CN112243249A

CN112243249A - LTE new access anchor point cell parameter configuration method and device under 5G NSA networking

Info

Publication number: CN112243249A
Application number: CN201910653862.7A
Authority: CN
Inventors: 李兵; 唐秋香
Original assignee: Datang Mobile Communications Equipment Co Ltd
Current assignee: Datang Mobile Communications Equipment Co Ltd
Priority date: 2019-07-19
Filing date: 2019-07-19
Publication date: 2021-01-19
Anticipated expiration: 2039-07-19
Also published as: CN112243249B

Abstract

The embodiment of the invention provides a method and a device for configuring parameters of an anchor point cell of LTE new access network under 5G NSA networking, wherein the method comprises the following steps: acquiring cell characteristic data of an anchor cell; the cell characteristic data at least comprises work parameter data; inputting the cell characteristic data into a scene classification model, and acquiring a scene classification result output by the scene classification model; the scene classification model is obtained based on sample cell characteristic data of a sample cell and sample scene class training; configuring the anchor cell based on the cell parameter of the preferred cell corresponding to the scene classification result; the preferred cell is selected from the sample cells corresponding to the scene classification result. The method and the device provided by the embodiment of the invention realize the accurate classification of the anchor point cells so as to meet the parameter configuration requirements under different scene categories, fully utilize the value of the existing data of the existing network and effectively ensure the stability of the cell access to the network.

Description

LTE new access anchor point cell parameter configuration method and device under 5G NSA networking

Technical Field

The invention relates to the technical field of mobile communication, in particular to a method and a device for configuring parameters of an anchor point cell of LTE new access under 5G NSA networking.

Background

Under a Non-independent Networking (NSA) architecture of a 5G (5th-Generation, fifth Generation mobile communication network), a modified and upgraded LTE (Long Term Evolution) base station of an existing network or a newly-built LTE base station is used as a control plane anchor point based on a networking mode of Option3x to provide a signaling service. How to ensure the stable indexes of the LTE anchor point network of the current network and the user perception are not influenced is the premise of stable and rapid deployment of the 5G NSA network.

Fig. 1 is a schematic diagram of a networking mode of a 5G NSA Option3x in the prior art, as shown in fig. 1, an epc (evolved Packet Core) is a 4G Core network, a 5G Core is a 5G Core network, and a ue (user equipment) is a user terminal. In the 5G NSA Option3x scenario, an LTE eNB (evolved node b) is used as a main base station, control plane signaling is forwarded through the eNB, and an NR (National Instruments, new air interface) gNB (5G base station) may offload data to the eNB. The LTE eNB and the NR gbb provide high data rate services to users in the form of dual links.

Under the Option3x networking mode, a 5G gNB new cell is accessed, and an LTE new cell after upgrading and reconstruction is also required to be newly accessed. In order to ensure that the current network index of the LTE is stable and the user perception is not affected, a parameter configuration research needs to be performed on an anchor cell for LTE new network access under 5G NSA networking.

The traditional new network access cell parameter configuration is carried out according to the inherent station-opening template and planning data, the parameter configuration requirements under different scenes cannot be fully considered, and the value of the existing data of the existing network cannot be mined by the current unified configuration template, so that the network access stability of the cell is difficult to guarantee.

Disclosure of Invention

The embodiment of the invention provides a method and a device for configuring parameters of an anchor point cell of LTE new access under 5G NSA networking, which are used for solving the problems that the existing unified template configuration method cannot meet the requirements of different scenes and the stability of cell access is poor.

In a first aspect, an embodiment of the present invention provides a method for configuring parameters of an anchor point cell for LTE new access under 5G NSA networking, including:

acquiring cell characteristic data of an anchor cell; the cell characteristic data at least comprises work parameter data;

inputting the cell characteristic data into a scene classification model, and acquiring a scene classification result output by the scene classification model; the scene classification model is obtained based on sample cell characteristic data of a sample cell and sample scene class training;

configuring the anchor cell based on the cell parameter of the preferred cell corresponding to the scene classification result; the preferred cell is selected from the sample cells corresponding to the scene classification result.

Preferably, the scene classification model comprises a first classification model and a second classification model;

correspondingly, the inputting the cell feature data into a scene classification model to obtain a scene classification result output by the scene classification model specifically includes:

inputting the working parameter data in the cell characteristic data into the first classification model, and obtaining a first classification result output by the first classification model; the first classification model is obtained by training based on sample work parameter data of a sample cell and the sample scene category;

configuring the anchor point cell based on the cell parameter of the optimal cell corresponding to the first classification result, and acquiring busy hour characteristic data of the anchor point cell; the busy hour characteristic data comprises the work parameter data and at least one of cell performance index data, wireless environment data and coverage scene data;

inputting the busy hour characteristic data into the second classification model to obtain the scene classification result output by the second classification model; the second classification model is obtained by training based on sample busy hour feature data of the sample cell and the sample scene category.

Preferably, the configuring the anchor cell based on the cell parameter of the preferred cell corresponding to the scene classification result specifically includes:

acquiring the similarity between each preferred cell corresponding to the scene classification result and the anchor cell;

and selecting the cell parameter of the preferred cell with the highest similarity to configure the anchor cell.

Preferably, the obtaining the similarity between each preferred cell corresponding to the scene classification result and the anchor cell specifically includes:

and calculating the similarity between any preferred cell and the anchor cell based on the cell performance index data, the wireless environment data and the coverage scene data of any preferred cell and the anchor cell.

Preferably, the inputting the cell feature data into a scene classification model and obtaining a scene classification result output by the scene classification model further include:

acquiring a sample cell characteristic data set of a sample cell; the sample cell characteristic data set comprises sample cell characteristic data of each of the sample cells; the sample cell characteristic data comprises working parameter data, cell performance index data, wireless environment data and coverage scene data;

and clustering the sample cell characteristic data sets to obtain a sample scene category corresponding to each sample cell.

Preferably, the clustering the sample cell feature data sets to obtain a sample scene category corresponding to each sample cell specifically includes:

acquiring contour coefficients of the sample cell characteristic data set under the quantity of a plurality of cluster types based on a contour coefficient method;

and selecting the cluster number corresponding to the maximum value of the outline coefficient as a K value, carrying out K-means clustering on the sample cell characteristic data set, and obtaining the sample scene category corresponding to each sample cell.

Preferably, the clustering the sample cell feature data sets to obtain a sample scene category corresponding to each sample cell further includes:

preprocessing the sample cell characteristic data set; the pre-processing includes data cleansing and/or normalization processing.

Preferably, the configuring the anchor cell based on the cell parameter of the preferred cell corresponding to the scene classification result includes:

acquiring a key performance index set comprising key performance indexes of a plurality of sample cells;

training a scoring card model based on the key performance index set to obtain an automatic scoring system;

acquiring the comprehensive performance score of the sample cell based on the automatic scoring system;

and selecting a preferred cell corresponding to any sample scene type based on the comprehensive performance score of each sample cell corresponding to any sample scene type.

Preferably, the training of the scoring card model based on the set of key performance indicators to obtain an automatic scoring system specifically includes:

setting a cell label of the sample cell based on the key performance index of the sample cell in the key performance index set and a preset index threshold;

performing data cleaning on the key performance index set;

continuously variable binning is carried out on the key performance index set to obtain binning results, the prediction capability of each key performance index is calculated based on the binning results, and key performance indexes are screened;

training a logistic regression model based on the screened key performance index set;

and constructing the automatic scoring system based on the independent variable coefficient in the logistic regression model and the scoring card model.

Preferably, the configuring the anchor cell based on the cell parameter of the preferred cell corresponding to the scene classification result further includes:

acquiring a time sequence of key performance indexes of the anchor cell;

performing stationarity detection on the time sequence to obtain a stationarity detection result of the time sequence;

performing randomness detection on the time sequence to obtain a randomness detection result of the time sequence;

and acquiring a dynamic detection result of the anchor point cell according to the stationarity detection result and the randomness detection result of the time sequence and a preset dynamic early warning threshold.

Preferably, the acquiring a dynamic detection result of the anchor point cell according to the stationarity detection result and the randomness detection result of the time sequence and a preset dynamic early warning threshold further includes:

obtaining a historical time sequence of the key performance indexes of the optimal cell corresponding to the scene classification result;

performing stationarity detection on the historical time sequence to obtain a stationarity detection result of the historical time sequence;

performing randomness detection on the historical time sequence to obtain a randomness detection result of the historical time sequence;

and acquiring the preset dynamic early warning threshold based on the stationarity detection result and the randomness detection result of the historical time sequence.

Preferably, the key performance index includes at least one of a radio connection rate, a radio drop rate, a handover success rate, an eSRVCC handover success rate, a VOLTE voice connection rate, and a VOLTE voice drop rate.

In a second aspect, an embodiment of the present invention provides an apparatus for configuring parameters of an anchor cell for LTE new access under 5G NSA networking, including:

a cell characteristic acquiring unit, configured to acquire cell characteristic data of an anchor cell; the cell characteristic data at least comprises work parameter data;

the scene classification unit is used for inputting the cell characteristic data into a scene classification model and acquiring a scene classification result output by the scene classification model; the scene classification model is obtained based on sample cell characteristic data of a sample cell and sample scene class training;

a parameter configuration unit, configured to configure the anchor cell based on the cell parameter of the preferred cell corresponding to the scene classification result; the preferred cell is selected from the sample cells corresponding to the scene classification result.

correspondingly, the scene classification unit is specifically configured to:

Preferably, the parameter configuration unit specifically includes:

the similarity calculation operator unit is used for acquiring the similarity between each preferred cell corresponding to the scene classification result and the anchor cell;

a parameter configuration subunit, configured to select the cell parameter of the preferred cell with the highest similarity to configure the anchor cell.

Preferably, the similarity operator unit is specifically configured to:

Preferably, the method further comprises the following steps:

a sample feature acquisition unit, configured to acquire a sample cell feature data set of a sample cell; the sample cell characteristic data set comprises sample cell characteristic data of each of the sample cells; the sample cell characteristic data comprises working parameter data, cell performance index data, wireless environment data and coverage scene data;

and the sample cell clustering unit is used for clustering the sample cell characteristic data sets to obtain a sample scene category corresponding to each sample cell.

Preferably, the sample cell clustering unit is specifically configured to:

Preferably, the method further comprises the following steps:

the preprocessing unit is used for preprocessing the sample cell characteristic data set; the pre-processing includes data cleansing and/or normalization processing.

Preferably, the method further comprises the following steps:

a sample KPI obtaining unit, configured to obtain a key performance indicator set including key performance indicators of a plurality of sample cells;

the scoring system training unit is used for training a scoring card model based on the key performance index set to obtain an automatic scoring system;

the scoring unit is used for acquiring the comprehensive performance score of the sample cell based on the automatic scoring system;

and the cell selecting unit is used for selecting a preferred cell corresponding to any sample scene type based on the comprehensive performance score of each sample cell corresponding to any sample scene type.

Preferably, the scoring system training unit is specifically configured to:

performing data cleaning on the key performance index set;

Preferably, the system further comprises a dynamic early warning unit; the dynamic early warning unit is used for:

acquiring a time sequence of key performance indexes of the anchor cell;

Preferably, the system further comprises an early warning threshold acquisition unit; the early warning threshold acquisition unit is used for:

In a third aspect, an embodiment of the present invention provides an electronic device, including a processor, a communication interface, a memory, and a bus, where the processor and the communication interface, the memory complete communication with each other through the bus, and the processor may call a logic instruction in the memory to perform the steps of the method provided in the first aspect.

In a fourth aspect, an embodiment of the present invention provides a non-transitory computer readable storage medium, on which a computer program is stored, which when executed by a processor, implements the steps of the method as provided in the first aspect.

According to the method and the device for configuring the parameters of the anchor point cell for LTE new access under 5G NSA networking, provided by the embodiment of the invention, the scene classification result of the anchor point cell is obtained based on the scene classification model, so that the anchor point cell is accurately classified, and the parameter configuration requirements under different scene categories are met. And configuring an anchor cell based on the cell parameter of the preferred cell corresponding to the scene classification result, and compared with the issuing of a unified configuration template, fully utilizing the value of the existing data of the existing network and effectively ensuring the stability of cell network access.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.

FIG. 1 is a schematic diagram of a networking scheme of a 5G NSA Option3x in the prior art;

fig. 2 is a schematic flowchart of a method for configuring parameters of an anchor point cell for LTE new access in 5G NSA networking according to an embodiment of the present invention;

FIG. 3 is a graph showing the relationship between the contour coefficient and the number of clusters according to an embodiment of the present invention;

FIG. 4 is a schematic diagram of ROC curve fitting provided by an embodiment of the present invention;

fig. 5 is a schematic flowchart of a method for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to another embodiment of the present invention;

fig. 6 is a schematic structural diagram of an LTE new access anchor point cell parameter configuration device in 5G NSA networking according to an embodiment of the present invention;

fig. 7 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The traditional new network access cell parameter configuration is carried out according to the inherent station opening template and the planning data, the parameter configuration requirements under different scenes cannot be fully considered, and the stability of cell network access is difficult to guarantee. In view of the above, the embodiment of the present invention provides a method for configuring parameters of an anchor point cell for LTE new access under 5G NSA networking. Fig. 2 is a schematic flowchart of a method for configuring parameters of an anchor cell for LTE new access under 5G NSA networking, where as shown in fig. 2, the method includes:

step 210, obtaining cell characteristic data of an anchor cell; the cell characteristic data at least comprises work parameter data.

Here, in the networking mode of the anchor cell, i.e., 5G NSA Option3x, the LTE cell to be configured is newly accessed. The cell feature data includes engineering parameter data, and may further include KPI (Key Performance Indicator) data, MR (Measurement Report) sampling point data, coverage simulation result data, raster data, and the like. The operating parameter may be a base station height corresponding to the anchor cell, a base station antenna direction angle, a base station antenna electronic downtilt, a base station antenna mechanical downtilt, a base station antenna transmission power, a base station position, and the like, which is not specifically limited in this embodiment of the present invention.

Step 220, inputting the cell characteristic data into a scene classification model, and acquiring a scene classification result output by the scene classification model; the scene classification model is obtained by training based on sample cell feature data of the sample cell and a sample scene category.

Specifically, the scene classification model is used for selecting a scene category to which the anchor cell belongs from a plurality of preset scene categories based on the input cell feature data of the anchor cell, and outputting the scene category as a scene classification result.

Before step 220 is executed, the scene classification model may also be obtained by training in advance, and specifically, the scene classification model may be obtained by training in the following manner: firstly, collecting sample cell characteristic data and sample scene categories of a large number of sample cells; the sample cell is a cell applied during model training, the sample cell characteristic data is cell characteristic data of the sample cell, and similarly, the sample cell characteristic at least comprises working parameter data; the sample scene type is a scene type corresponding to the sample cell, the sample scene type is obtained by classifying the sample cell based on the sample cell feature data, and all types of sample scene types constitute the preset multiple scene types. And training the initial model based on the sample cell characteristic data and the sample scene category to obtain a scene classification model. The initial model may be a single neural network model or a combination of a plurality of neural network models, and the embodiment of the present invention does not specifically limit the type and structure of the initial model.

Step 230, configuring an anchor cell based on the cell parameter of the preferred cell corresponding to the scene classification result; the preferred cell is selected from sample cells corresponding to the scene classification result.

Specifically, for any scene category, corresponding to a plurality of sample cells, based on the cell performance of each sample cell, one or more sample cells with better cell performance are selected from the plurality of sample cells as the preferred cells of the scene category. And after the scene classification result of the anchor cell is obtained, acquiring a corresponding preferred cell under the scene classification according to the scene classification result, and configuring parameters of the anchor cell according to the cell parameters of the preferred cell.

The method provided by the embodiment of the invention obtains the scene classification result of the anchor point cell based on the scene classification model, and realizes the accurate classification of the anchor point cell so as to meet the parameter configuration requirements under different scene categories. And configuring an anchor cell based on the cell parameter of the preferred cell corresponding to the scene classification result, and compared with the issuing of a unified configuration template, fully utilizing the value of the existing data of the existing network and effectively ensuring the stability of cell network access.

The existing new cell parameter configuration needs to invest manpower to perform parameter adaptation optimization after a unified configuration template is issued, which wastes time and labor and has poor effect. To this end, based on the above embodiment, in the method, the scene classification model includes a first classification model and a second classification model, and correspondingly, step 220 specifically includes:

step 221, inputting the work parameter data in the cell characteristic data into a first classification model, and obtaining a first classification result output by the first classification model; the first classification model is obtained by training based on sample parameter data and sample scene categories of the sample cell.

Specifically, the first classification model is configured to select a scene category to which the anchor cell belongs from a plurality of preset scene categories based on input work parameter data, implement scene classification of the anchor cell, and output an obtained scene category, that is, a first classification result.

Before step 221 is executed, the first classification model may be obtained by training in advance, and specifically, the first classification model may be obtained by training in the following manner: firstly, collecting sample parameter data and sample scene categories of a large number of sample cells; wherein, the sample work parameter data is the work parameter data of the sample cell; and training the initial model based on the sample parameter data and the sample scene category to obtain a first classification model. The initial model may be a single neural network model or a combination of a plurality of neural network models, and the embodiment of the present invention does not specifically limit the type and structure of the initial model.

Step 222, configuring an anchor point cell based on the cell parameter of the optimal cell corresponding to the first classification result, and acquiring busy hour characteristic data of the anchor point cell; the busy hour characteristic data comprises work parameter data and at least one of cell performance index data, MR sampling point data and ground feature data.

Specifically, for any scene category, corresponding to a plurality of sample cells, based on the cell performance of each sample cell, a sample cell with the best cell performance is selected from the plurality of sample cells as the best cell of the scene category. And after the first classification result is obtained, acquiring the cell parameter of the optimal cell under the scene type indicated by the first classification result, and configuring the anchor cell based on the cell parameter of the optimal cell. Here, the cell parameter of the optimal cell is a temporary configuration parameter of the anchor cell.

And after the parameter configuration of the anchor point cell is completed, acquiring busy hour characteristic data in the operation process of the anchor point cell. Here, the busy hour feature data is feature data acquired by the anchor cell running in the busy hour under the cell parameter configuration of the optimal cell corresponding to the first classification result. The busy hour characteristic data not only comprises the work parameter data of the anchor point cell, but also comprises at least one of cell performance index data, wireless environment data and coverage scene data.

The cell performance index data may include uplink and downlink traffic of the cell, a handover success rate between base stations, a handover attempt number between base stations, a maximum number of effective RRC (Radio Resource Control) connections, an average number of effective RRC connections, an average uplink and downlink PRB (Physical Resource Block) utilization rate, an uplink IOT (Interference over Thermal), an uplink and downlink BLER (Block error rate), and the like; the wireless environment data may include RSRP (Reference Signal Receiving Power) values of the respective sampling points, or an average coverage rate of RSRP greater than a preset RSRP threshold; the coverage scenario data may include the height, area, and location of the cell coverage building, etc.

Step 223, inputting the busy hour feature data into a second classification model, and obtaining a scene classification result output by the second classification model; and the second classification model is obtained by training based on the sample busy hour characteristic data of the sample cell and the sample scene category.

Specifically, the second classification model is configured to select a scene category to which the anchor cell belongs from a plurality of preset scene categories based on the input busy hour feature data, implement scene classification of the anchor cell, and output the obtained scene category, that is, a scene classification result finally output by the scene classification model.

Before step 223 is executed, the second classification model may be obtained by training in advance, and specifically, the second classification model may be obtained by training in the following manner: firstly, collecting sample busy hour characteristic data and sample scene categories of a large number of sample cells; the sample busy hour characteristic data is the busy hour characteristic data of the sample cell; and training the initial model based on the sample busy hour feature data and the sample scene category to obtain a second classification model. The initial model may be a single or two neural network model, or may be a combination of a plurality of neural network models, and the embodiment of the present invention does not specifically limit the type and structure of the initial model.

According to the method provided by the embodiment of the invention, the anchor point cell is temporarily configured after scene classification is carried out based on the working parameter data, and the busy hour characteristic data under the temporary configuration is obtained for scene classification, so that the accuracy of cell scene classification can be effectively improved.

Based on any of the above embodiments, in the method, step 230 specifically includes: 231, obtaining similarity between each preferred cell corresponding to the scene classification result and the anchor cell; step 232, selecting the cell parameter configuration anchor cell of the preferred cell with the highest similarity.

Specifically, there may be multiple preferred cells corresponding to any scene category. When the scene classification result corresponds to a plurality of preferred cells, it is necessary to calculate the similarity between the anchor cell and each preferred cell, select a preferred cell with the highest similarity from the plurality of preferred cells, and perform parameter configuration of the anchor cell according to the cell parameter of the preferred cell.

Here, the calculation of the similarity between the anchor cell and any preferred cell may be obtained according to the parameters of the anchor cell and the preferred cell, or may be obtained according to cell performance index data, wireless environment data, or coverage scenario data of the anchor cell and the preferred cell, which is not specifically limited in this embodiment of the present invention.

Based on any of the above embodiments, in the method, step 231 specifically includes: and calculating the similarity between any preferred cell and the anchor cell based on the cell performance index data, the wireless environment data and the coverage scene data of any preferred cell and the anchor cell.

Here, there may be a plurality of methods for calculating the similarity, for example, a manhattan distance calculation method, a euclidean distance calculation method, a pearson correlation coefficient calculation method, and the like, which is not particularly limited in this embodiment of the present invention.

According to any of the above embodiments, the method further includes, before the step 220: step 201, obtaining a sample cell characteristic data set of a sample cell; the sample cell characteristic data set comprises sample cell characteristic data of each sample cell; the sample cell characteristic data comprises work parameter data, cell performance index data, wireless environment data and coverage scene data; step 202, clustering the sample cell feature data sets to obtain a sample scene category corresponding to each sample cell.

Specifically, before training a scene classification model based on sample cell feature data and sample scene categories of sample cells, the sample cells need to be subjected to scene classification in advance, and a sample scene category corresponding to each sample cell is obtained. First, sample cell characteristic data of each sample cell needs to be acquired, and a sample cell characteristic data set is constructed accordingly. After the sample cell characteristic data set is obtained, clustering is carried out on the sample cell characteristic data set to obtain a plurality of clusters, each cluster corresponds to one scene type, and therefore the scene type corresponding to the sample cell, namely the sample scene type of the sample cell, is determined according to the cluster to which any sample cell belongs.

Here, there are various clustering methods, such as K-Means (K-Means) clustering, mean shift clustering, density clustering (DBSCAN), etc., and this is not particularly limited by the embodiment of the present invention.

Based on any of the above embodiments, in the method, step 202 specifically includes: acquiring contour coefficients of the sample cell characteristic data set under the quantity of a plurality of cluster types based on a contour coefficient method; and selecting the cluster number corresponding to the maximum value of the contour coefficient as a K value, carrying out K-means clustering on the sample cell characteristic data set, and obtaining the sample scene category corresponding to each sample cell.

Specifically, before performing K-means clustering, a parameter of K-means clustering, i.e., a value of the number of clusters K, needs to be determined. In the embodiment of the invention, the K value is obtained by a contour coefficient method. The contour Coefficient (Silhouette Coefficient) is an evaluation method for evaluating the clustering effect. The value of the contour coefficient is between-1, and the closer to 1, the better the cohesion and separation of the clusters. The method can obtain the profile coefficient corresponding to the clustering result obtained by clustering the sample cell characteristic data set under different clustering quantities, and select the clustering quantity corresponding to the maximum profile coefficient from the profile coefficients, namely the clustering quantity corresponding to the best clustering effect as the K value, so as to perform K-means clustering based on the K value.

The method for obtaining the contour coefficient specifically comprises the following steps:

1) taking the sample cell characteristic data of any sample cell in the sample cell characteristic data set as a sample i, and calculating the average distance a between the sample i and other samples in the same cluster_i。a_iIs the intra-cluster dissimilarity of sample i. A of all samples in Cluster C_iThe mean is the cluster dissimilarity of cluster C.

2) Calculating sample i to some other cluster C_jAverage distance b of all samples_ij，b_ijFor sample i and cluster C_jDegree of dissimilarity. Define inter-cluster dissimilarity for sample i: b_i＝min{b_i1,b_i2,...,b_ikWhere k is the number of clusters.

3) According to the intra-cluster dissimilarity a of the sample i_iDegree of dissimilarity with clusters b_iDefining the contour coefficients of the sample i as:

where s (i) is the contour coefficient of sample i, a (i) is the intra-cluster dissimilarity of sample i, and b (i) is the inter-cluster dissimilarity of sample i.

And taking the mean value of the contour coefficients of all the samples as the contour coefficient of the sample cell characteristic data set.

For example, fig. 3 is a graph of relationship between the contour coefficients and the number of clusters provided by the embodiment of the present invention, in fig. 3, the abscissa number of clusters, i.e., the number of clusters, and the ordinate Silhouette coeffient score, i.e., the contour coefficients, show that the corresponding contour Coefficient is the largest when the number of clusters is 3, so the K value is set to 3.

The method for performing K-means clustering based on the K value specifically comprises the following steps:

1) inputting a K value.

2) From the sample cell feature dataset, K samples are selected as initial centroids.

3) Assigning each sample to the nearest centroid, forming K clusters;

4) recalculating the mass center of each cluster until the cluster is unchanged or the maximum iteration number is reached, and outputting a cluster division result C ═ C₁,C₂,C₃,……,C_K}. Wherein, C_iIs the ith cluster, C_iContains each sample in the ith cluster.

According to any of the above embodiments, the method further includes, before step 202: preprocessing a sample cell characteristic data set; the pre-processing includes data cleansing and/or normalization processing.

In particular, data cleansing is a process used to re-examine and verify data with the purpose of deleting duplicate information, correcting existing errors, e.g., performing null processing, and rejecting outliers. When null value processing is carried out, the null value can be directly deleted when the proportion of the null value sample number is small, and when the proportion of the null value sample number is large and filling is needed, mode or median filling can be adopted, and random forest filling can also be adopted; outliers include outliers and other unreasonable values, such as measurements that deviate more than three standard deviations from the mean, or values that are more than 100% or less than 0% by weight, that need to be rejected.

In the sample cell characteristic data set, different sample cell characteristic data often have different dimensions and dimension units, and the direct application can affect the result of data analysis. Normalizing an expression which is about to have dimensions, converting the expression into a dimensionless expression through transformation, converting the expression into a scalar, and converting the scalar by adopting a linear function, wherein the expression is (x-MinValue)/(MaxValue-MinValue), x and y are sample cell characteristic data before and after conversion respectively, and MinValue and MaxValue are the minimum value and the maximum value of the sample cell characteristic data respectively.

According to any of the above embodiments, the method further includes, before step 230: step 203, obtaining a key performance index set comprising key performance indexes of a plurality of sample cells; step 204, training a scoring card model based on the key performance index set to obtain an automatic scoring system; step 205, acquiring comprehensive performance scores of sample cells based on an automatic scoring system; and step 206, selecting a preferred cell corresponding to any sample scene type based on the comprehensive performance score of each sample cell corresponding to the sample scene type.

Specifically, before configuring the anchor cell based on the cell parameter of the preferred cell, the sample cell needs to be scored, and the preferred cell is selected from the sample cell.

First, the key performance indicators of each sample cell are obtained, and a set of key performance indicators is constructed accordingly. Here, the type of the key performance indicator is preset, such as a wireless connection rate, a wireless disconnection rate, and the like.

And then training a scoring card model based on the key performance index set to obtain an automatic scoring system for scoring the comprehensive performance of the cell. Here, the scorecard model is a well-established prediction method, and the principle thereof is a generalized linear model of two classification variables by discretizing the feature weight encoding method of model variables and then applying a logistic regression model.

After the automatic scoring system is obtained, the key performance indexes of any sample cell in the key performance index set are input into the automatic scoring system, and the comprehensive performance score of the sample cell output by the automatic scoring system is obtained. Here, the overall performance score is used to reflect the quality of the overall performance of the sample cell, and the higher the overall performance score is, the better the overall performance of the sample cell is.

For any scene category, based on the comprehensive performance score, selecting a preferred cell from each sample cell with the sample scene category as the scene category, for example, selecting a sample cell with the comprehensive performance score larger than a preset score threshold value as the preferred cell, and for example, selecting a preset number of sample cells with the highest comprehensive performance score as the preferred cell. In addition, the sample cell with the highest comprehensive performance score can be directly selected as the optimal cell.

It should be noted that, in the embodiment of the present invention, the execution sequence of steps 201 and 202 and steps 203 to 205 is not specifically limited, and steps 201 and 202 may be executed before or after steps 203 to 205, or may be executed synchronously with steps 203 to 205.

According to the method provided by the embodiment of the invention, the automatic scoring system is obtained by training the scoring card model so as to realize the comprehensive performance evaluation of the sample cell and the selection of the preferred cell, and the stability of the newly-accessed cell is favorably ensured.

Based on any of the above embodiments, in the method, step 204 specifically includes: setting a cell label of a sample cell based on a key performance index of the sample cell in the key performance index set and a preset index threshold; performing data cleaning on the key performance index set; carrying out continuous variable binning on the key performance index set to obtain binning results, calculating the prediction capability of each key performance index based on the binning results, and screening the key performance indexes; training a logistic regression model based on the screened key performance index set; and constructing an automatic scoring system based on the independent variable coefficient in the logistic regression model and the scoring card model.

Specifically, for any sample cell, the cell label is used to indicate whether each key performance index of the sample cell meets a preset index threshold, and the cell label reflects the overall performance condition of the cell, for example, if the cell label is 1, the overall performance of the cell is good, and if the cell label is 0, the overall performance of the cell is bad.

For example, the key performance indicators and corresponding preset indicator thresholds of the sample cell are shown in the following table:

according to the preset index threshold shown in the table, if each key performance does not meet the corresponding preset index threshold, the cell label is determined to be 1, otherwise, the cell label is determined to be 0.

And then, performing data cleaning on the key performance index set, including null value processing and/or abnormal value elimination.

And then, carrying out continuous variable binning on the key performance index set to obtain binning results, calculating the prediction capability of each key performance index based on the binning results, and carrying out key performance index screening.

The continuous variable binning refers to segmenting continuous variables by adopting an optimal segmentation algorithm, wherein the variables are the key performance indexes. Optimal segmentation (Optimal pairing), namely supervised discretization, uses Recursive Partitioning (Recursive Partitioning) to segment continuous variables, namely to obtain a Binning result, wherein the Binning result comprises a plurality of bins for each variable.

After the binning results are obtained, for any variable, the Evidence Weight (Weight of Evidence, WoE) of each bin corresponding to the variable is calculated, and the tendency of WoE value to change with the variable is observed. WoE the mathematical definition is: WoE ═ ln (goodpattette/badattribute). goodtribute is the number of good cells in the box/total number of good cells in the data set; badattribute is the number of bad cells in the box/total number of bad cells in the dataset.

After obtaining WoE, the predictive power (IV) of the variable is calculated. IV is used to compare the predictive capabilities of the features, IV sum (goodpattette-basic) woe).

After obtaining the IV of each variable, the key performance index is screened. Generally, if the IV is greater than 0.1, the corresponding variable is considered to have the prediction capability, and if the IV is greater than 0.2, the corresponding variable is considered to have the strong prediction capability. In the embodiment of the invention, the key performance index with IV larger than 0.1 is selected, and the sample index of the key performance index is converted into the corresponding WoE value.

Training a logistic regression model based on the screened key performance index set to obtain independent coefficient and intercept (COE), and evaluating the fitting capability of the logistic regression model based on ROC (receiver operating regression) curve and AUC (area Under curve). Here, ROC refers to receiver operation characteristics, each point on the ROC curve reflects sensitivity to the same signal stimulus, fig. 4 is a schematic diagram of ROC curve fitting provided by an embodiment of the present invention, as shown in fig. 4, a horizontal axis represents negative and positive rate (FPR) specificity, and a vertical axis represents true rate (TPR) sensitivity. In fig. 4, AUC is the area under the ROC curve and is the area formed by the connection of the broken lines. Usually, the AUC is greater than 0.8, and the fitting degree of the logistic regression model is confirmed to meet the requirement.

Based on the independent variable coefficient in the logistic regression model and the scoring card model, an automatic scoring system is constructed as follows:

score＝A+B*(θ^Tx)＝A+B*(ω₀+ω₁x₁+…+ω_nx_n)＝(A+B*ω₀)+B*ω₁x₁+…+B*ω_nx_n

where score is the overall performance score, A and B are predetermined scores, θ is the variable weight, n is the number of variables, ω is₀Is a predetermined coefficient, ω₁，ω₂，…，ω_nIs the coefficient of independent variable in the logistic regression model, (A + B ω₀) Is the base score, B ω₁x₁，…，B*ω_nx_nIs the score assigned to each variable, as shown in the following table:

according to any of the above embodiments, the method further includes, after step 230: step 241, obtaining a time sequence of key performance indexes of the anchor cell; step 242, performing stationarity detection on the time sequence to obtain a stationarity detection result of the time sequence; performing randomness detection on the time sequence to obtain a randomness detection result of the time sequence; and 243, acquiring a dynamic detection result of the anchor point cell according to the stationarity detection result and the randomness detection result of the time sequence and a preset dynamic early warning threshold.

Specifically, after the parameter configuration of the anchor cell is completed, dynamic detection may also be performed on the anchor cell. Here, stationarity detection and randomness detection are respectively performed for a time series of any key performance index of the anchor cell:

the stationarity detection is used for detecting whether the time sequence is a stationary sequence or not, and if not, the time sequence is differentiated until the time sequence is stationary. Here, stationary means that the time series fluctuates around a constant up and down and the fluctuation range is limited, i.e. there is a constant mean and a constant variance. If there is a significant trend or periodicity, then the 1 st, 2 nd, … th, K th order differences are made in sequence until stationary, where the K th order difference is the subtraction between two sequence values that are K periods apart. The stationarity detection result of the time sequence of any key performance index obtained by the method is whether the time sequence is a stable sequence or not, and if the time sequence is not a stable sequence, the time sequence also comprises a difference hierarchy and an order.

The randomness test is used to detect whether the time sequence is a random sequence, i.e., a white noise sequence. There is no correlation between the values in a random sequence, and randomness tests are usually performed by building a linear model to fit the evolution of the sequence. And obtaining the randomness detection result of the time sequence of any key performance index, namely whether the time sequence is a random sequence.

The preset dynamic early warning threshold is the interval of results obtained by performing stationarity detection and randomness detection on a preset time sequence of key performance indexes in the normal running state of an anchor point cell. If the stationarity detection result or the randomness detection result of the time sequence of any key performance index does not meet the preset dynamic early warning threshold, the dynamic detection result of the anchor point cell is abnormal; otherwise, the dynamic detection result of the anchor cell is normal.

According to any of the above embodiments, the method further includes, before step 243: acquiring a historical time sequence of key performance indexes of an optimal cell corresponding to a scene classification result; performing stationarity detection on the historical time sequence to obtain a stationarity detection result of the historical time sequence; performing randomness detection on the historical time sequence to obtain a randomness detection result of the historical time sequence; and acquiring a preset dynamic early warning threshold based on the stationarity detection result and the randomness detection result of the historical time sequence.

Specifically, the preset dynamic early warning threshold is obtained by performing stationarity detection and randomness detection on historical time sequences of key performance indexes of the optimal cells belonging to the same scene category as the anchor cell.

Based on any of the above embodiments, in the method, the key performance index includes at least one of a wireless Call completing rate, a wireless drop rate, a handover success rate, an esvcc (Enhanced Single Radio Voice Call Continuity, Enhanced Single wireless Voice Call Continuity) handover success rate, a Voice over Long-Term Evolution (VOLTE) Voice Call completing rate, and a VOLTE Voice drop rate.

Based on any of the above embodiments, fig. 5 is a schematic flow chart of a method for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to another embodiment of the present invention, and as shown in fig. 5, the method includes:

s1, classifying the current LTE network cell:

acquiring work parameter data, KPI data, MR sampling point data, coverage simulation result data and raster data of an LTE current network cell, and calculating the following cell characteristic data based on the data, wherein the LTE current network cell is a sample cell, and the cell characteristic data comprises the work parameter data, cell performance index data, wireless environment data and coverage scene data, and specifically comprises the indexes shown in the following table:

based on a contour coefficient method, contour coefficients of cell characteristic data of LTE existing network cells under the number of a plurality of clusters are obtained, the number of clusters corresponding to the maximum value of the contour coefficients is selected as a K value, K-means clustering is carried out on the LTE existing network cells, and the sample scene category corresponding to each LTE existing network cell is obtained. Here the current network cell clusters are divided into J1, J2, …, Jn, where n equals the value of K.

S2: the comprehensive performance of the LTE current network cell is preferred:

the method comprises the steps of obtaining key performance indexes of an LTE current network cell, wherein the key performance indexes comprise access performance, maintenance performance and mobility KPI indexes, the access KPI indexes comprise a wireless call-on rate and a VoLTE voice call-on rate, the maintenance KPI indexes comprise a wireless call-off rate and a VoLTE voice call-off rate, and the mobility KPI indexes comprise a switching success rate and an eSRVCC success rate.

And training a logistic regression model based on key performance indexes of the LTE current network cell, and constructing an automatic scoring system based on independent variable coefficients in the logistic regression model and a scoring card model. And based on an automatic scoring system, acquiring the comprehensive performance score of each LTE current network cell, and selecting an optimal cell and an optimal cell corresponding to any sample scene type based on the comprehensive performance score of each LTE current network cell corresponding to the sample scene type.

S3: the 5G NSA network access LTE anchor point cell parameter self-learning configuration comprises the following steps:

the method comprises the following steps of taking working parameter data of an LTE current network cell as characteristics, taking a sample scene category as a label, and building a first classification model, wherein the flow is as follows:

inputting: training set T { (x)₁,y₁),(x₂,y₂),…,(x_N,y_N) }; the working parameter data is x, and the sample scene category is y;

and (3) outputting: lifting tree f_K(x)；

a) Initialization f₀(x) (ii) a Here, f₀(x) An initial lifting tree;

b) for K ═ 1,2, …, K do; wherein K is the total number of iterations;

c) calculating residual error

Wherein f is_k-1(x) Is the lifting tree for the k-1 iteration; l () is a loss function; r is_kiIs the residual error of the kth iteration;

d) fitting residual r_kiLearning a regression tree to obtain T (x; theta)_k) (ii) a Wherein T (x; theta)_k) For the kth decision tree, θ_kIs a decision tree parameter;

e) updating the tree f_k(x)＝f_k-1(x)+T(x,θ_k) (ii) a Wherein f is_k(x) And f_k-1(x) Respectively the kth and the kth-1 lifting tree;

f)end for

g) obtain a lifting tree

Wherein f is_K(x) I.e. the lifting tree obtained in the K-th iteration.

Inputting the work parameters of the anchor point cell into a first classification model, and acquiring a first classification result output by the first classification model; and taking the cell parameter of the optimal cell corresponding to the first classification result as temporary configuration, and configuring the anchor point cell.

Judging whether a service alarm exists in the anchor point cell in real time, if so, carrying out manual intervention analysis, and otherwise, collecting busy hour characteristic data of the anchor point cell; here, the busy hour characteristic data includes work parameter data, cell performance index data, wireless environment data, and coverage scenario data.

And inputting the busy hour characteristic data into a second classification model obtained by training the busy hour characteristic data of the LTE current network cell as a characteristic and the sample scene category as a label, and obtaining a scene classification result output by the second classification model. And calculating the similarity between each preferred cell and the anchor cell by applying an Euclidean distance algorithm based on the cell performance index data, the wireless environment data and the coverage scene data of each preferred cell and the anchor cell corresponding to the scene classification result, and selecting the cell parameter corresponding to the preferred cell with the highest similarity from the similarity to configure the anchor cell.

S4: the 5G network access LTE anchor point cell performance can be dynamically monitored and early-warned:

and acquiring a historical time sequence of key performance indexes of the optimal cell corresponding to the scene classification result of the anchor point cell, performing stability detection and randomness detection on the historical time sequence, and setting a preset dynamic early warning threshold by using the obtained stability detection result and randomness detection result of the historical time sequence as a preset stability detection characteristic and a preset randomness detection characteristic.

Acquiring a time sequence of key performance indexes of an anchor point cell, performing stability detection and randomness detection on the time sequence, comparing the obtained stability detection result and randomness detection result with a preset dynamic early warning threshold, performing manual intervention analysis if the preset dynamic early warning threshold is triggered, and otherwise, solidifying the cell parameter configuration of the anchor point cell.

The method provided by the embodiment of the invention fully excavates the value of the LTE data of the current network, and performs parameter configuration analysis on the LTE new access anchor point cell under the 5G NSA networking by taking the LTE cell of the current network as a research object, thereby better ensuring the stability of the current network and the perception of users. In addition, the method provided by the embodiment of the invention can directly research the parameter configuration of the 5G cell based on the KPI, the user distribution and the wireless environment of the 5G cell. Moreover, the embodiment of the invention fully exerts the machine learning efficiency, improves the self-configuration accuracy of the LTE new access anchor point cell parameters under the 5G NSA networking and reduces the manual parameter configuration investment. Finally, the embodiment of the invention comprehensively considers the different service models and different KPI performance requirements under the cells of different scene types, and combines the current network data to design a dynamic monitoring and early warning mechanism, thereby improving the LTE network access cell performance monitoring accuracy under 5G NSA and ensuring the stable network access of the cell.

Based on any of the above embodiments, fig. 6 is a schematic structural diagram of a device for configuring parameters of an LTE new access anchor cell under 5G NSA networking according to an embodiment of the present invention, and as shown in fig. 6, the device includes a cell feature obtaining unit 610, a scene classification unit 620, and a parameter configuration unit 630;

the cell characteristic acquiring unit 610 is configured to acquire cell characteristic data of an anchor cell; the cell characteristic data at least comprises work parameter data;

the scene classification unit 620 is configured to input the cell feature data into a scene classification model, and obtain a scene classification result output by the scene classification model; the scene classification model is obtained based on sample cell characteristic data of a sample cell and sample scene class training;

the parameter configuration unit 630 is configured to configure the anchor cell based on the cell parameter of the preferred cell corresponding to the scene classification result; the preferred cell is selected from the sample cells corresponding to the scene classification result.

The device provided by the embodiment of the invention obtains the scene classification result of the anchor point cell based on the scene classification model, and realizes the accurate classification of the anchor point cell so as to meet the parameter configuration requirements under different scene categories. And configuring an anchor cell based on the cell parameter of the preferred cell corresponding to the scene classification result, and compared with the issuing of a unified configuration template, fully utilizing the value of the existing data of the existing network and effectively ensuring the stability of cell network access.

According to any one of the above embodiments, in the apparatus, the scene classification model includes a first classification model and a second classification model;

correspondingly, the scene classification unit 620 is specifically configured to:

Based on any of the above embodiments, in the apparatus, the parameter configuration unit 630 specifically includes:

Based on any of the above embodiments, in the apparatus, the similarity calculation subunit is specifically configured to:

Based on any embodiment above, the apparatus further comprises:

Based on any of the above embodiments, in the apparatus, the sample cell clustering unit is specifically configured to:

Based on any embodiment above, the apparatus further comprises:

Based on any of the above embodiments, in the apparatus, the scoring system training unit is specifically configured to:

performing data cleaning on the key performance index set;

Based on any one of the above embodiments, the device further comprises a dynamic early warning unit; the dynamic early warning unit is used for:

acquiring a time sequence of key performance indexes of the anchor cell;

Based on any of the above embodiments, the apparatus further comprises an early warning threshold acquisition unit; the early warning threshold acquisition unit is used for:

Based on any of the above embodiments, in the apparatus, the key performance index includes at least one of a radio access rate, a radio drop rate, a handover success rate, an eSRVCC handover success rate, a VOLTE voice access rate, and a VOLTE voice drop rate.

Fig. 7 is a schematic entity structure diagram of an electronic device according to an embodiment of the present invention, and as shown in fig. 7, the electronic device may include: a processor (processor)701, a communication Interface (Communications Interface)702, a memory (memory)703 and a communication bus 704, wherein the processor 701, the communication Interface 702 and the memory 703 complete communication with each other through the communication bus 704. The processor 701 may invoke a computer program stored in the memory 703 and executable on the processor 701 to execute the method for configuring the cell parameter of the LTE new access anchor under 5G NSA networking provided in the foregoing embodiments, including: acquiring cell characteristic data of an anchor cell; the cell characteristic data at least comprises work parameter data; inputting the cell characteristic data into a scene classification model, and acquiring a scene classification result output by the scene classification model; the scene classification model is obtained based on sample cell characteristic data of a sample cell and sample scene class training; configuring the anchor cell based on the cell parameter of the preferred cell corresponding to the scene classification result; the preferred cell is selected from the sample cells corresponding to the scene classification result.

In addition, the logic instructions in the memory 703 can be implemented in the form of software functional units and stored in a computer readable storage medium when the software functional units are sold or used as independent products. Based on such understanding, the technical solutions of the embodiments of the present invention may be essentially implemented or make a contribution to the prior art, or may be implemented in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the methods described in the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.

An embodiment of the present invention further provides a non-transitory computer-readable storage medium, where a computer program is stored on the non-transitory computer-readable storage medium, and when executed by a processor, the computer program is implemented to execute a method for configuring a cell parameter of an anchor point of LTE new access under 5G NSA networking, where the method includes: acquiring cell characteristic data of an anchor cell; the cell characteristic data at least comprises work parameter data; inputting the cell characteristic data into a scene classification model, and acquiring a scene classification result output by the scene classification model; the scene classification model is obtained based on sample cell characteristic data of a sample cell and sample scene class training; configuring the anchor cell based on the cell parameter of the preferred cell corresponding to the scene classification result; the preferred cell is selected from the sample cells corresponding to the scene classification result.

The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.

Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.

Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.

Claims

1. A method for configuring parameters of an anchor point cell for LTE new access under 5G NSA networking is characterized by comprising the following steps:

2. The method for configuring parameters of anchor cell for LTE new access under 5G NSA networking according to claim 1, wherein the scene classification model comprises a first classification model and a second classification model;

3. The method for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to claim 1, wherein configuring the anchor cell based on the cell parameters of the preferred cell corresponding to the scene classification result specifically includes:

4. The method for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to claim 3, wherein the obtaining of the similarity between each preferred cell corresponding to the scene classification result and the anchor cell specifically includes:

and calculating the similarity of any preferred cell and the anchor cell based on the cell performance index data, the wireless environment data and the coverage scene data of any preferred cell and the anchor cell.

5. The method for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to claim 1, wherein the step of inputting the cell feature data into a scene classification model to obtain a scene classification result output by the scene classification model further comprises:

6. The method for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to claim 5, wherein the clustering the sample cell feature data sets to obtain a sample scene category corresponding to each sample cell specifically comprises:

7. The method for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to claim 5, wherein the clustering the sample cell feature data sets to obtain the sample scene category corresponding to each of the sample cells further comprises:

8. The method for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to claim 1, wherein the configuring the anchor cell based on the cell parameters of the preferred cell corresponding to the scene classification result comprises:

9. The method for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to claim 8, wherein the training of a scoring card model based on the set of key performance indicators to obtain an automatic scoring system specifically comprises:

performing data cleaning on the key performance index set;

10. The method for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to claim 1, wherein the configuring the anchor cell based on the cell parameters of the preferred cell corresponding to the scene classification result further comprises:

acquiring a time sequence of key performance indexes of the anchor cell;

11. The method for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to claim 10, wherein the method for acquiring the dynamic detection result of the anchor cell according to the stationarity detection result and the randomness detection result of the time sequence and a preset dynamic early warning threshold further comprises:

12. The method for configuring cell parameters of an anchor point of LTE new access under 5G NSA networking according to any of claims 8 to 11, wherein the key performance indicators include at least one of a radio connection rate, a radio drop rate, a handover success rate, an eSRVCC handover success rate, a VOLTE voice connection rate, and a VOLTE voice drop rate.

13. The utility model provides a LTE new access anchor point district parameter configuration device under 5G NSA networking which characterized in that includes:

14. The device for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to claim 13, wherein the scene classification model includes a first classification model and a second classification model;

correspondingly, the scene classification unit is specifically configured to:

15. The device for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to claim 13, wherein the parameter configuring unit specifically includes:

16. The device for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to claim 15, wherein the similarity operator unit is specifically configured to:

17. The device for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to claim 13, further comprising:

18. The device for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to claim 17, wherein the sample cell clustering unit is specifically configured to:

19. The device for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to claim 17, further comprising:

20. The device for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to claim 13, further comprising:

21. The device for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to claim 20, wherein the scoring system training unit is specifically configured to:

performing data cleaning on the key performance index set;

22. The device for configuring parameters of an anchor cell for LTE new access under 5G NSA networking according to claim 13, further comprising a dynamic early warning unit; the dynamic early warning unit is used for:

acquiring a time sequence of key performance indexes of the anchor cell;

23. The device for configuring parameters of an anchor point cell for LTE new access under 5G NSA networking according to claim 22, further comprising an early warning threshold acquisition unit; the early warning threshold acquisition unit is used for:

24. The device for configuring cell parameter of anchor point of LTE new access under 5G NSA networking according to any of claims 20 to 23, wherein the key performance index includes at least one of radio connection rate, radio drop rate, handover success rate, eSRVCC handover success rate, VOLTE voice connection rate, and VOLTE voice drop rate.

25. 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 steps of the method for configuring the parameters of the LTE new access anchor cell under 5G NSA networking according to any one of claims 1 to 12 when executing the program.

26. A non-transitory computer readable storage medium, having a computer program stored thereon, wherein the computer program, when being executed by a processor, implements the steps of the method for configuring the LTE new access anchor cell parameters under 5G NSA networking according to any one of claims 1 to 12.