CN111368384B - Method and equipment for predicting antenna engineering parameters - Google Patents

Method and equipment for predicting antenna engineering parameters Download PDF

Info

Publication number
CN111368384B
CN111368384B CN201811502499.0A CN201811502499A CN111368384B CN 111368384 B CN111368384 B CN 111368384B CN 201811502499 A CN201811502499 A CN 201811502499A CN 111368384 B CN111368384 B CN 111368384B
Authority
CN
China
Prior art keywords
antenna
data
parameter
engineering
model
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201811502499.0A
Other languages
Chinese (zh)
Other versions
CN111368384A (en
Inventor
喻国军
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Huawei Technologies Co Ltd
Original Assignee
Huawei Technologies Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Huawei Technologies Co Ltd filed Critical Huawei Technologies Co Ltd
Priority to CN201811502499.0A priority Critical patent/CN111368384B/en
Publication of CN111368384A publication Critical patent/CN111368384A/en
Application granted granted Critical
Publication of CN111368384B publication Critical patent/CN111368384B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Abstract

The application provides a method and equipment for predicting or correcting antenna engineering parameters, wherein the method comprises the following steps: the method comprises the steps of firstly carrying out model training based on an engineering parameter set containing engineering parameters of a first equipment antenna in a first geographic area, a configuration data set containing configuration data of the first equipment antenna and a measurement report set uploaded to the first equipment antenna by a terminal to obtain an antenna engineering parameter prediction model, then inputting the configuration data of a second equipment antenna in a second geographic area and measurement report data uploaded to the second equipment antenna by the terminal to the antenna engineering parameter prediction model to obtain a prediction result of the engineering parameters of the second equipment antenna or correct the engineering parameters of the second equipment antenna. The method can realize more accurate prediction and correction of the antenna engineering parameters with lower cost.

Description

Method and equipment for predicting antenna engineering parameters
Technical Field
The invention relates to the technical field of communication, in particular to a training method and equipment of an antenna engineering parameter prediction model and a method and equipment for predicting engineering parameters based on the antenna engineering parameter prediction model.
Background
Long Term Evolution (LTE) is a Long Term Evolution of the Universal Mobile Telecommunications System (UMTS) technology standard established by the 3rd Generation Partnership Project (3 GPP) organization. With the continuous expansion of the construction scale of the LTE network and the rapid increase of LTE users, the accuracy of the engineering parameters (herein, "engineering parameters" may be referred to as "engineering parameters") of the base station becomes more and more important in the optimization and adjustment of the daily network. The engineering parameters of the base station refer to parameters related to the radio frequency antenna of the base station in the wireless network planning, such as longitude and latitude, azimuth angle, downtilt angle and the like of the antenna. The accuracy of the engineering parameters of the base station is related to the accuracy of network data and network coverage, and has an extremely important influence on user terminal perception and network problem analysis.
However, the traditional checking of the accuracy of the engineering parameters is always a big short board in network optimization, and the existing method is to test, analyze and verify the accuracy of the engineering parameters in a manual standing mode. However, the manual checking of the engineering parameters may have some errors introduced manually, and it is difficult to ensure the timeliness and integrity of the engineering parameters, so the checking efficiency is low and the accuracy is poor. Another conventional method is to install a Positioning device, such as a Global Positioning System (GPS) device, at a base station, obtain position information of a radio-frequency antenna of the base station through the Positioning device, send the position information to a network management device, and then generate a working parameter according to the position information of the radio-frequency antenna of the base station by the network management device. However, this method may cause a large equipment overhead, and some base station equipment may be installed indoors, and in this case, the positioning equipment has a low indoor positioning accuracy, resulting in a poor accuracy of the work parameter.
Disclosure of Invention
The embodiment of the invention provides a method and equipment for predicting antenna engineering parameters, which can overcome the defects of the prior art and realize a relatively accurate and low-cost working parameter generation scheme.
In a first aspect, an embodiment of the present invention provides a method for predicting antenna engineering parameters, where the method includes a training method for an antenna engineering parameter prediction model, and the method specifically includes: acquiring a first engineering parameter set, a first configuration data set and a first measurement report data set uploaded to a first equipment antenna by a terminal; wherein the first set of engineering parameters includes engineering parameters of the first device antenna including at least one of position data (e.g., longitude, latitude, altitude, etc. of the first device antenna) and attitude data (e.g., downtilt, azimuth, etc. of the first device antenna) of the first device antenna; the first set of configuration data comprises configuration data of the first device antenna, the configuration data of the first device antenna representing configuration information of network parameters of the first device antenna; measurement Report (MR) data in the first Measurement Report data set includes location data (e.g., longitude, latitude, altitude, etc. of the terminal) and Signal Received Power (RSRP) data of the terminal; wherein the first device antenna may be any of a plurality of device antennas within a first geographic area; performing model training according to the engineering parameters of the first equipment antenna, the configuration data of the first equipment antenna and the first measurement report data set to obtain an antenna parameter prediction model (which may be referred to as a parameter prediction model herein for short); and the antenna engineering parameter prediction model is used for outputting the engineering parameters of the second equipment antenna according to a second configuration data set and a second measurement report data set uploaded to the second equipment antenna by the terminal. Wherein the second device antenna may be a device antenna within a second geographic area, which may be different from the first geographic area.
It can be seen that the embodiment of the present invention is capable of constructing a model for predicting the parameters of the device antenna based on the existing sample data (such as MR data, configuration data, parameter data, and the like) by means of model training. In this way, subsequent application of the model will enable prediction parameters to be obtained with a high degree of confidence in the device antenna based on MR data and configuration data. Therefore, the technical scheme of the embodiment of the invention can overcome the defects of the prior art, effectively reduce the acquisition cost of the working parameters of the equipment antenna and improve the accuracy of the working parameters.
In the embodiment of the present invention, the first geographic area represents a geographic position range where one or more device antennas corresponding to sample data used for model training are located. If the sample data used for model training corresponds to multiple device antennas, the multiple device antennas may be referred to as multiple device antennas in the first geographic area, and so on, the first device antenna in this embodiment may be referred to as a first device antenna in the first geographic area, and other device antennas except the first device antenna in the multiple device antennas may be referred to as other device antennas in the first geographic area, and so on.
Based on the first aspect, in a first implementation, the antenna parameter prediction model includes an antenna parameter generating model (which may be referred to herein simply as a parameter generating model); the performing model training according to the engineering parameters of the first device antenna, the configuration data of the first device antenna, and the first measurement report data set to obtain an antenna engineering parameter prediction model includes: obtaining first sample characteristic data of the first device antenna from the configuration data of the first device antenna and the first measurement report data set; the first sample characteristic data includes a plurality of signal reception power data of a cell or a Remote Radio Unit (RRU) belonging to the first device antenna, and position data of a terminal corresponding to each signal reception power data in the plurality of signal reception power data; obtaining the antenna type of the first equipment antenna according to the configuration data of the first equipment antenna; performing model training according to the engineering parameters and the first characteristic set of the first equipment antenna to obtain an antenna engineering parameter generation model; the first feature set comprises the first sample feature data and an antenna type of the first device antenna, and the antenna parameter generation model is used for outputting engineering parameters according to the input first feature set.
The antenna parameter generation model is, for example, a Neural Networks (NN) algorithm model. The training process of the antenna parameter generation model can be represented by the following formula:
(Latitude,Longtitude)＝NN(Feature1056,AntennaType,Wnn1)
wherein Latitude represents Latitude value in working parameter of equipment antenna, Longtitude represents longitude value in working parameter of equipment antenna, NN represents neural network algorithm, Feature1056Represents the first sample characteristic data, antenna type represents the type of the device antenna, Wnn1Representing model parameters in the parameter generation model.
For different device antennas, their corresponding Latitude, Longtitude and Feature1056The AntennaType data are different, and the W can be calculated by training the model by taking the AntennaType data as the input data of the engineering parameter generation modelnn1(for example, in this example, W can be calculated by a gradient descent methodnn1) And obtaining the trained parameter generation model.
It can be seen that the embodiment of the present invention can perform data extraction based on the existing sample data (e.g. MR data, configuration data, working parameter data, etc.), obtain input data of the working parameter generation model (e.g. first sample feature data, type data of the device antenna, working parameter data of the device antenna, etc., therefore, a working parameter prediction model for predicting the device antenna is trained based on the data (the working parameter prediction model can be regarded as a working parameter generation model), and the working parameter of the device antenna can be generated by applying the model, therefore, the method and the device can be suitable for various sample data scenes to effectively screen data, and improve the efficiency and accuracy of model training, so that the accuracy of subsequent working parameter prediction of the equipment antenna based on the model can be improved, and the acquisition cost of the working parameters of the equipment antenna can be effectively reduced.
Based on the first implementation manner of the first aspect, in a possible embodiment, when the amount of MR data in the training set is large, the occupied memory is large, which may result in a large amount of MR data of the common antenna cell. The common antenna cell lists corresponding to different device antennas are also different, and the number of cells is not fixed. In order to better train the model (e.g., avoid overfitting and improve the operation speed and efficiency), the embodiment of the present invention may design a uniform training data template for the MR data corresponding to the common antenna cells of different device antennas, so that the MR data corresponding to the common antenna cells of each device antenna may be screened and merged based on the training data template to obtain sample feature data (which may be referred to as first sample feature data) of the common antenna cells of each device antenna.
In obtaining first sample characterization data for a device antenna, the obtaining first sample characterization data from configuration data for the first device antenna and the first measurement report data set comprising: determining a cell or RRU (radio remote unit) belonging to the first equipment antenna according to the configuration data of the first equipment antenna; determining measurement report data corresponding to the cell or the RRU from the measurement report data set; and performing feature extraction according to the measurement report data corresponding to the cell or the RRU to obtain the first sample feature data.
For example, in a specific implementation, a range of RSRP values of each cell is, for example, {1,4,7, …,97}, and for a single device antenna, for example, the device antenna 1, RSRP values of all User Equipments (UEs) of MR data of a common antenna cell of the device antenna 1, where the RSRP value of a serving cell (for example, the cell 1) is a predetermined value (for example, the predetermined value is 7, although any other value may be selected) may be added together to obtain a mean value, so as to obtain a central point, and the central point may be approximately regarded as a possible longitude and latitude position of the base station device. Then, starting from the center point, the light beam is uniformly directed in a plurality of directions (for example8 directions) extend the exit line. Then, when the RSRP values of the serving cells are 1,4,7, …, and 97 (33) respectively, from the MR data of each of the co-antenna cells of the device antenna 1, the MR data closest to the position point with the same value of each directional ray can be searched (if there is no such MR data, one MR data can be constructed with all 0 s instead). Thus, a total of 8 × 33 — 264 sets of MR data are found. Then, for each of the 264 sets of low-dimensional MR data, the first sample feature data of the co-antenna cell is selected by extracting two or more of features such as the longitude where the UE is located, the latitude where the UE is located, the altitude where the UE is located, and the Angle of arrival (AOA) of the cell 1. For example, when 4 features, i.e., the longitude of the UE, the latitude of the UE, the altitude of the UE, and the AOA of cell 1, are extracted at the same time, 4 × 264 sub-features, i.e., 1056 sub-features are generated, and the first sample Feature data composed of the 1056 sub-features is referred to as "Feature 1056”。
Therefore, the embodiment can adapt to various sample data scenes, and the operation speed and efficiency of the model training process are improved, so that a better engineering parameter generation model is trained.
Based on the first aspect, in a second implementation manner, the antenna parameter prediction model further includes an antenna parameter correction model (also referred to as an "parameter correction model" for short herein) in addition to the antenna parameter generation model; in this case, the set of engineering parameters further comprises engineering parameters of at least one other device antenna; the configuration data set further comprises configuration data of the at least one other device antenna; the at least one other device antenna represents a device antenna of the plurality of device antennas other than the first device antenna;
correspondingly, after the model training is performed according to the engineering parameters and the first feature set of the first device antenna to obtain the antenna engineering parameter generation model, the method further includes: obtaining second sample characteristic data of the first device antenna according to the configuration data set and the measurement report data set, where the second sample characteristic data includes multiple pieces of signal reception power data of the cell or the RRU of the at least one other device antenna, and location data of a terminal corresponding to each piece of signal reception power data in the multiple pieces of signal reception power data and signal reception power data of the cell or the RRU of the first device antenna; obtaining a prediction result of the engineering parameters of the first equipment antenna and a prediction result of the engineering parameters of the at least one other equipment antenna according to the antenna engineering parameter generation model; performing model training according to the engineering parameter set and the second characteristic set to obtain the antenna engineering parameter correction model; the second feature set comprises the second sample feature data, the predicted result of the engineering parameter of the first equipment antenna and the predicted result of the engineering parameter of the at least one other equipment antenna, and the antenna engineering parameter correction model is used for outputting the engineering parameter according to the input second feature set.
The antenna parameter correction model is, for example, a Neural Network (NN) algorithm model. The training process of the working parameter correction model can be represented by the following formula:
(Latitude,Longtitude)＝NN((Featurejoin_i,Featurebasic_i),Wnn2)
wherein Latitude represents Latitude value in the working parameter of the device antenna i, Longtitude represents longitude value in the working parameter of the device antenna i, NN represents neural network algorithm, Featurejoin_iSecond sample characteristic data, Feature, representing a top N antenna of the device antennabasic_iPredicted working parameter, W, for the (1+ Top N) antenna group representing device antenna inn2And representing model parameters in the working parameter correction model.
Therefore, for different device antennas, their corresponding Latitude, Longtitude and Featurejoin_i、Featurebasic_iThe data are different, and the W can be calculated by training the model according to the data as the input data of the working parameter correction modelnn2(for example, in this example, W can be calculated by a gradient descent methodnn2) And obtaining the trained parameter correction model.
It can be seen that the embodiments of the present invention can perform data extraction based on off-the-shelf sample data (such as MR data, configuration data, engineering parameter data, etc.), and obtain the prediction result of the engineering parameter of the first device antenna and the prediction result of the engineering parameter of the at least one other device antenna according to the engineering parameter generation model trained by the first implementation manner of the first aspect, thereby forming input data for an antenna parameter correction model (e.g., the second sample characteristic data, the predicted result of the engineering parameter of the first device antenna and the predicted result of the engineering parameter of the at least one other device antenna, a set of engineering parameters, etc.), thereby further training the antenna parameter correction model based on these data, the working parameter correction model can further correct the predicted working parameters of the working parameter generation model, so that the predicted working parameters with higher reliability are obtained. The antenna parameter prediction model of the embodiment comprises an antenna parameter generation model and an antenna parameter correction model, and by applying the model, the generation and correction of the parameter of the device antenna can be realized, so that the prediction parameter with high reliability is obtained. The embodiment of the invention can be suitable for various sample data scenes to effectively screen data and improve the efficiency and accuracy of model training. Therefore, the accuracy of subsequent work parameter prediction of the equipment antenna based on the model can be improved, and the cost for acquiring the work parameters of the equipment antenna is effectively reduced.
Based on the second implementation manner of the first aspect, in a possible embodiment, the at least one other device antenna is top N device antennas around the first device antenna, where the top N device antennas represent N device antennas most relevant to the first device antenna among the multiple device antennas, and N is an integer greater than or equal to 1.
For example, in some embodiments, the top N antenna is the N most peripheral device antennas in the plurality of device antennas peripheral to the first device antenna that are spaced the greatest distance from the first device antenna. (ii) a In still other embodiments, the top N antenna is the N device antennas with the largest signal overlapping degree with the first device antenna among the plurality of device antennas around the first device antenna; in still other embodiments, the top N antenna is the N device antennas with the largest number of terminal switching times among the peripheral multiple device antennas of the first device antenna, and so on.
The embodiment of the invention trains the working parameter correction model based on the N most relevant equipment antennas of the first equipment antenna, and is beneficial to correcting the predicted working parameters of the first equipment antenna based on the working parameters of the N most relevant equipment antennas of the first equipment antenna, thereby training a better working parameter correction model and improving the budget speed and the prediction accuracy of the working parameter correction model. Therefore, the implementation of the embodiment is beneficial to integrally improving the reliability of the prediction result of the work parameter prediction model (i.e. including the work parameter generation model and the work parameter correction model).
In a possible embodiment, the obtaining second sample feature data according to the configuration data set and the measurement report data set includes: determining a cell or RRU (radio remote unit) belonging to the first equipment antenna and a cell or RRU belonging to the top N equipment antennas according to the configuration data set; setting at least one piece of measurement report data corresponding to a cell of the top N device antennas or a cell of any one of the RRUs according to the first measurement report data set, wherein each piece of measurement report data in the at least one piece of measurement report data comprises signal receiving power data of the cell of any one of the device antennas or the RRU, signal receiving power data of the cell of the first device antenna or the RRU and position data of the terminal; and performing feature extraction according to the cell of the top N equipment antennas or the cell of each equipment antenna in the RRUs or the measurement report data corresponding to the RRU to obtain the second sample feature data.
The second sample characteristic data characterizes measurement characteristics (or antenna joint measurement characteristics) respectively presented by different device antennas in the same measurement of the UE (i.e. in the same UE geographical location).
For example, in a specific implementation, for a top N antenna of any device antenna, for example, the top N antenna of the device antenna a, a cell is determined according to a common antenna cell list of each neighboring antenna of the device antenna a, and the cell is referred to as a neighboring cell of the device antenna a, then, N neighboring antennas respectively determine N neighboring cells, and such N neighboring cells may be referred to as top N neighboring cells of the device antenna a. For example, for a first neighboring antenna in the top N antenna of the device antenna a, if there are multiple cells in the co-antenna cell of the first neighboring antenna, a cell with the largest occurrence number in the MR data of the low latitude of the first neighboring antenna may be exemplarily selected as a neighboring cell corresponding to the first neighboring antenna. By analogy, the adjacent cells corresponding to each adjacent antenna in the top N antennas, that is, the top N adjacent cells of the device antenna a, can be determined respectively. Then, for any neighboring cell in the top N neighboring cells of any device antenna, for example, any neighboring cell in the top N neighboring cells of the device antenna a, M MR data may be selected from the multiple MR data of the device antenna a, and the M MR data may be associated with the neighboring cell. Where each of the M MR data includes measurement characteristic information of the neighboring cell (e.g., RSRP of the neighboring cell, AOA of the neighboring cell, etc.), for example, RSRP values of neighboring cells in M low-dimensional MR data may be different, where M is an integer greater than or equal to 1. In this way, the second sample feature data of the device antenna a may be extracted from each of the M pieces of low-dimensional MR data, and each sample feature data may include two or more of positioning information of the UE (e.g., longitude of the UE, latitude of the UE, altitude of the UE, etc.), RSRP of the serving cell, AOA of the serving cell, RSRP of the neighbor cell, AOA of the neighbor cell, etc. That is, based on the above description, M low-dimensional MR data associated with any one of the top N neighbor cells may be determined, and M second sample feature data may be determined based on the M low-dimensional MR data. For convenience of description, it may be noted that the M second sample characteristic data of the top N neighbor cell of the device antenna i is "Featurejoin _ i", and the device antenna i is any device antenna in the plurality of device antennas.
Therefore, the embodiment can adapt to various sample data scenes, and the operation speed and efficiency of the model training process are improved, so that a better working parameter correction model is trained.
Based on the first implementation manner and the second implementation manner of the first aspect, in a possible embodiment, before performing model training, feature information selection may also be performed on MR data corresponding to a common antenna cell of an antenna of a device, so as to obtain low-dimensional MR data of the common antenna cell.
That is to say, in an optional embodiment, K dimensions of feature information are selected for MR data corresponding to each cell of the common antenna, so as to obtain low-dimension MR data, where K is an integer greater than or equal to 2.
For example, in some embodiments, the feature information of K dimensions that needs to be selected includes two or more of the following: the location information of the UE includes, for example, a longitude where the UE is located, a latitude where the UE is located, optionally an altitude where the UE is located, and the like; the respective IDs of the co-antenna cells, for example, the ID of cell 1 (for example, cell 1 is a serving cell), the ID of cell 2 is …, and the ID of cell J is an integer of 1 or more. The RSRPs of the co-antenna cells, e.g., RSRP of cell 1, RSRP of cell 2 …, RSRP of cell N, etc. The AOAs of the co-antenna cells, e.g., AOA of cell 1, AOA of cell 2, AOA of cell …, AOA of cell N, etc.
Correspondingly, in this case, the obtaining the first sample characteristic data according to the configuration data of the first device antenna and the measurement report data set specifically includes: determining a cell or RRU (radio remote unit) belonging to the first equipment antenna according to the configuration data of the first equipment antenna; determining low-dimensional measurement report data corresponding to the cell or the RRU from the measurement report data set; and performing feature extraction according to the low-dimensional measurement report data corresponding to the cell or the RRU to obtain the first sample feature data.
Correspondingly, in this case, the obtaining second sample characteristic data according to the configuration data set and the measurement report data set specifically includes: determining a cell or RRU (radio remote unit) belonging to the first equipment antenna and a cell or RRU belonging to the top N equipment antennas according to the configuration data set; and setting at least one piece of low-dimensional measurement report data corresponding to the cell of the top N device antennas or the cell of any one of the RRUs according to the first measurement report data set, wherein each piece of low-dimensional measurement report data in the at least one piece of low-dimensional measurement report data comprises the signal reception power data of the cell of any one of the device antennas or the RRU, the signal reception power data of the cell of the first device antenna or the RRU and the position data of the terminal. And performing feature extraction according to the cell of the top N device antennas or the cell of each device antenna in the RRUs or the measurement report data corresponding to the RRU, and obtaining the second sample feature data.
Therefore, the embodiment of the invention is beneficial to reducing the computational complexity and improving the efficiency of model training.
In a second aspect, an embodiment of the present invention provides a method for predicting antenna engineering parameters, where the method includes a method for predicting working parameters based on an antenna working parameter prediction model, and the method specifically includes: acquiring a second configuration data set and a second measurement report data set uploaded to the second equipment antenna by the terminal; wherein the second configuration data set comprises configuration data of the second device antenna, the configuration data of the second device antenna representing configuration information of network parameters of the second device antenna; the measurement report data in the second measurement report data set comprises position data (e.g. longitude, latitude, altitude, etc. of the terminal) and signal received power data (RSRP data for short) of the terminal; wherein the second device antenna may be any of a plurality of device antennas within a second geographic area; inputting the second configuration data set and the second measurement report data set into an antenna parameter prediction model (which may be referred to as a parameter prediction model herein for short), and obtaining a prediction result of the engineering parameter of the second device antenna; the antenna engineering parameter prediction model is obtained by training according to a first engineering parameter set, a first configuration data set and a first measurement report data set uploaded to a first equipment antenna by a terminal; the engineering parameters of the second device antenna include at least one of position data (e.g., longitude, latitude, altitude, etc. of the second device antenna) and attitude data (e.g., downtilt, azimuth, etc. of the second device antenna) of the first device antenna; the first device antenna may be any of a plurality of device antennas within a first geographic area, and the second geographic area may be different from the first geographic area.
Wherein the first set of engineering parameters comprises engineering parameters of the first device antenna including at least one of position data and attitude data of the first device antenna; the first set of configuration data comprises configuration data of the first device antenna, the configuration data of the first device antenna representing configuration information of network parameters of the first device antenna; measurement report data in the first measurement report data set comprises position data and signal received power data of the terminal; the first device antenna is any of a plurality of device antennas within the first geographic area.
It can be seen that, according to the embodiment of the present invention, the pre-trained antenna parameter prediction model is input to the model based on the ready-made sample data (e.g., MR data, configuration data, etc.), so that the more accurate parameter of the device antenna can be obtained. Therefore, the technical scheme of the embodiment of the invention can overcome the defects of the prior art, effectively reduce the acquisition cost of the working parameters of the equipment antenna and improve the accuracy of the working parameters.
In the embodiment of the present invention, the second geographical area indicates a geographical location range where one or more device antennas of the working parameter need to be predicted in practical application. If the multiple device antennas correspond to the sample data used for the engineering parameter prediction, the multiple device antennas may be referred to as multiple device antennas in the second geographic area, and in the similar way, the second device antenna in this embodiment may be referred to as the second device antenna in the second geographic area, and the other device antennas except the second device antenna in the multiple device antennas may be referred to as other device antennas in the second geographic area. The geographical location range represented by the second geographical area may be different from the geographical location range represented by the first geographical area, that is, the second device antenna of the second geographical area may be different from the first device antenna of the first geographical area, and the plurality of device antennas of the second geographical area may be different from the plurality of device antennas of the first geographical area.
Based on the second aspect, in a first embodiment, the antenna parameter prediction model includes an antenna parameter generative model (which may also be referred to herein as simply an antenna parameter generative model); inputting the second configuration data set and the second measurement report data set into an antenna engineering parameter prediction model to obtain a prediction result of the engineering parameters of the second equipment antenna, including: obtaining first sample characteristic data of the second device antenna according to the second measurement report data set and the configuration data of the second device antenna; the first sample characteristic data of the second device antenna comprises a plurality of signal receiving power data of a cell or a Remote Radio Unit (RRU) which is subordinate to the second device antenna and position data of a terminal corresponding to each signal receiving power data in the plurality of signal receiving power data; obtaining the antenna type of the second equipment antenna according to the configuration data of the second equipment antenna; and inputting the first sample characteristic data and the antenna type of the second equipment antenna into the antenna engineering parameter generation model to obtain a first prediction result of the engineering parameter of the second equipment antenna.
The antenna engineering parameter generation model is obtained by performing model training according to engineering parameters and a first feature set of the first equipment antenna in the first geographic area; the first feature set includes first sample feature data of the first device antenna and an antenna type of the first device antenna, where the first sample feature data of the first device antenna includes multiple signal reception power data of a cell or an RRU that belongs to the first device antenna, and location data of a terminal corresponding to each signal reception power data in the multiple signal reception power data.
The antenna parameter generation model is, for example, a Neural Network (NN) algorithm model. The training process of the antenna parameter generation model can be represented by the following formula:
(Latitude,Longtitude)＝NN(Feature1056,AntennaType,Wnn1)
wherein L isatitude represents latitude value in working parameter of equipment antenna, Longtitude represents longitude value in working parameter of equipment antenna, NN represents neural network algorithm, Feature1056Represents the first sample characteristic data, antenna type represents the type of the device antenna, Wnn1And representing model parameters in the engineering parameter generation model.
For different device antennas, their corresponding features 1056And AntennaType data are different, and the working parameter prediction results of different equipment antennas can be obtained by inputting the data into the working parameter generation model.
Therefore, the embodiment of the invention can be applied to various sample data scenes to effectively extract data, improve the accuracy of working parameter prediction of the second equipment antenna based on the model, and effectively reduce the acquisition cost of the working parameters of the second equipment antenna.
In a possible embodiment, similar to the process of obtaining the first sample characteristic data in the first implementation manner of the first aspect, the process of obtaining the first sample characteristic data according to the second measurement report data set and the configuration data of the second device antenna includes: determining a cell or RRU (radio remote unit) belonging to the second equipment antenna according to the configuration data of the second equipment antenna; determining measurement report data corresponding to the cell or the RRU from the second measurement report data set; and performing feature extraction according to the measurement report data corresponding to the cell or the RRU to obtain the first sample feature data.
Therefore, the embodiment can adapt to various sample data scenes, and the operation speed and efficiency of the model prediction process are improved, so that the prediction parameters with higher accuracy are obtained.
Based on the second aspect, in a second embodiment, the antenna parameter prediction model further includes an antenna parameter correction model (which may also be referred to as simply an antenna parameter correction model herein) in addition to the antenna parameter generation model; the second set of configuration data further comprises configuration data of at least one other device antenna; the at least one other device antenna represents a device antenna of the plurality of device antennas other than the second device antenna; after obtaining the first prediction result of the engineering parameter of the second device antenna, the method further includes: obtaining second sample characteristic data of the second device antenna according to the second configuration data set and the second measurement report data set, where the second sample characteristic data of the second device antenna includes multiple pieces of signal reception power data of the cell or the RRU of the at least one other device antenna, and location data of a terminal corresponding to each piece of signal reception power data in the multiple pieces of signal reception power data and signal reception power data of the cell or the RRU of the second device antenna; obtaining a prediction result of the engineering parameters of the at least one other equipment antenna according to the antenna engineering parameter generation model; and inputting the second sample characteristic data, the first prediction result of the engineering parameter of the second equipment antenna and the prediction result of the engineering parameter of the at least one other equipment antenna into the antenna engineering parameter correction model to obtain a second prediction result of the engineering parameter of the second equipment antenna.
The antenna engineering parameter correction model is obtained by performing model training according to a first engineering parameter set and a second characteristic set in the first geographic area; wherein the first set of engineering parameters comprises engineering parameters of the first device antenna and engineering parameters of at least one other device antenna within the first geographic area; the second feature set includes second sample feature data for the first device antenna, a prediction of an engineering parameter for the first device antenna, and a prediction of an engineering parameter for at least one other device antenna within the first geographic area; the second sample characteristic data includes multiple pieces of signal reception power data of a cell or an RRU of at least one other device antenna in the first geographic area, and location data of a terminal corresponding to each piece of signal reception power data in the multiple pieces of signal reception power data and signal reception power data of the cell or the RRU of the first device antenna; the predicted result of the engineering parameter of the first device antenna and the predicted result of the engineering parameter of at least one other device antenna in the first geographic area are obtained according to the antenna engineering parameter generation model.
The antenna parameter correction model is, for example, a Neural Network (NN) algorithm model. The training process of the parameter correction model can be represented by the following formula:
(Latitude,Longtitude)＝NN((Featurejoin_i,Featurebasic_i),Wnn2)
wherein Latitude represents Latitude value in the working parameter of the device antenna i, Longtitude represents longitude value in the working parameter of the device antenna i, NN represents neural network algorithm, Featurejoin_iSecond sample characteristic data, Feature, representing a top N antenna of the device antennabasic_iPredicted working parameter, W, for the (1+ Top N) antenna group representing device antenna inn2Representing model parameters in the parameter correction model.
Therefore, for different device antennas, their corresponding featuresjoin_i、Featurebasic_iAnd the data are also different, and the predicted working parameters of the equipment antenna can be further corrected through the working parameter correction model according to the data input into the working parameter correction model, so that a working parameter prediction result with higher reliability is obtained.
It can be seen that, according to the embodiment of the present invention, data extraction can be performed based on off-the-shelf sample data (e.g., MR data, configuration data, etc.), and according to the first implementation manner of the second aspect, the prediction result of the engineering parameter of the second device antenna and the prediction result of the engineering parameter of the at least one other device antenna can be obtained, so as to form input data of an antenna engineering parameter correction model (e.g., the second sample feature data, the prediction result of the engineering parameter of the first device antenna, the prediction result of the engineering parameter of the at least one other device antenna, etc.), so that the predicted engineering parameter of the engineering parameter generation model is further corrected based on the input data into the antenna engineering parameter correction model, and thus the predicted engineering parameter with higher reliability is obtained. The antenna parameter prediction model of the embodiment includes an antenna parameter generation model and an antenna parameter correction model, and by applying the model, the generation and correction of the parameter of the second device antenna can be realized, so that the predicted parameter with high reliability is obtained. The embodiment of the invention can be suitable for various sample data scenes to effectively screen data and improve the efficiency and accuracy of model training. Therefore, the accuracy of subsequent work parameter prediction of the second equipment antenna based on the model can be improved, and the acquisition cost of the work parameters of the second equipment antenna is effectively reduced.
Based on the second implementation manner of the second aspect, in a possible embodiment, the at least one other device antenna is top N device antennas around the second device antenna, where the top N device antennas represent N device antennas most relevant to the second device antenna among the plurality of device antennas, and N is an integer greater than or equal to 1.
For example, in some embodiments, the top N antenna is the N most peripheral device antennas in spatial distance from the second device antenna among the plurality of device antennas peripheral to the second device antenna. (ii) a In still other embodiments, the top N antenna is the N device antennas with the largest signal overlapping degree with the second device antenna among the plurality of device antennas around the second device antenna; in some further embodiments, the top N antenna is the N device antennas with the largest number of terminal switching times among the plurality of device antennas around the second device antenna, and so on.
The embodiment of the invention can predict the working parameters of the second equipment antenna by using the working parameter correction model based on the N most relevant equipment antennas of the second equipment antenna, is favorable for correcting the predicted working parameters of the first equipment antenna based on the working parameters of the N most relevant equipment antennas of the second equipment antenna, and improves the budget speed and the prediction accuracy of the working parameter correction model. Therefore, the implementation of the embodiment is beneficial to integrally improving the reliability of the prediction result of the work parameter prediction model (i.e. including the work parameter generation model and the work parameter correction model).
In a possible embodiment, similar to the process of obtaining the second sample characteristic data in the second implementation manner of the first aspect, the process of obtaining the second sample characteristic data according to the second configuration data set and the second measurement report data set may include: determining a cell or RRU (radio remote unit) belonging to the second equipment antenna and a cell or RRU belonging to the top N equipment antennas according to the second configuration data set; setting at least one piece of measurement report data corresponding to the cell of the top N device antennas or the cell of any one of the RRUs according to the second measurement report data set, wherein each piece of measurement report data in the at least one piece of measurement report data comprises signal receiving power data of the cell of any one device antenna, signal receiving power data of the cell of the second device antenna and position data of the terminal; and performing feature extraction according to the cell of the top N equipment antennas or the cell of each equipment antenna in the RRUs or the measurement report data corresponding to the RRU to obtain the second sample feature data.
Therefore, the implementation of the method and the device can adapt to various sample data scenes, and the operation speed and efficiency of the working parameter prediction process based on the model are improved, so that the working parameter prediction result with high reliability is obtained.
Based on the first implementation manner and the second implementation manner of the first aspect, in a possible embodiment, when performing the parameter prediction based on the model, feature information selection may also be performed on MR data corresponding to a common antenna cell of the device antenna, so as to obtain low-dimensional MR data of the common antenna cell.
That is to say, in an optional embodiment, K dimensions of feature information are selected for MR data corresponding to each cell of the common antenna, so as to obtain low-dimension MR data, where K is an integer greater than or equal to 2.
Correspondingly, in this case, the obtaining the first sample characteristic data according to the second measurement report data set and the configuration data of the second device antenna specifically includes: determining a cell or RRU (remote radio unit) affiliated to the second equipment antenna according to the configuration data of the second equipment antenna; determining low-dimensional measurement report data corresponding to the cell or the RRU from the second measurement report data set; and performing feature extraction according to the low-dimensional measurement report data corresponding to the cell or the RRU to obtain the first sample feature data.
Correspondingly, in this case, the obtaining second sample feature data according to the second configuration data set and the second measurement report data set specifically includes: determining a cell or RRU (radio remote unit) belonging to the second equipment antenna and a cell or RRU belonging to the top N equipment antennas according to the second configuration data set; setting at least one low-dimensional measurement report data corresponding to the cell of the top N device antennas or the cell of any one of the RRUs according to the second measurement report data set, wherein each measurement report data in the at least one low-dimensional measurement report data comprises signal receiving power data of the cell of any one device antenna, signal receiving power data of the cell of the second device antenna and position data of the terminal; and performing feature extraction according to the low-dimensional measurement report data corresponding to the cells of the top N equipment antennas or the cells of each equipment antenna in the RRUs or the RRUs to obtain the second sample feature data.
Therefore, the implementation of the embodiment of the invention is beneficial to reducing the calculation complexity and improving the efficiency of carrying out the I-shaped parameter prediction based on the model.
In a third aspect, an embodiment of the present invention provides a computing device, including: the device comprises a data acquisition module and a model training module. Wherein:
the data acquisition module is used for acquiring a first engineering parameter set, a first configuration data set and a first measurement report data set uploaded by a terminal to a first equipment antenna in a plurality of equipment antennas in a first geographical area; wherein the first set of engineering parameters comprises engineering parameters of the first device antenna including at least one of position data and attitude data of the first device antenna; the first set of configuration data comprises configuration data of the first device antenna, the configuration data of the first device antenna representing configuration information of a network parameter of the first device antenna; measurement report data in the first measurement report data set comprises position data and signal received power data of the terminal; the first device antenna is any device antenna in the plurality of device antennas;
The model training module is used for carrying out model training according to the engineering parameters of the first equipment antenna, the configuration data of the first equipment antenna and the measurement report data set to obtain an antenna engineering parameter prediction model; and the antenna engineering parameter prediction model is used for outputting the engineering parameters of the second equipment antenna according to a second configuration data set from the terminal to a second geographical area and a second measurement report data set uploaded by the terminal to a second equipment antenna in the second geographical area.
The functional modules of the computing device are particularly useful for implementing the method described in the first aspect.
In a fourth aspect, an embodiment of the present invention provides another computing device, including: the system comprises a data acquisition module and a work parameter prediction module. Wherein:
the data acquisition module is used for acquiring a second configuration data set in a second geographical area and a second measurement report data set uploaded by the terminal to a second equipment antenna in the second geographical area; wherein the second configuration data set comprises configuration data of the second device antenna, the configuration data of the second device antenna representing configuration information of network parameters of the second device antenna; the measurement report data in the second measurement report data set comprises position data and signal received power data of the terminal; the second device antenna is any device antenna of a plurality of device antennas in the second geographic area;
The power engineering parameter prediction module is used for inputting the second configuration data set and the second measurement report data set into an antenna power engineering parameter prediction model to obtain a prediction result of the engineering parameters of the second equipment antenna; the antenna engineering parameter prediction model is obtained by training according to a first engineering parameter set, a first configuration data set and a first measurement report data set uploaded by a terminal to a first device antenna in a plurality of device antennas in a first geographic area; the engineering parameters of the second device antenna include at least one of position data and attitude data of the first device antenna.
The functional modules of the computing device may be specifically adapted to implement the method described in the second aspect.
In a fifth aspect, an embodiment of the present invention provides a computing device for training an antenna parameter prediction model, where the computing device includes: a memory and a processor. The memory is for storing sample data and program code, and the processor is for executing the program code to implement the method described in the first aspect.
In a sixth aspect, an embodiment of the present invention provides a computing device for predicting an antenna parameter based on an antenna parameter prediction model, where the computing device includes: a memory and a processor. The memory is used for storing sample data and program codes, and the processor is used for executing the program codes to realize the method described in the second aspect
In a seventh aspect, an embodiment of the present invention provides a system, where the system includes the apparatus described in the third aspect and the apparatus described in the fourth aspect; alternatively, the system comprises a device as described in the fifth aspect and a device as described in the sixth aspect. Wherein:
the computing device for training the antenna engineering parameter prediction model is specifically configured to obtain a first engineering parameter set, a first configuration data set and a first measurement report data set uploaded by a terminal to a first device antenna of multiple device antennas in a first geographic area; performing model training according to the engineering parameters of the first equipment antenna, the configuration data of the first equipment antenna and the first measurement report data set to obtain an antenna engineering parameter prediction model; inputting the antenna parameters prediction model to the computing device for predicting the parameters based on the antenna parameters prediction model;
the computing device for predicting the power parameters based on the antenna power parameter prediction model is specifically configured to obtain a second configuration data set in a second geographic area and a second measurement report data set uploaded by the terminal to a second device antenna in the second geographic area; and inputting the second configuration data set and the second measurement report data set into an antenna engineering parameter prediction model to obtain a prediction result of the engineering parameters of the second equipment antenna.
In an eighth aspect, embodiments of the present invention provide a non-volatile computer-readable storage medium; the computer-readable storage medium is used for storing code for implementing the method of the first aspect (or the second aspect). The program code is for a method provided by any one of the possible designs of the first aspect (or the second aspect) when executed by a computing device.
In a ninth aspect, an embodiment of the present invention provides a computer program product. The computer program product comprising program instructions which, when executed by a computing device, perform the method provided by any one of the possible designs of the first aspect (or the second aspect) described above. The computer program product may be a software installation package, which, in case it is required to use the method provided by any of the possible designs of the aforementioned first aspect (or second aspect), may be downloaded and executed on a processor of the computing device to implement the method of the first aspect (or second aspect).
It can be seen that, in the embodiment of the present invention, a model for engineering parameter prediction can be trained, and through inputting the trained model for engineering parameter prediction (such as the engineering parameter prediction model in the embodiment) to the model based on ready-made sample data (such as MR data, configuration data, etc.), engineering parameters of the device antenna can be generated and corrected, so as to obtain a predicted engineering parameter with high reliability. Therefore, the technical scheme of the embodiment of the invention can overcome the defects of the prior art, effectively reduce the acquisition cost of the working parameters of the equipment antenna and improve the accuracy of the working parameters.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments or the background art of the present invention, the drawings required to be used in the embodiments or the background art of the present invention will be described below.
FIG. 1 is a diagram illustrating a system architecture according to an embodiment of the present invention;
FIG. 2 is a schematic structural diagram of a computing device according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of an engineering parameter determination system according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of a model training process provided by an embodiment of the invention;
FIG. 5 is a schematic diagram of a process for predicting a work parameter based on a model according to an embodiment of the present invention;
FIG. 6 is a schematic structural diagram of another embodiment of an engineering parameter determination system provided in the present invention;
FIG. 7 is a schematic diagram of another model training process provided by an embodiment of the invention;
FIG. 8 is a schematic diagram of another model-based parameter prediction process provided by an embodiment of the present invention;
FIG. 9 is a schematic structural diagram of another embodiment of an employee parameter determination system;
FIG. 10 is a schematic diagram of yet another model training process provided by an embodiment of the invention;
FIG. 11 is a schematic diagram of another model-based parameter prediction process provided by an embodiment of the present invention;
FIG. 12 is a schematic flow chart diagram of a model training method according to an embodiment of the present invention;
FIG. 13 is a schematic flowchart of a method for predicting work parameters based on a model according to an embodiment of the present invention;
FIG. 14 is a schematic flowchart of a method for training a prediction model of an employee parameter according to an embodiment of the present invention;
FIG. 15 is a schematic illustration of some MR data provided by an embodiment of the invention;
fig. 16 is a diagram of MR data of a co-antenna cell of some base station antennas provided by an embodiment of the present invention;
FIG. 17 is a schematic diagram illustrating dimension selection of low-dimensional MR data according to an embodiment of the present invention;
FIG. 18 is a schematic diagram of a scenario for obtaining first sample feature data according to an embodiment of the present invention;
FIG. 19 is a schematic flow chart of a method for predicting an employee parameter based on an employee parameter prediction model according to an embodiment of the present invention;
FIG. 20 is a schematic flow chart of another method for predicting an employee parameter based on an employee parameter prediction model according to an embodiment of the present invention;
FIG. 21 is a schematic structural diagram of another computing device provided by an embodiment of the invention;
fig. 22 is a schematic structural diagram of another computing device according to an embodiment of the present invention.
Detailed Description
The embodiments of the present invention will be described below with reference to the drawings. The terminology used in the description of the embodiments of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. In the present invention, "at least one" means one or more, "a plurality" means two or more. "and/or" describes the association relationship of the associated objects, meaning that there may be three relationships, e.g., a and/or B, which may mean: a exists alone, A and B exist simultaneously, and B exists alone, wherein A and B can be singular or plural. The character "/" generally indicates that the former and latter associated objects are in an "or" relationship. "at least one of the following" or similar expressions refer to any combination of these items, including any combination of the singular or plural items. For example, at least one (one) of a, b, or c, may represent: a, b, c, a-b, a-c, b-c, or a-b-c, wherein a, b, c may be single or multiple.
Referring to fig. 1, fig. 1 is an exemplary schematic diagram of a system architecture of an embodiment of the present invention. Fig. 1 shows a terminal, a network element device, and a computing device according to an embodiment of the present invention, where the terminal and the network element device may communicate with each other through a communication network using an air interface technology. The communication network of the certain air interface technology may include: existing 2G (2 d Generation, such as Global System for Mobile Communications (GSM)) networks, Worldwide Interoperability for Microwave Access (WiMAX) communication networks, 3G (3rd Generation, such as UMTS) networks, Wideband Code Division Multiple Access (Wideband Code Division Multiple Access, WCDMA) networks, Time Division Synchronous Code Division Multiple Access (TD-SCDMA) networks, 4G (4th Generation, such as FDD LTE, TDD LTE) networks, and New Radio Access Technology (New RAT) networks, such as future 4.5G (e.g., Advanced LTE Pro) networks, 5G (5th Generation) networks, and so on.
The Network Element (NE) device is a device for communicating with one or more terminals (e.g., terminal 1, terminal 2, and terminal 3 in the figure), and specifically, the Network Element device may include a radio frequency antenna for performing communication interaction with the terminal through a cell belonging to the radio frequency antenna, where the radio frequency antenna of the Network Element device may be simply referred to as a device antenna, and it should be noted that in different application scenarios, the device antenna may also be referred to as a base station antenna, a cell antenna, and an RRU antenna. The Network element device may be a bts (base Transceiver station) in GSM or cdma (code Division Multiple access), an nb (nodeb) in WCDMA, an evolved Node B (eNB) in LTE, a relay station, a vehicle-mounted device, an access Network device (gnnodeb) in a future 5G Network, an access Network device in a future evolved Public Land Mobile Network (PLMN) Network, or the like. For convenience of description, in the specific embodiment, a network element device is taken as a base station device for example to describe the technical solution of the present invention.
The terminal may include a Relay (Relay), and what the network element device may perform data communication may be regarded as a terminal, and the present invention will be described with reference to the terminal in a general sense. A terminal can also be called a mobile station, an access terminal, a user equipment, a subscriber unit, a subscriber station, a mobile station, a remote terminal, a mobile device, a user terminal, a wireless communication device, a user agent, or a user equipment, among others. The terminal may be a mobile phone, a cordless phone, a Session Initiation Protocol (SIP) phone, a Wireless Local Loop (WLL) station, a tablet, a Personal Digital Assistant (PDA), a handheld device with Wireless communication function or other processing device connected to a Wireless modem, a vehicle mounted device, a wearable device, a mobile station in a future 5G network or a terminal in a future evolved PLMN network, etc. The terminal may detect signals (e.g., cell signals) of one or more network element devices in the environment, and perform communication interaction with a device antenna (which may be referred to as a target device antenna) of a target network element device through the serving cell. For convenience of description, a terminal is taken as User Equipment (UE) to describe the technical solution of the present invention.
The cell (cell) mentioned in the following embodiments may be a cell (e.g., a serving cell, a neighboring cell) corresponding to a base station device, and the cell may belong to a macro base station, or may belong to a base station corresponding to a Small cell (Small cell), where the Small cell may include: urban cells (Metro cells), Micro cells (Micro cells), Pico cells (Pico cells), Femto cells (Femto cells), and the like, wherein the small cells have the characteristics of small coverage area and low transmission power, and are suitable for providing high-rate data transmission services.
The computing device is configured to perform the model training method described in the embodiments of the present invention, and/or the model-based engineering parameter prediction method (abbreviated as "engineering parameter prediction method"). The computing device may be an independent physical server, or may be a server cluster (such as a cloud computing service center) formed by a plurality of physical servers. The computing device may also be deployed at the base station device in the form of a processing chip or a controller, or may be deployed in a Network Management System (NMS) or an Element Management System (EMS). The input data for the computing device to perform the model training method and/or the model-based prediction method may include data provided by the terminal, the base station device, and the data may be data uploaded by the terminal to the base station device, and/or data configured in the terminal, and/or data obtained by measuring the base station device, and the like. For example, the data includes, for example, sample data for model training (data in a training set, such as configuration data, measurement report data, session data, engineering parameters, etc.), for example, input data for model prediction (data in a prediction set, such as configuration data, measurement report data, session data, engineering parameters, etc.), and so on.
Referring to fig. 2, fig. 2 is a schematic structural diagram of a computing device 102 according to an embodiment of the present invention, and as shown in fig. 2, the computing device 102 may include a communication interface 1023, a memory 1022, and a processor 1021 coupled to the memory 1022. Processor 1021, memory 1022, and communication interface 1023 may be connected by a bus or other means (as exemplified by a bus connection in FIG. 2). Wherein:
communication interface 1023 may be used to obtain data for model training and/or data for model prediction. In one embodiment, communication interface 1023 may be used to receive data sent by terminals and/or base station devices. In yet another embodiment, communication interface 1023 may be used to receive data transmitted over a wired or wireless network. In yet another embodiment, communication interface 1023 may be used to obtain data in a non-volatile memory (e.g., hard disk, USB flash drive, magnetic disk, flash memory, etc.).
The processor 1021 may be one or more Central Processing Units (CPUs), and in the case that the processor 1021 is a single CPU, the CPU may be a single-core CPU or a multi-core CPU, taking a single processor as an example in fig. 2. The processor has computing functionality and functionality to control the operation of the computing device 102, and may be configured with one or more of the following: executing the model training method described in the embodiment of the invention; performing the model-based prediction method described in the embodiments of the present invention; operating the engineering parameter determination system (which may be referred to as an engineering parameter determination system or an antenna engineering parameter determination system) described in the embodiment of the present invention; operating training modules described in the embodiments of the present invention, for example, an engineering parameter generation model training module (which may be referred to as an engineering parameter generation model training module for short herein), an engineering parameter correction model training module (which may be referred to as an engineering parameter correction model training module for short herein), an engineering parameter prediction model training module (which may be referred to as an engineering parameter prediction model training module for short herein), and the like, to implement the model training method described in the embodiments of the present invention; the prediction module described in the embodiment of the present invention, for example, the engineering parameter generation module prediction module (may be referred to as an engineering parameter generation module prediction module for short herein), the engineering parameter correction model prediction module (may be referred to as an engineering parameter correction model prediction module for short herein), the engineering parameter prediction model prediction module (may be referred to as an engineering parameter prediction model prediction module for short herein), and the like are operated to implement the model-based engineering parameter prediction method described in the embodiment of the present invention.
The Memory 1022 includes, but is not limited to, a Random Access Memory (RAM), a Read-Only Memory (ROM), an Erasable Programmable Read-Only Memory (EPROM), or a portable Read-Only Memory (CD-ROM), and is used to store relevant program codes and data, such as code instructions for implementing the model training method and/or the method for predicting parameters based on the model according to the embodiment of the present invention, and the data includes data of a training set and/or sample data of a prediction set; and also for storing a parameter determination system that can be used to perform learning feature combinations through machine learning, training correlation models (e.g., parameter generation models, parameter correction models, parameter prediction models, etc.), performing parameter generation or correction based on the correlation models, and so forth.
It should be noted that fig. 2 is only one specific implementation manner of the embodiment of the present invention, and in practical applications, the computing device 102 may further include more or less components, which is not limited herein.
Various engineering parameter determination systems (which may be referred to as "engineering parameter determination systems" for short, or "antenna engineering parameter determination systems") related to the embodiments of the present invention are further described below.
Referring to fig. 3, fig. 3 illustrates a parameter determining system 201 according to an embodiment of the present invention, where the parameter determining system 201 may include at least one of a parameter generating model training module 2011 and a parameter generating model predicting module 2012. The engineering parameter generation model training module 2011 is configured to train an engineering parameter generation model (which may be referred to as an engineering parameter generation model or an antenna engineering parameter generation model herein for short) based on sample data of a training set, so as to obtain a trained engineering parameter generation model. In a possible embodiment, the trained work parameter generation model may also be tested to verify whether the work parameter generation model reaches the training index. The worker parameter generation model training module 2011 may input the trained worker parameter generation model to the worker parameter generation model prediction module 2012, and the worker parameter generation model training module 2011 may further send the related information of the sample feature combination to the worker parameter generation model prediction module 2012. The working parameter generation model prediction module 2012 is configured to perform working parameter prediction based on sample data of the prediction set, relevant information of the sample feature combination, a trained working parameter generation model, and the like, and generate working parameters of the target device antenna.
Referring to fig. 4, for the industrial parameter generation model training module 2011, in some embodiments, sample data of a training set includes feature data X and tag data Y for a first geographic region, X including, for example, a Measurement Report dataset and a configuration dataset, wherein the Measurement Report dataset includes a plurality of Measurement Report (MR) data. Herein, the MR data in the sample data may be MR data reported by the UE to the base station device, where the MR data is measured by periodic or specific event triggering, and the network environment characteristics at a certain point in time during the call process are recorded in units of certain measurement contents (such as frequency measurement/inter-system measurement/traffic measurement/quality measurement/UE internal measurement/UE location measurement/AOA measurement, etc.). For example, each MR data may include one or more of location information of the UE (geographical information such as longitude and latitude information, altitude information, etc. of the UE), Signal Received Power (RSRP) data of a serving cell detected by the UE, RSRP data of neighbor cells, time advance of the serving cell, UE transmit Power headroom (UE), antenna Angle of arrival (AOA), etc.; the configuration data set includes at least one configuration data, and the configuration data may include configuration information of network parameters of the base station device, such as an antenna type of a device antenna, a Cell list, a Physical Cell Identifier (PCI), a Physical Random Access Channel (PRACH), and so on. Y comprises a set of engineering parameters of the radio frequency antenna of the at least one base station device, said set of engineering parameters comprising at least one engineering parameter (may be referred to as an engineering parameter for short), herein each engineering parameter for example comprising at least one of a longitude and latitude, an azimuth, a downtilt, etc. of the antenna.
As shown in fig. 4, the employee parameter generation Model training module 2011 may pre-construct a basic Model of the employee parameter generation Model (unknown Model parameters W1 exist), and the employee parameter generation Model may be characterized by Y ═ Model (X, W1), where Model represents a Model function and W1 represents a Model parameter. Then, the parameter generation model training module 2011 performs model training on the parameter generation model by using sample data (MR data, configuration data, and parameters), and calculates a model parameter W1, thereby obtaining the trained parameter generation model.
Referring to fig. 5, for the parameter generation model prediction module 2012, in some embodiments, the sample data of the prediction set includes MR data (including location information of the UE) and configuration data of a second geographic region (which may be different from the first geographic region). The work parameter generation model prediction module 2012 includes a work parameter generation model trained via a work parameter generation model training module 2011. As shown in fig. 5, the parameter generation model prediction module 2012 may extract data in the prediction set and input the data into the trained parameter generation model, so as to output a prediction result of the parameter of the target device antenna in the second geographic area.
Referring to FIG. 6, FIG. 6 illustrates yet another embodiment of an artifact determination system 202, wherein the artifact determination system 202 may include at least one of an artifact correction model training module 2021 and an artifact correction model prediction module 2022. The engineering parameter correction model training module 2021 is configured to train an engineering parameter correction model (which may be referred to as an engineering parameter correction model for short, or referred to as an antenna engineering parameter correction model for short) based on sample data of a training set, so as to obtain a trained engineering parameter correction model. In a possible embodiment, the trained parameter correction model may be tested to verify whether the parameter generation model meets the training index. The worker parameter correction model training module 2021 may input the trained worker parameter correction model to the worker parameter correction model prediction module 2022, and the worker parameter correction model training module 2021 may further send the related information of the sample feature combination to the worker parameter correction model prediction module 2022. The working parameter correction model prediction module 2022 is configured to perform prediction of working parameters based on sample data of the prediction set, relevant information of the sample feature combination, and a trained working parameter correction model, so as to correct the working parameters of the target device antenna.
Referring to fig. 7, for the working parameter correction model training module 2021, in some embodiments, the sample data of the training set includes feature data X and tag data Y for the first geographic region, X including, for example, a measurement report data set, a configuration data set, and a low confidence engineering parameter set, and Y including a high confidence engineering parameter set for at least one device antenna. The working parameters in the low-confidence engineering parameter set are, for example, working parameters acquired in a relatively coarse manner (for example, working parameters acquired through one-time manual measurement), and the working parameters in the high-confidence engineering parameter set are, for example, working parameters acquired in a relatively precise manner (for example, working parameters acquired through multiple GPS measurements and multiple manual measurements).
As shown in fig. 7, the employee parameter correction Model training module 2021 may pre-construct a basic Model of the employee parameter correction Model (there are unknown Model parameters W2), and the employee parameter correction Model may be characterized by Y ═ Model (X, W2), where Model represents a Model function and W1 represents a Model parameter. Then, the working parameter correction model training module 2021 may perform model training on the working parameter correction model by using sample data (MR data, configuration data, working parameters with low confidence level, and working parameters with high confidence level), and calculate a model parameter W2, thereby obtaining a trained working parameter correction model.
Referring to fig. 8, for the working parameter correction model prediction module 2022, in some embodiments, the sample data of the prediction set includes MR data (including positioning information of the UE), configuration data, and engineering parameters of the second geographic region (which may be different from the first geographic region). The employee parameter correction model prediction module 2022 includes an employee parameter correction model trained via the employee parameter correction model training module 2021. As shown in fig. 8, the working parameter correction model prediction module 2022 may extract data in the prediction set, and input the data into the trained working parameter correction model, so as to output a prediction result of the working parameter of the target device antenna in the second geographic area, so that the prediction result may be compared with the working parameter in the prediction set, so as to correct the working parameter in the prediction set.
In still other embodiments of the present invention, the above two model training and model prediction processes may be integrated. For example, the parameter generation model training module 2011 and the parameter correction model training module 2012 may be implemented in a combined manner, and the parameter generation model prediction module 2012 and the parameter correction model prediction module 2022 may be implemented in a combined manner.
Referring to fig. 9, fig. 9 illustrates another embodiment of the present invention, which is an embodiment of the present invention, wherein the working parameter determination system 203 may include at least one of a working parameter prediction model training module 2031 and a working parameter prediction model prediction module 2032. The engineering parameter prediction model training module 2031 is configured to train an engineering parameter prediction model (which may be referred to as an engineering parameter prediction model herein or as an antenna engineering parameter prediction model) based on sample data of a training set, so as to obtain a trained engineering parameter prediction model. The work parameter prediction model training module 2031 may include a work parameter generation model training module and a work parameter correction model training module, and the work parameter prediction model prediction module 2032 may include a work parameter generation model prediction module and a work parameter correction model prediction module, that is, the work parameter prediction model may be regarded as a combination of the work parameter generation model and the work parameter correction model. Therefore, the worker parameter prediction model training module 2031 may input the trained worker parameter prediction model into the worker parameter prediction model prediction module 2032, and the worker parameter prediction model training module 2031 may further send the relevant information of the sample feature combination to the worker parameter prediction model prediction module 2032. The work parameter prediction model prediction module 2032 is configured to perform work parameter prediction based on sample data of the prediction set, relevant information of the sample feature combination, a trained work parameter prediction model, and the like, and obtain a prediction result of the work parameter of the target device antenna.
In a possible implementation, the parameter correction model training module may further implement training of the self model by using output data of the model (parameter generation model) trained by the parameter generation model training module in a process of training the self model (parameter correction model).
Similarly, in the working parameter prediction process, the working parameter generation model prediction module may generate working parameter data, further input the generated working parameter data to the working parameter correction model prediction module, and then output the prediction result of the working parameter of the target device antenna through the working parameter correction model prediction module. Therefore, the embodiment based on fig. 9 is beneficial to obtaining the high-reliability employee parameter prediction result.
Referring to fig. 10, for the work parameter prediction model training module 2031, in some embodiments, the sample data of the training set includes feature data X and tag data Y for the first geographic region, X including, for example, a measurement report data set and a configuration data set, and Y including an engineering parameter set for at least one device antenna.
As shown in fig. 10, the employee parameter prediction Model training module 2031 may pre-construct a basic Model of the employee parameter prediction Model (there is unknown Model parameter W3), and the employee parameter prediction Model may be characterized by Y ═ Model (X, W3), where Model represents a Model function, W3 represents a Model parameter of the employee parameter prediction Model, and W3 may be determined by the Model parameter W1 of the employee parameter generation Model and the Model parameter W2 of the employee parameter correction Model. That is, in some embodiments, the i.e. predicted Model Y ═ Model (X, W3) can be regarded as a combination of the i.e. generated Model Y ═ Model (X, W1) and the i.e. corrected Model Y ═ Model (X, W2). Then, the parameter prediction model training module 2031 performs model training on the parameter prediction model by using sample data (MR data, configuration data, parameter), and calculates the model parameter W3. In some embodiments, the sample data may be input into the employee parameter generation model training module of the employee parameter prediction model training module 2031, the employee parameter generation model is trained, so as to determine the model parameter W1 of the employee parameter generation model, and the trained employee parameter generation model is obtained according to the W1. Therefore, the engineering parameter correction model training module of the engineering parameter prediction model training module 2031 can output the prediction result of the engineering parameter according to the trained engineering parameter generation model. The engineering parameter correction model training module trains the engineering parameter correction model based on the sample data and the prediction result of the engineering parameters to obtain model parameters W2 of the engineering parameter correction model. Therefore, the training of the worker parameter prediction model is realized through the process, and the trained worker parameter prediction model is obtained.
Referring to fig. 11, for the parameter prediction model prediction module 2032, in some embodiments, the sample data of the prediction set includes MR data (including positioning information of the UE) and configuration data of a second geographic region (which may be different from the first geographic region). The employee parameter prediction model prediction module 2032 includes an employee parameter prediction model trained by the employee parameter prediction model training module 2031 (i.e., may be considered to include a trained employee parameter generation model and a trained employee parameter correction model). As shown in fig. 11, the worker parameter prediction model prediction module 2012 may extract data in the prediction set, input the data to the worker parameter generation model prediction module of the worker parameter prediction model prediction module 2032, output the prediction result of the worker parameter to the worker parameter correction model prediction module of the worker parameter prediction model training module 2031 by using the trained worker parameter generation model, and output the worker parameter prediction result of the target device antenna in the second geographic area with high reliability by using the trained worker parameter correction model.
It will be appreciated that in some of the model training and model prediction processes described above, the engineering parameters may include at least one of position data and attitude data of the device antenna, such as longitude and/or latitude of the device antenna, attitude data of the device antenna, such as azimuth and/or downtilt of the device antenna, and so on, so that in practical applications, different models may be trained based on different engineering parameters (e.g., based on longitude and latitude of the antenna, or based on longitude and latitude and azimuth of the antenna, or based on longitude and latitude, azimuth and downtilt of the antenna, and so on) as desired.
Based on the above-described working parameter determination system, a model training method provided by the embodiment of the invention is described below. Referring to fig. 12, fig. 12 is a schematic flowchart of a model training method according to an embodiment of the present invention, which includes, but is not limited to, the following steps:
step 301, acquiring an engineering parameter set and a configuration data set of at least one device antenna in a first geographic area, and a measurement report data set uploaded by a terminal to a target device antenna in the at least one device antenna in the first geographic area.
In the embodiment of the present invention, the first geographic area represents a geographic position range where one or more device antennas corresponding to sample data used for model training are located. If the multiple device antennas correspond to sample data used for model training, the multiple device antennas may be referred to as multiple device antennas in the first geographic area, and so on, the target device antenna in this embodiment may be referred to as a target device antenna in the first geographic area, and the other device antennas except the target device antenna in the multiple device antennas may be referred to as other device antennas in the first geographic area.
In some embodiments, the model to be trained by the method embodiment includes an employee parameter generating model, which may be considered as an employee parameter generating model described in the foregoing fig. 4 embodiment, for example, in which case:
The set of engineering parameters includes operational parameters of a target device antenna within the first geographic area (the target device antenna within the first geographic area may also be referred to as the first device antenna). The configuration data set may include configuration data of the target device antenna, and specific content of the configuration data includes configuration information of network parameters of the target device antenna, such as antenna type of the target device antenna, cell list, and so on.
The measurement report data set includes a plurality of measurement report data (i.e., MR data) uploaded by the terminal to the target device antenna in the first geographic area, and specific content of the MR data may include location information of the UE (geographic information such as longitude and latitude information and altitude information of the UE), RSRP data of a serving cell detected by the UE, and RSRP data of neighboring cells detected by the UE. Optionally, at least one of the AOA of the serving cell, the AOA of the neighbor cell, the serving cell time advance, the UE transmit power margin, and the like may also be included.
In still other embodiments, when the models to be trained by the embodiment of the method include an employee parameter generation model and an employee parameter correction model, the models may be regarded as employee parameter prediction models as described in the foregoing embodiment of fig. 10, in which case:
The set of engineering parameters includes parameters of device antennas of a plurality of base station devices of a first geographic area, and the parameters of device antennas of the plurality of base station devices include parameters of a target device antenna and parameters of at least one other device antenna, where the target device antenna may be understood as any one of the plurality of base station devices.
The configuration data set may include configuration data of one or more base station devices (i.e. configuration data of target device antennas are included), and specific content of the configuration data includes configuration information of network parameters of the base station devices, such as antenna types of device antennas, cell lists, and so on.
The measurement report data set includes a plurality of MR data uploaded to the target device antenna by the terminal in the first geographic area, and optionally, MR data uploaded to the at least one other device antenna by other terminals.
Also, the specific content of the MR data may include location information of the UE (geographical information such as latitude and longitude information, altitude information, etc. of the UE), RSRP data of a serving cell detected by the UE, RSRP data of a neighbor cell detected by the UE. Optionally, at least one of the AOA of the serving cell, the AOA of the neighbor cell, the serving cell time advance, the UE transmit power margin, and the like may also be included.
And 302, performing model training according to the engineering parameters, the configuration data and the measurement report data set of the target equipment antenna to obtain an engineering parameter prediction model.
In some embodiments, where the model to be trained in the method embodiments includes an engineering parameter generation model, the training process for the model may include the following: from the configuration data and the measurement report data set, first sample Feature data (which may be, for example, Feature as described in the embodiment of FIG. 14) is obtained1056) The first sample characteristic data comprises a plurality of signal receiving power data of a cell which is subordinate to the target equipment antenna and position data of a terminal corresponding to each signal receiving power data in the plurality of signal receiving power data; obtaining the antenna type of the target equipment antenna according to the configuration data; performing model training according to the engineering parameters and the first characteristic set of the target equipment antenna to obtain the engineering parameter generation model; wherein the first feature set comprises the first sample feature data and an antenna type of the target device antenna, and the engineering parameter generation model is used for outputting engineering parameters according to the input first feature set.
For the specific training process of the reference generation model, reference may also be made to the detailed description of steps 501 to 507 in the foregoing embodiment of fig. 4 and the following embodiment of fig. 14, and for brevity of description, details are not repeated here.
In some embodiments, where the model to be trained by the method embodiment includes an employee parameter generation model and an employee parameter correction model, the training process for the model may include training the employee parameter generation model and training the employee parameter correction model.
The training process of the engineering parameter generation model can comprise the following steps: from the configuration data and the measurement report data set, first sample Feature data (which may be, for example, Feature as described in the embodiment of FIG. 14) is obtained1056) The first sample characteristic data comprises a plurality of signal receiving power data of a cell which is subordinate to the target equipment antenna and position data of a terminal corresponding to each signal receiving power data in the plurality of signal receiving power data; obtaining the antenna type of the target equipment antenna according to the configuration data; performing model training according to the engineering parameters and the first characteristic set of the target equipment antenna to obtain the engineering parameter generation model; wherein the first feature set comprises the first sample feature data and an antenna type of the target device antenna The engineering parameter generation model is used for outputting engineering parameters according to the input first feature set.
Similarly, the detailed training process for the worker reference generation model can also refer to the detailed description of steps 501-507 in the foregoing fig. 4 embodiment and the following fig. 14 embodiment.
The training process of the engineering parameter correction model can comprise the following steps: from the configuration dataset and the measurement report dataset, second sample Feature data (which may be, for example, Feature as described in the embodiment of FIG. 14) is obtainedjoin_i) The second sample characteristic data includes a plurality of pieces of signal reception power data of the cell of the at least one other device antenna, and position data of a terminal corresponding to each piece of signal reception power data among the plurality of pieces of signal reception power data and signal reception power data of the cell of the target device antenna; obtaining a prediction result of the engineering parameters of the target equipment antenna and a prediction result of the engineering parameters of the at least one other equipment antenna according to the engineering parameter generation model; performing model training according to the engineering parameter set and the second feature set to obtain the engineering parameter correction model; the second feature set comprises the second sample feature data, the predicted result of the engineering parameter of the target device antenna and the predicted result of the engineering parameter of the at least one other device antenna, and the engineering parameter correction model is used for outputting the engineering parameter according to the input second feature set.
For the specific training process of the worker reference correction model, reference may also be made to the detailed description of step 508-step 512 in the foregoing embodiment of fig. 7 and the following embodiment of fig. 14, and for brevity of description, no further description is given here.
It can be seen that, in the embodiment of the present invention, a model for predicting the working parameters of the device antenna can be constructed based on ready-made sample data (for example, MR data, configuration data, working parameter data, and the like) by a model training mode, and the working parameters of the device antenna can be generated and corrected by applying the model, so as to obtain the predicted working parameters with higher reliability. Therefore, the technical scheme of the embodiment of the invention can overcome the defects of the prior art, effectively reduce the acquisition cost of the working parameters of the equipment antenna and improve the accuracy of the working parameters.
Based on the above-described employee parameter determination system, the following describes a model-based employee parameter prediction method provided by the embodiment of the present invention. Referring to fig. 13, fig. 13 is a schematic flowchart of a method for predicting a worker's parameters based on a model according to an embodiment of the present invention, where the method includes, but is not limited to, the following steps:
step 401, obtaining a measurement report data set uploaded to the target device antenna by the terminal in the second geographic area, and a configuration data set.
In the embodiment of the present invention, the second geographical area indicates a geographical location range where one or more device antennas of the engineering parameter need to be predicted in practical application. If the multiple device antennas correspond to the sample data used for the engineering parameter prediction, the multiple device antennas may be referred to as multiple device antennas in the second geographic area, and in this way, the target device antenna in this embodiment may be referred to as a target device antenna in the second geographic area, and the other device antennas except the target device antenna in the multiple device antennas may be referred to as other device antennas in the second geographic area. The geographic location range represented by the second geographic area may be different from the geographic location range represented by the first geographic area, that is, the target device antenna of the second geographic area may be different from the target device antenna of the first geographic area, and the device antennas of the second geographic area may be different from the device antennas of the first geographic area.
In some embodiments, the model required by the embodiment of the method for work parameter prediction includes a pre-trained work parameter generating model, which may be considered as the work parameter generating model described in the foregoing embodiment of fig. 5, for example, in this case:
The configuration data set may include configuration data of the target device antenna, and specific content of the configuration data includes configuration information of network parameters of the target device antenna (the target device antenna in the second geographic area may also be referred to as a second device antenna), such as antenna type of the target device antenna, cell list, and so on.
The measurement report data set includes a plurality of MR data uploaded to the target device antenna by the terminal in the second geographic area, and specific contents of the MR data may include location information of the UE (geographic information such as latitude and longitude information and altitude information of the UE), RSRP data of a serving cell detected by the UE, and RSRP data of a neighboring cell detected by the UE. Optionally, at least one of AOA of the serving cell, AOA of the neighboring cell, serving cell time advance, UE transmit power headroom, and the like may also be included.
In still other embodiments, when the model required by the embodiment of the method for predicting the working parameter includes a pre-trained working parameter generation model and a working parameter correction model, the model may be regarded as the working parameter prediction model described in the embodiment of fig. 11, for example, in this case:
the configuration data set may include configuration data of one or more base station devices (i.e. configuration data of target device antennas are included), and specific content of the configuration data includes configuration information of network parameters of the base station devices, such as antenna types of device antennas, cell lists, and so on.
The measurement report data set includes multiple MR data uploaded by the terminal to the target device antenna in the second geographic area, and optionally, may further include MR data uploaded by other terminals to the at least one other device antenna.
Also, the specific content of the MR data may include location information of the UE (geographical information such as latitude and longitude information, altitude information, etc. of the UE), RSRP data of a serving cell detected by the UE, RSRP data of a neighbor cell detected by the UE. Optionally, at least one of AOA of the serving cell, AOA of the neighboring cell, serving cell time advance, UE transmit power headroom, and the like may also be included.
Step 402, inputting the measurement report data set and the configuration data set into the trained employee parameter prediction model to obtain a prediction result of the engineering parameters of the target equipment antenna.
In some embodiments, in the case that the model for work parameter prediction required by the embodiment of the method includes a pre-trained work parameter generation model, the process of performing work parameter prediction based on the model may include the following steps: obtaining first sample characteristic data according to the measurement report data set and the configuration data of the target equipment antenna; the first sample characteristic data comprises a plurality of signal receiving power data of a cell belonging to the target equipment antenna and position data of a terminal corresponding to each signal receiving power data in the plurality of signal receiving power data; obtaining the antenna type of the target equipment antenna according to the configuration data; inputting the first sample characteristic data and the antenna type of the target equipment antenna into the trained engineering parameter generation model to obtain a first prediction result of the engineering parameters of the target equipment antenna.
For the specific process of performing the engineering parameter prediction based on the engineering parameter generation model, reference may also be made to the foregoing detailed description of step 601 to step 607 in the embodiment of fig. 5 and the following detailed description of the embodiment of fig. 19, and for brevity of the description, details are not repeated here.
In some embodiments, in the case that the model required for the work parameter prediction by the embodiment of the method includes a pre-trained work parameter generation model and a work parameter correction model, performing the work parameter prediction process based on the model may include generating a work parameter based on the work parameter generation model and further correcting the generated work parameter based on the work parameter correction model to obtain a final work parameter prediction result.
The process of generating the engineering parameters based on the engineering parameter generation model can comprise the following steps: obtaining first sample characteristic data according to the measurement report data set and the configuration data of the target equipment antenna; the first sample characteristic data comprises a plurality of signal receiving power data of a cell belonging to the target equipment antenna and position data of a terminal corresponding to each signal receiving power data in the plurality of signal receiving power data; obtaining the antenna type of the target equipment antenna according to the configuration data; inputting the first sample characteristic data and the antenna type of the target equipment antenna into the trained engineering parameter generation model to obtain a first prediction result of the engineering parameters of the target equipment antenna.
Similarly, for the specific process of performing the work parameter prediction based on the work parameter generating model, reference may be made to the foregoing detailed description of the embodiment in fig. 5 and the following detailed description of steps 601 to 607 in fig. 19, and for the sake of brevity of the description, no further description is given here.
The specific process of further correcting the generated working parameters based on the working parameter correction model to obtain the final working parameter prediction result may include: obtaining second sample characteristic data according to the configuration data set and the measurement report data set, where the second sample characteristic data includes multiple pieces of signal received power data of the cell of the at least one other device antenna, and location data of a terminal corresponding to each piece of signal received power data in the multiple pieces of signal received power data and signal received power data of the cell of the target device antenna; obtaining a prediction result of the engineering parameters of at least one other equipment antenna in the second geographical area according to the engineering parameter generation model; inputting the second sample characteristic data, the first prediction result of the engineering parameter of the target equipment antenna and the prediction result of the engineering parameter of at least one other equipment antenna in the second geographic area to the trained engineering parameter correction model to obtain a second prediction result of the engineering parameter of the target equipment antenna.
For the specific process of performing the work reference prediction based on the work reference correction model, reference may also be made to the foregoing detailed description of step 608 to step 611 in the embodiment of fig. 8 and the following detailed description of step 608 to step 611 in the embodiment of fig. 19, and for the sake of brevity of the description, no further description is given here.
It can be seen that, according to the embodiment of the present invention, the pre-trained model for working parameter prediction can be input to the model based on the ready-made sample data (e.g., MR data, configuration data, etc.), so that the working parameters of the device antenna can be generated and corrected, and the predicted working parameters with high reliability can be obtained. Therefore, the technical scheme of the embodiment of the invention can overcome the defects of the prior art, effectively reduce the acquisition cost of the working parameters of the equipment antenna and improve the accuracy of the working parameters.
Based on the above-mentioned employee parameter determination system 203, the following describes a training method of an employee parameter prediction model provided in an embodiment of the present invention, and the employee parameter includes longitude and latitude of an antenna as an example for description. Referring to fig. 14, fig. 14 is a schematic flowchart illustrating a method for training a parameter prediction model according to an embodiment of the present invention, where the method includes, but is not limited to, the following steps:
step 501, an engineering parameter set, a configuration data set and an MR data set of a first geographical area are obtained.
Specifically, the model training module (e.g., the employee parameter generation model training module in the employee parameter prediction model training module, or a separate employee parameter generation model training module) may obtain the engineering parameter set, the configuration data set, and the MR data set of the first geographic area from sample data of the training set.
The set of engineering parameters may include parameters of device antennas of one or more base station devices of the first geographic area, where each parameter includes, for example, a longitude and a latitude of an antenna. The configuration data set comprises configuration data of one or more base station devices, which may comprise configuration information of network parameters of the base station devices, such as antenna types of device antennas, cell lists, etc. The MR data set comprises a plurality of MR data. The plurality of MR data may be reported to the one or more base station devices by one or more UEs, and each MR data may include at least two or more of location information of the UE (geographical information such as longitude and latitude information and altitude information of the UE), RSRP data of a serving cell detected by the UE, RSRP data of a neighbor cell detected by the UE, AOA of the serving cell, AOA of the neighbor cell, serving cell time advance, UE transmit power headroom, and so on. For example, a certain UE may periodically (e.g., every 10 seconds) upload MR data to a target device antenna in a first geographic area, then multiple MR data inputs over a preset time period may be acquired from the target device antenna into the training set.
Step 502, a cell list of the common antenna is obtained according to the configuration data.
In a particular embodiment, the plurality of MR data in the MR data set may contain MR data relating to radio frequency antennas of different base station devices, which may be subjected to data extraction and data classification in order to extract MR data from these MR data, which are respectively associated with the respective base station devices.
In some embodiments, the number of cells owned by different device antennas is different according to the antenna types of the device antennas and different site situations. The model training module may obtain configuration data from the prediction set and classify the plurality of cells according to the equipment antennas, thereby obtaining a cell list belonging to the same equipment antenna, where the co-antenna cell list may include IDs of one or more cells. For example, the Cell list of the common antennas of the radio frequency antenna of the base station apparatus 1 (simply referred to as apparatus antenna 1) may include Cell-1, Cell-2, and so on. The co-antenna Cell list of the radio frequency antennas of base station device 2 (simply referred to as device antenna 2) may include Cell-3, Cell-4, Cell-5, and so on.
It should be noted that in other possible embodiments of the present invention, the data extraction and data classification may be performed on the plurality of MR data in other manners. For example, different device antennas have different numbers of Remote Radio Units (RRUs). If a cell can correspond to multiple RRUs, the model training module can obtain configuration data from the prediction set, and classify the multiple RRUs according to the device antenna, thereby obtaining an RRU list belonging to the same device antenna, where the RRU list with the common antenna may include IDs of one or more RRUs. For example, the common-antenna cell list of the device antenna of the base station device a may include RRU-1, RRU-2, RRU-3, RRU-4, and so on, in these embodiments, the MR report corresponding to the RRU may be used in the subsequent steps instead of the MR data corresponding to the cell, and the specific implementation process for performing the model training may similarly refer to the implementation process of the cell, which will not be described in detail later.
Step 503, obtaining MR data corresponding to the co-antenna cell according to the MR data set and the co-antenna cell list.
Specifically, the model training module may perform data classification on a plurality of MR data in the MR data set, and obtain MR data (for short, cell MR data) corresponding to each cell of a common antenna (i.e., the same antenna) by associating a serving cell ID of the MR data with a cell ID of the common antenna.
For example, in some embodiments, the MR data set includes MR data 1, MR data 2, MR data 3, and MR data 4 as shown in fig. 15, where the serving cell related to MR data 1 is cell 1, the serving cell related to MR data 2 is cell 2, the serving cell related to MR data 3 is cell 3, and the serving cell related to MR data 4 is cell 4. If the co-antenna cell list of the radio frequency antenna of the base station apparatus 1 (i.e. the apparatus antenna 1) includes the cell 1 and the cell 2, and the co-antenna cell list of the radio frequency antenna of the base station apparatus 2 (i.e. the apparatus antenna 2) includes the cell 3 and the cell 4, the MR data may be associated with the cell ID in the co-antenna cell list through the serving cell ID of the MR data to obtain the MR data corresponding to the co-antenna cell. As shown in fig. 16, MR data 1 may be associated to cell 1 of device antenna 1 and MR data 2 may be associated to cell 2 of device antenna 1, so the MR data of the co-antenna cell of device antenna 1 may include MR data 1 and MR data 2. The MR data 3 may be associated to a cell 3 of the device antenna 2 and the MR data 4 may be associated to a cell 4 of the device antenna 2, so the MR data of the co-antenna cell of the device antenna 2 may comprise the MR data 3 and the MR data 4.
In particular, in a possible embodiment, if the number of pieces of MR data of the co-antenna cell of a certain antenna is lower than a certain value, the MR data of the antenna can be discarded to avoid affecting the accuracy of the trained model and the accuracy of the prediction parameters.
And 504, selecting characteristic information of the MR data corresponding to the common antenna cell to obtain the low-dimensional MR data of the common antenna cell.
In an optional embodiment, in order to reduce computational complexity and improve model training and prediction efficiency, feature information of K dimensions is selected for MR data corresponding to each cell of the common antenna, so as to obtain MR data of low dimensions, where K is an integer greater than or equal to 2.
For example, in some embodiments, the feature information of K dimensions that needs to be selected includes two or more of the following:
the location information of the UE includes, for example, a longitude where the UE is located, a latitude where the UE is located, optionally an altitude where the UE is located, and the like;
the ID of each cell of the co-antenna, for example, the ID of cell 1 (e.g., cell 1 is the serving cell), the ID of cell 2 is ID... the ID of cell J, and the like, J is an integer of 1 or more.
RSRP of the co-antenna cells, respectively, e.g. RSRP of cell 1, RSRP of cell 2.
The AOAs of the co-antenna cells, respectively, e.g. AOA of cell 1, AOA of cell 2.
For example, as shown in fig. 17, fig. 17 exemplarily shows low-dimensional MR data 1 obtained after feature information selection is performed on MR data 1 depicted in fig. 15, and it can be seen that the low-dimensional MR data 1 depicted in fig. 17 reduces feature information of UE transmit power margin, serving cell time advance, and the like compared to the MR data 1 depicted in fig. 15, so that the low-dimensional MR data 1 depicted in fig. 17 has fewer data dimensions than the MR data 1 depicted in fig. 15. Therefore, the embodiment is beneficial to reducing the amount of stored data, improving the computational complexity of subsequent model training and model prediction and improving the computational efficiency.
It should be noted that the example shown in fig. 17 is only used for explaining the technical solution of the present invention and is not limited. In practical applications, the MR data with lower dimension can be selected as required, for example, in an alternative embodiment, the MR data with lower dimension can be designed not to include at least one of the altitude where the UE is located and the AOA of the cell, and so on.
And 505, calculating first sample characteristic data of the common antenna cell according to the low-dimensional MR data of the common antenna cell.
In an embodiment, when the amount of MR data in the training set is large, the occupied memory is large, which results in a lot of MR data in the co-antenna cell. The common antenna cell lists corresponding to different device antennas are also different, and the number of cells is not fixed. In order to better train the model (e.g., avoid overfitting and improve the operation speed and efficiency), the embodiment of the present invention may design a uniform training data template for the MR data corresponding to the common antenna cells of different device antennas, so that the MR data corresponding to the common antenna cells of each device antenna may be screened and merged based on the training data template to obtain sample feature data (which may be referred to as first sample feature data) of the common antenna cells of each device antenna.
For example, a method for screening and merging data is described below, so that each antenna obtains sample feature data of the same dimension.
Referring to fig. 18, in a specific implementation, a range of RSRP values of each cell is, for example, {1,4,7, …,97}, for a single device antenna, for example, the device antenna 1, MR data (optionally, low-dimensional MR data) of a common antenna cell of the device antenna 1 may be added, and longitude and latitude of all UEs of MR data in which the RSRP value of a serving cell (for example, the cell 1) is a predetermined value (for example, the predetermined value is 7, and of course, any other value may be selected) are averaged to obtain a central point, and then the central point may be approximately regarded as a possible longitude and latitude position of the base station device. Then, the ray extends from the central point uniformly to multiple directions (for example, 8 directions of east, west, south, north, south, east, north, west, and north, and of course, other numbers of directions are also possible in the drawing), and the ray in each direction is exemplarily marked with a position point corresponding to RSRP of 1,4,7, …,97 value (of course, other values are also possible, and this is not limited). Then, in the MR data (optionally, low-dimensional MR data) of the common antenna cell of the device antenna 1, the UE longitude and latitude corresponding to the RSRP value of the serving cell (the serving cell of the device antenna 1 includes, for example, cell 1) is 1,4,7, …,97 are mapped into the area shown in fig. 18, and each circle in the drawing represents the UE longitude and latitude corresponding to a certain value RSRP cell.
Then, in an example, when the RSRP values of the serving cells are 1,4,7, …, and 97 (33) respectively, from the low-dimensional MR data of each of the co-antenna cells of the device antenna 1, the MR data closest to the position point having the same value of each directional ray respectively can be searched (if there is no such MR data, one MR data can be constructed by all 0 s instead). That is to say, for a cell with RSRP value of 1, 8 most suitable MR data are found corresponding to 8 directions respectively; for the cell with the RSRP value of 4, searching for 8 most appropriate MR data corresponding to 8 directions respectively; for a cell with an RSRP value of 7, 8 most suitable MR data are found corresponding to 8 directions, respectively. Thus, a total of 8 × 33 — 264 sets of low-dimensional MR data are found.
Then, for each of the 264 sets of low-dimensional MR data, two or more of the longitude of the UE, the latitude of the UE, the altitude of the UE, the AOA of the cell 1, and the like are selected to be extracted to form first sample feature data of the co-antenna cell. For example, when 4 features, that is, the longitude of the UE, the latitude of the UE, the altitude of the UE, and the AOA of cell 1, are extracted at the same time, 4 × 264 sub-features (for the sake of description, the first sample Feature data composed of 1056 sub-features is referred to as "Feature 1056"); for another example, when only 2 features, that is, the longitude where the UE is located and the latitude where the UE is located are extracted, 2 × 264-528 sub-features are generated.
For example, the first sample characteristic data table of the co-antenna cell of the device antenna 1 generated in the above manner is shown in table 1:
TABLE 1
Figure BDA0001897193090000231
Figure BDA0001897193090000241
It should be noted that the above examples are only used for explaining the technical solution of the present invention and are not limited.
Step 506, obtaining the antenna type of the device antenna according to the configuration data.
Specifically, the model training module may obtain the type (antenna type) of the device antenna to which the common antenna cell belongs according to the configuration data in the configuration data set, for example, obtain the antenna type of the device antenna 1 according to the configuration data of the device antenna 1, obtain the antenna type of the device antenna 2 according to the configuration data of the device antenna 2, and so on.
And step 507, training the engineering parameter generation model according to the engineering parameter set, the first sample characteristic data and the antenna type.
In a specific embodiment, the model training module may use a machine learning algorithm to construct an engineering parameter generation model (e.g., a neural network algorithm model), and perform model training on the engineering parameter generation model according to the engineering parameters of the device antennas in the training set, the first sample feature data obtained through step 505, and the type of the device antenna obtained through step 506.
The training process of the engineering parameter generation model can be represented by the following formula:
(Latitude,Longtitude)＝NN(Feature1056,AntennaType,Wnn1)
wherein Latitude represents Latitude value in working parameter of equipment antenna, Longtitude represents longitude value in working parameter of equipment antenna, NN represents neural network algorithm, Feature1056Represents the first sample characteristic data obtained by the manner as shown in the above Table 1, antenna type represents the type of the antenna of the device, Wnn1And representing model parameters in the engineering parameter generation model.
Therefore, the model training module carries out model training based on a large amount of sample data of the training set, and for different equipment antennas, the corresponding Latitude, Longtitude and Feature1056The AntennaType data are different, and the W can be calculated by training the model by taking the AntennaType data as the input data of the engineering parameter generation modelnn1(for example, in this example, W can be calculated by a gradient descent methodnn1) And obtaining the trained worker parameter generation model.
As can be seen, through the steps 501 to 507, the worker parameter generation model training module in the worker parameter prediction model training module can train the worker parameter generation model according to sample data. Subsequently, the worker parameter correction model training module in the worker parameter prediction model training module performs related model training on the worker parameter correction model through the following steps 508 to 512.
And step 508, obtaining predicted working parameters of a plurality of equipment antennas.
In some embodiments, after the trained working parameter generating model is obtained in step 507, a predicted working parameter of the device antenna may be obtained according to the sample data and the trained working parameter generating model, for example, a result of predicting a longitude and Latitude (lathitude) of the device antenna may be obtained by taking longitude and Latitude prediction as an example.
It can be understood that, when the sample data of the training set includes sample data of multiple device antennas, such as sample data of the device antenna 1 (specifically, the sample data includes configuration data of the device antenna 1, working parameters of the device antenna 1, and MR data reported by the terminal to the device antenna 1), sample data of the device antenna 2 (specifically, the sample data of the device antenna 2, working parameters of the device antenna 2, and MR data reported by the terminal to the device antenna 2), and so on, then a model is generated according to the sample data of each device antenna and the trained working parameters, so that a result of predicting the longitude and latitude of each device antenna can be obtained.
For example, the result of predicting the latitude and longitude of each device antenna obtained through the trained parameter generation model is shown in table 2:
TABLE 2
Figure BDA0001897193090000242
Figure BDA0001897193090000251
It should be noted that the above-mentioned examples are only for explaining the embodiments of the present invention and are not limitative.
Step 509, calculating a top N antenna of the device antenna according to the common antenna cell list of the device antenna and the low-dimensional MR data of the device antenna.
For the target device antenna, the top N antenna represents N device antennas most related to the target device antenna based on a preset rule among peripheral device antennas of the target device antenna, and N is an integer greater than or equal to 1. For example, in some embodiments, the top N antenna is the N device antennas that are closest to the target device antenna among the plurality of device antennas that are closest to the target device antenna in spatial distance; in still other embodiments, the top N antenna is the N device antennas with the largest signal overlapping degree with the target device antenna among the plurality of device antennas around the target device antenna; in still other embodiments, the top N antenna is the N device antennas with the largest number of terminal switching times among the peripheral multiple device antennas of the target device antenna, and so on.
A method of determining top N antennas for respective device antennas is given below.
Through the above step 502, a co-antenna cell list of any of the plurality of device antennas may be obtained. By the above-described step 504, low-dimensional MR data of a common antenna cell of any of the plurality of device antennas can be obtained. Then, the IDs of the device antennas corresponding to other cells except the serving cell in the low-dimensional MR data of the target device antenna may be determined according to the common antenna cell list of any device antenna in the multiple device antennas. For example, for the low-dimensional MR data 1 of the device antenna 1 shown in the above-described embodiment of fig. 17, it may be determined that the cell 2 rate belongs to the device antenna 1, the cell 3 rate belongs to the device antenna 2, the cell 4 rate belongs to the device antenna 2, and so on, according to the co-antenna cell list of any device antenna in the multiple device antennas. If the low-dimensional MR data 1 has only two cells (e.g., cell 3 and cell 4) belonging to the device antenna 2, the device antenna 2 is said to appear 2 times in the low-dimensional MR data 1 of the device antenna 1, and so on for other device antennas (e.g., device antenna 3, etc.). Similarly, for other low-dimensional MR data of the device antenna 1 (e.g., the low-dimensional MR data 2, etc.), IDs of device antennas corresponding to other cells except the serving cell may also be found, and then times of occurrence of other device antennas (e.g., the device antenna 2, the device antenna 3, etc., which may also be referred to as neighboring antennas of the device antenna 1) in the other low-dimensional MR data of the device antenna 1 are counted. In this way, the number of occurrences of each neighboring antenna in all the low-dimensional MR data of the device antenna 1 can be counted.
It can be understood that, based on the above description, the number of occurrences of each neighboring antenna of any device antenna of the plurality of base station devices in all low-dimensional MR data of the device antenna can be further counted.
For example, the number of times that each neighboring antenna of any device antenna of the multiple device antennas appears in all low-dimensional MR data of the base station may be as shown in table 3:
TABLE 3
Figure BDA0001897193090000252
Then, the N neighboring antennas with the largest number of occurrences of each device antenna are selected as Top N antennas of the device antenna.
For example, in table 3 above, the neighboring antennas of the device Antenna have been sorted sequentially based on the occurrence frequency, and it can be seen that, for the device Antenna 1, the IDs of the neighboring antennas sorted according to the occurrence frequency are Antenna-2 and Antenna-13. For the device Antenna 2, the IDs of the neighboring antennas sorted according to the number of occurrences are Antenna-5, Antenna-1.
Then, in one possible embodiment, the neighboring antennas for each device antenna may be as shown in table 4 below:
TABLE 4
Figure BDA0001897193090000261
It should be noted that the above examples are only used for explaining the technical solutions of the embodiments of the present invention, and are not limited.
Step 510, obtaining the predicted power parameters of each adjacent antenna in the top N antennas according to the predicted power parameters of the multiple device antennas and the top N antennas of each device antenna.
It can be understood that, through the above step 508, the predicted parameters of each device antenna (for example, table 2) have been obtained, so that the predicted parameters of any device antenna can be obtained based on the ID of any device antenna; based on the ID of each adjacent antenna in the top N antenna of any device antenna, the predicted parameters of each adjacent antenna in the top N antenna of any device antenna can be obtained. As shown in table 5 below:
TABLE 5
Figure BDA0001897193090000262
Table 5 shows the predicted parameters of any device antenna (or target device antenna) and its corresponding Top N antenna, so that the Top N antenna corresponding to any device antenna can be collectively referred to as a (1+ Top N) antenna group. Therefore, through this step 510, the predicted working parameters of the (1+ top N) antenna group of any device antenna can be obtained. For convenience of description, in the embodiment of the present invention, the predicted working parameter of the (1+ Top N) antenna group of the device antenna i may be recorded as "Featurebasic_i", a device antenna i is any of the plurality of device antennas.
Step 511, obtaining second sample characteristic data of the top N antenna according to the low-dimensional MR data of the base station device and the predicted parameters of each adjacent antenna in the top N antenna of the base station device.
The second sample characteristic data characterizes measurement characteristics (or antenna joint measurement characteristics) respectively presented by different device antennas in the same measurement of the UE (i.e. in the same UE geographical location).
A method of obtaining second sample characteristic data for a top N antenna is described below.
Firstly, for a top N antenna of any device antenna, for example, the top N antenna of the device antenna a, a cell is determined according to a common antenna cell list of each neighboring antenna of the device antenna a, and the cell is called as a neighboring cell of the device antenna a, then, N neighboring antennas respectively and correspondingly determine N neighboring cells, and such N neighboring cells may be called as top N neighboring cells of the device antenna a. For example, for a first neighboring antenna in the top N antenna of the device antenna a, if there are multiple cells in the co-antenna cell of the first neighboring antenna, a cell with the largest occurrence number in the MR data of the low latitude of the first neighboring antenna may be exemplarily selected as a neighboring cell corresponding to the first neighboring antenna. By analogy, the adjacent cells corresponding to each adjacent antenna in the top N antenna, that is, the top N adjacent cells of the device antenna a, can be determined respectively. That is, based on the above description, a top N neighbor cell for any device antenna may be determined.
Then, for any neighboring cell in the top N neighboring cells of any device antenna, for example, any neighboring cell in the top N neighboring cells of the device antenna a, M pieces of low-dimensional MR data may be selected from the multiple pieces of low-dimensional MR data of the device antenna a, and the M pieces of low-dimensional MR data are associated with the neighboring cell. Wherein, any one of the M pieces of low-dimensional MR data contains measurement characteristic information of the neighboring cell (e.g., RSRP of the neighboring cell, AOA of the neighboring cell, etc.), for example, RSRP values of neighboring cells in the M pieces of low-dimensional MR data may be different, and M is an integer greater than or equal to 1. In this way, the second sample feature data of the device antenna a may be extracted from each of the M pieces of low-dimensional MR data, and each sample feature data may include two or more of positioning information of the UE (e.g., longitude of the UE, latitude of the UE, altitude of the UE, etc.), RSRP of the serving cell, AOA of the serving cell, RSRP of the neighbor cell, AOA of the neighbor cell, etc. That is, based on the above description, M low-dimensional MR data associated with any one of the top N neighbor cells may be determined, and M second sample feature data may be determined based on the M low-dimensional MR data. For convenience of description, the M second sample Feature data of the top N neighbor cell of the device antenna i may be recorded as "Feature join_i", a device antenna i is any of the plurality of device antennas.
For example, the second sample characteristic data of the top N antenna of the device antenna i is shown in table 6 below:
TABLE 6
Figure BDA0001897193090000271
Figure BDA0001897193090000281
It should be noted that the above examples are only used for explaining the technical solutions of the embodiments of the present invention, and are not limited.
And step 512, training a working parameter correction model according to the engineering parameter set, the predicted working parameters of the (1+ top N) antenna group and the second sample characteristic data of the top N antenna.
In a specific embodiment, the model training module (i.e., the working parameter correction model training module in the working parameter prediction model training module) may use a machine learning algorithm to construct a working parameter correction model (e.g., a neural network algorithm model), and perform model training on the working parameter correction model according to the working parameters of the device antennas in the training set, the predicted working parameters of the (1+ top N) antenna group obtained in step 510, and the second sample feature data of the top N antenna obtained in step 511.
The training process of the parameter correction model can be represented by the following formula:
(Latitude,Longtitude)＝NN((Featurejoin_i,Featurebasic_i),Wnn2)
wherein Latitude represents Latitude value in the working parameter of the device antenna i, Longtitude represents longitude value in the working parameter of the device antenna i, NN represents neural network algorithm, Feature join_iSecond sample characteristic data representing a top N antenna of the device antenna, Featurebasic_iPredicted operating parameters, W, of the (1+ Top N) antenna group representing the device antenna inn2And representing model parameters in the working parameter correction model.
Therefore, the model training module carries out model training based on a large amount of sample data of the training set, and for different equipment antennas, the corresponding Latitude, Longtitude and Featurejoin_i、Featurebasic_iThe data are also different and are used as the output of an engineering parameter correction modelInputting data, the model can be trained to calculate Wnn2(for example, in this example, W can be calculated by a gradient descent methodnn2) And obtaining the trained parameter correction model.
It can be seen that, in this embodiment, through steps 501 to 507, training of the parameter generation model according to the sample data can be realized, and the model parameter W of the parameter generation model is calculatednn1Through the steps 508 to 512, the engineering parameter correction model can be trained, and the model parameter W of the engineering parameter correction model is calculatednn2. It is understood that the parameter prediction model described in the embodiments of the present invention may be regarded as including a parameter generating model and a parameter correcting model, and the model parameters of the parameter prediction model may be regarded as including the model parameters W of the parameter generating model nn1Model parameter W of harmonic parameter correction modelnn2Therefore, based on the steps 501 to 512, the training of the employee parameter prediction model is completed.
It should be further noted that, in some possible embodiments, if the parameter-generating model is trained only by using the parameter-generating model training module shown in fig. 4 (in this case, the parameter-predicting model training module may be regarded as including only the parameter-generating model training module), the implementation process of the model training may also be similar to the description of steps 501 to 507 above, and will not be described in detail herein for the sake of brevity of description.
It should be further noted that, in some possible embodiments, if the parameter correction model is trained only by using the parameter correction model training module shown in fig. 7 (in this case, it can also be considered that the parameter prediction model training module only includes the parameter correction model training module), the model training process can be similar to the process described with reference to the above steps 508-512, except that (as different from the above step 508), the Feature in the training processbasic_iLow confidence engineering parameter sets, features, in sample data (X) from a training setbasic_iRepresenting the engineering parameters of the (1+ Top N) antenna group of the equipment antenna i in the engineering parameter set with low reliability; (Latitude, Longtitude) high confidence in sample data (Y) from the training set The engineering parameter set of (1), namely, Latitude represents the Latitude value in the working parameter of the high-reliability engineering parameter set equipment antenna i, and Longtitude represents the longitude value in the working parameter of the high-reliability engineering parameter set equipment antenna i. The working parameters in the low-confidence engineering parameter set are, for example, working parameters acquired in a relatively coarse manner (for example, working parameters acquired through one manual measurement), and the working parameters in the high-confidence engineering parameter set are, for example, working parameters acquired through a relatively accurate manner (for example, working parameters acquired through multiple GPS measurements and multiple manual measurements). Then, based on the above description of steps 508-512, the method for training the employee parameter correction model by the employee parameter correction model training module shown in fig. 7 will be similarly understood by those skilled in the art, and for the sake of brevity of the description, the detailed description will not be repeated herein.
It can be seen that, in the embodiment of the present invention, a model (such as an employee parameter prediction model in the present embodiment) for predicting an employee parameter of a device antenna can be constructed based on existing sample data (e.g., MR data, configuration data, employee parameter data, etc.) through a model training mode, and application of the model can realize generation and correction of the employee parameter of the device antenna, so as to obtain a predicted employee parameter with higher reliability. Therefore, the technical scheme of the embodiment of the invention can overcome the defects of the prior art, effectively reduce the acquisition cost of the working parameters of the equipment antenna and improve the accuracy of the working parameters.
Based on the above-mentioned working parameter determining system 203, a method for predicting working parameters based on a working parameter prediction model provided in the embodiment of the present invention is described below, where the working parameters include longitude and latitude of an antenna as an example for description. Referring to fig. 19, fig. 19 is a schematic flow chart of a method for predicting a parameter based on a parameter prediction model according to an embodiment of the present invention, where the method includes, but is not limited to, the following steps:
step 601, acquiring an MR data set and a configuration data set of the second geographical area.
Specifically, the model prediction module (e.g., the parameter generation model training module in the parameter prediction model prediction module, or a separate parameter generation model prediction module) may obtain the MR data set in the second geographic area and the configuration data set of the base station device from the data of the prediction training set.
Wherein the second geographical area may be different from the first geographical area in step 501. That is, the base station apparatus located in the second geographical area and the base station apparatus located in the first geographical area are not the same base station apparatus.
Wherein the configuration data set comprises configuration data of one or more base station devices, and the configuration data may comprise configuration information of network parameters of the base station devices, such as antenna types of device antennas, cell lists, and the like. The MR data set comprises a plurality of MR data. The plurality of MR data may be reported to the one or more base station devices by one or more UEs, and each MR data may include at least two or more of location information of the UE (geographical information such as longitude and latitude information and altitude information of the UE), RSRP data of a serving cell detected by the UE, RSRP data of a neighbor cell detected by the UE, AOA of the serving cell, AOA of the neighbor cell, serving cell time advance, UE transmit power headroom, and so on. For example, a certain UE may periodically (e.g., every 10 seconds) upload MR data to the target device antenna in the second geographic area, then multiple MR data inputs over a preset time period may be acquired from the target device antenna into the prediction set.
Step 602, obtaining a cell list of the common antenna according to the configuration data. The specific implementation process may be similar to the description of step 502 in the embodiment of fig. 14, and for brevity of the description, no further description is provided here.
Step 603, obtaining MR data corresponding to the co-antenna cell according to the MR data set and the co-antenna cell list. The specific implementation process may be similar to the description of step 503 in the embodiment with reference to fig. 14, and is not described herein again for brevity of the description.
And step 604, selecting characteristic information of the MR data corresponding to the common antenna cell to obtain the low-dimensional MR data of the common antenna cell. The specific implementation process may be similar to the description of step 504 in the embodiment in fig. 14, and for the brevity of the description, no further description is provided here.
Step 605, calculating first sample characteristic data of the common antenna cell according to the low-dimensional MR data of the common antenna cell. The specific implementation process may be similar to the description of step 505 in the embodiment in fig. 14, and for the brevity of the description, the detailed description is omitted here.
And 606, obtaining the antenna type of the equipment antenna according to the configuration data. The specific implementation process may be similar to the description of step 506 in the embodiment of fig. 14, and for the brevity of the description, the detailed description is omitted here.
And step 607, inputting the first sample characteristic data and the antenna type into the trained work parameter generation model, thereby obtaining the predicted work parameter of the equipment antenna.
The trained parameter generating model is, for example, the parameter generating model (e.g., neural network algorithm model) trained in the foregoing steps 501 to 507 of the embodiment of fig. 14, and the model predicting module uses the first sample Feature data (Feature) corresponding to the device antenna1056) And inputting the antenna type (antenna type) into the trained parameter generation model, so as to obtain the predicted parameter of the device antenna. For example, by taking longitude and Latitude prediction as an example, a result of predicting the longitude and Latitude (Latitude, Longtitude) of the device antenna may be obtained.
It can be understood that, when the input data of the training set includes sample data of multiple device antennas, such as sample data of the device antenna 1 (specifically, configuration data of the device antenna 1, MR data reported by the terminal to the device antenna 1), sample data of the device antenna 2 (specifically, configuration data of the device antenna 2, MR data reported by the terminal to the device antenna 2), and so on, for different device antennas, their corresponding features1056And AntennaType data are also different, and features corresponding to equipment antennas are adopted 1056And AntennaType data are used as input data of the parameter generating model, and then the predicted parameters of each equipment antenna can be obtained.
It can be seen that through the foregoing steps 601 to 607, the power parameter generation model prediction module in the power parameter prediction model prediction module can obtain the predicted power parameters of the device antenna by using the power parameter generation model according to the sample data of the prediction set. Subsequently, the power parameter correction model prediction module in the power parameter prediction model prediction module further processes the predicted power parameters of the equipment antenna by using the power parameter correction model through the following steps 608 to 611, so as to obtain the predicted power parameters with higher reliability (accuracy).
And step 608, calculating a top N antenna of the equipment antenna according to the common antenna cell list of the equipment antenna and the low-dimensional MR data of the equipment antenna. The specific implementation process may be similar to the description of step 509 in the embodiment of fig. 14, and for brevity of the description, no further description is provided here.
Step 609, obtaining the predicted working parameters of each adjacent antenna in the top N antenna according to the predicted working parameters of the multiple device antennas and the top N antenna of each device antenna. The specific implementation process may be similar to the description of step 510 in the embodiment of fig. 14, and for brevity of the description, no further description is provided here.
And step 610, obtaining second sample characteristic data of the top N antenna according to the low-dimensional MR data of the base station equipment and the predicted working parameters of each adjacent antenna in the top N antenna of the base station equipment. The specific implementation process may be similar to the description of step 511 in the embodiment of fig. 14, and for the brevity of the description, the detailed description is omitted here.
Step 611, inputting the predicted working parameters of the (1+ top N) antenna group and the second sample characteristic data of the top N antenna into the trained working parameter correction model, so as to obtain the predicted working parameters with high reliability of the device antenna.
The trained parameter correction model is, for example, a parameter correction model (e.g., a neural network algorithm model) obtained through the training in steps 508 to 512 in the embodiment of fig. 14, and the model prediction module (a parameter correction model prediction module in the parameter prediction model prediction module) uses the second sample Feature data (Feature) corresponding to the device antenna ijoin_i) And predicted parameters (Feature) of (1+ Top N) antenna group of device antenna ibasic_i) And inputting the parameters into the trained parameter correction model to obtain the high-reliability predicted parameters of the equipment antenna i. For example, by taking longitude and Latitude prediction as an example, a highly reliable prediction result of the longitude and Latitude (Latitude) of the device antenna i can be obtained.
It can be understood that, when the input data of the training set includes sample data of multiple device antennas, for example, sample data of the device antenna 1 (specifically, configuration data of the device antenna 1, MR data reported by the terminal to the device antenna 1), sample data of the device antenna 2 (specifically, configuration data of the device antenna 2, MR data reported by the terminal to the device antenna 2), and so on, for different device antennas, there are differences in the second sample characteristic data corresponding to the different device antennas, and the (1+ Top N) antenna group are used as the input data of the working parameter correction model, so that the high-reliability predicted working parameters of the different device antennas can be obtained.
It can be seen that, in this embodiment, through steps 601 to 607, the predicted parameters of the device antenna can be obtained by using the parameter generation model according to the sample data of the prediction set, and through steps 608 to 611, the predicted parameters of the device antenna can be further corrected by using the parameter correction model, so that the predicted parameters with high reliability can be obtained. It can be understood that the parameter prediction model described in the embodiment of the present invention may be regarded as including a parameter generation model and a parameter correction model, so that the prediction of the device antenna parameter based on the parameter prediction model is completed based on the above steps 601 to 611.
It should be further noted that, in some possible embodiments, if the parameter of the device antenna is predicted based on the parameter-generating model only by using the parameter-generating model prediction module shown in fig. 5 (in this case, the parameter-generating model prediction module can be regarded as including only the parameter-generating model prediction module), the implementation process of the model training can be similar to the description of steps 601-607, and will not be described in detail herein for the sake of brevity of the description.
It should be further noted that, in some possible embodiments, if the power parameters of the device antenna are predicted based on the power parameter correction model only by using the power parameter correction model prediction module shown in fig. 8 (in this case, it can also be considered that the power parameter prediction model prediction module only includes the power parameter correction model prediction module), the implementation process of the model training may also be similar to the description of step 608 to step 611 described above, except that the predicted power parameters of the (1+ Top N) antenna group in the prediction process are not obtained through the power parameter generation model, but are the engineering parameter set in the sample data from the prediction set, and the power parameters in the engineering parameter set may be regarded as low-reliability power parameters, for example, power parameters obtained in a coarser manner. Based on the above steps 608 to 611, those skilled in the art will similarly understand the method for the power reference correction model prediction module to perform power reference prediction based on the power reference correction model to obtain the high-confidence power reference of the device antenna, as shown in fig. 8, and for the sake of brevity of the description, the detailed description will not be repeated herein.
It can be seen that, in the embodiment of the present invention, the pre-trained model for parameter prediction (such as the parameter prediction model in the embodiment) is input to the model based on the ready-made sample data (such as MR data, configuration data, and the like), so that the parameter of the device antenna can be generated and corrected, and the predicted parameter with high reliability can be obtained. Therefore, the technical scheme of the embodiment of the invention can overcome the defects of the prior art, effectively reduce the acquisition cost of the working parameters of the equipment antenna and improve the accuracy of the working parameters.
In order to better understand the technical solution of the embodiment of the present invention, a method for correcting the original parameters of the device antenna in some application scenarios is described below. Referring to fig. 20, the method includes, but is not limited to, the steps of:
step 701, the method starts, setting variable i to 0.
In step 702, if i is equal to 0, the method jumps to step 703, otherwise, if i is not equal to 0, the method jumps to step 708.
Step 703, the method executes the selected sample data.
In some embodiments, if the model prediction module in this method embodiment is an engineering parameter generation model prediction module, the sample data includes MR data (including positioning information of the UE) and configuration data of the base station device. The sample data and the original working parameters of the equipment antenna in the original working parameter list have corresponding relations.
In some embodiments, if the model prediction module in this method embodiment is an operating parameter correction model, the sample data includes MR data (including positioning information of the UE) of the base station device, configuration data, and an original operating parameter of the device antenna (i.e., an operating parameter to be corrected of the device antenna) in an original engineering parameter list (which may be referred to herein simply as an original operating parameter list).
In some embodiments, if the model prediction module in this method embodiment is a duty cycle prediction model, the sample data includes MR data (including positioning information of the UE) and configuration data of the base station device. The sample data and the original working parameters of the equipment antenna in the original working parameter list have corresponding relations.
Wherein the original parameters list includes original parameters of one or more device antennas.
Step 704, inputting the sample data into a model prediction module for work parameter prediction.
It can be understood that, if the model prediction module in the embodiment of the method is an engineering parameter generation model prediction module, the engineering parameter generation model prediction module may generate the predicted engineering parameters of the device antenna by using the engineering parameter generation model according to the sample data. For a specific implementation process, reference may be made to the foregoing description in the embodiment of fig. 5 and the related descriptions of step 601 to step 607 in the embodiment of fig. 19, and for brevity of description, no further description is given here.
It can be understood that, if the model prediction module in the embodiment of the method is an operating parameter correction model prediction module, the operating parameter correction model prediction module may obtain the predicted operating parameters of the device antenna by using the operating parameter correction model according to the sample data. For a specific implementation process, reference may be made to the foregoing description of the step 608 to the step 611 in the embodiment in fig. 8 and the embodiment in fig. 19, and for brevity of the description, no further description is provided here.
It can be understood that, if the model prediction module in the embodiment of the method is an engineering parameter prediction model prediction module, the engineering parameter prediction model prediction module may obtain the predicted engineering parameters of the device antenna by using the engineering parameter prediction model according to the sample data. For a specific implementation process, reference may be made to the foregoing description of the step 601-step 611 in the embodiment in fig. 11 and the embodiment in fig. 19, and for brevity of the description, no further description is provided here.
Step 705, setting the variable i to i + ═ 1.
Step 706, the method executes to judge whether the average difference between the original working parameters of the predicted working parameters of the equipment antenna and the original working parameters of the equipment antenna meets the preset condition, if not, the method continues to execute step 707; if the preset condition is not met, the step 709 is executed.
For example, in some embodiments, the number of the device antennas is multiple, and the predicted parameters of each device antenna are respectively denoted as (Px1, Py1), (Px2, Py2), …, and (Pxn, Pyn), where any of the predicted parameters (Pxi, Pyi) represents the predicted parameter of the ith device antenna (or antenna feeder).
The original parameters of each device antenna are respectively denoted as (Rx1, Ry1), (Rx2, Ry2), …, (Rxn, Ryn), wherein any original parameter (Rxi, Ryi) represents the original parameter of the ith device antenna.
Inputting the predicted working parameters and the original working parameters of each equipment antenna into the following formula:
Figure BDA0001897193090000321
if diff < the preset value (e.g., 0.0002, the preset value is not limited by the present invention), the accuracy of the parameter is considered to meet the requirement. That is, if diff < the preset value, the average difference between the predicted work parameters of the device antenna and the original work parameters of the device antenna is said to satisfy the preset condition. If diff is larger than or equal to a preset value, the average difference between the original working parameters of the predicted working parameters of the equipment antenna and the original working parameters of the equipment antenna does not meet the preset condition.
It should be noted that, in practical application, the method may further set more preset conditions, for example, a relationship between i and a maximum iteration number (maxtitertimes) may be determined, for example, the maximum iteration number is set to 5 (which is only an example, and the present invention is not limited thereto). If i is greater than the maximum number of iterations, the method jumps to execute step 708; if i is less than or equal to the maximum number of iterations, the method continues at step 707.
And 707, updating the worker parameters in the original worker parameter list by the method, and putting the original worker parameters with larger difference into an abnormal worker parameter list. Then a jump is made to execute step 702.
Specifically, in some embodiments, an abnormal working parameter list is set for placing the original working parameters with larger difference between the preset working parameter value and the original working parameter (for example, greater than or equal to the preset value described in step 706). Then, in step 706, if there are some estimated work parameter values of the equipment antenna that are different from the original work parameter values (for example, greater than or equal to the predetermined values described in step 706), the original work parameters of the equipment antenna are put into the abnormal work parameter list. If the working parameter predicted values of some device antennas are smaller than the original working parameter difference values (for example, smaller than the preset values described in step 706), the original working parameters of the device antennas are replaced by the predicted working parameters of such device antennas in the original working parameter list, so as to update the original working parameter list.
It should be noted that, in a possible implementation, only the ID of the device antenna corresponding to the abnormal operation parameter may be stored in the abnormal operation parameter list.
And step 708, judging whether the abnormal working parameter list is empty. If the abnormal task parameter list is empty, the method continues with step 709; if the abnormal working parameter list is not empty, the method jumps to execute step 703.
And 709, using the preset work parameter which is predicted in the step 704 and meets the preset condition in the step 706 as a final corrected work parameter and outputting the final corrected work parameter. That is, by executing steps 701 to 709, the original parameters of each device antenna can be corrected, and more accurate predicted parameters of each device antenna can be obtained.
It can be seen that, in the embodiment of the present invention, a pre-trained model (such as an employee parameter prediction model in this embodiment) for predicting an employee parameter can be input to the model based on ready-made sample data (e.g., MR data, configuration data, etc.), so that an employee parameter prediction result can be obtained by generating and correcting the employee parameter of an equipment antenna, and an original employee parameter of a base station equipment can be updated based on a comparison between the employee parameter prediction result and the original employee parameter of the base station equipment, so that an employee parameter value with a high reliability of the base station equipment is obtained. Therefore, the technical scheme of the embodiment of the invention can overcome the defects of the prior art, effectively reduce the acquisition cost of the working parameters of the equipment antenna and improve the accuracy of the working parameters.
The system architecture and method of the embodiment of the present invention are described in detail above, and the following provides related devices of the embodiment of the present invention based on the same inventive concept.
Referring to fig. 21, fig. 21 is a schematic structural diagram of a computing device 80 according to an embodiment of the present invention, where the computing device 80 includes: a data acquisition module 801 and a model training module 802, wherein:
a data obtaining module 801, configured to obtain a first engineering parameter set, a first configuration data set in a first geographic area, and a first measurement report data set uploaded by a terminal to a first device antenna of multiple device antennas in the first geographic area; wherein the first set of engineering parameters comprises engineering parameters of the first device antenna including at least one of position data and attitude data of the first device antenna; the first set of configuration data comprises configuration data of the first device antenna, the configuration data of the first device antenna representing configuration information of network parameters of the first device antenna; measurement report data in the first measurement report data set comprises position data and signal received power data of the terminal; the first device antenna is any device antenna in the plurality of device antennas;
a model training module 802, configured to perform model training according to the engineering parameters of the first device antenna, the configuration data of the first device antenna, and the measurement report data set, to obtain an antenna engineering parameter prediction model; and the antenna engineering parameter prediction model is used for outputting the engineering parameters of the second equipment antenna according to a second configuration data set from the terminal to a second geographical area and a second measurement report data set uploaded by the terminal to a second equipment antenna in the second geographical area.
In some embodiments, the antenna parameter prediction model comprises an antenna parameter generation model; the model training module 802 is specifically configured to: obtaining first sample characteristic data according to the configuration data of the first device antenna and the measurement report data set; the first sample characteristic data comprises a plurality of signal receiving power data of a cell or a remote radio unit RRU belonging to the first equipment antenna and position data of a terminal corresponding to each signal receiving power data in the plurality of signal receiving power data; obtaining the antenna type of the first equipment antenna according to the configuration data of the first equipment antenna; performing model training according to the engineering parameters and the first characteristic set of the first equipment antenna to obtain an antenna engineering parameter generation model; the first feature set comprises the first sample feature data and an antenna type of the first device antenna, and the antenna parameter generation model is used for outputting engineering parameters according to the input first feature set.
In a possible embodiment, the model training module 802 is specifically configured to: determining a cell or RRU (radio remote unit) belonging to the first equipment antenna according to the configuration data of the first equipment antenna; determining measurement report data corresponding to the cell or the RRU from the measurement report data set; and performing feature extraction according to the measurement report data corresponding to the cell or the RRU to obtain the first sample feature data.
In some embodiments, the antenna parameter prediction model includes both an antenna parameter generation model and an antenna parameter correction model; the engineering parameter set further comprises engineering parameters of at least one other device antenna; the configuration data set further comprises configuration data of the at least one other device antenna; the at least one other device antenna represents a device antenna of the plurality of device antennas other than the first device antenna; the model training module 802 is further configured to: obtaining second sample characteristic data according to the configuration data set and the measurement report data set, where the second sample characteristic data includes multiple pieces of signal reception power data of the cell or the RRU of the at least one other device antenna, and location data of a terminal corresponding to each piece of signal reception power data in the multiple pieces of signal reception power data and signal reception power data of the cell or the RRU of the first device antenna; obtaining a prediction result of the engineering parameters of the first equipment antenna and a prediction result of the engineering parameters of the at least one other equipment antenna according to the antenna engineering parameter generation model; performing model training according to the engineering parameter set and the second characteristic set to obtain the antenna engineering parameter correction model; the second feature set comprises the second sample feature data, the predicted result of the engineering parameter of the first equipment antenna and the predicted result of the engineering parameter of the at least one other equipment antenna, and the antenna engineering parameter correction model is used for outputting the engineering parameter according to the input second feature set.
In a possible embodiment, the at least one other device antenna is top N device antennas around the first device antenna, where the top N device antennas represent N device antennas most related to the first device antenna among the plurality of device antennas, and N is an integer greater than or equal to 1.
In a possible embodiment, the model training module 802 is specifically configured to: determining a cell or RRU (radio remote unit) belonging to the first equipment antenna and a cell or RRU belonging to the top N equipment antennas according to the configuration data set; setting at least one measurement report data corresponding to a cell of the top N device antennas or a cell of any one of RRUs according to the first measurement report data set, wherein each measurement report data in the at least one measurement report data comprises signal receiving power data of the cell of any one of the device antennas or the RRUs, signal receiving power data of the cell of the first device antenna or the RRU and position data of the terminal; and performing feature extraction according to the cell of the top N device antennas or the cell of each device antenna in the RRUs or the measurement report data corresponding to the RRU, and obtaining the second sample feature data.
The computing device 80 may be used to implement the model training process described in the embodiments of the present invention.
In a specific implementation, the program codes of the data obtaining module 801 and the model training module 802 may be stored in the memory 1022 described in the embodiment of fig. 2, and may be called by the processor 1021 to execute the model training method described in the embodiment of the present invention.
In a specific implementation, the data obtaining module 801 and the model training module 802 may be used together to implement the function of the parameter generation model training module described in the above embodiment of fig. 4, or together to implement the function of the parameter correction model training module described in the above embodiment of fig. 7, or together to implement the function of the parameter prediction model training module described in the above embodiment of fig. 10. For a specific implementation process, reference may be made to the description related to the embodiment in fig. 12 or the embodiment in fig. 14, and for brevity of the description, no further description is provided here.
Referring to fig. 22, fig. 22 is a schematic structural diagram of a computing device 90 according to an embodiment of the present invention, where the computing device 90 includes: a data acquisition module 901 and a work parameter prediction module 902, wherein:
a data obtaining module 901, configured to obtain a second configuration data set in a second geographic area and a second measurement report data set uploaded by a terminal to a second device antenna in the second geographic area; wherein the second configuration data set comprises configuration data of the second device antenna, the configuration data of the second device antenna representing configuration information of network parameters of the second device antenna; measurement report data in the second measurement report data set comprises position data and signal received power data of the terminal; the second device antenna is any device antenna of a plurality of device antennas in the second geographic area;
An engineering parameter prediction module 902, configured to input the second configuration data set and the second measurement report data set to an antenna engineering parameter prediction model, so as to obtain a prediction result of an engineering parameter of the second device antenna; the antenna engineering parameter prediction model is obtained by training according to a first engineering parameter set, a first configuration data set and a first measurement report data set uploaded by a terminal to a first device antenna in a plurality of device antennas in a first geographic area; the engineering parameters of the second device antenna include at least one of position data and attitude data of the first device antenna.
In some embodiments, the antenna parameter prediction model comprises an antenna parameter generation model; the work parameter prediction module 902 is specifically configured to: obtaining first sample characteristic data according to the second measurement report data set and the configuration data of the second equipment antenna; the first sample characteristic data comprises a plurality of signal receiving power data of a cell belonging to the second equipment antenna or a remote radio unit RRU, and position data of a terminal corresponding to each signal receiving power data in the plurality of signal receiving power data; obtaining the antenna type of the second equipment antenna according to the configuration data of the second equipment antenna; and inputting the first sample characteristic data and the antenna type of the second equipment antenna into the antenna engineering parameter generation model to obtain a first prediction result of the engineering parameter of the second equipment antenna.
In a possible embodiment, the parameter prediction module 902 is specifically configured to: determining a cell or RRU (radio remote unit) belonging to the second equipment antenna according to the configuration data of the second equipment antenna; determining measurement report data corresponding to the cell or the RRU from the second measurement report data set; and performing feature extraction according to the measurement report data corresponding to the cell or the RRU to obtain the first sample feature data.
In some embodiments, the antenna parameter prediction model includes both an antenna parameter generation model and an antenna parameter correction model; the second set of configuration data further comprises configuration data of at least one other device antenna; the at least one other device antenna represents a device antenna of the plurality of device antennas other than the second device antenna; the parameter prediction module 902 is further configured to: obtaining second sample characteristic data according to the second configuration data set and the second measurement report data set, where the second sample characteristic data includes multiple pieces of signal reception power data of the cell or the RRU of the at least one other device antenna, and location data of a terminal corresponding to each piece of signal reception power data in the multiple pieces of signal reception power data and signal reception power data of the cell or the RRU of the second device antenna; obtaining a prediction result of the engineering parameters of the at least one other equipment antenna according to the antenna engineering parameter generation model; inputting the second sample characteristic data, the first prediction result of the engineering parameter of the second equipment antenna and the prediction result of the engineering parameter of the at least one other equipment antenna to the antenna engineering parameter correction model to obtain a second prediction result of the engineering parameter of the second equipment antenna.
In a possible embodiment, the at least one other device antenna is top N device antennas around the second device antenna, where the top N device antennas represent N device antennas most related to the second device antenna among the plurality of device antennas, and N is an integer greater than or equal to 1.
In a possible embodiment, the work parameter prediction module 902 is specifically configured to: determining a cell or RRU (radio remote unit) belonging to the second equipment antenna and a cell or RRU belonging to the top N equipment antennas according to the second configuration data set; setting at least one piece of measurement report data corresponding to the cell of the top N device antennas or the cell of any one of the RRUs according to the second measurement report data set, wherein each piece of measurement report data in the at least one piece of measurement report data comprises signal receiving power data of the cell of any one device antenna, signal receiving power data of the cell of the second device antenna and position data of the terminal; and performing feature extraction according to the cell of the top N device antennas or the cell of each device antenna in the RRUs or the measurement report data corresponding to the RRU, and obtaining the second sample feature data.
The computing device 90 is particularly useful for implementing the model-based engineering parameter prediction process described in the embodiments of the present invention.
In a specific implementation, the program codes of the data obtaining module 901 and the employee parameter prediction module 902 may be stored in the memory 1022 described in the embodiment of fig. 2, and may be called by the processor 1021 to execute the model-based employee parameter prediction method described in the embodiment of the present invention.
In a specific implementation, the data obtaining module 901 and the parameter prediction module 902 may be used together to implement the function of the parameter generation model prediction module described in the above embodiment of fig. 5, or used together to implement the function of the parameter correction model prediction module described in the above embodiment of fig. 8, or used together to implement the function of the parameter prediction model prediction module described in the above embodiment of fig. 11. For a specific implementation process, reference may be made to the related description of the embodiment in fig. 13, the embodiment in fig. 19, or the embodiment in fig. 20, and for brevity of description, details are not repeated here.
Based on the same inventive concept, embodiments of the present invention provide an employee parameter determination system that may include a computing device 80 as described in FIG. 21 and a computing device 90 as described in FIG. 22. In a specific implementation, the work parameter determining system is, for example, the work parameter determining system 201 described in the embodiment of fig. 3, the work parameter determining system 202 described in the embodiment of fig. 6, or the work parameter determining system 203 described in the embodiment of fig. 9.
In the above embodiments, all or part may be implemented by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer program instructions which, when loaded and executed on a computer, cause a process or function according to an embodiment of the invention to be performed, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, the computer instructions may be transmitted from one network site, computer, server, or data center to another network site, computer, server, or data center by wire (e.g., coaxial cable, fiber optic, digital subscriber line) or wirelessly (e.g., infrared, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer and can be a data storage device, such as a server, a data center, etc., that includes one or more of the available media. The usable medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape, etc.), an optical medium (e.g., Digital Video Disc (DVD), etc.), a semiconductor medium (e.g., solid state disk), etc., and so forth.
In the foregoing embodiments, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.

Claims (18)

1. A method of predicting antenna engineering parameters, comprising:
acquiring a first engineering parameter set, a first configuration data set and a first measurement report data set uploaded to a first equipment antenna by a terminal; wherein the first set of engineering parameters comprises engineering parameters of the first device antenna including at least one of position data and attitude data of the first device antenna; the first set of configuration data comprises configuration data of the first device antenna, the configuration data of the first device antenna representing configuration information of network parameters of the first device antenna; measurement report data in the first measurement report data set comprises position data and signal received power data of the terminal;
performing model training according to the engineering parameters of the first equipment antenna, the configuration data of the first equipment antenna and the first measurement report data set to obtain an antenna engineering parameter prediction model; the antenna working parameter prediction model is used for outputting the engineering parameters of a second equipment antenna according to a second configuration data set and a second measurement report data set uploaded to the second equipment antenna by a terminal, wherein the second configuration data set comprises the configuration data of the second equipment antenna;
The antenna working parameter prediction model comprises an antenna working parameter generation model and an antenna working parameter correction model;
the performing model training according to the engineering parameters of the first device antenna, the configuration data of the first device antenna, and the first measurement report data set to obtain an antenna engineering parameter prediction model specifically includes:
performing model training according to the engineering parameters of the first equipment antenna, the configuration data of the first equipment antenna and the first measurement report data set to obtain an antenna working parameter generation model;
obtaining a prediction result of the engineering parameters of the first equipment antenna and a prediction result of the engineering parameters of the at least one other equipment antenna according to the antenna engineering parameter generation model;
performing model training according to the first engineering parameter set and the second characteristic set to obtain the antenna engineering parameter correction model; the second feature set comprises second sample feature data, a prediction result of the engineering parameters of the first device antenna and a prediction result of the engineering parameters of the at least one other device antenna, the antenna engineering parameter correction model is used for outputting the engineering parameters according to the input second feature set, and the first engineering parameter set further comprises the engineering parameters of the at least one other device antenna around the first device antenna; the second sample characteristic data includes multiple pieces of signal reception power data of the cell or the RRU of the at least one other device antenna, and position data of the terminal corresponding to each piece of signal reception power data in the multiple pieces of signal reception power data and signal reception power data of the cell or the RRU of the first device antenna.
2. The method of claim 1,
the performing model training according to the engineering parameters of the first device antenna, the configuration data of the first device antenna, and the first measurement report data set to obtain the antenna working parameter generation model includes:
obtaining first sample characteristic data of the first device antenna from the configuration data of the first device antenna and the first measurement report data set; the first sample characteristic data comprises a plurality of signal receiving power data of a cell or a remote radio unit RRU belonging to the first equipment antenna and position data of a terminal corresponding to each signal receiving power data in the plurality of signal receiving power data;
obtaining the antenna type of the first equipment antenna according to the configuration data of the first equipment antenna;
performing model training according to the engineering parameters and the first characteristic set of the first equipment antenna to obtain an antenna engineering parameter generation model; the first feature set comprises the first sample feature data and an antenna type of the first device antenna, and the antenna parameter generation model is used for outputting engineering parameters according to the input first feature set.
3. The method of claim 2, wherein obtaining first sample characterization data for the first device antenna based on the configuration data for the first device antenna and the first measurement report data set comprises:
determining a cell or RRU (radio remote unit) belonging to the first equipment antenna according to the configuration data of the first equipment antenna;
determining measurement report data corresponding to the cell or the RRU from the first measurement report data set;
and performing feature extraction according to the measurement report data corresponding to the cell or the RRU to obtain first sample feature data of the first equipment antenna.
4. A method according to claim 2 or 3, wherein the first set of engineering parameters further comprises engineering parameters of at least one other device antenna peripheral to the first device antenna; the first set of configuration data further comprises configuration data of the at least one other device antenna;
after the model training is performed according to the engineering parameters and the first feature set of the first device antenna to obtain the antenna engineering parameter generation model, the method further includes:
obtaining second sample characteristic data of the first device antenna from the first configuration data set and the first measurement report data set.
5. The method according to claim 4, wherein the at least one other device antenna is top N device antennas around the first device antenna, the top N device antennas representing N device antennas most correlated to the first device antenna among the plurality of device antennas, and N being an integer equal to or greater than 1.
6. The method of claim 5, wherein obtaining second sample characterization data for the first device antenna based on the first configuration data set and the first measurement report data set comprises:
determining a cell or RRU (radio remote unit) belonging to the first equipment antenna and a cell or RRU belonging to the top N equipment antennas according to the first configuration data set;
setting at least one measurement report data corresponding to a cell of the top N device antennas or a cell of any one of RRUs according to the first measurement report data set, wherein each measurement report data in the at least one measurement report data comprises signal receiving power data of the cell of any one of the device antennas or the RRUs, signal receiving power data of the cell of the first device antenna or the RRU and position data of the terminal;
And performing feature extraction according to the cell of the top N device antennas or the cell of each device antenna in the RRUs or the measurement report data corresponding to the RRU, and obtaining second sample feature data of the first device antenna.
7. A method of predicting antenna engineering parameters, comprising:
acquiring a second configuration data set and a second measurement report data set uploaded to a second equipment antenna by the terminal; wherein the second configuration data set comprises configuration data of the second device antenna, the configuration data of the second device antenna representing configuration information of network parameters of the second device antenna; measurement report data in the second measurement report data set comprises position data and signal received power data of the terminal;
inputting the second configuration data set and the second measurement report data set into an antenna engineering parameter prediction model to obtain a prediction result of the engineering parameters of the second equipment antenna; the antenna engineering parameter prediction model is obtained by training according to a first engineering parameter set, a first configuration data set and a first measurement report data set uploaded to a first equipment antenna in a first antenna set by a terminal; the engineering parameters of the second device antenna comprise at least one of position data and attitude data of the first device antenna;
The antenna working parameter prediction model comprises an antenna working parameter generation model and an antenna working parameter correction model;
inputting the second configuration data set and the second measurement report data set into an antenna engineering parameter prediction model to obtain a prediction result of the engineering parameters of the second equipment antenna, including:
obtaining a first prediction result of the engineering parameters of the first equipment antenna according to the antenna engineering parameter generation model;
obtaining a prediction result of the engineering parameters of at least one other equipment antenna according to the antenna engineering parameter generation model;
inputting second sample characteristic data of the second equipment antenna, a first prediction result of the engineering parameters of the second equipment antenna and a prediction result of the engineering parameters of the at least one other equipment antenna into the antenna engineering parameter correction model to obtain a second prediction result of the engineering parameters of the second equipment antenna; the second sample characteristic data of the second device antenna includes multiple pieces of signal reception power data of the cell or the RRU of the at least one other device antenna, and location data of a terminal corresponding to each piece of signal reception power data in the multiple pieces of signal reception power data and the signal reception power data of the cell or the RRU of the second device antenna.
8. The method of claim 7, wherein the first set of engineering parameters comprises engineering parameters of the first device antenna including at least one of position data and attitude data of the first device antenna; the first set of configuration data comprises configuration data of the first device antenna, the configuration data of the first device antenna representing configuration information of network parameters of the first device antenna; the measurement report data in the first measurement report data set comprises position data and signal received power data of the terminal.
9. The method according to claim 7 or 8, wherein obtaining a first prediction of the engineering parameters of the first device antenna from the antenna engineering parameter generation model comprises:
obtaining first sample characteristic data of the second device antenna according to the second measurement report data set and the configuration data of the second device antenna; the first sample characteristic data of the second device antenna comprises a plurality of signal receiving power data of a cell or a Remote Radio Unit (RRU) which is subordinate to the second device antenna and position data of a terminal corresponding to each signal receiving power data in the plurality of signal receiving power data;
Obtaining the antenna type of the second equipment antenna according to the configuration data of the second equipment antenna;
and inputting the first sample characteristic data of the second equipment antenna and the antenna type of the second equipment antenna into the antenna engineering parameter generation model to obtain a first prediction result of the engineering parameters of the second equipment antenna.
10. The method of claim 9, wherein the antenna parameter generation model is obtained by model training according to engineering parameters and a first feature set of the first device antenna; the first feature set includes first sample feature data of the first device antenna and an antenna type of the first device antenna, where the first sample feature data of the first device antenna includes multiple signal reception power data of a cell or an RRU that belongs to the first device antenna, and location data of a terminal corresponding to each signal reception power data in the multiple signal reception power data.
11. The method of claim 9, wherein obtaining the first sample characteristic data of the second device antenna according to the second measurement report data set and the configuration data of the second device antenna comprises:
Determining a cell or RRU (radio remote unit) belonging to the second equipment antenna according to the configuration data of the second equipment antenna;
determining measurement report data corresponding to the cell or the RRU from the second measurement report data set;
and performing feature extraction according to the measurement report data corresponding to the cell or the RRU to obtain first sample feature data of the second equipment antenna.
12. The method according to any one of claims 8, 10 and 11, wherein the second configuration data set further comprises configuration data of at least one other device antenna peripheral to the second device antenna;
after obtaining the first prediction result of the engineering parameter of the second device antenna, the method further includes:
obtaining second sample characteristic data of the second device antenna from the second configuration data set and the second measurement report data set.
13. The method of claim 12, wherein the antenna engineering parameter correction model is obtained by model training according to the first engineering parameter set and the second feature set; wherein the first engineering parameter set comprises engineering parameters of the first equipment antenna and engineering parameters of at least one other equipment antenna around the first equipment antenna; the second feature set comprises second sample feature data of the first device antenna, a prediction result of the engineering parameter of the first device antenna, and a prediction result of the engineering parameter of at least one other device antenna around the first device antenna; the second sample characteristic data includes multiple pieces of signal reception power data of a cell of at least one other device antenna or an RRU around the first device antenna, and position data of a terminal corresponding to each piece of signal reception power data in the multiple pieces of signal reception power data and signal reception power data of the cell or the RRU of the first device antenna; the prediction result of the engineering parameter of the first equipment antenna and the prediction result of the engineering parameter of at least one other equipment antenna around the first equipment antenna are obtained according to the antenna engineering parameter generation model.
14. The method according to claim 12, wherein the at least one other device antenna around the second device antenna is top N device antennas around the second device antenna, the top N device antennas representing N device antennas most correlated to the second device antenna among the plurality of device antennas, where N is an integer greater than or equal to 1.
15. The method of claim 14, wherein obtaining second sample characterization data for the second device antenna based on the second configuration data set and the second measurement report data set comprises:
determining a cell or RRU (radio remote unit) belonging to the second equipment antenna and a cell or RRU belonging to the top N equipment antennas according to the second configuration data set;
setting at least one measurement report data corresponding to the cell of the top N device antennas or the cell of any one of the RRUs according to the second measurement report data set, wherein each measurement report data in the at least one measurement report data comprises signal receiving power data of the cell of any one device antenna, signal receiving power data of the cell of the second device antenna and position data of the terminal;
And performing feature extraction according to the cell of the top N device antennas or the cell of each device antenna in the RRUs or the measurement report data corresponding to the RRU, and obtaining second sample feature data of the second device antenna.
16. A computing device that predicts antenna engineering parameters, the computing device comprising a processor and a memory, wherein:
the memory is used for storing a program code, a first engineering parameter set, a first configuration data set and a first measurement report data set uploaded to a first equipment antenna by a terminal; wherein the first set of engineering parameters comprises engineering parameters of the first device antenna including at least one of position data and attitude data of the first device antenna; the first set of configuration data comprises configuration data of the first device antenna, the configuration data of the first device antenna representing configuration information of network parameters of the first device antenna; measurement report data in the first measurement report data set comprises position data and signal received power data of the terminal;
the processor is configured to execute the program code in the memory to implement the method of any of claims 1-6.
17. A computing device that predicts antenna engineering parameters, the computing device comprising a processor and a memory, wherein:
the memory is used for storing a program code, a second configuration data set and a second measurement report data set uploaded to a second equipment antenna by the terminal; wherein the second configuration data set comprises configuration data of the second device antenna, the configuration data of the second device antenna representing configuration information of network parameters of the second device antenna; measurement report data in the second measurement report data set comprises position data and signal received power data of the terminal;
the processor is configured to execute the program code in the memory to implement the method of any of claims 7-15.
18. A system for predicting antenna engineering parameters, the system comprising the computing device for training an antenna parameters prediction model of claim 16 and the computing device for predicting parameters based on the antenna parameters prediction model of claim 17, wherein:
the computing device for training the antenna engineering parameter prediction model is specifically configured to obtain a first engineering parameter set, a first configuration data set, and a first measurement report data set uploaded to the first device antenna by a terminal; performing model training according to the engineering parameters of the first equipment antenna, the configuration data of the first equipment antenna and the first measurement report data set to obtain an antenna engineering parameter prediction model; inputting the antenna parameters prediction model to the computing device for predicting the parameters based on the antenna parameters prediction model;
The computing device for predicting the power parameters based on the antenna power parameter prediction model is specifically configured to obtain a second configuration data set and a second measurement report data set uploaded to a second device antenna by the terminal; inputting the second configuration data set and the second measurement report data set into an antenna engineering parameter prediction model to obtain a prediction result of the engineering parameters of the second equipment antenna; the second set of configuration data includes configuration data for the second device antenna.
CN201811502499.0A 2018-12-07 2018-12-07 Method and equipment for predicting antenna engineering parameters Active CN111368384B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811502499.0A CN111368384B (en) 2018-12-07 2018-12-07 Method and equipment for predicting antenna engineering parameters

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811502499.0A CN111368384B (en) 2018-12-07 2018-12-07 Method and equipment for predicting antenna engineering parameters

Publications (2)

Publication Number Publication Date
CN111368384A CN111368384A (en) 2020-07-03
CN111368384B true CN111368384B (en) 2022-06-10

Family

ID=71209727

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811502499.0A Active CN111368384B (en) 2018-12-07 2018-12-07 Method and equipment for predicting antenna engineering parameters

Country Status (1)

Country Link
CN (1) CN111368384B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114826988A (en) * 2021-01-29 2022-07-29 中国电信股份有限公司 Method and device for anomaly detection and parameter filling of time sequence data
CN113642168A (en) * 2021-08-09 2021-11-12 佛山科学技术学院 Antenna design method based on structural parameter prediction and crowd sourcing optimization
CN113746572B (en) * 2021-08-31 2023-12-05 深圳市华信天线技术有限公司 Method for detecting GNSS signal receiving performance of base station antenna industrial parameter equipment
CN116249134A (en) * 2021-12-08 2023-06-09 中国电信股份有限公司 Base station azimuth correction method, device and system and storage medium

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101778409A (en) * 2009-12-23 2010-07-14 华为技术有限公司 Method, device and system for obtaining adjacent area antenna configuration parameters
EP2413520A1 (en) * 2010-07-26 2012-02-01 Alcatel Lucent A method for transmission and amplification of signals, a transmitting device and a receiving device therefor
CN102065445A (en) * 2010-11-30 2011-05-18 广州天越电子科技有限公司 Simulation system and method for realizing wireless network communication rate coverage
CN103983235B (en) * 2014-05-30 2016-05-18 西安融思电子信息技术有限公司 Antenna for base station engineering parameter measuring method
US9866993B2 (en) * 2015-02-27 2018-01-09 Qualcomm Incorporated Distribution and utilization of antenna information for location determination operations
CN105509704B (en) * 2016-01-04 2018-05-22 深圳市顶一精密五金有限公司 The longitude and latitude localization method of outdoor short-wave antenna
CN107830846B (en) * 2017-09-30 2020-04-10 杭州艾航科技有限公司 Method for measuring angle of communication tower antenna by using unmanned aerial vehicle and convolutional neural network
CN108375363B (en) * 2017-12-05 2021-06-04 中国移动通信集团福建有限公司 Antenna azimuth deflection checking method, device, equipment and medium
CN108920841B (en) * 2018-07-05 2023-02-14 中南大学 Antenna design method
CN111628833B (en) * 2020-06-10 2022-02-08 桂林电子科技大学 MIMO antenna number estimation method based on convolutional neural network

Also Published As

Publication number Publication date
CN111368384A (en) 2020-07-03

Similar Documents

Publication Publication Date Title
EP3661249B1 (en) Measurement-based wireless communications network design
CN111368384B (en) Method and equipment for predicting antenna engineering parameters
US9585031B2 (en) Method to guide the placement of new small cell
CN103369668B (en) Wireless communication system is utilized to carry out the method for location positioning, device and mobile terminal
US11012133B2 (en) Efficient data generation for beam pattern optimization
CN113661676B (en) Selecting an uplink transmission band in a wireless network
CN110741564A (en) Cell ranking in a multi-beam system
CN109428657B (en) Positioning method and device
US10986553B2 (en) Terminal device, communication system, and communication control method
CN112135291B (en) State detection method and device
US9813929B2 (en) Obtaining information for radio channel modeling
CN104205910A (en) Mobile communication terminals, method for using a communication service and method for determining information related to a geographical position of a mobile communication terminal
EP4047382A1 (en) Rf-fingerprinting map update
US20230071942A1 (en) Cellular network indoor traffic auto-detection
US11652538B2 (en) Reducing uplink interference induced by aerial user equipment
CN111405464B (en) Base station position detection method and device
CN114487995A (en) Method for determining cell antenna azimuth angle, related device and equipment
US20230084811A1 (en) Device positioning
CN113453334B (en) Positioning method and positioning device
WO2024079737A1 (en) First node and methods performed thereby for handling location of a network node in a geographical area for operation in a communications system
WO2023117205A1 (en) Sidelink positioning in cellular system
WO2024017516A1 (en) Bandwidth and/or scenario based feature selection
CN116782276A (en) Network quality evaluation method, device and storage medium
CN114630378A (en) Method, device, server and storage medium for determining network distribution
CN116193488A (en) Position determining method, device and storage medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant