Cell function deployment method, device, medium and program product

Abstract

The invention provides a cell function deployment method and device, a medium and a program product. The method comprises the steps that first feature data of a target cell and a function strategy model corresponding to a scene of the target cell are acquired; inputting the first feature data into the function strategy model to obtain performance prediction values corresponding to multiple function strategies; and determining a target function strategy deployed to the target cell from the multiple function strategies according to the performance prediction value. According to the method provided by the embodiment of the invention, the target function strategy matched with the scene where the cell is located can be determined, so that the comprehensive performance of different cells can be optimal in different scenes, the throughput of the cell is improved, and the user experience is improved.

Description

Technical Field

[0001] This application relates to the field of communication technology, and in particular to a method, device, medium and program product for deploying cell functions. Background Technology

[0002] In wireless communication systems, the functional deployment of a physical cell is often determined after the cell is launched or after a version upgrade. Except for a small number of functions with adaptive adjustment capabilities, most functions adopt the same configuration strategy in different scenarios. However, due to factors such as user tidal effects, shifts between busy and idle periods, and changes in user channel statistical characteristics, the cell scenario may change significantly. The focus of cell performance observation will also change accordingly in different scenarios. If an unchanging functional configuration strategy is adopted, it will limit cell performance, leading to a decrease in cell spectral efficiency and a deterioration in user experience. Summary of the Invention

[0003] This application provides a cell function deployment method, electronic device, computer-readable storage medium, and computer program product, which aims to at least solve the technical problem of limiting cell performance by using a single-function configuration strategy.

[0004] In a first aspect, embodiments of this application provide a method for determining cell function strategies, including:

[0005] Obtain the first feature data of the target cell and the functional strategy model corresponding to the scenario of the target cell;

[0006] The first feature data is input into the functional strategy model to obtain performance prediction values ​​corresponding to various functional strategies.

[0007] Based on the performance prediction values, a target functional strategy for deployment to the target cell is determined from a variety of the aforementioned functional strategies.

[0008] Secondly, embodiments of this application provide an electronic device, including:

[0009] One or more processors;

[0010] A memory having stored one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method for determining a cell function policy as described in the first aspect of this application.

[0011] Thirdly, embodiments of this application provide a computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program, which, when executed by a processor, implements the method for determining cell function strategies as described in the first aspect of this application.

[0012] Fourthly, embodiments of this application provide a computer program product, characterized in that it includes a computer program, which, when executed by a processor, implements the method for determining cell function strategies as described in the first aspect of this application.

[0013] In this embodiment, first feature data and a functional strategy model of the target cell are obtained. The first feature data is input into the obtained functional strategy model to obtain the corresponding performance prediction value. Then, based on the performance prediction value, a target functional strategy to be deployed to the target cell is determined from multiple functional strategies. The method provided in this embodiment can determine a target functional strategy that is compatible with the scenario in which the cell is located, so that the overall performance of different cells can reach the optimal level in different scenarios, thereby improving cell throughput and enhancing user experience. Attached Figure Description

[0014] Figure 1 A flowchart illustrating a method for determining cell function strategies provided in an embodiment of this application;

[0015] Figure 2 A flowchart illustrating a method for determining a target cell scenario, provided in an embodiment of this application;

[0016] Figure 3 A schematic diagram illustrating the training process of a scene classification model provided in an embodiment of this application;

[0017] Figure 4 A schematic diagram of a process for obtaining a second training set provided in an embodiment of this application;

[0018] Figure 5 A flowchart illustrating a functional strategy model training method provided in an embodiment of this application;

[0019] Figure 6 A flowchart illustrating a method for determining cell function strategies provided in an embodiment of this application;

[0020] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0021] To enable those skilled in the art to better understand the technical solutions of this application, the technical solutions provided in this application will be described in detail below with reference to the accompanying drawings.

[0022] Exemplary embodiments will be described more fully below with reference to the accompanying drawings; however, the described exemplary embodiments may be embodied in different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this application will be thorough and complete, and will enable those skilled in the art to fully understand the scope of this application.

[0023] As used herein, the term “and / or” includes any and all combinations of one or more related enumerated entries.

[0024] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application. As used herein, the singular forms “a” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will also be understood that when the terms “comprising” and / or “made of” are used in this specification, the presence of a feature, integral, step, operation, element, and / or component is specified, but the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or groups thereof is not excluded.

[0025] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0026] Unless otherwise specified, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art. It will also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant art and this application, and will not be interpreted as having an idealized or overly formal meaning, unless expressly so defined in the embodiments of this application.

[0027] To facilitate a better understanding of the solutions in the embodiments of this application, the relevant technologies will be introduced first below.

[0028] The functional deployment of physical cells in a wireless communication system mainly revolves around the following aspects:

[0029] (1) Automatic configuration of physical cell identifier:

[0030] The Physical Cell Identifier (PCI) is a fundamental parameter in wireless network configuration used to distinguish wireless signals from different cells.

[0031] Due to the limited number of available PCI slots, PCI reuse will inevitably occur.

[0032] During the deployment of new base stations, it is necessary to use the Physical Cell Identifier (PCI) autoconfiguration capability to assign a PCI to each cell controlled by the base station.

[0033] Ensure that there are no identical PCIs within the coverage area of ​​the relevant cells, and that neighboring cells use different PCIs.

[0034] (2) Automatic neighbor relationship configuration:

[0035] When the wireless network topology changes (such as the addition / removal of a new base station or a base station malfunctioning and unable to operate), the base station needs to be able to automatically maintain neighbor relationships with surrounding base stations.

[0036] This is of great significance for inter-cell handover, load balancing, and interference coordination.

[0037] (3) Inter-cell interference coordination:

[0038] Due to the limited nature of wireless resources, especially frequency resources, it is necessary to reuse wireless resources within a small area.

[0039] Inter-cell interference can occur when neighboring cells use the same wireless resources.

[0040] Inter-cell interference coordination involves coordinating the use of radio resources by neighboring cells to avoid neighboring cells using the same frequency resources during the same time period, or to allow neighboring cells to use the same frequency resources when interference is acceptable.

[0041] (4) Wireless resource management:

[0042] As a unit of wireless resource management, a physical cell meets two conditions:

[0043] The cell is allocated channels to users as a whole, and users can only see one logical node providing services to them.

[0044] Regardless of where a user is located in the area, the probability of each channel providing service to the user is equal among all available channels in that area.

[0045] (5) Service portfolio and function decomposition:

[0046] When deploying in a community, it is also necessary to consider the impact of service composition on function splitting and deployment.

[0047] Different splitting methods can provide more optimization opportunities. For example, splitting method B allows different services to use different encoding technologies and allows joint decoding algorithms to effectively overcome interference.

[0048] (6) Deployment method and coverage:

[0049] Depending on the application scenario (such as hotspots, stadiums, large shopping malls, and airports), the deployment method and coverage of physical communities will vary.

[0050] For example, wide-area coverage using fiber optic deployments can enable centralized deployment of all wireless access network functions, achieving co-location with core network functions and maximizing cooperative diversity gain.

[0051] In summary, the functional deployment of physical cells in a wireless communication system is a comprehensive process that requires consideration of multiple aspects, including automatic configuration of physical cell identifiers, automatic neighbor relationship configuration, inter-cell interference coordination, radio resource management, service composition and function decomposition, as well as deployment methods and coverage.

[0052] Artificial intelligence (AI) models are models built using artificial intelligence techniques. They can improve the accuracy of their predictions and decisions through learning and training. Specifically, an AI model is a mathematical model based on artificial intelligence technology. It can learn and be trained on large amounts of data, enabling it to make predictions and decisions about new data. Such models can process various types of data, including text, images, and sound, and extract useful information from them to make corresponding predictions and decisions.

[0053] In wireless communication systems, the functional deployment of a physical cell is often determined after the cell is opened or after a version upgrade. Except for a small number of functions with adaptive adjustment, most functions adopt the same configuration strategy in different scenarios. However, due to factors such as the tidal effect of users in the cell, the transition between busy and idle periods, and changes in the statistical characteristics of user channels, the cell scenario may change significantly. The focus of cell performance observation will also change in different scenarios. If the unchanging functional configuration strategy is continued, it will limit cell performance, resulting in a decrease in cell spectrum efficiency and a deterioration in user experience.

[0054] To address the technical problem of limited cell performance due to uniform and fixed configuration strategies, this application provides a cell function deployment method, equipment, media, and program product. Based on scenario recognition, it intelligently deploys cell-level functions, enabling different cells to achieve optimal overall performance under different scenarios, thereby improving cell throughput and enhancing user experience.

[0055] It should be noted that the preset AI model provided in the embodiments of this application may include, but is not limited to, mainstream machine learning algorithm models, deep learning algorithm models, and large artificial intelligence models, such as decision tree models, multilayer perceptrons, support vector machines, radial basis function networks, recurrent neural networks, and convolutional neural networks (CNNs). The training methods include, but are not limited to, supervised learning, unsupervised learning, and reinforcement learning.

[0056] The method provided in this application embodiment can be applied to wireless access network equipment, including but not limited to: base stations (BTS) in Global System for Mobile Communications (GSM) or Code Division Multiple Access (CDMA), base stations (NodeB, NB) in Wideband Code Division Multiple Access (WCDMA), evolved NodeBs (eNBs), access points (APs), or relay stations in LTE networks, and base stations (gNBs) in NR networks, etc., without limitation.

[0057] Please refer to Figure 1 This is a flowchart illustrating a method for determining cell function strategies provided in an embodiment of this application. Figure 1 As shown, the method in this application embodiment may include, but is not limited to, the following steps S101 to S103:

[0058] Step S101: Obtain the first feature data of the target cell and the functional strategy model corresponding to the scenario of the target cell.

[0059] For example, the aforementioned first feature data can be determined based on the cell's associated feature information. This associated feature information may include, but is not limited to: a certain number of user channel measurement information within the cell, Reference Significant Receiving Power (RSRP), user distribution information such as user Direction of Arrival (DOA) distribution information, user energy projection distribution information, the number of users accessing the cell, the utilization rate of the cell's Physical Resource Block (PRB), neighboring cell interference information, functional policy information, and user historical scheduling information. In specific implementation, the cell's associated feature information can be subjected to joint statistical characteristic analysis to obtain the joint statistical characteristic analysis results, and then the first feature data can be constructed based on these results. The first feature data can specifically be in vector form. The aforementioned joint statistical characteristic analysis includes, but is not limited to: central tendency analysis, dispersion analysis, correlation analysis, normality test, regression analysis, variance analysis, cluster analysis, and time series analysis.

[0060] In one possible implementation of this application, the first feature data of the target cell is obtained by the following method: acquiring at least one of the following information of the target cell over a period of time (e.g., 15-minute granularity): the number of Radio Resource Control (RRC) access users, Physical Resource Block (PRB) utilization, average Channel Quality Indicator (CQI), Spectral Efficiency (SE), User-Aware Rate (RAR), or Neighbor Interference (NI) information; and constructing the first feature data based on the above information, for example, represented as data1 = [RRC_std, PRB_Ratio_std, CQI_std, NI_std].

[0061] In another possible implementation of this application, the first feature data of the target cell is obtained by the following method: obtaining an energy projection vector characterizing the degree of line-of-sight (LoS) / non-line-of-sight (NLOS) of users in the target cell, for example, data1 = [energy projection vector of 64 beams], and constructing the first feature data based on the energy projection vector of the target cell.

[0062] It should be noted that, in the actual implementation, the information used to construct the first feature data can be flexibly selected and adjusted according to the actual situation.

[0063] In this embodiment of the application, a functional strategy model corresponding to the scenario of the target cell is used to predict the performance of the target cell after deploying a preset functional strategy, wherein the functional strategy and the functional strategy model are corresponding.

[0064] In this embodiment of the application, the scene of the target cell can be set manually, determined by a preset algorithm, or predicted by an AI model based on the feature data of the cell.

[0065] Understandably, performance prediction can begin after obtaining the first feature data and the functional strategy model corresponding to the scenario of the target cell.

[0066] Step S102: Input the first feature data into the functional strategy model to obtain the performance prediction values ​​corresponding to various functional strategies.

[0067] Step S103: Based on the performance prediction value, determine the target functional strategy to be deployed to the target cell from multiple functional strategies.

[0068] Understandably, by inputting the first feature data into the functional strategy model, the model can simulate and predict the deployment of functional strategies based on the first feature data, obtaining performance prediction values. These performance prediction values ​​represent the cell performance when running the given functional strategy; a higher performance prediction value indicates better cell performance under that functional strategy. The performance prediction values ​​can then be used to select the most suitable functional strategy for the target cell in the current scenario.

[0069] In this embodiment, first feature data and a functional strategy model of the target cell are obtained. The first feature data is input into the obtained functional strategy model to obtain the corresponding performance prediction value. Then, based on the performance prediction value, the target functional strategy to be deployed to the target cell is determined from multiple functional strategies. The method provided in this embodiment enables intelligent deployment of cell-level functions based on specific scenarios, ensuring that different cells achieve optimal overall performance in different scenarios, thereby improving cell throughput and enhancing user experience.

[0070] It is understood that in this embodiment, the functional strategy needs to be determined based on the scenario of the target cell in order to achieve the optimal overall performance of the cell. Therefore, the scenario of the target cell needs to be determined before determining the functional strategy. Please refer to [reference needed]. Figure 2 This is a flowchart illustrating a method for determining a target cell scenario according to an embodiment of this application. Figure 2 As shown, the method for determining the target cell in this application embodiment may include, but is not limited to, steps S201 to S202:

[0071] Step S201: Obtain the second feature data of the target cell.

[0072] Step S202: Input the second feature data into the scene classification model to obtain the scene corresponding to the target cell.

[0073] Through such Figure 2 The scenario determination method for the target cell shown can determine the scenario in which the target cell is located, and then determine the target function strategy based on the scenario in which it is located.

[0074] For example, the second feature data of the target cell can be obtained as follows: Information such as the number of Radio Resource Control (RRC) users accessing the target cell over a period of time (e.g., 15-minute granularity), Physical Resource Block (PRB) utilization, Average Channel Quality (CQI), Cell Spectral Efficiency (SE), User-Aware Rate (Rate), Neighbor Interference (NI), and Functional Policy Labels (FPRs) is acquired. All or part of this information is combined into an input feature vector for scene classification. The input features are then standardized, and the second feature data is determined based on the standardized scene classification input feature vector. For instance, the number of RRC users accessing the target cell, PRB utilization, average CQI, and cell SE are acquired and denoted as data2 = [RRC_std, PRB_Rate_std, CQI_std], thus obtaining the second feature data.

[0075] For example, the second feature data of the target cell can also be obtained by: obtaining the energy projection vector characterizing the degree of line-of-sight (LoS) / non-line-of-sight (NLOS) of users in the target cell, for example, data2 = [energy projection vector of 64 beams], and constructing the second feature data based on the energy projection vector of the target cell.

[0076] For example, the scene classification model in this application embodiment is obtained through pre-training; please refer to... Figure 3 This is a schematic diagram illustrating the training process of a scene classification model provided in an embodiment of this application, as shown below. Figure 3 As shown, the training steps for the scene classification model in this embodiment may include, but are not limited to, steps S301 to S302:

[0077] Step S301: Obtain the first training set.

[0078] Step S302: Obtain the initial scene classification model. Perform the first training on the initial scene classification model based on the first training set to obtain the scene classification model.

[0079] For example, the first training set includes multiple first training sample data, which include, but are not limited to, at least one of the following data: number of Radio Resource Control (RRC) access users, Physical Resource Block (PRB) utilization, average Channel Quality (CQI), Cell Spectral Efficiency (SE), User-Aware Rate (Rate), Neighbor Interference (NI), Functional Policy Labels (FPRs), and Energy Projection Vectors.

[0080] For example, in one embodiment of this application, the training method for the scene classification model is as follows:

[0081] (1) Obtain a certain amount of user channel measurement information, RSRP, user DOA distribution information, user energy projection distribution information, number of users accessing the cell, cell PRB utilization rate, neighboring cell interference information, functional policy information, user historical scheduling information, and other related feature data within the cell. Perform joint statistical characteristic analysis on the obtained information, and screen and construct the first training set for training the scenario classification model. The joint statistical characteristic analysis may include, but is not limited to: central tendency analysis, dispersion analysis, correlation analysis, normality test, regression analysis, variance analysis, cluster analysis, time series analysis, etc. The specific joint statistical characteristic analysis method can be selected according to actual needs. This application does not impose too many restrictions on the selection of joint statistical characteristic analysis methods.

[0082] (2) Based on the constructed first training set, the initial scene classification model is trained for the first time to obtain the trained scene classification model.

[0083] It should be noted that the scene classification model used in the embodiments of this application can be an unsupervised clustering model (such as Kmeans clustering), a supervised scene recognition model (requiring data to have some scene labels), or a scene classification model that incorporates expert experience, etc.

[0084] It should be noted that the community scene classification can be a large-scale network-level classification of community scenes, or a single community-level scene classification.

[0085] In one possible implementation of this application, the training of the scene classification model can be performed as follows: Information such as the number of Radio Resource Control (RRC) users accessing the cell within a 15-minute granularity period, PRB utilization, average channel quality (CQI), cell spectral efficiency (SE), user perceived rate (Rate), neighboring cell interference (NI), and functional policy ID (FQI ID) is obtained. All or part of this information is combined into the first training sample data for scene classification. The input features are then standardized. For example, the standardized first training sample data for scene classification is denoted as: data1 = [RRC_std, PRB_Rate_std, CQI_std] (the scene classification input features can also be energy projection vectors used to characterize the Los / NLos level of users within the cell, such as data1 = [64-beam energy projection vector], which can be flexibly selected and adjusted according to the actual situation). A certain amount of data is collected to construct a dataset containing the scene classification input features, for example, 90,000 data points are collected to form a scene classification dataset. The 90,000 datasets are then used to train the cell scene classification using a clustering algorithm (e.g., K-means algorithm) to obtain a qualified scene classification model. The scene classification model required for the embodiments of this application can be trained using the above method.

[0086] After obtaining the trained scene classification model, a second training set can be constructed using the scene classification model and the first training set to train the functional policy model. Please refer to [link / reference]. Figure 4 This is a schematic diagram of a process for obtaining a second training set provided in an embodiment of this application, such as... Figure 4 As shown, obtaining the second training set may include, but is not limited to, the following steps S401 to S403:

[0087] Step S401: Determine the scene label corresponding to each first training sample data based on the scene classification model, wherein the scene label is used to characterize the type of scene.

[0088] Step S402: Determine the performance label corresponding to the first training sample data based on the scene label corresponding to the first training sample data.

[0089] Step S403: Based on all the first training sample data and the performance labels corresponding to the first training sample data, construct a second training set, wherein the second training set is used to train the functional policy model.

[0090] For example, taking the acquisition of a dataset of 90,000 data points as described above, after acquiring the dataset, it is input into a scene classification model for classification. Assume that the total number of classified scene labels is three: Scene 1, Scene 2, and Scene 3. The obtained scene labels are used to determine the performance labels corresponding to the first training sample data in subsequent step S402.

[0091] For example, in step S402, determining the performance label corresponding to the first training sample data based on the scene label corresponding to the first training sample data may include: obtaining the class center point vector corresponding to the scene label based on the scene label corresponding to the first training sample data; and determining the performance label corresponding to the first training sample data based on the class center point vector.

[0092] Specifically, in one embodiment of this application, 90,000 data points (i.e., the first training sample data) are divided into three categories according to scenarios: Scenario 1, Scenario 2, and Scenario 3. Assuming each scenario corresponds to 30,000 data points, the class center vectors after classification are denoted as follows: Scenario 1: K1 = [0.25, 0.15, 0.35]; Scenario 2: K2 = [0.5, 0.4, 0.55]; Scenario 3: K3 = [0.8, 0.9, 0.75]. The data in each scenario is then weighted according to the class center vector of the current scenario, with the cell spectral efficiency (SE) and user perceived rate (Rate) of each data point weighted to obtain the performance label New_SE corresponding to the first training sample data. An example of the weighting method is as follows:

[0093] Taking scenario 1 as an example, the class center point vector is: K1=[w1,w2,w3]=[0.25,0.15,0.35], then New_SE is: New_SE=w1*SE+w2*SE+(1-w1)*Rate+(1-w2)*Rate=0.25*SE+0.15*SE+0.75*Rate+0.85*Rate=0.4*SE+1.6*Rate.

[0094] For example, constructing a second training set based on all the first training sample data and the performance labels corresponding to the first training sample data may include: determining feature sample data based on the first training sample data, and forming second training sample data based on the feature sample data and performance labels corresponding to the first training sample data; determining the scene label and function policy label corresponding to the second training sample data based on the first training sample data corresponding to the second training sample data; classifying all the second training sample data according to the scene label and function policy label to obtain multiple second training sets, wherein the second training sample data in the second training set corresponds to the same scene label and the same function policy label.

[0095] Specifically, in one embodiment of this application, all first training sample data are first classified according to scenarios to obtain first training sample data for multiple scenarios. For the first training sample data of each scenario, the corresponding functional policy label is obtained. For example, if the functional policy label of the first training sample data is 0, it indicates that the first training sample data was obtained when the cell MU function was off in the corresponding scenario; if the functional policy label of the first training sample data is 1, it indicates that the first training sample data was obtained when the cell MU function was on in the corresponding scenario. All first training sample data in each scenario are classified according to the functional policy label, thus obtaining second training sample data by scenario and by functional policy. Second training sample data of the same scenario and the same functional policy constitute a second training set.

[0096] For example, the first training sample data corresponding to scenario 1 is represented as [RRC_std, PRB_Rat io_std, CQI_std, NI_std, function policy label]. Feature sample data [RRC_std, PRB_Rat io_std, CQI_std, NI_std] is extracted from the first training sample data. The feature sample data and the performance label New_SE of the first training sample data are combined to form a second training sample data, which can be denoted as data2 = [RRC_std, PRB_Rat io_std, CQI_std, NI_std, New_SE]. Specifically, when the function policy label is 0, it indicates that the cell MU function is off; when the function policy label is 1, it indicates that the cell MU function is on. For example, when data2 = [RRC_std = 20%, PRB_Rat io_std = 10%, CQI_std = 80%, NI_std = 13%, 0, 1.8], it means that in the preset scenario, the cell MU function is off, and the actual cell performance is 1.8. When data2 = [RRC_std = 19%, PRB_Rat io_std = 9%, CQI_std = 80%, NI_std = 13%, 1, 1.6], it means that in the preset scenario, the cell MU function is on, and the actual cell performance is 1.6.

[0097] Assuming there are three scenarios and two different functional strategies, namely functional strategy 1 and functional strategy 2, and the second training sample data in the second training set corresponds to the same scenario label and the same functional strategy label, then:

[0098] The training sample data for functional strategy 1 in scenario 1 is denoted as: data_S1_F1 = [RRC_std, PRB_Rat io_std, CQI_std, NI_std, New_SE];

[0099] The training sample data for functional strategy 2 in scenario 1 is denoted as: data_S1_F2=[RRC_std,PRB_Rat io_std,CQI_std,NI_std,New_SE];

[0100] The training sample data for functional strategy 1 in scenario 2 is denoted as: data_S2_F1=[RRC_std,PRB_Rat io_std,CQI_std,NI_std,New_SE];

[0101] The training sample data for functional strategy 2 in scenario 2 is denoted as: data_S2_F2=[RRC_std,PRB_Rat io_std,CQI_std,NI_std,New_SE];

[0102] The training sample data for functional strategy 1 in scenario 3 is denoted as: data_S3_F1=[RRC_std,PRB_Rat io_std,CQI_std,NI_std,New_SE];

[0103] The training sample data for functional strategy 2 in scenario 3 is denoted as: data_S3_F2 = [RRC_std, PRB_Rat io_std, CQI_std, NI_std, New_SE].

[0104] A second training set is formed by the second training sample data of the same scenario and the same functional strategy. For example, multiple training sample data data data_S1_F1 form the second training set data_S1_F1 Set. This training set can be used to train the functional strategy model corresponding to scenario 1 and functional strategy 1.

[0105] For example, each scenario has multiple corresponding functional strategy models, and each functional strategy model has a corresponding functional strategy label. For instance, scenario 1 includes functional strategy model 1A and functional strategy model 1B. The functional strategy label corresponding to functional strategy model 1A is 0, which indicates the cell MU function is disabled; the functional strategy label corresponding to functional strategy model 1B is 1, which indicates the cell MU function is enabled.

[0106] For example, the training method of the functional policy model in this application embodiment is as follows: An initial functional policy model and a second training set are obtained based on scene labels and functional policy labels; the initial functional policy model is trained a second time based on the second training set to obtain a functional policy model used to predict the performance of a target cell under the corresponding functional policy, thereby obtaining a performance prediction value. The second training set includes multiple second training sample data, which include feature sample data and performance labels.

[0107] The scene label corresponding to the functional policy model 1A is "Scene 1" and the corresponding functional policy label is "0". Then, the corresponding second training set data_S1_F1 Set is obtained.

[0108] Please refer to Figure 5 This is a flowchart illustrating a functional strategy model training method provided in an embodiment of this application. Figure 5 As shown in the embodiments of this application, the initial functional policy model is trained a second time based on the second training set to obtain the functional policy model, which may include, but is not limited to, steps S501 to S503:

[0109] Step S501: Input the feature sample data into the initial functional strategy model to obtain the performance prediction value corresponding to the feature sample data.

[0110] Step S502: Determine the prediction error between the performance prediction value and the performance label corresponding to the feature sample data based on the preset objective function.

[0111] Step S503: Train the initial functional strategy model based on the prediction error values ​​corresponding to each feature sample data to obtain the functional strategy model.

[0112] For example, in one embodiment of this application, an initial functional strategy model for the corresponding scenario is first selected, and a preset objective function for the initial functional strategy model is constructed. A neural network model can be selected as the initial functional strategy model for each functional strategy in each scenario. The objective function can be constructed by the following formula:

[0113]

[0114] Where G represents the prediction error value, New_SEi represents the true performance value, P_New_SE_i represents the performance prediction value, N represents the number of feature sample data, and W is a preset parameter.

[0115] After constructing the initial functional strategy model, the prediction error between the performance prediction value and the performance label corresponding to the feature sample data is determined based on the preset objective function.

[0116] Subsequently, the initial functional policy model is trained based on the prediction error values ​​corresponding to each feature sample data to obtain a functional policy model. For example, in the three scenarios and two different functional policies mentioned above, six different second training sets are obtained through classification. After training, six different functional policy models can be obtained, denoted as: AI_S1_F1, AI_S1_F2, AI_S2_F1, AI_S2_F2, AI_S3_F1, and AI_S3_F2. Among them, AI_S3_F1 is taken as an example, which represents the functional policy model of functional policy 1 in scenario 3.

[0117] In the embodiments of this application, each scenario has multiple corresponding functional strategy models, and each functional strategy model corresponds to a functional strategy.

[0118] In this embodiment of the application, the first feature data is input into the functional strategy model to obtain the performance prediction values ​​corresponding to various functional strategies, including: inputting the first feature data into each functional strategy model to obtain the performance prediction values ​​output by each functional strategy model.

[0119] It is understandable that multiple different functional strategy models are used in each different scenario to adapt to different time periods or conditions for functional strategy deployment. Among them, the performance prediction value is used to determine which functional strategy is most suitable for the target cell under the current condition in order to obtain the optimal performance.

[0120] In this embodiment of the application, determining the target functional strategy to be deployed to the target cell from multiple functional strategies based on the performance prediction value may include: determining the maximum performance prediction value from the performance prediction values ​​output by each functional strategy model; and determining the target functional strategy to be deployed to the target cell based on the functional strategy model corresponding to the maximum performance prediction value and the functional strategy corresponding to the functional strategy model.

[0121] For example, after model training is completed, the model is deployed and enters the model inference stage. Assume that three functional strategy models for different scenarios are deployed, with two functional strategy models for each scenario. Within an application cycle, the second feature data at a 15-minute granularity is obtained and input into the scenario classification model. First, the scenario classification model infers the cell scenario for the current application cycle; for example, the scenario classification model determines the current scenario to be scenario 2. Then, the two functional strategy models AI_S2_F1 and AI_S2_F2 under scenario 2 are applied. After inference by the two strategy sub-models, the performance labels after strategy deployment and the performance prediction values ​​P_New_SE1 and P_New_SE2 are obtained respectively. The one with the larger performance prediction value is selected as the optimal functional strategy; for example, if P_New_SE2 > P_New_SE1, then functional strategy 2 should be applied under scenario 2 within the current application cycle.

[0122] The solutions of this application embodiment are described below through specific application examples.

[0123] Please refer to Figure 6 This is a flowchart illustrating a method for determining cell function strategies provided in an embodiment of this application. Figure 6 As shown, the specific process of the method for determining cell function strategy in this embodiment includes a model training part and a model inference part, and may include, but is not limited to, the following steps:

[0124] Step S601: Within a preset time period, acquire the feature data of the cell at a preset time granularity (e.g., 15 min) as the first training sample data, and construct the first training set based on the acquired first training sample data.

[0125] The first training sample data includes at least one of the following: number of Radio Resource Control (RRC) access users, Physical Resource Block (PRB) utilization rate, average Channel Quality (CQI), Cell Spectral Efficiency (SE), User-Aware Rate (Rate), Neighbor Interference (NI), Functional Policy Labels (FPRs), and Energy Projection Vectors.

[0126] For example, the first training sample data is denoted as: data1 = [RRC_std, PRB_Ratio_std, CQI_std, NI_std, Functional Policy Label]. Where RRC_std represents the number of Radio Resource Control (RRC) access users, PRB_Ratio_std represents PRB utilization, CQI_std represents average channel quality, and NI_std represents neighboring cell interference level.

[0127] Step S602: Use the first training set to train the initial scene classification model for cell scene classification. The training method can be unsupervised training. After completing the scene classification training, the scene classification model is obtained and used for scene classification model inference in model inference.

[0128] In this example, we assume that there are a total of 3 categorized scene types: Scene 1, Scene 2, and Scene 3.

[0129] Step S603: The data in the first training set is split into scenarios. Assume the first training set has 90,000 data points, with 30,000 data points allocated to each scenario. Then, based on the segmented data for each scenario, the class centroid vector corresponding to each scenario is determined. Here, we assume the class centroid vectors for each scenario are:

[0130] Scenario 1: K1 = [0.25, 0.15, 0.35];

[0131] Scenario 2: K2 = [0.5, 0.4, 0.55];

[0132] Scenario 3: K3 = [0.8, 0.9, 0.75];

[0133] For each first training sample data, a weighted operation is performed based on the class center point vector of the scene corresponding to the first training sample data to determine the performance label corresponding to the first training sample data.

[0134] Step S604: Construct a second training set based on the first training set for training the functional policy model.

[0135] Obtain the first training sample data for each scenario. Based on the performance label New_SE corresponding to the first training sample data, convert the first training sample data data1 into data2 = [RRC_std, PRB_Ratio_std, CQI_std, NI_std, function policy label, New_SE].

[0136] When the function policy label is 0, it indicates that the cell MU function is off; when the function policy label is 1, it indicates that the cell MU function is on. For example, when data2 = [RRC_std = 20%, PRB_Rat io_std = 10%, CQI_std = 80%, NI_std = 13%, 0, 1.8], it means that in this scenario, the cell MU function is off, and the actual cell performance is 1.8. When the input vector data2 = [RRC_std = 19%, PRB_Rat io_std = 9%, CQI_std = 80%, NI_std = 13%, 1, 1.6], it means that in this scenario, the cell MU function is on, and the actual cell performance is 1.6.

[0137] Then, all data2 were classified according to scene labels and functional strategy labels to obtain multiple second training sets.

[0138] Suppose there are three scenarios and two different functional strategies, namely functional strategy 1 and functional strategy 2. The second training sample data in the second training set corresponds to the same scenario label and the same functional strategy label, for example:

[0139] The second training sample data for functional strategy 1 in scenario 1 is denoted as: data_S1_F1 = [RRC_std, PRB_Ratio_std, CQI_std, NI_std, New_SE];

[0140] The second training sample data for functional strategy 2 in scenario 1 is denoted as: data_S1_F2=[RRC_std,PRB_Ratio_std,CQI_std,NI_std,New_SE];

[0141] The second training sample data of functional strategy 1 in scenario 2 is denoted as: data_S2_F1=[RRC_std,PRB_Ratio_std,CQI_std,NI_std,New_SE];

[0142] The second training sample data for functional strategy 2 in scenario 2 is denoted as: data_S2_F2=[RRC_std,PRB_Ratio_std,CQI_std,NI_std,New_SE];

[0143] The second training sample data of functional strategy 1 in scenario 3 is denoted as: data_S3_F1=[RRC_std,PRB_Ratio_std,CQI_std,NI_std,New_SE];

[0144] The second training sample data for functional strategy 2 in scenario 3 is denoted as: data_S3_F2 = [RRC_std, PRB_Ratio_std, CQI_std, NI_std, New_SE].

[0145] Step S605: Select the initial functional strategy model for the corresponding scenario, and construct a preset objective function for the initial functional strategy model. A neural network model can be selected as the initial functional strategy model for each functional strategy in each scenario.

[0146] Step S606: Train the initial functional strategy model based on the prediction error value corresponding to each feature sample data to obtain the functional strategy model. For example, in the case of the three scenarios and two different functional strategies mentioned above, six different second training sets are obtained by classification. After training, six different functional strategy models can be obtained, which are denoted as: AI_S1_F1, AI_S1_F2, AI_S2_F1, AI_S2_F2, AI_S3_F1, AI_S3_F2. Taking AI_S3_F1 as an example, it represents the functional strategy model of functional strategy 1 in scenario 3.

[0147] Step 607: After the model training is completed, deploy the model by merging the trained scene classification model and functional strategy model. After deploying the model, you can start model inference for strategy prediction and intelligent deployment.

[0148] The model merging methods can include, but are not limited to, model cascading and model paralleling. Taking model cascading as an example, AI sub-model 1 (scene classification model) provides scene-specific label IDs for the current cell. After obtaining the label IDs, AI sub-model 2 (functional strategy model) for the corresponding scene is activated to perform model inference, thereby obtaining the optimal strategy for the current cell. Taking model paralleling as an example, input vectors are constructed separately. AI sub-model 1 and AI sub-model 2 first perform independent inference calculations, and finally, the inference results of the two models are combined to make the optimal functional strategy decision and deployment.

[0149] Step S608: Within an application cycle, acquire feature data at a 15-minute granularity for the current application cycle, and perform scenario classification model inference to obtain the cell scenario for the current application cycle. The feature data includes, but is not limited to: Radio Resource Control (RRC) access user count, Physical Resource Block (PRB) utilization rate, Average Channel Quality (CQI), Cell Spectral Efficiency (SE), User Aware Rate (Rate), Neighbor Interference (NI), Functional Policy Labels (FPR), and Energy Projection Vector.

[0150] Step S609: Select a functional strategy model based on the scenario type determined by the scenario classification, and determine the optimal functional strategy for the cell based on the performance prediction value output by the functional strategy model.

[0151] For example, if the current scenario is determined to be Scenario 2, then two functional strategy models, AI_S2_F1 and AI_S2_F2, are selected under Scenario 2. After inference, the two functional strategy models obtain the performance prediction values ​​P_New_SE1 and P_New_SE2 corresponding to the performance tags after strategy deployment. The one with the larger prediction value is the optimal functional strategy. For example, if P_New_SE2 > P_New_SE1, then functional strategy 2 should be applied under Scenario 2 within the current application cycle.

[0152] Step S610: Deploy the optimal function strategy for the cell.

[0153] Following the previous example, the current cycle is cell deployment function strategy 2.

[0154] The method for determining cell function strategies provided in this application embodiment can determine the target function strategy that is compatible with the scenario in which the cell is located, so that the comprehensive performance of different cells can reach the optimal level in different scenarios, thereby improving cell throughput and enhancing user experience.

[0155] This application also provides an electronic device, such as... Figure 7 As shown, the electronic device 1400 includes:

[0156] One or more processors 1410;

[0157] The memory 1420 stores one or more programs that, when executed by one or more processors 1410, cause the one or more processors 1410 to implement: the method for determining cell function policies provided in any embodiment of this application.

[0158] Memory 1420, as a non-transitory network system, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory 1420 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory 1420 may optionally include remotely located memories 1420 relative to processor 1410, which can be connected to processor 1410 via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0159] The memory 1420 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 1420 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1420 and is called and executed by the processor 1410.

[0160] The processor 1410 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application.

[0161] In some embodiments, the electronic device further includes:

[0162] Input / output interfaces are used to implement information input and output;

[0163] The communication interface is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0164] The bus transmits information between various components of the device (e.g., processor 1410, memory 1420, input / output interface, and communication interface);

[0165] The processor 1410, memory 1420, input / output interface, and communication interface can communicate with each other within the device via a bus.

[0166] An embodiment of this application also provides a computer-readable storage medium storing computer-executable instructions for performing the method for determining cell function strategies provided in any embodiment of this application.

[0167] An embodiment of this application also provides a computer program product, including a computer program or computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer program or computer instructions from the computer-readable storage medium and executes the computer program or computer instructions, causing the computer device to perform a method for determining a cell function strategy, as provided in any embodiment of this application.

[0168] The system architecture and application scenarios described in this application are intended to more clearly illustrate the technical solutions of this application and do not constitute a limitation on the technical solutions provided in this application. Those skilled in the art will understand that as system architectures evolve and new application scenarios emerge, the technical solutions provided in this application are also applicable to similar technical problems.

[0169] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0170] It will be understood by those skilled in the art that all or some of the steps and systems in the methods disclosed above can be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, as is known to those skilled in the art, communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0171] The above description, with reference to the accompanying drawings, illustrates some embodiments of this application, but does not limit the scope of the invention. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and spirit of this invention should be considered within the scope of this application.

Claims

1. A method for determining cell function strategies, comprising: Obtain the first feature data of the target cell and the functional strategy model corresponding to the scenario of the target cell; The first feature data is input into the functional strategy model to obtain performance prediction values ​​corresponding to various functional strategies. Based on the performance prediction values, a target functional strategy for deployment to the target cell is determined from a variety of the aforementioned functional strategies.

2. The method for determining cell function strategies according to claim 1, characterized in that, The scenario of the target cell is determined through the following steps: Obtain the second feature data of the target cell; The second feature data is input into the scene classification model to obtain the scene corresponding to the target cell.

3. The method for determining cell function strategies according to claim 2, characterized in that, The scene classification model is trained through the following steps: Obtain a first training set, wherein the first training set includes multiple first training sample data; An initial scene classification model is obtained, and the initial scene classification model is trained based on the first training set to obtain the scene classification model.

4. The method for determining cell function strategies according to claim 3, characterized in that, The first training sample data includes at least one of the following: number of Radio Resource Control (RRC) access users, Physical Resource Block (PRB) utilization, average Channel Quality (CQI), Cell Spectral Efficiency (SE), User-Aware Rate (Rate), Neighbor Interference (NI), Functional Policy Labels (FPRs), and Energy Projection Vectors.

5. The method for determining cell function strategies according to claim 3, characterized in that, After obtaining the scene classification model, the method includes: Based on the scene classification model, a scene label is determined for each of the first training sample data, wherein the scene label is used to characterize the type of the scene; Based on the scene label corresponding to the first training sample data, determine the performance label corresponding to the first training sample data; Based on all the first training sample data and the performance labels corresponding to the first training sample data, a second training set is constructed, wherein the second training set is used to train the functional policy model.

6. The method for determining cell function strategies according to claim 5, characterized in that, The construction of a second training set based on all the first training sample data and the performance labels corresponding to the first training sample data includes: Feature sample data is determined based on the first training sample data, and second training sample data is formed based on the feature sample data corresponding to the first training sample data and the performance label; Based on the first training sample data corresponding to the second training sample data, determine the scene label and function strategy label corresponding to the second training sample data; All the second training sample data are classified according to the scene label and the function policy label to obtain multiple second training sets, wherein the second training sample data in the second training set corresponds to the same scene label and the same function policy label.

7. The method for determining cell function strategies according to claim 5, characterized in that, The step of determining the performance label corresponding to the first training sample data based on the scene label corresponding to the first training sample data includes: Based on the scene label corresponding to the first training sample data, obtain the class center point vector corresponding to the scene label; The performance label corresponding to the first training sample data is determined based on the class center point vector.

8. The method for determining cell function strategies according to claim 1, characterized in that, Each scenario has multiple corresponding functional strategy models, and each functional strategy model has a corresponding functional strategy label. The functional strategy model is trained through the following steps: Obtain the initial functional strategy model and the second training set based on the scene labels and the functional strategy labels; The initial functional policy model is trained a second time based on the second training set to obtain the functional policy model.

9. The method for determining cell function strategies according to claim 8, characterized in that, The second training set includes multiple second training sample data, which include feature sample data and performance labels. The second training of the initial functional policy model based on the second training set to obtain the functional policy model includes: The feature sample data is input into the initial functional strategy model to obtain the performance prediction value corresponding to the feature sample data; The prediction error value between the performance prediction value and the performance label corresponding to the feature sample data is determined based on a preset objective function. The initial functional strategy model is trained based on the prediction error value corresponding to each of the aforementioned feature sample data to obtain the functional strategy model.

10. The method for determining cell function strategies according to claim 1, characterized in that, Each scenario has multiple corresponding functional strategy models, and each functional strategy model corresponds to a functional strategy. The step of inputting the first feature data into the functional strategy model to obtain performance prediction values ​​corresponding to multiple functional strategies includes: The first feature data is input into each of the functional strategy models to obtain the performance prediction value output by each of the functional strategy models.

11. The method for determining cell function strategies according to claim 10, characterized in that, The step of determining the target function strategy to be deployed to the target cell from a variety of function strategies based on the performance prediction value includes: The maximum performance prediction value is determined from the performance prediction values ​​output by each of the functional strategy models; Based on the functional strategy model corresponding to the maximum performance prediction value and the functional strategy corresponding to the functional strategy model, the target functional strategy to be deployed to the target cell is determined.

12. An electronic device, characterized in that, include: One or more processors; A memory having stored one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method for determining a cell function policy as described in any one of claims 1-11.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method for determining a cell function strategy as described in any one of claims 1-11.

14. A computer program product, characterized in that, Includes a computer program, which, when executed by a processor, implements the method for determining a cell function strategy as described in any one of claims 1-11.

Images