Cell configuration parameter optimization method, device, and electronic device

Abstract

The present application discloses a method and device for optimizing cell configuration parameters, which belongs to the field of communication technology. The method includes: clustering cells based on user distribution characteristics, and determining the cells included in each cluster obtained by clustering; learning the relationship between the configuration data and performance index data of the cells included in the target cluster, and obtaining a relationship model between the configuration data and performance index data corresponding to the target cluster, wherein the target cluster is a target cell belonging cluster, and the target cell matches a first load state; according to the relationship model, determining the configuration parameter optimization scheme of the target cell. This method, by combining big data processing methods such as machine learning, classifies the cell scenarios of the cells, and then mines the intrinsic relationship between the cell configuration parameters and performance indicators for each cell scenario through machine learning methods, and searches for the optimized configuration parameters of the high-load cells based on the mined intrinsic relationship, thereby improving the efficiency of cell configuration parameter optimization.

Description

Technical Field

[0001] The present application relates to the field of communication technology, and in particular to a method and apparatus for optimizing cell configuration parameters, as well as electronic equipment and a computer-readable storage medium. Background Art

[0002] Optimizing cell configuration parameters is crucial for improving network resource utilization and user experience. For example, in 4G networks, some cells with persistently high loads suffer from poor user experience due to limited network resources, necessitating timely reduction of the load in these cells. Existing techniques typically rely on manual verification to identify issues in these cells and provide targeted solutions. However, this manual optimization approach suffers from low efficiency.

[0003] It can be seen that the cell configuration parameter optimization method in the prior art still needs to be improved. Summary of the Invention

[0004] The embodiments of the present application provide a method and apparatus for optimizing cell configuration parameters, which can improve the efficiency of optimizing cell configuration parameters.

[0005] In a first aspect, an embodiment of the present application provides a method for optimizing cell configuration parameters, including:

[0006] Clustering cells based on user distribution characteristics and determining cells included in each cluster obtained by clustering;

[0007] Learning a relationship between configuration data and performance indicator data of cells included in a target cluster to obtain a relationship model between the configuration data and the performance indicator data corresponding to the target cluster, wherein the target cluster is a target cell belonging cluster, and the target cell matches a first load state;

[0008] A configuration parameter optimization scheme for the target cell is determined based on the relationship model.

[0009] Optionally, determining a configuration parameter optimization scheme for the target cell according to the relationship model includes:

[0010] Predicting performance indicators corresponding to the adjusted configuration parameters of similar cells with reference to the target cell using the relationship model, searching for local optimal configuration parameters using a hill climbing algorithm, and determining optimized configuration parameters of the target cell, wherein the similar cells are cells in the target cluster that match the second load state;

[0011] A configuration parameter optimization scheme for the target cell is determined based on the optimized configuration parameters.

[0012] Optionally, estimating, by using the relationship model, performance indicators corresponding to the adjusted configuration parameters of similar cells with reference to the target cell, and using a hill climbing algorithm to search for local optimal configuration parameters to determine the optimized configuration parameters of the target cell includes:

[0013] Determining a current node based on a target cell, and determining a neighbor node based on similar cells to the target cell;

[0014] Obtaining the current node performance indicator corresponding to the configuration parameters of the current node through the relationship model;

[0015] In response to the current node performance indicator being the optimal performance indicator corresponding to the neighboring node, determining the configuration parameters of the current node as the optimized configuration parameters of the target cell;

[0016] In response to the current node performance indicator not being the optimal performance indicator corresponding to the neighboring node, the current node is replaced by the target neighboring node, and the process jumps to the step of obtaining the current node performance indicator corresponding to the configuration parameters of the current node through the relationship model, and performing iterative climbing until the number of iterative climbing reaches a preset iteration threshold, and determining that the configuration parameters corresponding to the current node are the optimized configuration parameters of the target cell, wherein the target neighboring node is: the neighboring node corresponding to the similar cell having the highest similarity with the cell operation scenario corresponding to the current node.

[0017] Optionally, before determining the current node according to the target cell and determining the neighbor node according to similar cells of the target cell, the method further includes:

[0018] Determining the similarity of the operating scenarios of the target cell and each of the similar cells;

[0019] After determining the configuration parameter optimization solution of the target cell according to the relationship model, the method further includes:

[0020] Outputting a configuration parameter optimization solution for the target cell, and / or outputting the similarity between the target cell and the operation scenarios of each of the similar cells.

[0021] Optionally, the learning of the relationship between the configuration data and the performance indicator data of the cells included in the target cluster to obtain a relationship model between the configuration data and the performance indicator data corresponding to the target cluster includes:

[0022] constructing an input matrix according to eigenvalues ​​of preset configuration parameters of cells included in the target cluster, and constructing an output matrix according to eigenvalues ​​of preset performance indicators of cells included in the target cluster;

[0023] A machine learning model for solving multivariate regression problems is used to learn the relationship between the input matrix and the output matrix, and obtain a relationship model between the configuration data and performance indicator data corresponding to the target cluster.

[0024] Optionally, clustering cells based on user distribution characteristics and determining cells included in each cluster obtained by clustering includes:

[0025] Obtaining preset user distribution characteristics of each of the cells;

[0026] Based on the preset user distribution characteristics, similarity distances between corresponding cells are calculated, and K-means clustering is performed on the cells based on the similarity distances to determine multiple clusters, and the cells included in each cluster are determined.

[0027] Optionally, obtaining preset user distribution characteristics of each cell includes:

[0028] Based on the measurement report data of each cell within a specified time period, the preset user distribution characteristics of the corresponding cell are obtained, wherein the preset user distribution characteristics include one or more of the following: the arithmetic mean of the cell uplink utilization rate, the arithmetic mean of the cell downlink utilization rate, the arithmetic mean of the maximum number of users, the variance of the number of users, the variance of the flow rate, the kurtosis coefficient of the number of users, the kurtosis coefficient of the cell uplink utilization rate, the kurtosis coefficient of the cell downlink utilization rate, the flow rate kurtosis coefficient, the skewness coefficient of the number of users, the skewness coefficient of the cell uplink utilization rate, the skewness coefficient of the cell downlink utilization rate, and the flow rate skewness coefficient.

[0029] In a second aspect, an embodiment of the present application provides a cell configuration parameter optimization device, including:

[0030] A cell clustering module is used to cluster cells based on user distribution characteristics and determine the cells included in each cluster obtained by clustering;

[0031] a configuration parameter and performance relationship model training module, configured to learn the relationship between configuration data and performance indicator data of cells included in a target cluster, and obtain a relationship model between the configuration data and performance indicator data corresponding to the target cluster, wherein the target cluster is a target cell belonging cluster, and the target cell matches a first load state;

[0032] The configuration parameter optimization solution determination module is used to determine the configuration parameter optimization solution of the target cell according to the relationship model.

[0033] Optionally, the configuration parameter optimization solution determination module is further configured to:

[0034] Predicting performance indicators corresponding to the adjusted configuration parameters of similar cells with reference to the target cell using the relationship model, searching for local optimal configuration parameters using a hill climbing algorithm, and determining optimized configuration parameters of the target cell, wherein the similar cells are cells in the target cluster that match the second load state;

[0035] A configuration parameter optimization scheme for the target cell is determined based on the optimized configuration parameters.

[0036] Optionally, estimating, by using the relationship model, performance indicators corresponding to the adjusted configuration parameters of similar cells with reference to the target cell, and using a hill climbing algorithm to search for local optimal configuration parameters to determine the optimized configuration parameters of the target cell includes:

[0037] Determining a current node based on a target cell, and determining a neighbor node based on similar cells to the target cell;

[0038] Obtaining the current node performance indicator corresponding to the configuration parameters of the current node through the relationship model;

[0039] In response to the current node performance indicator being the optimal performance indicator corresponding to the neighboring node, determining the configuration parameters of the current node as the optimized configuration parameters of the target cell;

[0040] In response to the current node performance indicator not being the optimal performance indicator corresponding to the neighboring node, the current node is replaced by the target neighboring node, and the process jumps to the step of obtaining the current node performance indicator corresponding to the configuration parameters of the current node through the relationship model, and performing iterative climbing until the number of iterative climbing reaches a preset iteration threshold, and determining that the configuration parameters corresponding to the current node are the optimized configuration parameters of the target cell, wherein the target neighboring node is: the neighboring node corresponding to the similar cell having the highest similarity with the cell operation scenario corresponding to the current node.

[0041] Optionally, before determining the current node according to the target cell and determining the neighbor node according to similar cells of the target cell, the method further includes:

[0042] Determining the similarity of the operating scenarios of the target cell and each of the similar cells;

[0043] After determining the configuration parameter optimization solution of the target cell according to the relationship model, the method further includes:

[0044] Outputting a configuration parameter optimization solution for the target cell, and / or outputting the similarity between the target cell and the operation scenarios of each of the similar cells.

[0045] Optionally, the configuration parameter and performance relationship model training module is further used to:

[0046] constructing an input matrix according to eigenvalues ​​of preset configuration parameters of cells included in the target cluster, and constructing an output matrix according to eigenvalues ​​of preset performance indicators of cells included in the target cluster;

[0047] A machine learning model for solving multivariate regression problems is used to learn the relationship between the input matrix and the output matrix, and obtain a relationship model between the configuration data and performance indicator data corresponding to the target cluster.

[0048] Optionally, the cell clustering module is further configured to:

[0049] Obtaining preset user distribution characteristics of each of the cells;

[0050] Based on the preset user distribution characteristics, similarity distances between corresponding cells are calculated, and K-means clustering is performed on the cells based on the similarity distances to determine multiple clusters, and the cells included in each cluster are determined.

[0051] Optionally, obtaining preset user distribution characteristics of each cell includes:

[0052] Based on the measurement report data of each cell within a specified time period, the preset user distribution characteristics of the corresponding cell are obtained, wherein the preset user distribution characteristics include one or more of the following: the arithmetic mean of the cell uplink utilization rate, the arithmetic mean of the cell downlink utilization rate, the arithmetic mean of the maximum number of users, the variance of the number of users, the variance of the flow rate, the kurtosis coefficient of the number of users, the kurtosis coefficient of the cell uplink utilization rate, the kurtosis coefficient of the cell downlink utilization rate, the flow rate kurtosis coefficient, the skewness coefficient of the number of users, the skewness coefficient of the cell uplink utilization rate, the skewness coefficient of the cell downlink utilization rate, and the flow rate skewness coefficient.

[0053] In a third aspect, an embodiment of the present application further discloses an electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor implements the cell configuration parameter optimization method described in the embodiment of the present application when executing the computer program.

[0054] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps of the cell configuration parameter optimization method disclosed in the embodiment of the present application are performed.

[0055] The cell configuration parameter optimization method disclosed in the embodiment of the present application clusters cells based on user distribution characteristics to determine the cells included in each cluster obtained by clustering; learns the relationship between the configuration data and performance indicator data of the cells included in the target cluster to obtain a relationship model between the configuration data and performance indicator data corresponding to the target cluster, wherein the target cluster is a target cell belonging cluster, and the target cell matches a first load state; based on the relationship model, determines a configuration parameter optimization plan for the target cell, and can automatically output a configuration parameter optimization plan for the target cell, thereby improving the optimization efficiency of the cell configuration parameters.

[0056] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0058] Figure 1 This is one of the flow charts of the cell configuration parameter optimization method in the embodiment of the present application;

[0059] Figure 2 This is the second flow chart of the cell configuration parameter optimization method in the embodiment of the present application;

[0060] Figure 3 It is a flowchart of the steps of determining the configuration parameter optimization solution in the cell configuration parameter optimization in the embodiment of the present application;

[0061] Figure 4 Schematic diagram of the structure of the cell configuration parameter optimization device in the embodiment of the present application;

[0062] Figure 5 A block diagram schematically shows an electronic device for executing the method according to the present application; and

[0063] Figure 6 The figure schematically shows a storage unit for storing or carrying a program code for implementing the method according to the present application. DETAILED DESCRIPTION

[0064] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0065] The cell configuration parameter optimization method disclosed in the embodiment of the present application is applied to the identification of high-load cells, and automatically provides targeted solutions for the performance problems existing in high-load cells. For example, for high-load cells of the indoor type, a solution for cell splitting or expansion is provided; for macro-station cells, when the proportion of access outside the TA of one times the station spacing of the cell is greater than the preset proportion threshold, a solution for lowering the tilt angle is provided; when the number of base station cells is less than the preset number threshold, a solution for increasing sectors is provided, etc. For some complex situations, for example, for macro stations with different frequencies in the same direction at the same station, it is necessary to further analyze the causes of the high load of the cell based on the specific configuration parameters of the cell, and provide targeted solutions. Based on this demand, a cell configuration parameter optimization method disclosed in the embodiment of the present application combines big data processing and machine learning technology to provide a configuration parameter optimization solution for high-load cells.

[0066] like Figure 1 As shown, a cell configuration parameter optimization method disclosed in an embodiment of the present application includes: steps 110 to 130.

[0067] Step 110: cluster the cells based on the user distribution characteristics, and determine the cells included in each cluster obtained by clustering.

[0068] In an embodiment of the present application, cells are first clustered based on user distribution characteristics, and cells with similar cell scenarios are clustered into a cluster (i.e., a cell cluster), so as to facilitate subsequent analysis of cell data in the cluster corresponding to the cell scenario for different cell scenarios and obtain the relationship between the cell configuration parameters and performance indicators.

[0069] The user distribution characteristics are characteristics that can express the size of access traffic in a cell, the distribution trend of access volume based on time, the change trend, etc.

[0070] The distribution range of cells used for clustering is determined based on service requirements and includes cells matching various load states within a specified area. For example, these include: high-load cells (i.e., cells matching the first load state described below) and non-high-load cells (i.e., cells matching the second load state described below).

[0071] For example, when it is necessary to analyze the load status of cells in a certain province in the past week, all cells in the province can be used as cells to be clustered.

[0072] After clustering the cells, the cells are divided into multiple clusters based on similarity of user distribution characteristics, and each cluster may include one or more cells.

[0073] Step 120 , learning the relationship between the configuration data and performance indicator data of the cells included in the target cluster, and obtaining a relationship model between the configuration data and performance indicator data corresponding to the target cluster, wherein the target cluster is a target cell belonging cluster, and the target cell matches the first load state.

[0074] In the embodiments of the present application, the cell that needs to be optimized is referred to as a "target cell." In the embodiments of the present application, a high load state is defined as a "first load state." The target cell in the embodiments of the present application is a cell in a high load state, i.e., a target cell that matches the first load state.

[0075] In the embodiment of the present application, the configuration parameters of the corresponding high-load cell are automatically optimized by referring to the configuration data of non-high-load cells with similar cell scenarios as the high-load cell, thereby achieving automatic optimization of the high-load cell. Therefore, when it is necessary to determine the optimized configuration parameters of the target cell, it is further necessary to mine the inherent relationship between the configuration data and performance indicators of non-high-load cells with similar cell scenarios as the target cell, so as to optimize the configuration parameters of the target cell based on the mined relationship.

[0076] Among them, cells with similar cell scenarios to the high-load cell are cells that are clustered into the same cluster as the target cell.

[0077] In an embodiment of the present application, the cells clustered into the same cluster based on user distribution characteristics may include both cells matching the first load state (i.e., high-load cells) and cells matching the second load state (i.e., normal-load cells, or low-load cells).

[0078] In some embodiments of the present application, machine learning technology can be used to automatically learn the inherent relationship between the configuration parameters and performance indicators of the cell that matches the second load state (such as the low load state) by training a machine learning model. In this way, the trained machine learning model can be used as a relationship model between the preferred configuration parameters and performance indicators to express the relationship between the cell configuration parameters and performance indicators.

[0079] In some embodiments of the present application, the inherent relationship between the cell configuration parameters and performance indicators can also be automatically learned by training a machine learning model based on the relationship between the configuration data and performance indicator data of all cells included in a cluster. During the learning process, cells matching the first load state are used as negative samples, and cells matching the second load state are used as positive samples. In this way, the ability of the trained relationship model to express the inherent relationship between the configuration parameters and performance indicators of cells matching the second load state (e.g., low load state) can be effectively verified.

[0080] Step 130: Determine a configuration parameter optimization solution for the target cell based on the relationship model.

[0081] After obtaining a determined relationship model based on the configuration parameters and performance indicators of the cells in the cluster to which the target cell belongs, the corresponding performance indicators can be estimated for a given set of configuration parameters. Therefore, by adjusting the configuration parameters of the target cell, the performance indicators corresponding to the adjusted configuration parameters can be estimated, thereby determining the preferred configuration parameters for the target cell to enter the second load state (such as a low load state).

[0082] Afterwards, according to the determined preferred configuration parameters, the corresponding configuration parameter optimization scheme is output.

[0083] The cell configuration parameter optimization method disclosed in the embodiment of the present application clusters cells based on user distribution characteristics to determine the cells included in each cluster obtained by clustering; learns the relationship between the configuration data and performance indicator data of the cells included in the target cluster to obtain a relationship model between the configuration data and performance indicator data corresponding to the target cluster, wherein the target cluster is a target cell belonging cluster, and the target cell matches a first load state; based on the relationship model, determines a configuration parameter optimization plan for the target cell, and can automatically output a configuration parameter optimization plan for the target cell, thereby improving the optimization efficiency of the cell configuration parameters.

[0084] The cell configuration parameter optimization method disclosed in the embodiment of the present application, by combining big data processing methods such as machine learning, classifies cells into cell scenarios, then uses machine learning methods to mine the intrinsic relationship between cell configuration parameters and performance indicators for each cell scenario, and searches for optimized configuration parameters for high-load cells based on the mined intrinsic relationship, and automatically outputs a configuration parameter optimization solution. This can quickly output an optimization solution for high-load cells, thereby improving the efficiency of cell configuration parameter optimization. Furthermore, by quickly and accurately reducing the pressure for high-load cells, network resource utilization is improved.

[0085] To facilitate readers' understanding of this solution, the specific implementation methods of each step of the cell configuration parameter optimization method disclosed in the embodiment of this application are further explained below.

[0086] The cells that need to be clustered in the aforementioned step 110 are cells determined based on specific service requirements. Figure 2 As shown, before clustering cells based on user distribution characteristics and determining cells included in each cluster obtained by clustering, the process further includes: step 100, step 102 and step 104.

[0087] Step 100: Obtain association data of cells within a specified area within a specified time period.

[0088] In an embodiment of the present application, cells are clustered based on the user distribution characteristics of each cell. Before implementing this solution, it is first necessary to obtain cell-level data within a specified time period. For example, first obtain the original data such as the working parameter data (such as cell azimuth, piezoelectric downtilt, azimuth, cell user distribution distance, etc.), measurement report (MR) data (such as cell uplink utilization, downlink utilization, RRC user number, etc.), cell configuration data (such as access RRU information, etc.), network management performance data, etc. of each cell in a specified area within a specified time period.

[0089] Step 102: extract and convert the associated data to obtain formatted data.

[0090] The formatted data is used for cell clustering and relationship model training. Each piece of the formatted data corresponds to a cell. The formatted data includes: characteristic values ​​of each preset user distribution characteristic, characteristic values ​​of preset configuration parameters, characteristic values ​​of preset performance indicators, and the cell name, etc.

[0091] After obtaining the associated data of cells in a specified area within a specified time period, according to the cell configuration parameter optimization disclosed in the embodiment of the present application, data items required for operations such as cell clustering, relationship model training, and generation of configuration parameter optimization solutions are performed, and the raw data obtained above is subjected to feature extraction, conversion, and formatting through a data processing platform to obtain formatted data.

[0092] The specific technical means for extracting features, converting and formatting the cell-related data can adopt existing technologies and will not be described in detail here.

[0093] Step 104: Determine the load status and scenario type matching each cell based on the formatted data.

[0094] In the embodiments of the present application, the scenario types include, but are not limited to, high-traffic scenarios, low-traffic scenarios, multi-user scenarios, and low-user scenarios. In the embodiments of the present application, the load states include, but are not limited to, a first load state and a second load state. As previously mentioned, the first load state refers to a high-load state, and the second load state is a non-high-load state, such as a low-load state or a normal load state.

[0095] In some embodiments of the present application, the load status and scenario type that each cell matches can be determined based on the formatted data according to preset rules. For example, the load status and scenario type that each cell matches can be determined based on a load-related KPI (key performance indicator) in the formatted data.

[0096] For example, the judgment rules for traffic scenarios include: for high-frequency cells (for example, cells with a frequency of 1.8G or 2.1G), the PRB (Physical Resource Block) utilization is greater than or equal to a preset utilization threshold (such as 70%), and the PDCP (Packet Data Convergence Protocol) layer traffic is greater than or equal to a first preset traffic threshold (such as 11.2G); for low-frequency cells (for example, cells with a frequency of 800M), the PRB utilization is greater than or equal to the preset utilization threshold, and the PDCP layer traffic is greater than or equal to a second preset traffic threshold (such as 2.8G). For another example, the judgment rules for multi-user scenarios include: for high-frequency cells, the PRB (Physical Resource Block) utilization rate is greater than or equal to the preset utilization rate threshold, and the number of RRC (Radio Resource Control) connected users is greater than or equal to the first preset user number threshold (such as 280); for low-frequency cells (for example, cells with a frequency of 800M), the PRB utilization rate is greater than or equal to the preset utilization rate threshold, and the number of RRC connected users is greater than or equal to the second preset user number threshold (such as 70).

[0097] According to this method, it can be determined whether a cell is a high-traffic scenario or a multi-user scenario.

[0098] Furthermore, it can be configured that if a cell meets the aforementioned high-traffic scenario or multi-user scenario on one or a specified number of days (e.g., four days) during the busy hours each week, or meets both the high-traffic scenario and the multi-user scenario, the cell is considered to be a high-load cell, i.e., the cell matches the first load state. Otherwise, the cell is considered to match the second load state.

[0099] According to this method, the load status and the matching scenario type of each cell can be determined.

[0100] In the embodiment of the present application, a correspondence between cells, matching load states, scenario types, formatted data, etc. can be established by setting labels for the formatted data corresponding to each cell.

[0101] After the data preparation is completed, the cells are clustered based on the user distribution characteristics, and the cells included in each cluster obtained by clustering are determined.

[0102] In some embodiments of the present application, clustering cells based on user distribution characteristics and determining the cells included in each cluster obtained by clustering includes: obtaining preset user distribution characteristics of each of the cells; calculating the similarity distance between the corresponding cells based on the preset user distribution characteristics, and performing K-means clustering on the cells based on the similarity distance to determine multiple clusters, and determining the cells included in each of the clusters.

[0103] The MR data of a cell can better reflect the user distribution of the cell. By analyzing and processing the MR data, the potential similarity patterns of the cell can be mined.

[0104] In some embodiments of the present application, obtaining the preset user distribution characteristics of each of the cells includes: obtaining the preset user distribution characteristics of the corresponding cell based on the measurement report data of each cell within a specified time period, wherein the preset user distribution characteristics include one or more of the following: the arithmetic mean of the cell uplink utilization, the arithmetic mean of the cell downlink utilization, the arithmetic mean of the maximum number of users, the variance of the number of users, the variance of the traffic, the kurtosis coefficient of the number of users, the kurtosis coefficient of the cell uplink utilization, the kurtosis coefficient of the cell downlink utilization, the traffic kurtosis coefficient, the skewness coefficient of the number of users, the skewness coefficient of the cell uplink utilization, the skewness coefficient of the cell downlink utilization, and the traffic skewness coefficient.

[0105] The method for extracting the feature values ​​of each of the above-mentioned preset user distribution features is described in the prior art and will not be further described in detail in the embodiments of this application. Those skilled in the art will understand that the above-mentioned preset user distribution features are only some of the features that can be used to express the user distribution of a cell, and not all of them. In specific implementations, the preset user distribution features can be determined based on the regional characteristics of the cell distribution, and then the feature values ​​of the selected preset user distribution features can be calculated based on the MR data of the cell.

[0106] After determining the characteristic value of the preset user distribution feature of each cell, the similarity between two cells is further calculated based on the characteristic value of the preset user distribution feature, and the cells are iteratively clustered based on the calculated similarity.

[0107] In some embodiments of the present application, a K-means clustering algorithm can be used to cluster the acquired cells to achieve scenario segmentation based on the user distribution feature dimension. The clustering process is described below using an example where the number of cells is N and the preset user distribution features include M types.

[0108] In some embodiments of the present application, the number of cells to be clustered can be represented as a data matrix with M rows and N columns, where each row of the matrix corresponds to a cell, the matrix elements in each row represent the aforementioned user distribution characteristics of the corresponding cell, and each matrix element in each row is considered a data object. During the clustering process, K data objects are first randomly selected from the data matrix as initial cluster centers (i.e., K cells are randomly selected as cluster centers). Then, the distance between the data objects corresponding to the matrix elements in each row and the cluster centers is calculated, and the data objects are assigned to the cluster containing the cluster center closest to them.

[0109] In some embodiments of the present application, the distance between data objects may be calculated using the Euclidean distance. For example, the distance between data objects may be calculated using the following formula:

[0110]

[0111] Among them, x i and x j The labels represent the matrix elements in the i-th row and the j-th row in the data matrix, that is, the i-th data object and the j-th data object, respectively. i,d represents the dth user distribution feature of the ith data object, corresponding to the element value of the ith row and dth column in the data matrix, x j,d represents the dth user distribution feature of the jth data object, corresponding to the element value of the jth row and dth column in the data matrix, dist(x i , x j ) represents the Euclidean distance between the i-th and j-th data objects. Based on the above distance calculation method, we can conclude that the distance between the i-th and j-th data objects reflects the similarity between the cell corresponding to the i-th and j-th data objects. After a round of clustering, cells with high similarity to each cell serving as a cluster center are clustered into clusters centered on that cell.

[0112] Next, update the cluster center. For example, you can take the mean of all data objects in the corresponding cluster as the cluster center of the updated cluster. Taking a cluster including G data objects as an example, the cluster center of the kth cluster can be calculated by the following formula: k :

[0113]

[0114] Among them, Gk is the total number of data objects in the kth cluster (corresponding to the total number of cells), C k is the set of data objects in the kth cluster, x i is the data object in the kth cluster. The summation refers to the summation of the eigenvalues ​​of each feature dimension of all data objects in cluster k. Therefore, the cluster center is a vector containing M eigenvalues. For example, it is represented as Center k =(Center k,1 ,Center k,2 ,…,Center k,M ).

[0115] The clusters are redivided and the cluster centers are updated through continuous iteration until the set number of iterations is reached. The clusters obtained at this time are the final clustering results.

[0116] Next, based on the configuration parameters and performance indicator data of the cells included in each cluster, a relationship model between the configuration data and performance indicators of the cells in each cluster can be obtained.

[0117] In the aforementioned step 120, the relationship between the configuration data and performance indicator data of the cells included in the target cluster is learned to obtain a relationship model between the configuration data and performance indicator data corresponding to the target cluster, including: constructing an input matrix based on the eigenvalues ​​of the preset configuration parameters of the cells included in the target cluster, and constructing an output matrix based on the eigenvalues ​​of the preset performance indicators of the cells included in the target cluster; using a machine learning model for solving multivariate regression problems to learn the relationship between the input matrix and the output matrix to obtain a relationship model between the configuration data and performance indicator data corresponding to the target cluster.

[0118] In some embodiments of the present application, the preset configuration parameters include one or more of the following: resource configuration parameters, frequency configuration parameters, and boundary level configuration parameters. Specifically, the resource configuration parameters, frequency configuration parameters, and boundary level configuration parameters are determined based on specific service requirements, and the characteristic value of each specific parameter can be obtained from formatted data obtained by processing raw data such as the engineering parameter data, cell configuration data, MR data, and network management performance data as described above.

[0119] In some embodiments of the present application, the preset performance indicators include one or more of the following: the maximum number of RRC connected users, uplink PRB utilization, and downlink PRB utilization. The characteristic values ​​of the preset performance indicators can be obtained from the formatted data obtained by processing the raw data such as the engineering parameter data, cell configuration data, MR data, and network management performance data as described above.

[0120] Taking the example of a cluster containing N cells where the target cell resides, with P preset configuration parameters for cell coverage within the entire cluster and Q preset performance indicators, a parameter matrix with N rows and P columns and a performance indicator matrix with N rows and Q columns can be generated based on the configuration data of the N cells within the cluster containing the target cell. In the parameter matrix, each row of the matrix elements corresponds to a cell, and each column of the matrix elements corresponds to an eigenvalue of a configuration parameter. In the performance indicator matrix, each row of the matrix elements corresponds to a cell, and each column of the matrix elements corresponds to an eigenvalue of a performance indicator.

[0121] Next, based on the parameter matrix and the performance indicator matrix, a relationship model expressing the relationship between the configuration parameters and the performance indicators is trained.

[0122] In an embodiment of the present application, when studying the relationship between performance indicators and configuration parameters, this problem is modeled as a multivariate regression problem. An algorithmic model is used to learn the intrinsic relationship of the regression problem, and the model is trained using training data. For example, a parameter matrix is ​​used as an input matrix, and a performance indicator matrix is ​​used as an output matrix to train a multivariate regression model to obtain a relational model that expresses the relationship between the preset configuration parameters and the preset performance indicators. During the relational model training process, the data in the input matrix (i.e., the parameter matrix) can be converted into a column matrix as the input of the machine learning model. Similarly, the output matrix (i.e., the converted performance indicator matrix) is converted into a column matrix as the output of the machine learning, thereby learning the intrinsic mapping relationship between the combination of various configuration parameters and the combination of various performance indicators.

[0123] After training a relationship model between configuration parameters and performance indicators, the corresponding performance indicators can be estimated for a given set of configuration parameters using the mapping relationships learned from the relationship model. This allows the system to continuously adjust the configuration parameters of a cell and evaluate the corresponding performance indicators using the relationship model to find the configuration parameters that achieve the local optimal performance indicator, which is then used as the local optimal configuration parameters for that cell.

[0124] In some embodiments of the present application, Figure 3 As shown, the aforementioned step 130 further includes: sub-step 1301 and sub-step 1302.

[0125] In sub-step 1301, the performance indicators corresponding to the adjusted configuration parameters of similar cells with reference to the target cell are estimated through the relationship model, and a hill climbing algorithm is used to search for the locally optimal configuration parameters to determine the optimized configuration parameters of the target cell, wherein the similar cells are cells in the target cluster that match the second load state.

[0126] For example, the target cell's configuration parameters are first mapped to the current node. By iteratively adjusting these parameters, the team gradually searches for configuration parameters that improve the performance indicators output by the relationship model, gradually climbing to the peak. This cycle continues until the preset number of iterations is reached, resulting in the local optimal configuration parameters.

[0127] In some embodiments of the present application, the step of estimating the performance indicators corresponding to the adjusted configuration parameters of similar cells with reference to the target cell through the relationship model, searching for local optimal configuration parameters using a hill climbing algorithm, and determining the optimized configuration parameters of the target cell includes: determining a current node based on the target cell, and determining neighbor nodes based on similar cells of the target cell; obtaining the current node performance indicators corresponding to the configuration parameters of the current node through the relationship model; in response to the current node performance indicators being the optimal performance indicators corresponding to the neighbor nodes, determining the configuration parameters of the current node as the optimized configuration parameters of the target cell; in response to the current node performance indicators not being the optimal performance indicators corresponding to the neighbor nodes, replacing the current node with the target neighbor node, jumping to the step of obtaining the current node performance indicators corresponding to the configuration parameters of the current node through the relationship model, performing iterative climbing until the number of iterative climbing reaches a preset iteration number threshold, and determining that the configuration parameters corresponding to the current node are the optimized configuration parameters of the target cell, wherein the target neighbor node is: the neighbor node corresponding to the similar cell having the highest similarity with the cell operation scenario corresponding to the current node.

[0128] In some embodiments of the present application, before determining the current node according to the target cell and determining the neighbor node according to the similar cells of the target cell, it also includes: determining the similarity of the operating scenarios of the target cell and each of the similar cells. The operating scenario similarity can be a similarity determined based on the aforementioned user distribution characteristics. In other embodiments of the present application, the operating scenario similarity can be a similarity determined based on the configuration data, TA value, etc. of the cell (for example, the similarity distance between the characteristic value of the configuration parameter of cell A and the characteristic value of the configuration parameter of cell B is used as the operating scenario similarity of cell A and cell B).

[0129] For example, in the process of searching for the local optimal configuration parameters using the hill climbing algorithm, the current node is first determined based on the target cell, and the neighboring nodes are determined based on the non-high-load cells (i.e., cells matching the second load state) clustered in the same cluster as the target cell. After adjusting the configuration parameters of the target cell, the performance indicators of the corresponding adjusted configuration parameters are estimated through the relationship model obtained through training, and the estimated performance indicators are compared with the performance indicators corresponding to each neighboring node to determine whether the estimated performance indicators are locally optimal. If so, the currently adjusted configuration parameters are used as the optimized configuration parameters of the target cell; if not, the neighboring node corresponding to the non-high-load cell with the highest similarity to the target cell is used to replace the current node, and the search for better configuration parameters is continued to achieve the purpose of gradually climbing to the peak. This cycle is repeated until the preset number of iterations is reached, and the local optimal configuration parameters are obtained.

[0130] Sub-step 1302: determining a configuration parameter optimization solution for the target cell based on the optimized configuration parameters.

[0131] After the optimized configuration parameters of the target cell are determined, a configuration parameter optimization solution for the target cell may be generated according to the optimized configuration parameters.

[0132] In some embodiments of the present application, determining the configuration parameter optimization scheme for the target cell based on the optimized configuration parameters includes: determining the configuration parameter optimization scheme for the target cell based on the degree of match between the optimized configuration parameters and a preset configuration parameter optimization scheme. For example, in a system implementing this solution, several conventional configuration parameter optimization schemes may be preset. The preset configuration parameter optimization scheme may include: the type of solution, the type of optimized configuration parameters corresponding to the solution, and optimization scheme generation rules. When one or more configuration parameters with the greatest difference are determined by comparing the optimized configuration parameters with the original configuration parameters of the target cell, the one or more determined configuration parameters are then matched with the types of optimized configuration parameters in the preset configuration parameter optimization scheme, and the configuration parameter optimization scheme with the highest degree of match is determined. Thereafter, a configuration parameter optimization scheme is further generated and output according to the successful matching optimization scheme generation rules in the determined configuration parameter optimization scheme with the highest degree of match.

[0133] In some embodiments of the present application, the correspondence between the solution type and the type of optimized configuration parameter in the preset configuration parameter optimization solution, as well as the optimization solution generation rules, can be set based on expert experience. For example, the optimization configuration parameters corresponding to the load balancing type solution can be set to include: frequency, and the optimization solution generation rules can be set to compare the traffic volume of each frequency, and output the corresponding optimization solution text based on the comparison results (for example, the output configuration parameter optimization solution can be "The traffic volume of the station's XX frequency is low, and traffic balancing can be implemented").

[0134] In some embodiments of the present application, determining a configuration parameter optimization solution for the target cell based on the optimized configuration parameters includes determining the configuration parameter optimization solution for the target cell based on the optimized configuration parameters and the original configuration parameters of the target cell. For example, the optimized configuration parameters may be compared with the original configuration parameters of the target cell to determine a difference, and then a configuration parameter optimization solution may be generated based on the difference. For example, a list of configuration parameters whose difference values ​​meet preset optimization conditions and the comparison results may be output.

[0135] In some embodiments of the present application, after determining the configuration parameter optimization scheme for the target cell based on the relationship model, the method further includes: outputting the configuration parameter optimization scheme for the target cell, and / or outputting the similarity between the operating scenarios of the target cell and each of the similar cells. By outputting the configuration parameter optimization scheme and outputting the similarity between the operating scenarios of high-load cells and non-high-load cells, visualization of the data processing process and data processing results is achieved, providing effective data reference for communication network maintenance personnel.

[0136] The cell configuration parameter optimization method disclosed in the embodiment of the present application analyzes and processes load-related KPIs, TA (time advanced, maximum time advance), industrial parameter data, cell configuration data, MR data, and other data. It only takes 10 minutes to complete the load status analysis of tens of thousands of cells, and output a list of high-load cells, as well as output the configuration parameter optimization scheme corresponding to the high-load cells, etc., effectively saving manual optimization time and improving cell maintenance efficiency. After manual verification, when the cell configuration parameter optimization method disclosed in the embodiment of the present application is used to analyze the operating data of the cell, the accuracy of determining the high-load cell list is 100%, and the accuracy of the determined high-load cell configuration parameter optimization scheme can reach 80%.

[0137] The embodiment of the present application discloses a cell configuration parameter optimization device, such as Figure 4 As shown, the device includes:

[0138] The cell clustering module 410 is configured to cluster cells based on user distribution characteristics and determine the cells included in each cluster obtained by clustering;

[0139] a configuration parameter and performance relationship model training module 420, configured to learn the relationship between configuration data and performance indicator data of cells included in a target cluster, and obtain a relationship model between the configuration data and performance indicator data corresponding to the target cluster, wherein the target cluster is a target cell belonging cluster, and the target cell matches a first load state;

[0140] The configuration parameter optimization solution determination module 430 is configured to determine a configuration parameter optimization solution for the target cell according to the relationship model.

[0141] In some embodiments of the present application, the configuration parameter optimization solution determination module 430 is further configured to:

[0142] Predicting performance indicators corresponding to the adjusted configuration parameters of similar cells with reference to the target cell using the relationship model, searching for local optimal configuration parameters using a hill climbing algorithm, and determining optimized configuration parameters of the target cell, wherein the similar cells are cells in the target cluster that match the second load state;

[0143] A configuration parameter optimization scheme for the target cell is determined based on the optimized configuration parameters.

[0144] In some embodiments of the present application, estimating the performance indicators corresponding to the adjusted configuration parameters of similar cells with reference to the target cell using the relationship model, and using a hill climbing algorithm to search for local optimal configuration parameters to determine the optimized configuration parameters of the target cell include:

[0145] Determining a current node based on a target cell, and determining a neighbor node based on similar cells to the target cell;

[0146] Obtaining the current node performance indicator corresponding to the configuration parameters of the current node through the relationship model;

[0147] In response to the current node performance indicator being the optimal performance indicator corresponding to the neighboring node, determining the configuration parameters of the current node as the optimized configuration parameters of the target cell;

[0148] In response to the current node performance indicator not being the optimal performance indicator corresponding to the neighboring node, the current node is replaced by the target neighboring node, and the process jumps to the step of obtaining the current node performance indicator corresponding to the configuration parameters of the current node through the relationship model, and performing iterative climbing until the number of iterative climbing reaches a preset iteration threshold, and determining that the configuration parameters corresponding to the current node are the optimized configuration parameters of the target cell, wherein the target neighboring node is: the neighboring node corresponding to the similar cell having the highest similarity with the cell operation scenario corresponding to the current node.

[0149] In some embodiments of the present application, before determining the current node according to the target cell and determining the neighbor node according to similar cells of the target cell, the method further includes:

[0150] Determining the similarity of the operating scenarios of the target cell and each of the similar cells;

[0151] After determining the configuration parameter optimization solution of the target cell according to the relationship model, the method further includes:

[0152] Outputting a configuration parameter optimization solution for the target cell, and / or outputting the similarity between the target cell and the operation scenarios of each of the similar cells.

[0153] In some embodiments of the present application, the configuration parameter and performance relationship model training module 420 is further configured to:

[0154] constructing an input matrix according to eigenvalues ​​of preset configuration parameters of cells included in the target cluster, and constructing an output matrix according to eigenvalues ​​of preset performance indicators of cells included in the target cluster;

[0155] A machine learning model for solving multivariate regression problems is used to learn the relationship between the input matrix and the output matrix, and obtain a relationship model between the configuration data and performance indicator data corresponding to the target cluster.

[0156] In some embodiments of the present application, the cell clustering module 410 is further configured to:

[0157] Obtaining preset user distribution characteristics of each of the cells;

[0158] Based on the preset user distribution characteristics, similarity distances between corresponding cells are calculated, and K-means clustering is performed on the cells based on the similarity distances to determine multiple clusters, and the cells included in each cluster are determined.

[0159] In some embodiments of the present application, obtaining the preset user distribution characteristics of each cell includes:

[0160] Based on the measurement report data of each cell within a specified time period, the preset user distribution characteristics of the corresponding cell are obtained, wherein the preset user distribution characteristics include one or more of the following: the arithmetic mean of the cell uplink utilization rate, the arithmetic mean of the cell downlink utilization rate, the arithmetic mean of the maximum number of users, the variance of the number of users, the variance of the flow rate, the kurtosis coefficient of the number of users, the kurtosis coefficient of the cell uplink utilization rate, the kurtosis coefficient of the cell downlink utilization rate, the flow rate kurtosis coefficient, the skewness coefficient of the number of users, the skewness coefficient of the cell uplink utilization rate, the skewness coefficient of the cell downlink utilization rate, and the flow rate skewness coefficient.

[0161] The cell configuration parameter optimization device disclosed in the embodiment of the present application is used to implement the cell configuration parameter optimization method described in the embodiment of the present application. The specific implementation methods of each module of the device will not be repeated here. Please refer to the specific implementation methods of the corresponding steps in the method embodiment.

[0162] The cell configuration parameter optimization device disclosed in the embodiment of the present application clusters cells based on user distribution characteristics to determine the cells included in each cluster obtained by clustering; learns the relationship between the configuration data and performance indicator data of the cells included in the target cluster to obtain a relationship model between the configuration data and performance indicator data corresponding to the target cluster, wherein the target cluster is a target cell belonging cluster, and the target cell matches a first load state; determines a configuration parameter optimization plan for the target cell based on the relationship model, and can automatically output the configuration parameter optimization plan for the target cell, thereby improving the optimization efficiency of the cell configuration parameters.

[0163] The cell configuration parameter optimization device disclosed in the embodiment of the present application, by combining big data processing methods such as machine learning, classifies cells into cell scenarios, then uses machine learning methods to mine the intrinsic relationship between cell configuration parameters and performance indicators for each cell scenario, and searches for optimized configuration parameters for high-load cells based on the mined intrinsic relationship, and automatically outputs a configuration parameter optimization plan. This can quickly output an optimization plan for high-load cells, thereby improving the efficiency of cell configuration parameter optimization. Furthermore, by quickly and accurately reducing the pressure on high-load cells, network resource utilization is improved.

[0164] Each embodiment in this specification is described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between the various embodiments can be referred to in conjunction with each other. For the device embodiments, since they are generally similar to the method embodiments, their description is relatively simple, and for relevant parts, reference can be made to the description of the method embodiments.

[0165] The above is a detailed introduction to a cell configuration parameter optimization method and device provided by the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for general technical personnel in this field, based on the ideas of the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

[0166] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0167] The various component embodiments of the present application can be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. It will be appreciated by those skilled in the art that a microprocessor or digital signal processor (DSP) can be used in practice to implement some or all of the functions of some or all of the components in the electronic device according to the embodiment of the present application. The application can also be implemented as a device or apparatus program (for example, a computer program and a computer program product) for performing a part or all of the methods described herein. Such a program implementing the present application can be stored on a computer-readable medium, or can have the form of one or more signals. Such a signal can be downloaded from an Internet website, or provided on a carrier signal, or provided in any other form.

[0168] For example, Figure 5An electronic device that can implement the method according to the present application is shown. The electronic device can be a PC, a mobile terminal, a personal digital assistant, a tablet computer, etc. The electronic device conventionally includes a processor 510 and a memory 520, and program code 530 stored on the memory 520 and executable on the processor 510. When the processor 510 executes the program code 530, the method described in the above embodiments is implemented. The memory 520 can be a computer program product or a computer-readable medium. The memory 520 can be an electronic memory such as a flash memory, an EEPROM (Electrically Erasable Programmable Read-Only Memory), an EPROM, a hard disk, or a ROM. The memory 520 has a storage space 5201 for program code 530 of a computer program for executing any of the method steps described above. For example, the storage space 5201 for program code 530 can include individual computer programs for implementing various steps in the above method. The program code 530 is computer-readable code. These computer programs can be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, compact disks (CDs), memory cards, or floppy disks. The computer program includes a computer-readable code, and when the computer-readable code is run on an electronic device, the electronic device is caused to execute the method according to the above embodiment.

[0169] An embodiment of the present application further discloses a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps of the cell configuration parameter optimization method as described in the first embodiment of the present application are implemented.

[0170] Such a computer program product may be a computer-readable storage medium having a computer program product. Figure 5 The memory 520 in the electronic device shown is similarly arranged as a storage segment, storage space, etc. The program code can be compressed and stored in the computer readable storage medium in an appropriate form. The computer readable storage medium is generally as shown in FIG. Figure 6 The portable or fixed storage unit generally includes computer-readable code 530', which is a code read by a processor and implements the steps of the above-described method when executed by the processor.

[0171] References herein to "one embodiment," "an embodiment," or "one or more embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present application. Furthermore, please note that instances of the phrase "in one embodiment" do not necessarily all refer to the same embodiment.

[0172] In the description provided herein, a large number of specific details are described. However, it is understood that the embodiments of the present application can be practiced without these specific details. In some instances, well-known methods, structures, and techniques are not shown in detail so as not to obscure the understanding of this description.

[0173] In the claims, any reference signs placed between brackets shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in the claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present application may be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In a unit claim enumerating several means, several of these means may be embodied by one and the same item of hardware. The use of the words first, second, and third etc. does not indicate any order. These words may be interpreted as names.

[0174] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for optimizing cell configuration parameters, characterized in that: include: Clustering cells based on user distribution characteristics and determining cells included in each cluster obtained by clustering; The user distribution characteristics include one or more of the following: user number variance, traffic variance, user number kurtosis coefficient, cell uplink utilization kurtosis coefficient, cell downlink utilization kurtosis coefficient, traffic kurtosis coefficient, user number skewness coefficient, cell uplink utilization skewness coefficient, cell downlink utilization skewness coefficient, and traffic skewness coefficient; Learning a relationship between configuration data and performance indicator data of cells included in a target cluster to obtain a relationship model between the configuration data and the performance indicator data corresponding to the target cluster, wherein the target cluster is a target cell belonging cluster, and the target cell matches a first load state; Determining a configuration parameter optimization scheme for the target cell according to the relationship model; The step of determining the configuration parameter optimization solution of the target cell according to the relationship model includes: Predicting performance indicators corresponding to the adjusted configuration parameters of similar cells with reference to the target cell using the relationship model, searching for local optimal configuration parameters using a hill climbing algorithm, and determining optimized configuration parameters of the target cell, wherein the similar cells are cells in the target cluster that match the second load state; Determining the configuration parameter optimization scheme for the target cell according to a matching degree between the optimized configuration parameters and a preset configuration parameter optimization scheme; Determining one or more configuration parameters with the largest difference by comparing the optimized configuration parameters with original configuration parameters of the target cell; Matching the determined one or more configuration parameters with the optimized configuration parameter types in a preset configuration parameter optimization scheme to determine a configuration parameter optimization scheme with the highest matching degree; A configuration parameter optimization solution is generated according to a successfully matched optimization solution generation rule among the configuration parameter optimization solutions with the highest matching degree.

2. The method according to claim 1, characterized in that The step of estimating the performance indicators corresponding to the adjusted configuration parameters of similar cells with reference to the target cell using the relationship model, searching for local optimal configuration parameters using a hill climbing algorithm, and determining the optimized configuration parameters of the target cell includes: Determining a current node based on a target cell, and determining a neighbor node based on similar cells to the target cell; Obtaining the current node performance indicator corresponding to the configuration parameters of the current node through the relationship model; In response to the current node performance indicator being the optimal performance indicator corresponding to the neighboring node, determining the configuration parameters of the current node as the optimized configuration parameters of the target cell; In response to the current node performance indicator not being the optimal performance indicator corresponding to the neighboring node, the current node is replaced by the target neighboring node, and the process jumps to the step of obtaining the current node performance indicator corresponding to the configuration parameters of the current node through the relationship model, and performing iterative climbing until the number of iterative climbing reaches a preset iteration threshold, and determining that the configuration parameters corresponding to the current node are the optimized configuration parameters of the target cell, wherein the target neighboring node is: the neighboring node corresponding to the similar cell having the highest similarity with the cell operation scenario corresponding to the current node.

3. The method according to claim 2, characterized in that Before determining the current node according to the target cell and determining the neighbor node according to the similar cells of the target cell, the method further includes: Determining the similarity of the operating scenarios of the target cell and each of the similar cells; After determining the configuration parameter optimization solution of the target cell according to the relationship model, the method further includes: Outputting a configuration parameter optimization solution for the target cell, and / or outputting the similarity between the target cell and the operation scenarios of each of the similar cells.

4. The method according to claim 1, wherein The learning of the relationship between the configuration data and the performance indicator data of the cells included in the target cluster to obtain a relationship model between the configuration data and the performance indicator data corresponding to the target cluster includes: constructing an input matrix according to eigenvalues ​​of preset configuration parameters of cells included in the target cluster, and constructing an output matrix according to eigenvalues ​​of preset performance indicators of cells included in the target cluster; A machine learning model for solving multivariate regression problems is used to learn the relationship between the input matrix and the output matrix, and obtain a relationship model between the configuration data and performance indicator data corresponding to the target cluster.

5. The method according to claim 1, wherein The clustering of cells based on user distribution characteristics and determining cells included in each cluster obtained by clustering include: Obtaining preset user distribution characteristics of each of the cells; Based on the preset user distribution characteristics, similarity distances between corresponding cells are calculated, and K-means clustering is performed on the cells based on the similarity distances to determine multiple clusters, and the cells included in each cluster are determined.

6. A cell configuration parameter optimization device, characterized in that: include: A cell clustering module is used to cluster cells based on user distribution characteristics and determine the cells included in each cluster obtained by clustering; The user distribution characteristics include one or more of the following: user number variance, traffic variance, user number kurtosis coefficient, cell uplink utilization kurtosis coefficient, cell downlink utilization kurtosis coefficient, traffic kurtosis coefficient, user number skewness coefficient, cell uplink utilization skewness coefficient, cell downlink utilization skewness coefficient, and traffic skewness coefficient; a configuration parameter and performance relationship model training module, configured to learn the relationship between configuration data and performance indicator data of cells included in a target cluster, and obtain a relationship model between the configuration data and performance indicator data corresponding to the target cluster, wherein the target cluster is a target cell belonging cluster, and the target cell matches a first load state; a configuration parameter optimization solution determination module, configured to determine a configuration parameter optimization solution for the target cell based on the relationship model; The configuration parameter optimization solution determination module is further used to: Predicting performance indicators corresponding to the adjusted configuration parameters of similar cells with reference to the target cell using the relationship model, searching for local optimal configuration parameters using a hill climbing algorithm, and determining optimized configuration parameters of the target cell, wherein the similar cells are cells in the target cluster that match the second load state; Determining the configuration parameter optimization scheme for the target cell according to a matching degree between the optimized configuration parameters and a preset configuration parameter optimization scheme; Determining one or more configuration parameters with the largest difference by comparing the optimized configuration parameters with original configuration parameters of the target cell; Matching the determined one or more configuration parameters with the optimized configuration parameter types in a preset configuration parameter optimization scheme to determine a configuration parameter optimization scheme with the highest matching degree; A configuration parameter optimization solution is generated according to a successfully matched optimization solution generation rule among the configuration parameter optimization solutions with the highest matching degree.

7. An electronic device comprising a memory, a processor, and a program code stored in the memory and executable on the processor, wherein: When the processor executes the program code, the cell configuration parameter optimization method according to any one of claims 1 to 5 is implemented.

8. A computer-readable storage medium having program code stored thereon, characterized in that: When the program code is executed by a processor, the steps of the cell configuration parameter optimization method according to any one of claims 1 to 5 are implemented.

Images