station and neighboring base stations using the remaining traffic capacity and adjacency matrix, calculates the path selection probability of traffic transmission between the base station and neighboring base stations based on the pheromone and heuristic information, determines the target transmission path, and allocates traffic to the base stations on the target transmission path. This invention achieves spatiotemporal characteristic fusion of traffic data through a pre-trained network traffic prediction model, enabling traffic prediction that conforms to the spatiotemporal characteristics of network traffic. This improves the accuracy of service traffic prediction in complex and variable cellular network environments, provides reliable data support for load balancing, calculates pheromones and heuristic information based on traffic prediction results, obtains current and future load change trends, dynamically adjusts the load distribution among base stations, and more flexibly allocates network resources to base stations expected to have high loads to cope with future traffic fluctuations. Through network traffic prediction and monitoring, it ensures high performance under high traffic loads, while using load balancing to avoid network port overload, thus ensuring network stability and efficiency.

[0087] Furthermore, refer to Figure 2 , showed Figure 1 A flowchart of step 102 in a traffic load balancing method is provided. This method is basically the same as the traffic load balancing method provided in the first embodiment of the present invention. Step 102 may include:

[0088] Step 1021: Use a pre-trained network traffic prediction model to perform feature analysis on the traffic data to obtain the temporal and spatial feature representations of the traffic data.

[0089] Step 1022: The temporal feature representation and spatial feature representation are fused to obtain the spatiotemporal feature representation of the traffic data.

[0090] Step 1023: Map the spatiotemporal feature representation to obtain the predicted traffic data of the base station at the next moment.

[0091] It should be noted that, in this embodiment of the invention, the pre-trained network traffic prediction model is an adaptive hybrid spatiotemporal graph neural network model. The adaptive hybrid spatiotemporal learning module within the adaptive hybrid spatiotemporal graph neural network model processes periodic traffic data, including recent, daily, and weekly base station traffic. The temporal convolution module (TCM) uses a gated temporal convolutional network or a graph convolutional network (GCN) to capture the complex relationships and periodicity of the time-series data. The output of each temporal convolutional TCN module is a temporal feature representation of the traffic data. Where D represents the depth or dimension of the feature.

[0092] Specifically, the Adaptive Hybrid Graph Learning (AHGLM) module captures the spatial relationships between cell towers. It mainly uses Static Adaptive Graph Learning (SAGL) to capture relatively stable spatial relationships, and Dynamic Adaptive Graph Learning (DAGL) to capture the dynamic influence between neighbors and process dynamic spatial features. Spatial control gate fusion is used to fuse the outputs of SAGL and DGL to adaptively control the flow of static and dynamic spatial dependencies. The input of this module is the spatial relationship between cell towers, and the output is the learned spatial feature representation. The data dimension is T×N×D.

[0093] In this embodiment, the spatiotemporal adaptive module STAM is used to capture node-level spatiotemporal adaptation trends. The inputs are the outputs of TCM and AHGLM, which are temporal feature representations and spatial feature representations, respectively. The output is the spatiotemporal feature representation fused with spatiotemporal information obtained after STAM processing. The data dimensions are T×N×D.

[0094] In this embodiment, the output layer performs skip connections on the outputs of each adaptive hybrid graph learning module, directly connecting them to the output layer. As multiple adaptive hybrid graph learning modules are stacked, the outputs of all adaptive hybrid graph learning modules are fused using summation, and the result is represented as H. out ∈R T×N×D Finally, the output layer generates the final multi-step prediction.

[0095] Specifically, the formula for the predicted traffic data of the base station at the next time step output by the output layer is as follows:

[0096]

[0097] in, and These are learnable parameters. It is the final output of the entire AHSTGNN.

[0098] In some embodiments, the mean absolute error between the predicted and actual values ​​can be used as a loss function in multi-step base station traffic prediction tasks, and it is minimized through backpropagation to improve the accuracy of the predicted traffic data output by the model. The formula for the loss function is as follows:

[0099]

[0100] Where θ represents all the learnable parameters of the model. Y represents the model's prediction for all base stations at time step i. i This is the actual value.

[0101] The present invention provides a service traffic prediction method based on an adaptive hybrid spatiotemporal graph neural network model, which makes the model more sensitive to time changes, improves the accuracy of service traffic prediction in complex and variable cellular network environments, and provides reliable data support for load balancing.

[0102] Specifically, step 1021 uses a pre-trained network traffic prediction model to perform feature analysis on the traffic data to obtain the temporal and spatial feature representations of the traffic data, which may include the following steps:

[0103] First, the traffic data is convolved with a preset time embedding matrix to obtain a temporal feature representation of the traffic data;

[0104] Secondly, static adaptive graph learning and dynamic adaptive graph learning are performed on the traffic data to obtain static spatial relationships and dynamic spatial relationships;

[0105] Secondly, the static spatial relationship and the dynamic spatial relationship are fused to obtain the spatial feature representation of the traffic data.

[0106] It should be noted that in the above steps, the traffic data is convolved with a preset temporal embedding matrix to obtain the temporal feature representation of the traffic data. Static adaptive graph learning and dynamic adaptive graph learning are then performed on the traffic data to obtain static and dynamic spatial relationships, respectively. Finally, the static and dynamic spatial relationships are fused to obtain the spatial feature representation of the traffic data. The temporal embedding matrix is ​​a matrix used to represent temporal information in time series data. It uses sine and cosine functions to generate temporal embeddings to dynamically represent temporal information. It is commonly used in deep learning models, especially when dealing with time-dependent data. The temporal embedding matrix can encode temporal information into a high-dimensional vector, enabling the model to better capture patterns and trends in the time series.

[0107] Specifically, a temporal embedding matrix is ​​added to the model initialization part and updated in each iteration to represent the temporal relationship. During the forward propagation of the SAGL module in AHGLM, the temporal embedding matrix is ​​applied to the input data and combined with the graph convolutional neural network (GCN) calculation. In this embodiment, the GCN calculation formula is modified to include the temporal embedding matrix, enabling the model to dynamically consider temporal information. The modified GCN calculation formula is as follows:

[0108]

[0109] Where D is the degree diagonal matrix determined by the adjacency matrix A, X is the input, θ are the neural network parameters, and t emb This is the time embedding matrix.

[0110] This invention incorporates a time embedding matrix into traffic data, enabling the model to dynamically learn and update relationships in the time dimension, thereby adapting to time-related patterns in the task. The weights of this time embedding matrix are trained as part of the model, making the model more sensitive to time changes and improving the accuracy of traffic prediction.

[0111] Furthermore, refer to Figure 3 , showed Figure 1 A flowchart of step 104 in a traffic load balancing method is provided. This method is basically the same as the traffic load balancing method provided in the first embodiment of the present invention. Step 104 may include:

[0112] Step 1041: Calculate the traffic difference between the actual traffic data and the predicted traffic data, and determine the neighboring base stations of the base station based on the adjacency matrix of the base station.

[0113] Step 1042: Calculate the pheromone increment of adjacent base stations using the traffic difference and preset incremental parameters.

[0114] Step 1043: Adjust the current pheromone of the adjacent base stations using pheromone increment to obtain the pheromone between the base station and the adjacent base stations.

[0115] It should be noted that, in this embodiment of the invention, considering the entire network topology of the cellular network, the current base station load, and the predicted base station load in the next moment, the adjacency matrix A of the cellular network is... ij ∈R N×N The initial traffic data Y of the base station at time t t ={y t,1 ,y t,2 ,...,y t,N} and the predicted traffic data Y of the base station at time t+1 t+1 ={y t+1,1 ,y t+1,2,...,y t+1,N} as input to the model.

[0116] Specifically, neighboring base stations are determined based on the base station's adjacency matrix. The traffic difference between the base station and its actual and predicted traffic data is calculated. This traffic difference is multiplied by a set maximum increment parameter to calculate the pheromone increment. When the base station is expected to be under high load, the corresponding maximum increment parameter increases, significantly reducing its pheromone concentration. The pheromone increment is used to adjust the current pheromone levels of neighboring base stations, thus obtaining the pheromone levels between the base station and its neighbors. Specifically, the traffic difference is used to determine the preferred direction for path selection; the traffic difference can be represented as Y. t -Y t+1 If the load increases in the next time step, this value will be positive, and the pheromone increment of the transfer path of the outflow traffic from this base station will be less than 0. Conversely, if the load decreases, the pheromone increment of the transfer path of the outflow traffic from this base station will be greater than 0. For each pair of adjacent base stations, the pheromone value of the corresponding path is adjusted according to the pheromone increment.

[0117] For example, for each pair of adjacent base stations, the pheromone value of the corresponding path is adjusted according to the pheromone increment. The specific steps are: for each base station, obtain its list of adjacent base stations, iterate through the list, and for each adjacent base station, update the pheromone value on the path. The update rule is: add the increment to the original pheromone value, ensuring that the pheromone value is not less than 0. The expression is:

[0118] m ij =m ij +d i

[0119] m ij =max(0,m) ij )

[0120] Where, m ij It is the pheromone value along the path from base station i to j, d i This represents the pheromone increment for base station i.

[0121] This invention is based on an ant colony optimization algorithm. It calculates the pheromone increment by utilizing the difference between actual traffic and predicted traffic, and dynamically updates the pheromone to ensure that the pheromone value reflects the changing trend of current and future load.

[0122] Specifically, step 105 uses the remaining traffic capacity and the adjacency matrix to calculate the heuristic information between the base station and its neighboring base stations. This may include: calculating the remaining traffic capacity of the base station with the adjacency matrix, and adjusting the non-adjacent matrices in the adjacency matrix to zero to obtain the heuristic information between the base station and its neighboring base stations.

[0123] In this embodiment, heuristic information is calculated based on the remaining traffic capacity of the current base station and the adjacency matrix. For each base station, the remaining capacity is multiplied by the adjacency matrix, and the remaining capacity of non-adjacent base stations is set to zero. This yields a heuristic information matrix corresponding to the adjacency matrix. The heuristic information expression is as follows:

[0124] h ij =A ij ×L j

[0125] Among them, h ij It is heuristic information, representing the expected value from node i to node j, A ij This represents the connection relationship between node i and node j in the adjacency matrix. A value of 1 indicates a connection exists, while a value of 0 indicates no connection. j This indicates the remaining traffic capacity of the base station.

[0126] Furthermore, refer to Figure 4 , showed Figure 1 A flowchart of step 106 in a traffic load balancing method is provided. This method is basically the same as the traffic load balancing method provided in the first embodiment of the present invention. Step 106 may include:

[0127] Step 1061: Use pheromones and heuristic information to calculate the selection probability of each path connecting base stations to each other, and obtain the path selection probability of traffic transmission between base stations and adjacent base stations.

[0128] Step 1062: Select the target transmission path from each path connected to the base station based on the path selection probability.

[0129] It should be noted that in this embodiment of the invention, pheromones and heuristic information are used to calculate the selection probability of each path connected to the base station, so as to obtain the path selection probability of traffic transmission between the base station and the adjacent base station. The path selection probability is the probability value of selecting the path. The target transmission path is selected from each path connected to the base station according to the path selection probability. The target transmission path is randomly selected as the next node according to the probability for transmitting traffic.

[0130] Specifically, for each base station, the selection probability of each path connected to that base station is calculated based on pheromone concentration and heuristic information. For load balancing at each time step, the pheromone concentration matrix changes according to the traffic prediction data for the next time step. The selection probability formula for each path is as follows:

[0131]

[0132] Where α is a parameter controlling the influence of pheromone concentration on path selection, β is a parameter controlling the influence of heuristic information on path selection, v is a vector representing the visited nodes, and m ij It is the pheromone value along the path from base station i to j, h ij It is heuristic information, representing the expected value from node i to node j.

[0133] According to the embodiments of the present invention, the target transmission path is selected from each path connected to the base station based on the path selection probability, and the load distribution between the base stations is dynamically adjusted to allocate network resources more flexibly to the base stations with expected high load in order to cope with future traffic fluctuations.

[0134] Reference Figure 5 The diagram illustrates a flowchart of another traffic load balancing method provided by an embodiment of the present invention. This method is basically the same as the traffic load balancing method provided by the first embodiment of the present invention, except that the method may further include:

[0135] Step 101: Obtain traffic data for a preset period of the base station.

[0136] Step 102: Use a pre-trained network traffic prediction model to perform feature analysis on the traffic data to obtain the spatiotemporal feature representation of the traffic data, and obtain the predicted traffic data of the base station at the next moment based on the spatiotemporal feature representation.

[0137] Step 103: Obtain the adjacency matrix of the base station, the actual traffic data at the current moment, and the remaining traffic capacity.

[0138] Step 104: Calculate the pheromone between the base station and its neighboring base stations using actual traffic data, predicted traffic data, and adjacency matrix.

[0139] Step 105: Use the remaining traffic capacity and adjacency matrix to calculate the heuristic information between the base station and its neighboring base stations.

[0140] Step 106: Calculate the path selection probability of traffic transmission between the base station and neighboring base stations based on pheromones and heuristic information, and determine the target transmission path.

[0141] Step 107: Distribute traffic to base stations on the target transmission path.

[0142] Steps 101 to 107 described above are the same as those described previously and will not be repeated here.

[0143] Step 108: Store the target transmission path into the initial solution space matrix.

[0144] Step 109: Monitor the traffic allocation of the base station and update the remaining traffic capacity of the target transmission path.

[0145] Step 110: Calculate the resource utilization rate and variance of the target transmission path using the remaining traffic capacity.

[0146] Step 111: Determine the target transmission path with a variance less than a preset threshold as the optimal transmission path.

[0147] It should be noted that in steps 108 to 111 above, an empty solution space matrix ant_solutions is initialized. The solution space matrix is ​​used to store each path selection scheme. The target transmission path is stored in the initial solution space matrix. The traffic allocation of the base station is monitored. The remaining traffic capacity of the target transmission path is updated. The remaining traffic capacity is calculated by subtracting the initial traffic from the maximum capacity. The remaining traffic capacity represents the available resources of each node. The resource utilization rate and the variance of the resource utilization rate of the target transmission path are calculated using the remaining traffic capacity.

[0148] Specifically, for a resource reallocation, a starting node is selected, and path selection is performed without visiting all nodes: First, the next node is selected using path selection probability, and resources are allocated to the selected node. If the next node cannot be selected (all paths have been visited or are infeasible), the current resource reallocation is terminated. The solution space matrix and remaining traffic capacity are updated based on the path selection results. Then, the resource reallocation operation is iterated until the number of iterations reaches the set parameter. After all iterations are completed, the optimal value among all target transmission paths is calculated. The resource utilization variance is used as a load balancing evaluation index to measure the degree of balance of resource utilization in the network. The algorithm is designed with the goal of minimizing the resource utilization variance. The target transmission path with a variance less than a preset threshold is determined as the optimal transmission path. At this time, the network traffic transfer strategy can be obtained from the solution space matrix ant_solutions.

[0149] In this embodiment, a cellular network load balancing algorithm is designed with reference to the ant colony optimization algorithm. Based on simulating the foraging behavior of ants, it realizes traffic scheduling between base stations to optimize network resource utilization. The resource utilization rate and variance of the target transmission path of the base station are expressed as:

[0150]

[0151] Where, x i c is the actual resource allocation for the i-th base station. i is the total capacity of the i-th base station, and N is the number of base stations.

[0152] Compared with the prior art, the embodiments of the present invention, based on achieving the beneficial effects brought by the first embodiment, use the resource utilization variance as a load balancing evaluation index to measure the degree of balance of resource utilization in the network. The algorithm is designed with the goal of minimizing the resource utilization variance, and the target transmission path with a variance less than a preset threshold is determined as the optimal transmission path, thereby achieving load balancing optimization to cope with future traffic fluctuations.

[0153] Reference Figure 7 The diagram shows a structural schematic of a traffic load balancing device provided in an embodiment of the present invention. The device includes:

[0154] The first acquisition module 201 is used to acquire traffic data of the base station for a preset period;

[0155] Traffic prediction module 202 is used to perform feature analysis on the traffic data using a pre-trained network traffic prediction model to obtain the spatiotemporal feature representation of the traffic data, and to obtain the predicted traffic data of the base station at the next moment based on the spatiotemporal feature representation.

[0156] The second acquisition module 203 is used to acquire the adjacency matrix of the base station, the actual traffic data at the current time, and the remaining traffic capacity;

[0157] The pheromone calculation module 204 is used to calculate the pheromone between the base station and its neighboring base stations using the actual traffic data, the predicted traffic data and the adjacency matrix;

[0158] The heuristic information calculation module 205 is used to calculate the heuristic information between the base station and its neighboring base stations using the remaining traffic capacity and the adjacency matrix;

[0159] The path determination module 206 is used to calculate the path selection probability of traffic transmission between the base station and the adjacent base station based on the pheromone and the heuristic information, and to determine the target transmission path.

[0160] Traffic allocation module 207 is used to allocate traffic to base stations on the target transmission path.

[0161] Furthermore, the traffic prediction module 202 includes:

[0162] The feature analysis submodule is used to perform feature analysis on the traffic data using a pre-trained network traffic prediction model to obtain the temporal and spatial feature representations of the traffic data.

[0163] The feature fusion submodule is used to fuse the temporal feature representation and the spatial feature representation to obtain the spatiotemporal feature representation of the traffic data;

[0164] The mapping submodule is used to map the spatiotemporal feature representation to obtain the predicted traffic data of the base station at the next moment.

[0165] Furthermore, the feature analysis submodule includes:

[0166] The convolution processing unit is used to perform convolution processing on the traffic data and a preset time embedding matrix to obtain the temporal feature representation of the traffic data;

[0167] An adaptive learning unit is used to perform static adaptive graph learning and dynamic adaptive graph learning on the traffic data to obtain static spatial relationships and dynamic spatial relationships.

[0168] The fusion unit is used to fuse the static spatial relationship and the dynamic spatial relationship to obtain a spatial feature representation of the traffic data.

[0169] Furthermore, the pheromone calculation module 204 includes:

[0170] The first calculation submodule is used to calculate the traffic difference between the actual traffic data and the predicted traffic data, and to determine the neighboring base stations of the base station according to the adjacency matrix of the base station.

[0171] The second calculation submodule is used to calculate the pheromone increment of the adjacent base station using the traffic difference and the preset incremental parameters;

[0172] The adjustment submodule is used to adjust the current pheromone of the adjacent base station using the pheromone increment to obtain the pheromone between the base station and the adjacent base station.

[0173] Furthermore, the heuristic information computing module 205 includes:

[0174] The third calculation submodule is used to calculate the remaining traffic capacity of the base station with the adjacency matrix, and adjust the non-adjacent matrices in the adjacency matrix to zero to obtain heuristic information between the base station and its neighboring base stations.

[0175] Furthermore, the path determination module 206 includes:

[0176] The fourth calculation submodule is used to calculate the selection probability of each path connecting the base station and the base station using the pheromone and the heuristic information, so as to obtain the path selection probability of traffic transmission between the base station and the adjacent base station.

[0177] The filtering submodule is used to filter out the target transmission path from each path connected to the base station based on the path selection probability.

[0178] Furthermore, the device also includes:

[0179] The storage path module is used to store the target transmission path into the initial solution space matrix;

[0180] The capacity update module is used to monitor the traffic allocation of the base station and update the remaining traffic capacity of the target transmission path;

[0181] The resource utilization calculation module is used to calculate the resource utilization rate of the target transmission path and the variance of the resource utilization rate using the remaining traffic capacity;

[0182] The optimal path determination module is used to determine the target transmission path whose variance is less than a preset threshold as the optimal transmission path.

[0183] The traffic load balancing device provided in this embodiment of the invention acquires traffic data of a base station for a preset period, performs feature analysis on the traffic data using a pre-trained network traffic prediction model to obtain a spatiotemporal feature representation of the traffic data, obtains the predicted traffic data of the base station at the next moment based on the spatiotemporal feature representation, acquires the adjacency matrix of the base station, the actual traffic data at the current moment and the remaining traffic capacity, calculates the pheromone between the base station and neighboring base stations using the actual traffic data, predicted traffic data and adjacency matrix, calculates the heuristic information between the base station and neighboring base stations using the remaining traffic capacity and adjacency matrix, calculates the path selection probability of traffic transmission between the base station and neighboring base stations based on the pheromone and heuristic information, determines the target transmission path, and allocates traffic to the base stations on the target transmission path. This invention achieves spatiotemporal characteristic fusion of traffic data through a pre-trained network traffic prediction model, enabling traffic prediction that conforms to the spatiotemporal characteristics of network traffic. This improves the accuracy of service traffic prediction in complex and variable cellular network environments, provides reliable data support for load balancing, calculates pheromones and heuristic information based on traffic prediction results, obtains current and future load change trends, dynamically adjusts the load distribution among base stations, and more flexibly allocates network resources to base stations expected to have high loads to cope with future traffic fluctuations. Through network traffic prediction and monitoring, it ensures high performance under high traffic loads, while using load balancing to avoid network port overload, thus ensuring network stability and efficiency.

[0184] Reference Figure 8 The present invention also provides an electronic device, such as... Figure 8 As shown, it includes a processor 301, a communication interface 302, a memory 303, and a communication bus 304, wherein the processor 301, the communication interface 302, and the memory 303 communicate with each other through the communication bus 304.

[0185] Processor 301, memory 303 for storing processor-executable instructions;

[0186] The processor 301 is configured to execute the instructions to implement the traffic load balancing method described below:

[0187] Obtain traffic data from the base station at a preset period;

[0188] A pre-trained network traffic prediction model is used to perform feature analysis on the traffic data to obtain the spatiotemporal feature representation of the traffic data, and the predicted traffic data of the base station at the next moment is obtained based on the spatiotemporal feature representation.

[0189] Obtain the base station's adjacency matrix, the current actual traffic data, and the remaining traffic capacity;

[0190] The pheromone between a base station and its neighboring base stations is calculated using the actual traffic data, the predicted traffic data, and the adjacency matrix.

[0191] Heuristic information between a base station and its neighboring base stations is calculated using the remaining traffic capacity and the adjacency matrix.

[0192] The path selection probability of traffic transmission between the base station and its neighboring base stations is calculated based on the pheromone and the heuristic information to determine the target transmission path.

[0193] Traffic allocation is performed on the base stations along the target transmission path.

[0194] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.

[0195] The communication interface is used for communication between the aforementioned terminal and other devices.

[0196] The memory may include random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0197] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0198] In another embodiment of the present invention, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements any of the traffic load balancing methods described in the above embodiments.

[0199] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid state disk (SSD)).

[0200] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0201] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0202] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of protection of the present invention.

Claims

1. A traffic load balancing method, characterized in that, The method includes: Obtain traffic data from the base station at a preset period; A pre-trained network traffic prediction model is used to perform feature analysis on the traffic data to obtain the spatiotemporal feature representation of the traffic data, and the predicted traffic data of the base station at the next moment is obtained based on the spatiotemporal feature representation. Obtain the base station's adjacency matrix, the current actual traffic data, and the remaining traffic capacity; The method of calculating the pheromone between a base station and its neighboring base stations using the actual traffic data, the predicted traffic data, and the adjacency matrix includes: calculating the traffic difference between the actual traffic data and the predicted traffic data, determining the neighboring base stations of the base station according to the adjacency matrix of the base station, calculating the pheromone increment of the neighboring base stations using the traffic difference and a preset increment parameter, and adjusting the current pheromone of the neighboring base stations using the pheromone increment to obtain the pheromone between the base station and its neighboring base stations. The heuristic information between a base station and its neighboring base stations is calculated using the remaining traffic capacity and the adjacency matrix, including: calculating the remaining traffic capacity of the base station with the adjacency matrix, and adjusting the non-adjacent matrices in the adjacency matrix to zero, thereby obtaining the heuristic information between the base station and its neighboring base stations; The path selection probability of traffic transmission between the base station and its neighboring base stations is calculated based on the pheromone and the heuristic information to determine the target transmission path. Traffic allocation is performed on the base stations along the target transmission path.

2. The method according to claim 1, characterized in that, The step of using a pre-trained network traffic prediction model to perform feature analysis on the traffic data to obtain a spatiotemporal feature representation of the traffic data, and obtaining the predicted traffic data of the base station at the next time step based on the spatiotemporal feature representation, includes: A pre-trained network traffic prediction model is used to perform feature analysis on the traffic data to obtain the temporal and spatial feature representations of the traffic data. The temporal feature representation and the spatial feature representation are fused to obtain the spatiotemporal feature representation of the traffic data; The spatiotemporal feature representation is mapped to obtain the predicted traffic data of the base station at the next moment.

3. The method according to claim 2, characterized in that, The step of using a pre-trained network traffic prediction model to perform feature analysis on the traffic data to obtain the temporal and spatial feature representations of the traffic data includes: The traffic data is convolved with a preset time embedding matrix to obtain a temporal feature representation of the traffic data. Static and dynamic spatial relationships are obtained by performing static adaptive graph learning and dynamic adaptive graph learning on the traffic data. By fusing the static spatial relationship and the dynamic spatial relationship, a spatial feature representation of the traffic data is obtained.

4. The method according to claim 1, characterized in that, The step of calculating the path selection probability of traffic transmission between the base station and neighboring base stations based on the pheromone and the heuristic information, and determining the target transmission path, includes: The path selection probability of traffic transmission between the base station and its neighboring base stations is obtained by using the pheromone and the heuristic information to calculate the selection probability of each path connecting the base station and the base station. The target transmission path is selected from each path connected to the base station based on the path selection probability.

5. The method according to claim 1, characterized in that, After allocating traffic to the base stations on the target transmission path, the method further includes: Store the target transmission path into the initial solution space matrix; Monitor the traffic allocation of the base station and update the remaining traffic capacity of the target transmission path; The remaining traffic capacity is used to calculate the resource utilization rate of the target transmission path and the variance of the resource utilization rate; The target transmission path with a variance less than a preset threshold is determined as the optimal transmission path.

6. A flow load balancing device, characterized in that, The device includes: The first acquisition module is used to acquire traffic data of the base station for a preset period; The traffic prediction module is used to perform feature analysis on the traffic data using a pre-trained network traffic prediction model to obtain the spatiotemporal feature representation of the traffic data, and to obtain the predicted traffic data of the base station at the next moment based on the spatiotemporal feature representation. The second acquisition module is used to acquire the base station's adjacency matrix, the actual traffic data at the current moment, and the remaining traffic capacity; A pheromone calculation module is used to calculate the pheromone between a base station and its neighboring base stations using the actual traffic data, the predicted traffic data, and the adjacency matrix. The pheromone calculation module includes: a first calculation submodule, used to calculate the traffic difference between the actual traffic data and the predicted traffic data, and determine the neighboring base stations of the base station according to the adjacency matrix of the base station; a second calculation submodule, used to calculate the pheromone increment of the neighboring base station using the traffic difference and a preset increment parameter; and an adjustment submodule, used to adjust the current pheromone of the neighboring base station using the pheromone increment to obtain the pheromone between the base station and its neighboring base stations. A heuristic information calculation module is used to calculate heuristic information between a base station and its neighboring base stations using the remaining traffic capacity and the adjacency matrix. The heuristic information calculation module includes a third calculation submodule, which is used to calculate the remaining traffic capacity of the base station and the adjacency matrix, and adjust the non-adjacent matrices in the adjacency matrix to zero to obtain the heuristic information between the base station and its neighboring base stations. The path determination module is used to calculate the path selection probability of traffic transmission between the base station and neighboring base stations based on the pheromone and the heuristic information, and to determine the target transmission path. The traffic allocation module is used to allocate traffic to the base stations on the target transmission path.

7. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; The memory is used to store computer programs; When the processor executes a program stored in the memory, it implements the method as described in any one of claims 1-5.

8. A computer-readable storage medium having instructions stored thereon that, when executed by one or more processors, cause the processors to perform the method as described in any one of claims 1-5.

Images