Network resource configuration methods, devices, servers, and storage media

Abstract

This application provides a network resource configuration method, apparatus, server, and storage medium, relating to the field of communication technology. The method monitors the first communication parameters of a base station executing corresponding service types based on currently configured multiple network bandwidth resources within a target time period; it performs quality assessments on the services executed by the base station for each service type based on the first communication parameters, obtaining quality assessment values ​​for each service type; it sums the quality assessment values ​​for each service type to obtain a total quality assessment value; if the difference between the current total quality assessment value and the previous total quality assessment value is greater than a first assessment threshold, it uses a reinforcement learning model based on the current total quality assessment value to update the currently configured multiple network bandwidth resources of the base station, ensuring reasonable allocation of network bandwidth corresponding to each service type, maximizing the total quality of each service type, and avoiding waste of network resources.

Description

Technical Field

[0001] This application relates to the field of communication technology, and in particular to a network resource configuration method, apparatus, server and storage medium. Background Technology

[0002] With the rapid development of the 5G industry, various types of services (such as making phone calls, sending and receiving text messages, and accessing the Internet) can be executed on 5G networks.

[0003] Typically, each base station can slice its corresponding 5G network. Then, the bandwidth of each sliced ​​network is used to execute the corresponding service type.

[0004] Currently, the network bandwidth allocated to different service types in base stations is unreasonable. This results in poor service quality (such as latency and noise) when the network bandwidth allocation for a certain service type is too small, while a large network bandwidth allocation for a certain service type wastes network resources. Summary of the Invention

[0005] This application provides a network resource configuration method, apparatus, server, and storage medium to solve the problem of unreasonable network bandwidth corresponding to various service types configured in base stations.

[0006] In a first aspect, this application provides a network resource configuration method, comprising: monitoring the first communication parameters of a base station executing services of corresponding service types according to multiple currently configured network bandwidth resources within a target time period; performing quality assessments on the services executed by the base station for each service type according to the first communication parameters to obtain quality assessment values ​​for each service type; summing the quality assessment values ​​for each service type to obtain a total quality assessment value; and if the difference between the total quality assessment value obtained this time and the total quality assessment value obtained last time is greater than a first assessment threshold, then using a reinforcement learning model based on the total quality assessment value obtained this time to update the multiple network bandwidth resources currently configured by the base station.

[0007] In one alternative implementation, based on the obtained total quality assessment value, a reinforcement learning model is used to update multiple network bandwidth resources currently configured for the base station, including:

[0008] Based on the total quality assessment value obtained in this study, a reinforcement learning model is used to determine the network bandwidth resources to be allocated. The network bandwidth resources of the first service type are allocated to the second service type. The first service type is the service type with a quality assessment value greater than the second assessment threshold, and the second service type is the service type with a quality assessment value less than the second assessment threshold.

[0009] In this way, the network bandwidth corresponding to each service type configured by the base station can be reasonably allocated, so as to maximize the overall quality of each service type and avoid wasting network resources.

[0010] In an optional implementation, before the monitoring base station executes the corresponding service type service according to the currently configured multiple network bandwidth resources within the target time period using the first communication parameter, the method provided in this application further includes: performing quality assessment on the service of each service type according to the second communication parameter to obtain the quality assessment value of the service of each service type; and summing the quality assessment values ​​of the service of each service type to obtain the total quality assessment value.

[0011] In an optional implementation, before detecting that the base station executes the communication parameters of the corresponding service type according to the currently configured multiple initial network bandwidth resources within the target time period, the method provided in the application further includes: extracting the traffic characteristics of the service type executed by the base station in the previous period adjacent to the period in which the target time period is located; using a pre-trained traffic prediction model to predict the traffic size of each service type in the target time period based on the extracted traffic characteristics of each service type, wherein the pre-trained traffic prediction model is obtained by training an initial network to be trained using input data and output data, the output data being: the traffic size of each service type in the target time period within multiple historical periods, and the input data being the traffic characteristics of each service type in the previous period adjacent to each historical period; determining the initial network bandwidth resources corresponding to each service type according to the predicted traffic size of each service type; and configuring the initial network bandwidth resources corresponding to each service type in the base station.

[0012] In this way, the traffic characteristics of each service type in the target time period in history can be used to predict the traffic volume of each service type in the next target time period. The reliability of the predicted traffic volume configuration in the initial network bandwidth resources corresponding to each service type configured in the base station is relatively high. This can reduce the number of iterations of the reinforcement learning model, save computing resources, and reduce the number of unreasonable network resource allocations.

[0013] In one alternative implementation, the initial network to be trained is a gated recurrent neural network (GRU).

[0014] In one optional implementation, the traffic features corresponding to each service type executed by the base station in the previous period adjacent to the period of the target time are extracted, including: extracting the sub-traffic features of the traffic corresponding to each service type executed by the base station in the previous period adjacent to the period of the target time at each sampling time according to a dual-stream convolutional neural network (CNN); extracting the temporal dependencies of the sub-traffic features at each sampling time according to a deep belief network (DBN); and fusing the sub-traffic features at each sampling time and the temporal dependencies of the sub-traffic features to obtain the traffic features corresponding to each service type.

[0015] This allows for the precise extraction of traffic characteristics corresponding to each business type, and makes the content of traffic characteristics richer.

[0016] In an optional implementation, the method provided in this application may further include: using the Seq2Seq method to perform noise reduction processing on the traffic characteristics corresponding to each service type.

[0017] This makes the extraction of traffic characteristics corresponding to each business type more reliable.

[0018] Secondly, this application also provides a network resource configuration device, comprising: a data monitoring unit, used to monitor the communication parameters of a base station executing services of corresponding service types according to multiple currently configured network bandwidth resources within a target time period; a quality assessment unit, used to perform quality assessments on the services executed by the base station for each service type according to the communication parameters, and obtain quality assessment values ​​for each service type; a data summation unit, used to sum the quality assessment values ​​for each service type, and obtain a total quality assessment value; and a resource configuration unit, used to update the multiple network bandwidth resources currently configured by the base station by adopting a reinforcement learning model based on the total quality assessment value obtained this time if the difference between the total quality assessment value obtained this time and the total quality assessment value obtained last time is greater than a first assessment threshold.

[0019] Thirdly, this application also provides a server, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the server performs the method provided in the first aspect.

[0020] Fourthly, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, causes the computer to perform the method provided in the first aspect, as performed by a terminal device.

[0021] Fifthly, this application also provides a computer program product, including a computer program that, when run, causes a computer to perform the method provided in the first aspect.

[0022] This application provides a network resource configuration method, apparatus, server, and storage medium. It can perform quality assessments on the communication parameters of multiple currently configured network bandwidth resources when executing services of corresponding service types, obtaining a total quality assessment value. If the difference between the current total quality assessment value and the previous total quality assessment value is greater than a first assessment threshold, a reinforcement learning model is used based on the current total quality assessment value to update the multiple network bandwidth resources currently configured by the base station until the difference between the current and previous total quality assessment values ​​is less than the first assessment threshold. In this way, the network bandwidth corresponding to each service type configured by the base station can be rationally allocated, maximizing the overall service quality of each service type without wasting network resources. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is one of the flowcharts for a network resource configuration method provided in an embodiment of this application;

[0025] Figure 2 This is a schematic diagram of the structure of the DQN network provided in the embodiments of this application;

[0026] Figure 3 The second flowchart of the network resource configuration method provided in the embodiments of this application;

[0027] Figure 4 The third flowchart of the network resource configuration method provided in the embodiments of this application;

[0028] Figure 5 for Figure 4 The detailed flowchart of S401;

[0029] Figure 6 A functional block diagram of the network resource configuration device provided in the embodiments of this application. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments made by those skilled in the art under the guidance of these embodiments are within the scope of protection of this application.

[0031] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0032] First, let me explain the terms used in this application:

[0033] Network slicing: Network slicing is an on-demand networking method that allows operators to separate multiple virtual end-to-end networks on a unified infrastructure. Each network slice is logically isolated from the radio access network to the bearer network and then to the core network to adapt to various types of services.

[0034] Deep Belief Networks (DBNs) are a type of neural network model in deep learning. DBNs have multiple hidden layers. Hidden layers in a DBN are interconnected with each other, and the last hidden layer is connected to the visible layer, but units within the same layer are not connected. Multiple hidden layers allow for higher-level correlation operations on the data. This also allows for a more sequential training process, where each hidden layer must be fully trained before the next hidden layer can be trained using the data from the previous layer. This process is repeated until the final trained data is obtained.

[0035] Deep Q-Network (DQN) refers to a Q-learning network based on deep learning. Unlike traditional Q-learning, DQN does not require building a complete Q-matrix; instead, it uses a neural network to estimate the value of the Q-function. The neural network responsible for estimating the Q-function is called the main network, where ω represents the set of parameters of the main network. The target network outputs target values, which are used to update the parameters of the main network. The estimated values ​​of the main network and the target values ​​of the target network constitute a loss function. The Adam optimization algorithm is used to perform stochastic gradient descent to update the parameters of the main network. The parameters of the main network are continuously updated based on the gradient of the loss function, causing the loss function value to decrease and thus making the estimated values ​​of the main network more accurate. Both the main network and the target network contain an input layer, convolutional layers, fully connected layers, and an output layer. The neurons in the input layer are responsible for feeding data to the neurons in the convolutional layers, which extract local features from the input data. The fully connected layers integrate these local features into global features, and the output layer is responsible for outputting the estimated value of the Q-function for each action in the current state.

[0036] GRU network: Gated recurrent neural network (GRU) is better able to capture dependencies with large time step distances in time series. It controls the flow of information through learnable gates.

[0037] Currently, the network bandwidth allocated to different service types in base stations is unreasonable. This results in poor service quality (such as latency and noise) when the network bandwidth allocation for a certain service type is too small, while a large network bandwidth allocation for a certain service type wastes network resources.

[0038] Based on the above-mentioned technical problems, the inventive concept of this application is as follows: when monitoring the services of the corresponding service type, a first communication parameter is used; the quality of each service type is evaluated according to the first communication parameter to obtain a total quality evaluation value; if the difference between the total quality evaluation value obtained this time and the total quality evaluation value obtained last time is greater than a first evaluation threshold, then a reinforcement learning model is used to update the multiple network bandwidth resources currently configured by the base station according to the total quality evaluation value obtained this time.

[0039] In this way, the network bandwidth corresponding to each service type configured by the base station can be reasonably allocated, so as to maximize the overall quality of each service type and avoid wasting network resources.

[0040] The technical solutions of this application and how they solve the aforementioned technical problems will be described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0041] Please see Figure 1 This application provides a network resource configuration method that can be applied to a server. The server communicates with a base station to monitor the communication parameters of the base station performing services. Figure 1 As shown, the network resource configuration method provided in this application includes:

[0042] S101: The first communication parameter when the server monitors the base station to execute the corresponding service type according to the currently configured multiple network bandwidth resources within the target time period.

[0043] For example, the target time period could be 9:00-12:00 every day; or 1:00-24:00 every non-working day, etc., without limitation.

[0044] Multiple network bandwidth resources are obtained using a reinforcement learning model based on the network bandwidth resources corresponding to each service type obtained in the previous iteration and the total quality assessment value obtained in the previous iteration. The total quality assessment value obtained in the previous iteration is obtained by summing the quality assessments of the services of the corresponding service types executed according to the configured multiple network bandwidth resources during the target time period.

[0045] Furthermore, the currently configured network bandwidth resources can each execute corresponding service types as follows: network bandwidth resource A executes service type A, network bandwidth resource B executes service type B, and network bandwidth resource C executes service type C. Service type A can be a voice call, service type B can be SMS sending and receiving, and service type C can be internet access.

[0046] S102: The server performs quality assessments on the services of each service type executed by the base station according to the first communication parameters, and obtains the quality assessment value of each service type.

[0047] For example, the quality assessment value of business type A is Q1; the quality assessment value of business type B is Q2; and the quality assessment value of business type C is Q3.

[0048] S103: The server sums the quality assessment values ​​of each service type to obtain the total quality assessment value.

[0049] For example, if the services of each service type include service type A, service type B, and service type C, and the quality assessment value of service type A is Q1; the quality assessment value of service type B is Q2; and the quality assessment value of service type C is Q3, then the total quality value Q4 = Q1 + Q2 + Q3.

[0050] S104: The server determines whether the difference between the total quality assessment value obtained this time and the total quality assessment value obtained last time is greater than the first assessment threshold. If so, S105 is executed.

[0051] Understandably, if the difference between the current total quality assessment value and the previous total quality assessment value is greater than the preset first assessment threshold, it means that the current total quality assessment value is significantly higher than the previous total quality assessment value. This indicates that the current configuration of multiple network bandwidth resources can still be updated to further improve the total quality assessment value.

[0052] S105: The server uses a reinforcement learning model based on the total quality assessment value obtained this time to update the multiple network bandwidth resources currently configured by the base station, and then returns to execute S101.

[0053] S105 can be specifically implemented as follows: The server uses the reinforcement learning model to determine the network bandwidth resources to be allocated based on the total quality assessment value obtained this time. The network bandwidth resources to be allocated from the multiple network bandwidth resources are then allocated to the second service type, where the first service type is the service type with a quality assessment value greater than the second assessment threshold, and the second service type is the service type with a quality assessment value less than the second assessment threshold.

[0054] Understandably, the higher the overall quality assessment threshold obtained in this study, the larger the network bandwidth resources to be allocated. Furthermore, when the quality assessment value for one service type is greater than the second assessment threshold, it indicates that there may be surplus network bandwidth resources allocated to that service type; conversely, when the quality assessment value for another service type is less than the second assessment threshold, it indicates that there is insufficient network bandwidth resources allocated to that service type. Therefore, the allocated network bandwidth resources for service types with quality assessment values ​​greater than the second assessment threshold are allocated to service types with quality assessment values ​​less than the second assessment threshold. This ensures that the network bandwidth configured by the base station for each service type is rationally allocated, maximizing the overall quality of services across all service types without wasting network resources.

[0055] It should be noted that, as Figure 2As shown, the reinforcement learning model in this embodiment is a DQN network. The main network and target network of the DQN network each include an input layer, convolutional layer 1, convolutional layer 2, fully connected layer 1, fully connected layer 2, and an output layer connected in sequence. The input layer is responsible for feeding data into the convolutional layers, which extract local features from the input data. The fully connected layers integrate these local features into global features, and the output layer is responsible for outputting the estimated value of the Q-function corresponding to each action in the current state.

[0056] In summary, this application provides a network resource configuration method, apparatus, and server. It can perform quality assessments on the communication parameters of multiple currently configured network bandwidth resources when executing services of corresponding service types, obtaining a total quality assessment value. If the difference between the current total quality assessment value and the previous total quality assessment value is greater than a first assessment threshold, a reinforcement learning model is used based on the current total quality assessment value to update the multiple network bandwidth resources currently configured by the base station until the difference between the current and previous total quality assessment values ​​is less than the first assessment threshold. This allows for the rational allocation of network bandwidth corresponding to each service type configured by the base station, maximizing the overall quality of each service type without wasting network resources.

[0057] It should be noted that, in the above... Figure 1 Based on the corresponding embodiment, before S101, such as Figure 3 As shown, the method provided in this application embodiment may further include:

[0058] S301: The server detects that the base station executes the second communication parameters of the corresponding service type according to the multiple initial network bandwidth resources currently configured within the target time period.

[0059] S302: The server performs quality assessments on the services of each service type according to the second communication parameters, and obtains the initial quality assessment value of each service type.

[0060] S303: The server sums the initial quality assessment values ​​for each service type to obtain the initial total quality assessment value.

[0061] The principles of S301-S303 are the same as those of S101-S103 mentioned above, and will not be repeated here.

[0062] Furthermore, in Figure 3 Based on the corresponding embodiment, before S301, such as Figure 4 As shown, the method provided in this application embodiment may further include:

[0063] S401: The server extracts the traffic characteristics of each service type executed by the base station in the previous cycle adjacent to the target time period.

[0064] The period can be one day, one week, or one month, and is not limited here. For example, when the period is one day, the period in which the target time period is located can be today, and the previous period adjacent to the period in which the target time period is located can be yesterday. Thus, the specific implementation of S401 can be to extract the traffic characteristics corresponding to each service type executed by the base station yesterday.

[0065] Specifically, such as Figure 5 As shown, the specific implementation of S401 can be as follows:

[0066] S501: The server extracts the sub-traffic features of the traffic corresponding to each service type executed by the base station in the previous cycle adjacent to the target time period based on the dual-stream convolutional neural CNN network at each sampling time.

[0067] Sub-traffic characteristics may include, but are not limited to, user experience rate, connection density, end-to-end latency, mobility, traffic density, peak user rate, and energy efficiency.

[0068] S502: The server extracts the sub-traffic features at each sampling time based on the deep belief DBN network, and their temporal dependencies.

[0069] S503: The server integrates the sub-traffic features at each sampling time and the temporal dependencies of the sub-traffic features to obtain the traffic features corresponding to each service type.

[0070] For example, the server can fuse sub-traffic features and temporal dependencies between these features at each sampling time point based on an attention mechanism. This allows for the accurate extraction of traffic features corresponding to each service type, resulting in richer content in the traffic features obtained.

[0071] Furthermore, the server can employ the Seq2Seq method to denoise the traffic features corresponding to each business type. This improves the reliability of extracting the traffic features for each business type.

[0072] S402: The server uses a pre-trained traffic prediction model to predict the traffic volume of each business type during the target time period based on the extracted traffic characteristics of each business type.

[0073] The pre-trained traffic prediction model is obtained by training the initial network to be trained using input and output data. The output data is the traffic volume of each service type in the target time period within multiple historical periods, and the input data is the traffic characteristics of each service type in the previous period adjacent to each historical period.

[0074] The traffic volume of each business type during the target time period within multiple historical periods can be: the traffic volume of each business type during the target time period from 9:00 to 12:00 every day in history; the traffic characteristics of each business type in the previous period adjacent to each historical period are: the traffic characteristics of each business type on the previous day of each historical period.

[0075] The initial network to be trained can be, but is not limited to, a gated recurrent neural network (GRU).

[0076] S403: The server determines the initial network bandwidth resources corresponding to each service type based on the predicted traffic volume of each service type.

[0077] In this way, the traffic characteristics of each service type in the target time period in history can be used to predict the traffic volume of each service type in the next target time period. The reliability of the predicted traffic volume configuration in the initial network bandwidth resources corresponding to each service type configured in the base station is relatively high. This can reduce the number of iterations of the reinforcement learning model, save computing resources, and reduce the number of unreasonable network resource allocations.

[0078] Please see Figure 6 This application also provides a network resource configuration device. It should be noted that the basic principle and technical effects of the network resource configuration provided in this application are the same as those in the above embodiments. For the sake of brevity, any parts not mentioned in the embodiments of this application can be referred to the corresponding content in the above embodiments. The network resource configuration device includes a data monitoring unit, a quality assessment unit, a data summation unit, and a resource configuration unit, wherein...

[0079] The data monitoring unit is used to monitor the communication parameters of the base station when it executes the corresponding service type according to the multiple network bandwidth resources currently configured within the target time period.

[0080] The quality assessment unit is used to assess the quality of each service type executed by the base station and obtain the quality assessment value of each service type.

[0081] The data summation unit is used to sum the quality assessment values ​​of each business type to obtain the total quality assessment value.

[0082] The resource allocation unit is used to update multiple network bandwidth resources currently configured by the base station based on the current total quality assessment value if the difference between the current total quality assessment value and the previous total quality assessment value is greater than the first assessment threshold.

[0083] In one optional implementation, the resource allocation unit is specifically used to determine the network bandwidth resources to be allocated based on the total quality assessment value obtained in this study using the reinforcement learning model. The network bandwidth resources of a first service type from among the multiple network bandwidth resources are then allocated to a second service type, wherein the first service type is a service type with a quality assessment value greater than a second assessment threshold, and the second service type is a service type with a quality assessment value less than the second assessment threshold.

[0084] In one optional implementation, the quality assessment unit is specifically used to perform quality assessments on services of each service type according to the second communication parameters to obtain an initial quality assessment value for each service type; and to sum the quality assessment values ​​of each service type to obtain an initial total quality assessment value.

[0085] In an optional embodiment, the apparatus provided in this application may further include: a feature extraction unit, configured to extract traffic features corresponding to each service type executed by the base station in the previous period adjacent to the period in which the target time period is located; a traffic prediction unit, configured to predict the traffic volume of each service type in the target time period based on the extracted traffic features corresponding to each service type using a pre-trained traffic prediction model, wherein the pre-trained traffic prediction model is obtained by training an initial network to be trained using input data and output data, the output data being: the traffic volume of each service type in the target time period within multiple historical periods, and the input data being the traffic features corresponding to each service type in the previous period adjacent to each historical period; a resource determination unit, configured to determine the initial network bandwidth resources corresponding to each service type according to the predicted traffic volumes corresponding to each service type; and a resource configuration unit, configured to configure the initial network bandwidth resources corresponding to each service type at the base station.

[0086] In one alternative implementation, the initial network to be trained is a gated recurrent neural network (GRU).

[0087] In one optional implementation, the feature extraction unit is specifically configured to extract sub-traffic features of traffic corresponding to each service type executed by the base station in the previous period adjacent to the period in which the base station is located during the target time period, based on the dual-stream convolutional neural network (CNN); extract the temporal dependencies of the sub-traffic features at each sampling time based on the deep belief network (DBN); and fuse the sub-traffic features at each sampling time and the temporal dependencies of the sub-traffic features to obtain the traffic features corresponding to each service type.

[0088] In an optional embodiment, the apparatus provided in this application may further include: a noise reduction processing unit, used to perform noise reduction processing on the traffic characteristics corresponding to each service type using the Seq2Seq method.

[0089] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the computer to perform the method provided in the above embodiments.

[0090] This application also provides a server, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it causes the server to perform the method provided in the above embodiments.

[0091] This application also provides a computer program product, including a computer program that, when run, causes a computer to perform the methods provided in the above embodiments.

[0092] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0093] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A method for configuring network resources, characterized in that, include: The first communication parameter when the monitoring base station executes the corresponding service type according to the currently configured multiple network bandwidth resources within the target time period; Based on the first communication parameters, the base station performs quality assessments on services of various service types to obtain quality assessment values ​​for each service type. The total quality assessment value is obtained by summing the quality assessment values ​​of each business type. If the difference between the current total quality assessment value and the previous total quality assessment value is greater than the first assessment threshold, then a reinforcement learning model is used to update the multiple network bandwidth resources currently configured by the base station based on the current total quality assessment value; the reinforcement learning model is a dynamic resource allocation model based on deep neural networks, used to dynamically adjust network bandwidth resources according to the total quality assessment value; Before the monitoring base station executes the first communication parameters for the corresponding service type according to the currently configured multiple network bandwidth resources within the target time period, the method further includes: Extract the traffic characteristics of each service type executed by the base station in the previous period adjacent to the period of the target time period; A pre-trained traffic prediction model is used to predict the traffic volume of each service type in the target time period based on the traffic features extracted for each service type. The pre-trained traffic prediction model is obtained by training an initial network to be trained using input data and output data. The output data is the traffic volume of each service type in the target time period over multiple historical periods. The input data is the traffic features of each service type in the previous period adjacent to each historical period. Based on the predicted traffic volume corresponding to each service type, determine the initial network bandwidth resources corresponding to each service type. Configure the initial network bandwidth resources corresponding to each service type in the base station.

2. The method according to claim 1, characterized in that, The step of updating multiple network bandwidth resources currently configured by the base station using the reinforcement learning model based on the total quality assessment value obtained this time includes: The reinforcement learning model is used to determine the network bandwidth resources to be allocated based on the total quality assessment value obtained in this study. The network bandwidth resources of the first service type to be allocated from the plurality of network bandwidth resources are allocated to the second service type, wherein the first service type is the service type whose quality assessment value is greater than the second assessment threshold, and the second service type is the service type whose quality assessment value is less than the second assessment threshold.

3. The method according to claim 1, characterized in that, Before the monitoring base station executes the first communication parameters for the corresponding service type according to the currently configured multiple network bandwidth resources within the target time period, the method further includes: The monitoring base station executes the second communication parameters of the corresponding service type according to the multiple initial network bandwidth resources currently configured within the target time period; Based on the second communication parameters, the quality of services for each service type is evaluated to obtain the initial quality evaluation value for each service type. The initial quality assessment values ​​for each business type are summed to obtain the initial total quality assessment value.

4. The method according to claim 1, characterized in that, The initial network to be trained is a gated recurrent neural network (GRU).

5. The method according to claim 1, characterized in that, The step of extracting the traffic characteristics corresponding to each service type executed by the base station in the previous period adjacent to the period of the target time includes: Based on the dual-stream convolutional neural network, the sub-traffic features of the traffic corresponding to each service type executed by the base station in the previous period adjacent to the period in the target time period are extracted at each sampling time. The sub-traffic features at each sampling time are extracted based on the deep belief DBN network, and their temporal dependencies are determined. By integrating the sub-traffic features at each sampling time and the temporal dependencies of the sub-traffic features, the traffic features corresponding to each service type are obtained.

6. The method according to claim 5, characterized in that, The method further includes: The Seq2Seq method is used to denoise the traffic features corresponding to each of the aforementioned service types.

7. A network resource allocation device, characterized in that, The device includes: The data monitoring unit is used to monitor the first communication parameters of the base station when it executes the corresponding service type according to the multiple network bandwidth resources currently configured within the target time period; The quality assessment unit is used to perform quality assessments on the services of each service type executed by the base station according to the first communication parameters, and to obtain the quality assessment value of each service type. The data summation unit is used to sum the quality assessment values ​​of each business type to obtain the total quality assessment value. The resource allocation unit is used to update multiple network bandwidth resources currently configured by the base station based on the current total quality assessment value if the difference between the current total quality assessment value and the previous total quality assessment value is greater than a first assessment threshold. The reinforcement learning model is a dynamic resource allocation model based on a deep neural network, used to dynamically adjust network bandwidth resources according to the total quality assessment value. The feature extraction unit is used to extract the traffic features corresponding to each service type executed by the base station in the previous period adjacent to the period in the target time period. A traffic prediction unit is used to predict the traffic volume of each service type in the target time period based on the extracted traffic features corresponding to each service type using a pre-trained traffic prediction model. The pre-trained traffic prediction model is obtained by training an initial network to be trained using input data and output data. The output data is the traffic volume of each service type in the target time period over multiple historical periods, and the input data is the traffic features corresponding to each service type in the previous period adjacent to each historical period. The resource determination unit is used to determine the initial network bandwidth resources corresponding to each service type based on the predicted traffic volume of each service type. The resource configuration unit is used to configure the initial network bandwidth resources corresponding to each service type at the base station.

8. A server comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it causes the server to perform the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it causes the computer to perform the method as described in any one of claims 1 to 6.

Images