WO2022156226A1 - 天线配置参数优化方法、装置及存储介质 - Google Patents

天线配置参数优化方法、装置及存储介质 Download PDF

Info

Publication number
WO2022156226A1
WO2022156226A1 PCT/CN2021/116534 CN2021116534W WO2022156226A1 WO 2022156226 A1 WO2022156226 A1 WO 2022156226A1 CN 2021116534 W CN2021116534 W CN 2021116534W WO 2022156226 A1 WO2022156226 A1 WO 2022156226A1
Authority
WO
WIPO (PCT)
Prior art keywords
antenna configuration
configuration parameter
prediction model
simulation
model
Prior art date
Application number
PCT/CN2021/116534
Other languages
English (en)
French (fr)
Inventor
李晓雯
耿新力
吴珏莹
宋其涛
Original Assignee
华为技术有限公司
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 华为技术有限公司 filed Critical 华为技术有限公司
Priority to EP21920605.9A priority Critical patent/EP4280660A4/en
Publication of WO2022156226A1 publication Critical patent/WO2022156226A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/02—Arrangements for optimising operational condition
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/06—Power analysis or power optimisation

Definitions

  • the present application relates to the field of communications technologies, and in particular, to an antenna configuration parameter optimization method, device, and storage medium.
  • Wireless communication network planning is to design a reasonable and feasible wireless network layout (usually deploying sites according to the shape of a cell) based on the customer's requirements for network quality (coverage, interference, capacity), combined with the topography of the planned area, user distribution and other characteristics.
  • network quality coverage, interference, capacity
  • network quality problems such as weak coverage, overlapping coverage, and unbalanced load in real networks.
  • the operator needs to optimize the existing network parameters to solve the above-mentioned quality problems in the current network.
  • site coverage is usually controlled by adjusting radio frequency parameters, and indicators such as road section-level or full user stereo coverage quality, capacity, and speed are enhanced to fully guarantee user accessibility, mobility, and experience.
  • the new features of the 5th generation wireless systems (5G) scenario bring huge technical challenges to network parameter optimization.
  • Massive Multi-input Multi-output (Massive MIMO) antenna of a large-scale antenna array it provides broadcast beam weights for different coverage scenarios, so that the Radio Frequency (RF) parameters are changed from those of ordinary antennas.
  • 3 types physical azimuth, physical downtilt, electronic downtilt
  • 6 types of Massive MIMO antennas physical azimuth, physical downtilt, digital azimuth, digital downtilt, horizontal beamwidth, vertical beamwidth. Therefore, when the Massive MIMO antenna is introduced, the adjustable parameters are increased from 3 elements to 6 elements, and the RF parameter adjustable combination space increases exponentially.
  • a simulation model is constructed based on the data (electronic map, antenna file, engineering parameter data, Measurement Report (MR)/Road Test (Drive Test, DT) data fed back from the existing network; wherein, the simulation model will evaluate parameters When the combination changes, the network quality changes; according to the evaluation results of the current parameter combination by the simulation model, identify the areas with poor quality in the current network, and determine the problem cells that need to adjust the parameters; by using the genetic algorithm for the parameter combinations of these problem cells. Seeking for optimization, until the genetic algorithm converges or reaches the maximum number of iterations of the genetic algorithm, then the antenna parameters obtained by optimization are configured as the antenna parameters of the above-mentioned problem cell, and the delivery of the existing network scheme is carried out.
  • the prior art obtains a new round of measurement
  • the simulation model needs to be re-established and the optimization algorithm is initialized, that is to say, each time a new round of measurement data is obtained, the simulation model needs to be re-established, and then the optimization algorithm needs to be re-initialized, which makes the end-to-end running time long. , the optimization efficiency is low.
  • the present application discloses an antenna configuration parameter optimization method, device and storage medium, which can improve the efficiency of finding optimal antenna configuration parameters.
  • an embodiment of the present application provides an antenna configuration parameter optimization method, including: incrementally updating a first prediction model according to a first antenna configuration parameter corresponding to an optimization area and a score of the first antenna configuration parameter to Obtain a second prediction model; obtain second antenna configuration parameters according to the second prediction model; determine whether to use the second antenna configuration parameters as the first target configuration parameters; if the second antenna configuration parameters are used as the first target configuration parameters, the second antenna configuration parameters are delivered.
  • the above-mentioned incremental updating of the first prediction model according to the first antenna configuration parameter corresponding to the optimization area and the score of the first antenna configuration parameter to obtain the second prediction model can be understood as: according to the first antenna corresponding to the optimization area
  • the configuration parameter and the score of the first antenna configuration parameter update the calculation formula, parameters or attributes of the first prediction model, thereby obtaining the second prediction model.
  • the full update is to generate the second prediction model according to the first antenna configuration parameter, the score of the first antenna configuration parameter, and the antenna configuration parameter and score used before generating the first prediction model. That is, the full update does not generate the second prediction model based on the already generated first prediction model, which does not establish the association between the first prediction model and the second prediction model.
  • the incremental update updates the generated first prediction model based on the first antenna configuration parameter and the score of the first antenna configuration parameter to obtain a new prediction model.
  • the first prediction model is incrementally updated according to the first antenna configuration parameter and the score of the first antenna configuration parameter, and then new antenna configuration parameters are obtained according to the updated prediction model.
  • the updated prediction model can quickly converge to the antenna configuration parameters with higher scores, thereby improving the optimization efficiency of the antenna configuration parameters.
  • this solution can effectively improve the delivery quality and delivery efficiency.
  • whether to use the second antenna configuration parameter as the first target configuration parameter is determined by determining whether a preset condition is reached. For example, according to whether the number of iterations N1 is reached, or according to whether the second antenna configuration parameter is obtained for N2 consecutive times (that is, the same antenna configuration parameter is obtained by N2 iterations), and then it is determined whether to use the second antenna configuration parameter as the first target configuration parameter.
  • N1 and N2 are both positive integers. If the number of iterations N1 is currently reached, or the second antenna configuration parameter is obtained for N2 consecutive times (at this time, the first antenna configuration parameter is the same as the second antenna configuration parameter), then the second antenna configuration parameter is determined as the first antenna configuration parameter.
  • a target configuration parameter is determined according to whether the number of iterations N1 is reached, or according to whether the second antenna configuration parameter is obtained for N2 consecutive times (that is, the same antenna configuration parameter is obtained by N2 iterations), and then it is determined whether to use the second antenna configuration parameter as the first target configuration parameter.
  • the second prediction model is determined according to the second antenna configuration parameter and the score of the second antenna configuration parameter Incremental updating is performed to obtain a third prediction model; and third antenna configuration parameters are obtained according to the third prediction model.
  • the third prediction model is obtained by incrementally updating the second prediction model according to the second antenna configuration parameter and the score of the second antenna configuration parameter.
  • the method further includes: acquiring measurement data and simulation data, wherein the measurement data is data obtained after the configuration parameters of the second target are issued, and the simulation data is obtained according to the second target configuration parameters and data obtained from the first simulation model; correcting the first simulation model according to the measurement data and simulation data to obtain a second simulation model; according to the first antenna configuration parameters and the second simulation model A score for the first antenna configuration parameter is obtained.
  • the simulation model is corrected based on the measurement data obtained after delivery based on the second target configuration parameters and the simulation data obtained by simulation according to the second target configuration parameters, so that the error of the simulation model is reduced and the The actual implementation effect of the live network, thereby improving the accuracy of the simulation.
  • the above-mentioned calibrating the first simulation model according to the measurement data and the simulation data to obtain the second simulation model includes: linking the first simulation model with the measurement data and the simulation data. The path loss is corrected to obtain the second simulation model.
  • the above-mentioned calibrating the link loss of the first simulation model according to the measurement data and simulation data includes: obtaining a first reference of each grid in the plurality of grids according to the measurement data signal received power; obtain the second reference signal received power of each grid in the plurality of grids according to the simulation data of the first simulation model; according to the first reference signal received power of each grid and the first reference signal received power
  • the Kalman matrix of each grid is obtained from the received power of the two reference signals; the received power of the first reference signal, the received power of the second reference signal and the Kalman matrix of each grid are used for the calculation of the first simulation model. Link loss is corrected.
  • the embodiment of the present application corrects errors in the simulation model based on the coverage evaluation and correction technology of Kalman filtering, so that the simulation evaluation is more accurate.
  • the method further includes: acquiring the stored first prediction model.
  • the first prediction model can be directly obtained when incrementally updating the first prediction model, thereby improving the efficiency of antenna configuration parameter optimization.
  • the corrected link loss also includes storing the corrected link loss.
  • the corrected link loss can be directly obtained when the simulation model is used for simulation, which improves the simulation efficiency and further improves the efficiency of antenna configuration parameter optimization.
  • the method further includes storing a second simulation model.
  • the second simulation model can be directly acquired and used during the later simulation, which improves the simulation efficiency.
  • the method further includes: obtaining an initial prediction model according to a plurality of groups of historical antenna configuration parameters corresponding to the optimization area and the scores of each group of the historical antenna configuration parameters, so as to obtain an initial prediction model according to the initial prediction model Describe the first prediction model.
  • an embodiment of the present application provides an apparatus for optimizing antenna configuration parameters, including: a first model generation module configured to evaluate the first antenna configuration parameter according to the first antenna configuration parameter corresponding to the optimization area and the score of the first antenna configuration parameter The prediction model is incrementally updated to obtain a second prediction model; a first parameter generation module is used to obtain a second antenna configuration parameter according to the second prediction model; a judgment module is used to determine whether to configure the second antenna The parameter is used as the first target configuration parameter; the parameter determination module is configured to deliver the second antenna configuration parameter if the second antenna configuration parameter is used as the first target configuration parameter.
  • the first prediction model is incrementally updated according to the first antenna configuration parameter and the score of the first antenna configuration parameter, and then new antenna configuration parameters are obtained according to the updated prediction model.
  • the prediction accuracy of the prediction model can be improved, thereby improving the optimization efficiency of the antenna configuration parameters, and effectively improving the delivery quality and delivery efficiency.
  • the apparatus further includes: a second model generation module, configured to, if the second antenna configuration parameter is not used as the first target configuration parameter, The second prediction model is incrementally updated by the score of the second antenna configuration parameter to obtain a third prediction model; the second parameter generation module is configured to obtain the third antenna configuration parameter according to the third prediction model.
  • a second model generation module configured to, if the second antenna configuration parameter is not used as the first target configuration parameter, The second prediction model is incrementally updated by the score of the second antenna configuration parameter to obtain a third prediction model; the second parameter generation module is configured to obtain the third antenna configuration parameter according to the third prediction model.
  • the device further includes a score determination module, configured to: acquire measurement data and simulation data, wherein the measurement data is data obtained after the second target configuration parameter is issued, and the simulation data is According to the data obtained from the second target configuration parameters and the first simulation model; the first simulation model is corrected according to the measurement data and the simulation data to obtain a second simulation model; according to the first antenna configuration parameters and The second simulation model obtains a score for the first antenna configuration parameter.
  • a score determination module configured to: acquire measurement data and simulation data, wherein the measurement data is data obtained after the second target configuration parameter is issued, and the simulation data is According to the data obtained from the second target configuration parameters and the first simulation model; the first simulation model is corrected according to the measurement data and the simulation data to obtain a second simulation model; according to the first antenna configuration parameters and The second simulation model obtains a score for the first antenna configuration parameter.
  • the apparatus further includes an obtaining module, configured to: obtain the stored first prediction model.
  • the device further includes a third model generation module, configured to: obtain an initial prediction model according to the multiple groups of historical antenna configuration parameters corresponding to the optimized area and the scores of each group of the historical antenna configuration parameters, so as to obtain an initial prediction model.
  • the first prediction model is obtained according to the initial prediction model.
  • the present application provides an apparatus for optimizing antenna configuration parameters, including a processor and a memory; wherein the memory is used to store program codes, and the processor is used to call the program codes to execute the above method.
  • the present application provides a computer storage medium, including computer instructions, which, when the computer instructions are executed on an electronic device, cause the electronic device to perform the method provided by any possible implementation manner of the first aspect .
  • the embodiments of the present application provide a computer program product, which, when the computer program product runs on a computer, enables the computer to execute the method provided by any possible implementation manner of the first aspect.
  • an embodiment of the present application provides a chip system, which is applied to an electronic device; the chip system includes one or more interface circuits and one or more processors; the interface circuit and the The processors are interconnected by lines; the interface circuit is used for receiving signals from the memory of the electronic device and sending the signals to the processor, the signals including computer instructions stored in the memory; when the processing The electronic device executes the method when the computer executes the computer instructions.
  • the device described in the second aspect the device described in the third aspect, the computer storage medium described in the fourth aspect, the computer program product described in the fifth aspect, or the chip described in the sixth aspect
  • the systems are used to perform any of the methods provided in the first aspect. Therefore, for the beneficial effects that can be achieved, reference may be made to the beneficial effects in the corresponding method, which will not be repeated here.
  • FIG. 1a is a schematic diagram of a scenario of an antenna configuration parameter optimization system provided by an embodiment of the present application
  • FIG. 1b is a schematic diagram of an antenna configuration parameter optimization method provided by an embodiment of the present application.
  • FIG. 1c is a schematic diagram of a method for optimizing an antenna configuration parameter provided by an embodiment of the present application
  • FIG. 2 is a schematic flowchart of an antenna configuration parameter optimization method provided by an embodiment of the present application
  • FIG. 3 is a schematic flowchart of another method for optimizing antenna configuration parameters provided by an embodiment of the present application.
  • FIG. 5 is a schematic structural diagram of an antenna configuration parameter optimization apparatus provided by an embodiment of the present application.
  • FIG. 6 is a schematic structural diagram of another antenna configuration parameter optimization apparatus provided by an embodiment of the present application.
  • the embodiments of the present application may be applied to 5G or other future networks, such as 6G.
  • the antenna configuration parameter optimization system includes a computing server 101 , a network management server 102 and a base station 103 .
  • the network management server 102 may specifically be an operation and maintenance center (Operation and Maintenance Centre, OMC).
  • OMC operation and Maintenance Center
  • This application takes the network management server 102 as an OMC as an example for description. It should be understood that the network management server 102 may also be an operations support system (Operations Support System). System, OSS) and so on.
  • the computing server 101 sends a measurement data input request to the OMC. After the OMC receives the measurement data input request, it controls the base station 103 to start collecting data.
  • the base station 103 starts to measure and control the measurement data of the antenna in the communication network, and reports the measurement data to the OMC in real time. After the OMC collects the measurement data, it can be imported into the computing server 101 to report the measurement data.
  • the computing server 101 performs an optimization process on the antenna configuration parameters based on the measurement data, and sends a configuration command to the OMC after obtaining the optimized antenna configuration parameters.
  • the antenna configuration parameter may also be called an antenna radio frequency (Radio Frequency, RF) parameter.
  • RF antenna radio frequency
  • the OMC converts the optimized RF parameters into Man Machine Language (MML) commands, and adjusts the RF parameters of the base station antenna to realize the delivery of the adjusted parameters.
  • MML Man Machine Language
  • the above-mentioned calculation server 101 performs optimization processing of antenna configuration parameters based on the measurement data, as shown in FIG. 1b.
  • the calculation server 101 establishes a simulation model for the input data (measurement data) based on the electromagnetic wave propagation characteristics.
  • the changed reference signal received power (Reference Signal Received Power, RSRP) value can be simulated and evaluated; then according to The RSRP threshold and the simulated RSRP value determine the problem cell in the optimization area; then use the optimization algorithm to optimize the RF parameters of the above problem cell; in the process of optimization, the scores of different antenna configuration parameters can be evaluated by simulation ; Train the prediction model according to the antenna configuration parameters and the corresponding scores tried in the RF parameter optimization process, and obtain the currently predicted optimal antenna configuration parameters according to the prediction model. By inputting the optimal antenna configuration parameters into the above simulation The model can obtain the score of the antenna configuration parameter, and then incrementally update the prediction model according to the antenna configuration parameter and the score of the antenna configuration parameter.
  • the computing server 101 also stores the updated prediction model.
  • the optimal antenna configuration parameter corresponding to the current prediction model is used as the target configuration parameter, and then the network management server 102 lowers the target configuration parameter send.
  • the specific steps on how to use and update the prediction model can be seen in Figure 1c, as follows:
  • the calculation server 101 determines whether a prediction model is stored; if the prediction model is not stored, the calculation server 101 reads the stored multiple groups of historical antenna configuration parameters and the scores of each group of antenna configuration parameters, and generates a prediction model, which may be referred to as Initial prediction model. If a prediction model is stored, the stored prediction model is directly read.
  • the calculation server 101 obtains the currently predicted optimal antenna configuration parameter according to the prediction model, and can obtain the score of the antenna configuration parameter based on the simulation model; when the algorithm corresponding to the prediction model does not converge or does not reach the number of iterations, calculate The server 101 incrementally updates the prediction model according to the antenna configuration parameter and the score of the antenna configuration parameter, until the algorithm corresponding to the prediction model converges or reaches the number of iterations, the calculation server 101 calculates the optimal antenna configuration corresponding to the current prediction model parameters as target configuration parameters and store the prediction model.
  • the computing server 101 further corrects the above-mentioned simulation model according to the RSRP value obtained by simulation and the RSRP value obtained by measurement, so as to obtain a corrected simulation model.
  • the computing server 101 also stores the above-mentioned simulation model, the problem cell and the antenna configuration parameters tried in the above-mentioned RF parameter optimization process and the corresponding score.
  • the above system is described by taking the calculation server 101 performing antenna configuration parameter optimization and the network management server 102 performing antenna configuration parameter delivery as an example.
  • the above-mentioned computing server 101 and the network management server 102 may be integrated into one, so as to directly perform optimization and delivery of antenna configuration parameters.
  • FIG. 2 it is a schematic flowchart of an antenna configuration parameter optimization method provided by an embodiment of the present application.
  • the embodiments of the present application are described by taking the calculation server executing the method for optimizing the configuration parameters of the antenna as an example.
  • the antenna configuration parameter optimization method includes steps 201-204, which are as follows:
  • the above-mentioned optimization area may be an area corresponding to a plurality of communication cells controlled by the base station that is periodically optimized.
  • the above-mentioned first antenna configuration parameter may be any antenna configuration parameter obtained by the computing server in the process of determining the target configuration parameter.
  • the first antenna configuration parameter may include one or more of the following: physical azimuth, physical downtilt, digital azimuth, digital downtilt, horizontal beamwidth or vertical beamwidth, and the like.
  • the score of the above-mentioned first antenna configuration parameter is used to represent the excellent degree of network quality (specifically, signal quality) corresponding to the first antenna configuration parameter. It should be understood that the score may also be specifically called a label, etc., which is not specifically limited in this solution.
  • the above-mentioned incremental updating of the first prediction model according to the first antenna configuration parameter corresponding to the optimization area and the score of the first antenna configuration parameter to obtain the second prediction model can be understood as: according to the first antenna corresponding to the optimization area
  • the configuration parameter and the score of the first antenna configuration parameter update the calculation formula, parameters or attributes of the first prediction model, thereby obtaining the second prediction model.
  • the full update is to generate the second prediction model according to the first antenna configuration parameter, the score of the first antenna configuration parameter, and the antenna configuration parameter and score used before generating the first prediction model. That is, the full update does not generate the second prediction model based on the already generated first prediction model, which does not establish the association between the first prediction model and the second prediction model.
  • the incremental update updates the generated first prediction model based on the first antenna configuration parameter and the score of the first antenna configuration parameter to obtain a new prediction model.
  • the above-mentioned first prediction model is generated according to the antenna configuration parameter set X N and the corresponding score set FN of the antenna configuration parameters.
  • the above-mentioned antenna configuration parameter set X N ⁇ x i ,i ⁇ N ⁇ represents the set of multiple groups of antenna configuration parameters x i ;
  • the score set F N ⁇ f(x i ),i ⁇ N ⁇ , represents the multiple groups of antennas
  • the set of scores f( xi ) corresponding to each set of parameters in the configuration parameters.
  • the first prediction model is taken as an example of a probability prediction model for description.
  • the antenna configuration parameters x i , x j both belong to the antenna configuration parameter set X N , and ⁇ and ⁇ are both non-zero coefficients.
  • ⁇ , ⁇ ) is the probability of F N appearing in the case of corresponding hyperparameters ⁇ and ⁇ .
  • the hyperparameters ⁇ and ⁇ are calculated according to the above maximized edge log-likelihood, and then the covariance matrix K calculated according to the above k(x i , x j ) can be expressed as:
  • the gain matrix F N can be expressed as:
  • the parameters of the first prediction model include a covariance matrix K and a gain matrix F N .
  • the above-mentioned first prediction model is updated according to the first antenna configuration parameter x best and the score f(x best ) of the first antenna configuration parameter.
  • the first antenna configuration parameter x best is obtained according to the first prediction model. For details, please refer to the description in the embodiment shown in FIG. 4 , which will not be repeated here.
  • the updated covariance matrix K of the first prediction model can be expressed as:
  • the covariance matrix is obtained as shown in formula (3), and a new row is added on the basis of the original gain matrix F N as shown in formula (2) to obtain The gain matrix F N is shown in formula (4).
  • the incremental update of the above-mentioned first prediction model can be realized.
  • the above is only an example, and it may also be an update in other forms, which is not specifically limited in this solution.
  • the first prediction model is taken as an example of a neural network prediction model for description.
  • a sample pair is generated according to the antenna configuration parameter set XN and the corresponding antenna configuration parameter score set FN .
  • the sample pair takes the antenna configuration parameters as input data, and takes the scores of the antenna configuration parameters as the expected output, and then trains a neural network through learning.
  • the neural network satisfies the following conditions:
  • the parameters of the first prediction model are the weight matrix W and the deviation matrix B.
  • the above-mentioned first prediction model is updated according to the first antenna configuration parameter x best and the score f(x best ) of the first antenna configuration parameter.
  • the first antenna configuration parameter x best is obtained according to the first prediction model. For details, reference may be made to the description in the embodiment shown in FIG. 4 , which will not be repeated here.
  • the updated sample pair of the first prediction model can be expressed as:
  • the updated weight matrix W and bias matrix B of the above-mentioned first prediction model can be expressed as:
  • the incremental update of the model is described above by taking two different models as examples. Among them, it may also be an incremental update of other forms of models, which is not specifically limited in this solution.
  • the above-mentioned first prediction model can be obtained in the following ways:
  • An initial prediction model is obtained according to the multiple groups of historical antenna configuration parameters corresponding to the optimized area and the scores of each group of the historical antenna configuration parameters, so as to obtain the first prediction model according to the initial prediction model.
  • the initial prediction model may be the first prediction model.
  • the computing server obtains the scores of each group of historical antenna configuration parameters by simulating the multiple groups of historical antenna configuration parameters respectively, and obtains the first prediction model according to the multiple groups of historical antenna configuration parameters and the score of each group of the historical antenna configuration parameters.
  • the historical antenna configuration parameter with the highest score among the multiple groups of historical antenna configuration parameters is obtained according to the first prediction model, and the historical antenna configuration parameter with the highest score is the first antenna configuration parameter.
  • the above-mentioned multiple groups of historical antenna configuration parameters may be obtained from a preset antenna configuration parameter set.
  • the above-mentioned groups of historical antenna configuration parameters may also be randomly generated.
  • the initial prediction model may also be the first prediction model in time sequence corresponding to the optimization region.
  • the computing server incrementally updates the initial prediction model, and obtains the first prediction model through continuous iterative updating. Specifically, the computing server obtains the scores of each group of historical antenna configuration parameters by simulating multiple groups of historical antenna configuration parameters respectively, and obtains the initial prediction model according to the multiple groups of historical antenna configuration parameters and the scores of each group of the historical antenna configuration parameters .
  • the first prediction model is obtained by continuous iterative updating according to the initial prediction model.
  • the method further includes step 200: acquiring the stored first prediction model.
  • the first prediction model can be directly obtained when incrementally updating the first prediction model, thereby improving the efficiency of antenna configuration parameter optimization.
  • the method further includes step 2011, storing the second prediction model, so that the second prediction model can be incrementally updated subsequently.
  • the step 2011 may be after the step 201 and before the step 202, or it may be after the step 202, which is not specifically limited in this solution.
  • the computing server when obtaining any prediction model, the computing server also stores the prediction model for use in subsequent incremental updates.
  • the computing server may obtain the score of the first antenna configuration parameter in the foregoing step 201 by correcting the simulation model based on the measurement data and the simulation data.
  • the method further includes steps 3001-3003.
  • FIG. 3 a schematic flowchart of an antenna configuration parameter optimization method provided by an embodiment of the present application is as follows:
  • the above-mentioned measurement data includes one or more of the following: electronic map, antenna file, work parameter data representing current radio frequency parameters, measured MR data, or DT data, etc.
  • electronic maps are maps that are digitally stored and consulted using computer technology.
  • Antenna files are antenna lobe map files.
  • the MR data is the measurement report reported by the User Equipment (UE), including the ID of the primary service cell, the RSRP of the primary service cell, the ID of the adjacent cell, the RSRP of the adjacent cell, and the beam ID and Corresponds to RSRP.
  • DT data is road test data, which is similar to Minimization of Drive-Test (MDT) data.
  • the MDT data is a measurement report with latitude and longitude information reported by the UE, which can be considered as MR data with latitude and longitude.
  • the second target configuration parameter is the target configuration parameter obtained by the computing server last time relative to the current search target parameter (ie, the first target configuration parameter in step 204 below). Specifically, after the computing server issues the second target configuration parameters through the network management server, the above measurement data is acquired through the network management server. The computing server also obtains the above-mentioned simulation data by inputting the second target configuration parameter into the first simulation model.
  • the link loss of the first simulation model is corrected according to the measurement data and the simulation data to obtain a second simulation model.
  • the received power of the first reference signal of each grid of the plurality of grids is obtained according to the measurement data; the first reference signal received power of each grid of the plurality of grids is obtained according to the simulation data of the first simulation model; Two reference signal received powers; the Kalman matrix of each grid is obtained according to the first reference signal received power and the second reference signal received power of each grid; the Kalman matrix of each grid is obtained according to the first reference signal of each grid.
  • the received power of the signal, the received power of the second reference signal, and the Kalman matrix are used to correct the link loss of the first simulation model to obtain the corrected link loss, and then the second simulation model is obtained.
  • the calculation server performs grid processing on the MR/DT data according to the longitude and latitude information in each piece of MR/DT data in the measurement data, and then belongs into multiple grids.
  • simulation is performed according to the second target configuration parameter to obtain the simulation value RSRP' of the reference signal received power of each grid above;
  • the simulation variance P - of the grid is obtained by obtaining the simulated values of the grid surrounding such as 8 grids.
  • the simulated values of 8 grids around the grid can be expressed as RSRP' i , i ⁇ [1-8];
  • RSRP" RSRP'+Kal(RSRP-RSRP')
  • the Kalman filtering technology (multi-data source fusion) is adopted according to the simulated RSRP value and the measured RSRP value newly fed back from the existing network, and the reliability of the estimated value and the measured value is comprehensively considered, and weighted to obtain each corrected value.
  • the Linkloss of the grid reduces the simulation error. Simulation accuracy is improved by fusing simulated RSRP values with the latest measured RSRP.
  • a second simulation model is then obtained based on the corrected link loss.
  • the corrected reference signal received power of each grid is obtained, and then the score of the first antenna configuration parameter is obtained.
  • the ratio of the grids whose RSRP values are greater than the RSRP threshold in the entire optimized area to the total grids is recorded as the score of the first antenna configuration parameter.
  • the simulation model is corrected according to the measurement data obtained by issuing the second target configuration parameters and the simulation data obtained by simulating the second target configuration parameters, thereby obtaining a simulation model with higher simulation accuracy, so that In the process of optimizing the target configuration parameters, the accuracy and precision of the scoring of the antenna configuration parameters are improved, and the optimization efficiency of the antenna configuration parameters is further improved.
  • the computing server obtains the second antenna configuration parameter by inputting a plurality of arbitrary antenna configuration parameters into the second prediction model.
  • the second antenna configuration parameter is a better configuration parameter obtained by the second prediction model.
  • the second antenna configuration parameter is the configuration parameter with the highest score among the above-mentioned multiple arbitrary antenna configuration parameters.
  • the computing server determines whether to use the second antenna configuration parameter as the first target configuration parameter by determining whether a preset condition is met.
  • N1 and N2 are both positive integers. If the number of iterations N1 is currently reached, or the second antenna configuration parameter is currently obtained for N2 consecutive times (at this time, the first antenna configuration parameter is the same as the second antenna configuration parameter), the computing server determines to configure the second antenna parameter as the first target configuration parameter.
  • the second antenna configuration parameter is used as the first target configuration parameter, deliver the second antenna configuration parameter.
  • the computing server uses the second antenna configuration parameter as the first target configuration parameter
  • the computing server delivers the first target configuration parameter (ie, the second antenna configuration parameter) through the network management server. That is, the computing server sends the first target configuration parameter to the network management server, so that the network management server delivers the first target configuration parameter.
  • the computing server increases the second prediction model according to the second antenna configuration parameter and the score of the second antenna configuration parameter.
  • a quantitative update is performed to obtain a third prediction model; and a third antenna configuration parameter is obtained according to the third prediction model.
  • the second antenna configuration parameter is not used as the first target configuration parameter, that is, the preset condition is not met, for example, the number of iterations N1 is not currently reached, or the second antenna configuration parameter is not obtained for N2 consecutive times currently.
  • the computing server determines whether to use the third antenna configuration parameter as the first target configuration parameter; if the third antenna configuration parameter is used as the first target configuration parameter, the third antenna configuration parameter is processed by the network management server. Issued.
  • the computing server performs the above steps repeatedly until the first target configuration parameter is obtained, and issues it through the network management server.
  • the embodiment of the present application uses a computing server as an example for description.
  • the network management server and the computing server may also be integrated into one.
  • the computing server directly delivers the second antenna configuration parameter.
  • the first prediction model is incrementally updated according to the first antenna configuration parameter and the score of the first antenna configuration parameter, and then new antenna configuration parameters are obtained according to the updated prediction model.
  • the updated prediction model can quickly converge to the antenna configuration parameters with higher scores, thereby improving the optimization efficiency of the antenna configuration parameters.
  • this solution can effectively improve the delivery quality and delivery efficiency.
  • FIG. 4 it is a schematic flowchart of an antenna configuration parameter optimization method provided by an embodiment of the present application. It includes steps 401-406, as follows:
  • the computing server may obtain the first measurement data corresponding to the optimized area through the network management server.
  • the network management server collects the first measurement data by controlling the base station.
  • the first measurement data is the initial measurement data corresponding to the optimization area, that is, the first measurement data is the measurement data corresponding to when the optimization area has not yet started to be optimized.
  • the computing server determines, according to the first measurement data, a cell for which antenna configuration parameter optimization needs to be performed, and then determines a third target configuration parameter.
  • the third target configuration parameter is the second target configuration parameter in the foregoing embodiment.
  • step 402 may include steps 4021-4023, as follows:
  • the computing server uses the above-mentioned first measurement data to establish a simulation model according to the electromagnetic wave propagation characteristics.
  • the simulation model includes link loss Linkloss.
  • the electromagnetic wave propagation characteristics satisfy the following formula:
  • Reference signal received power RSRP antenna gain AntennaGain+power Power-link loss Linkloss(7);
  • the above-mentioned link loss Linkloss is the link loss Linkloss of each grid in the plurality of grids obtained by performing grid processing on the above-mentioned first measurement data.
  • the MR/DT data is rasterized to obtain a plurality of grids, and then the RSRP of each grid in the plurality of grids is obtained.
  • the antenna file and the engineering parameter data in the first measurement data the antenna gain AntennaGain and power Power of each grid are obtained.
  • the link loss Linkloss of each grid can be obtained.
  • the calculation server stores the link loss Linkloss of each grid for subsequent use of the simulation model.
  • the computing server obtains the RSRP of each grid according to the MR/DT data. Based on the reference signal received power RSRP corresponding to each grid, the grids with the reference signal received power less than the preset threshold are defined as weak coverage grids; the weak coverage grids are clustered to obtain weak coverage areas; The cells whose latitude and longitude belong to the weak coverage area are defined as problem cells.
  • the computing server randomly generates a set of antenna configuration parameters x i , x i ⁇ X all , where X all represents the global antenna configuration parameters, that is, the full set of optional antenna configuration parameters of the problem cell corresponding to the optimization area, that is, A set of all optional parameters.
  • the AntennaGain and Power corresponding to each grid are obtained according to the antenna file and the engineering parameter data.
  • the RSRP value of each grid corresponding to the antenna configuration parameter xi can be obtained according to the antenna configuration parameter xi and the above-mentioned simulation model.
  • the ratio of the grids whose RSRP values are greater than the RSRP threshold in the entire optimized area to the total grids is calculated, and is recorded as the score f( xi ).
  • the computing server stores the antenna configuration parameter parameter x i and the score corresponding to the antenna configuration parameter parameter x i .
  • the antenna configuration parameter x * with the highest score is the third target configuration parameter.
  • the third target configuration parameter includes the configuration of each RF parameter of each antenna in the problem cell.
  • the computing server downloads the third target configuration parameter to the existing network through the network management server, so as to adjust the RF parameters of the antenna in the actual communication network.
  • the computing server obtains, through the network management server, the second measurement data corresponding to the delivery of the third target configuration parameter, that is, the new measurement data.
  • the computing server may send the second measurement data after obtaining and delivering the above-mentioned third target configuration parameter to the network management server after a preset time interval.
  • the preset time may be several hours or several days, etc., which is not specifically limited in this solution.
  • the third target configuration parameter is the second target configuration parameter in the foregoing embodiment
  • the second measurement data may be the measurement data in the foregoing embodiment.
  • the second measurement data may only include MR/DT data. It may also include electronic maps, antenna files, work parameter data representing current radio frequency parameters, measured MR data and/or DT data, etc., which are not specifically limited in this solution.
  • the computing server rasterizes and assigns the MR/DT data to the multiple grids generated in step 4021, and then obtains the multiple grids.
  • the RSRP of each grid in each grid and then count the proportion of grids whose RSRP value is greater than the RSRP threshold in the entire optimized area to the total grid. If the ratio meets the preset requirement, go to step 406 to end the optimization; otherwise, if the ratio of the grids with RSRP values greater than the RSRP threshold to the total grids in the entire optimized area is smaller than the preset requirement, then go to step 405 .
  • the computing server obtains the first target configuration parameter according to the second prediction model; wherein, the computing server incrementally updates the first prediction model according to the first antenna configuration parameter and the score of the first antenna configuration parameter The second prediction model is obtained.
  • the computing server obtains according to the multiple groups of historical antenna configuration parameters tried in the process of obtaining the above-mentioned third target configuration parameters during the first optimization and the scores of each group of historical antenna configuration parameters in the multiple groups of historical antenna configuration parameters
  • the initial predictive model, and the initial predictive model is stored.
  • the computing server obtains the first prediction model by reading the stored initial prediction model and continuously iteratively updating, and stores the first prediction model.
  • the computing server obtains the first prediction model by reading the stored initial prediction model and continuously iteratively updating, and stores the first prediction model.
  • the computing server reads the stored first prediction model during subsequent updates, obtains the first antenna configuration parameter according to the first prediction model, and obtains the score of the first antenna configuration parameter by simulating the first antenna configuration parameter , and then incrementally update the first prediction model according to the first antenna configuration parameter and the score of the first antenna configuration parameter to obtain the second prediction model.
  • the computing server reads the stored first prediction model during subsequent updates, obtains the first antenna configuration parameter according to the first prediction model, and obtains the score of the first antenna configuration parameter by simulating the first antenna configuration parameter , and then incrementally update the first prediction model according to the first antenna configuration parameter and the score of the first antenna configuration parameter to obtain the second prediction model.
  • the embodiments of the present application take two implementation manners as examples for description.
  • the first prediction model in the foregoing embodiment is an example of a probability prediction model for description.
  • any set of antenna configuration parameters x k by inputting the antenna configuration parameters x k into the first prediction model, calculate the predicted value ⁇ (x k ) of its score and the variance ⁇ (x k ) of the predicted value:
  • the predicted value ⁇ (x k ) of the score corresponding to each antenna configuration parameter and the prediction value can be calculated based on the above formula.
  • the variance ⁇ (x k ) of the values it is also possible to perform sampling calculation on only part of the antenna configuration parameters in all the global antenna configuration parameters, which is not specifically limited in this solution.
  • the gain corresponding to the antenna configuration parameter can be calculated through the above predicted ⁇ (x k ) and ⁇ (x k ):
  • the f(x * ) is the score corresponding to the target configuration parameter x * (ie, the third target configuration parameter) found in the existing network from the last round of parameter optimization scheme.
  • the meaning of the gain EI(x k ) can be understood as: the score f(x k ) of the antenna configuration parameter x k is expected to be better than the score f(x * ) of the antenna configuration parameter x * previously delivered to the existing network.
  • the corresponding score f(x best ) is obtained by inputting x best to the simulation model.
  • Whether to stop the above algorithm is determined by judging whether the total number of iterations N1 is reached, or whether the xbest is obtained for N2 consecutive times. If the number of iterations N1 is reached, or if the x best is obtained for N2 consecutive times at present, the antenna configuration parameter x best and its score f(x best ) are output.
  • the above-mentioned N1 may be, for example, 100 or the like, and the above-mentioned N2 may be, for example, 10 or the like. This plan does not make any specific restrictions on this.
  • the first prediction model in the foregoing embodiment is an example of a neural network prediction model for description.
  • any set of antenna configuration parameters x k by inputting the antenna configuration parameters x k into the first prediction model, calculate the predicted value ⁇ (x k ) of its score and the variance ⁇ (x k ) of the predicted value:
  • G is a non-zero constant
  • the predicted value ⁇ (x k ) of the score corresponding to each antenna configuration parameter and the prediction value can be calculated based on the above formula.
  • the variance ⁇ (x k ) of the values may be sampled and calculated, which is not specifically limited in this solution.
  • the gain corresponding to the antenna configuration parameter can be calculated through the above predicted ⁇ (x k ) and ⁇ (x k ):
  • f(x * ) is the score corresponding to the target configuration parameter x * (ie, the third target configuration parameter) delivered to the existing network by the last round of parameter optimization scheme.
  • the corresponding score f(x best ) is obtained by inputting x best to the simulation model.
  • Whether to stop the above algorithm is determined by judging whether the total number of iterations N1 is reached, or whether the xbest is obtained for N2 consecutive times. If the number of iterations N1 is reached, or if the x best is obtained for N2 consecutive times at present, the antenna configuration parameter x best and its score f(x best ) are output.
  • the above-mentioned N1 may be, for example, 100 or the like, and the above-mentioned N2 may be, for example, 10 or the like. This plan does not make any specific restrictions on this.
  • the above-mentioned first prediction model is updated according to the x best and the score f(x best ) to obtain the above-mentioned second prediction model.
  • the incremental update reference may be made to the description of step 201 in the foregoing embodiment, and details are not repeated here.
  • steps 403-406 can be repeatedly executed until the measurement data corresponding to the optimized area meets the preset requirements, and then Stop optimizing.
  • the prediction model is incrementally updated according to the antenna configuration parameters and the scores of the antenna configuration parameters, and then new antenna configuration parameters are obtained according to the updated prediction model.
  • the updated prediction model can quickly converge to the antenna configuration parameters with higher scores, thereby improving the optimization efficiency of antenna configuration parameters, and effectively improving delivery quality and delivery efficiency.
  • the method for optimizing the antenna configuration parameters provided by this solution, when new measurement data is acquired, does not need to re-establish the simulation model based on the above-mentioned stored simulation model, which effectively improves the overall optimization efficiency compared with the prior art.
  • an embodiment of the present application provides an apparatus for optimizing an antenna configuration parameter.
  • the device includes a first model generation module 501, a first parameter generation module 502, a judgment module 503 and a parameter determination module 504, the details are as follows:
  • a first model generation module 501 configured to incrementally update the first prediction model according to the first antenna configuration parameter corresponding to the optimization area and the score of the first antenna configuration parameter to obtain the second prediction model;
  • a first parameter generation module 502 configured to obtain a second antenna configuration parameter according to the second prediction model
  • a judgment module 503, configured to determine whether to use the second antenna configuration parameter as the first target configuration parameter
  • the parameter determination module 504 is configured to deliver the second antenna configuration parameter if the second antenna configuration parameter is used as the first target configuration parameter.
  • the first prediction model is incrementally updated according to the first antenna configuration parameter and the score of the first antenna configuration parameter, and then new antenna configuration parameters are obtained according to the updated prediction model.
  • the prediction accuracy of the prediction model can be improved, thereby improving the optimization efficiency of the antenna configuration parameters, and effectively improving the delivery quality and delivery efficiency.
  • the apparatus further includes: a second model generation module, configured to, if the second antenna configuration parameter is not the first target configuration parameter, generate the second antenna configuration parameter according to the second antenna configuration parameter, the score of the second antenna configuration parameter and the The second prediction model obtains a third prediction model; the second parameter generation module is configured to obtain a third antenna configuration parameter according to the third prediction model.
  • a second model generation module configured to, if the second antenna configuration parameter is not the first target configuration parameter, generate the second antenna configuration parameter according to the second antenna configuration parameter, the score of the second antenna configuration parameter and the The second prediction model obtains a third prediction model; the second parameter generation module is configured to obtain a third antenna configuration parameter according to the third prediction model.
  • the device further includes a simulation model generation module for: acquiring measurement data and simulation data, wherein, The measurement data is the data obtained after the second target configuration parameter is issued, and the simulation data is the data obtained according to the second target configuration parameter and the second simulation model; The second simulation model is corrected to obtain the first simulation model.
  • the apparatus further includes an obtaining module, configured to: obtain the stored first prediction model.
  • the device further includes a third model generation module, configured to: obtain an initial prediction model according to the multiple groups of historical antenna configuration parameters corresponding to the optimization area and the scores of each group of the historical antenna configuration parameters, so as to obtain an initial prediction model according to the initial The prediction model obtains the first prediction model.
  • a third model generation module configured to: obtain an initial prediction model according to the multiple groups of historical antenna configuration parameters corresponding to the optimization area and the scores of each group of the historical antenna configuration parameters, so as to obtain an initial prediction model according to the initial The prediction model obtains the first prediction model.
  • the present application further provides an apparatus for optimizing antenna configuration parameters.
  • the apparatus for optimizing antenna configuration parameters includes at least one processor 601 , at least one memory 602 and at least one communication interface 603 .
  • the processor 601, the memory 602 and the communication interface 603 are connected through the communication bus and complete the mutual communication.
  • the processor 601 may be a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits used to control the execution of the above programs.
  • CPU central processing unit
  • ASIC application-specific integrated circuit
  • the communication interface 603 is used to communicate with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area network (Wireless Local Area Networks, WLAN) and the like.
  • RAN radio access network
  • WLAN Wireless Local Area Networks
  • the memory 602 may be read-only memory (ROM) or other type of static storage device that can store static information and instructions, random access memory (RAM) or other type of static storage device that can store information and instructions It can also be an electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, EEPROM), a compact disc read-only memory (CD-ROM) or other optical disk storage, optical disk storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or capable of carrying or storing desired program code in the form of instructions or data structures and capable of being executed by a computer Access any other medium without limitation.
  • the memory can exist independently and be connected to the processor through a bus.
  • the memory can also be integrated with the processor.
  • the memory 602 is used for storing the application code for executing the above solution, and the execution is controlled by the processor 601 .
  • the processor 601 is configured to execute the application code stored in the memory 602 .
  • the code stored in the memory 602 can execute one of the antenna configuration parameter optimization methods provided above.
  • An embodiment of the present application further provides a chip system, the chip system is applied to an electronic device; the chip system includes one or more interface circuits and one or more processors; the interface circuit and the processor pass through line interconnection; the interface circuit is used to receive signals from the memory of the electronic device and send the signals to the processor, the signals include computer instructions stored in the memory; when the processor executes the When executing the computer instructions, the electronic device performs the method.
  • Embodiments of the present application further provide a computer-readable storage medium, where instructions are stored in the computer-readable storage medium, when the computer or the processor is run on the computer, the computer or the processor causes the computer or the processor to execute any one of the above methods. or multiple steps.
  • Embodiments of the present application also provide a computer program product including instructions.
  • the computer program product when run on a computer or processor, causes the computer or processor to perform one or more steps of any of the above methods.
  • the computer program product includes one or more computer instructions.
  • the computer may be a general purpose computer, special purpose computer, computer network, or other programmable device.
  • the computer instructions may be stored in or transmitted over a computer-readable storage medium. The computer instructions can be sent from one website site, computer, server, or data center to another website site, computer, server or data center for transmission.
  • the computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that includes an integration of one or more available media.
  • the usable media may be magnetic media (eg, floppy disks, hard disks, magnetic tapes), optical media (eg, DVDs), or semiconductor media (eg, solid state disks (SSDs)), and the like.
  • “at least one” means one or more, and “plurality” means two or more.
  • “And/or”, which describes the association relationship of the associated objects, indicates that there can be three kinds of relationships, for example, A and/or B, which can indicate: the existence of A alone, the existence of A and B at the same time, and the existence of B alone, where A, B can be singular or plural.
  • the process can be completed by instructing the relevant hardware by a computer program, and the program can be stored in a computer-readable storage medium.
  • the program When the program is executed , which may include the processes of the foregoing method embodiments.
  • the aforementioned storage medium includes: ROM or random storage memory RAM, magnetic disk or optical disk and other mediums that can store program codes.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • General Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Molecular Biology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Geometry (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Computer Hardware Design (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Variable-Direction Aerials And Aerial Arrays (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

本申请实施例提供一种天线配置参数优化方法、装置及存储介质，包括：根据优化区域对应的第一天线配置参数、所述第一天线配置参数的评分对第一预测模型进行增量式更新以得到第二预测模型；根据所述第二预测模型得到第二天线配置参数；确定是否将所述第二天线配置参数作为第一目标配置参数；若将所述第二天线配置参数作为第一目标配置参数，则将所述第二天线配置参数进行下发。通过根据第一天线配置参数和第一天线配置参数的评分对第一预测模型进行增量式更新，进而根据更新后的预测模型得到新的天线配置参数。采用该手段，基于更新后的预测模型可以快速收敛至评分较高的天线配置参数，进而提高了天线配置参数优化效率。

Description

天线配置参数优化方法、装置及存储介质
本申请要求于2021年1月19日提交中国专利局、申请号为202110068483.9、申请名称为“天线配置参数优化方法、装置及存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及通信技术领域，尤其涉及一种天线配置参数优化方法、装置及存储介质。
背景技术
无线通信网络规划是根据客户对于网络质量(覆盖、干扰、容量)的需求，结合规划区域的地形地貌，用户分布等特征，设计合理可行的无线网络布局(通常是按照蜂窝形状部署站点)，以最小的投资满足客户的需求。在现实场景中，由于网络规划与实际物理环境存在差异，城市建设、用户发展变化，场景化参数差异化配置需求，导致现实网络中可能存在弱覆盖、重叠覆盖以及负荷不均衡等网络质量问题，运营商需要对现有网络参数进行优化，解决当前网络中上述的质量问题。
网络参数优化中通常通过调整射频参数控制站点覆盖，增强路段级或全量用户立体覆盖质量、容量、速率等指标，充分保障用户可接入性、移动性和体验。
第五代移动通信技术(5th generation wireless systems，5G)场景的新特性给网络参数优化带来巨大的技术挑战。针对大规模天线阵列的多天线形态(Massive Multi-input Multi-output，Massive MIMO)天线来说，其提供了不同覆盖场景的广播波束权值，使得射频(Radio Frequency，RF)参数从普通天线的3种(物理方位角、物理下倾角、电子下倾角)，变为Massive MIMO天线的6种(物理方位角、物理下倾角、数字方位角、数字下倾角、水平波束宽度、垂直波束宽度)。因此，当Massive MIMO天线引入后，可调参数从3要素增加为6要素，射频参数可调整组合空间指数级增长。此外，在5G场景下，由于宏站和微站相结合，站与站之间距离更近，耦合更强，导致网络环境更复杂；由于MIMO、多小区协作(Coordinated multi-point processing，CoMP)组网等新特性，导致网络结构更复杂；在这种场景下，仅根据基本电磁波传播公式建立的仿真平台，存在仿真模型精度低的问题，难以精准识别网络中存在的质量问题和评估出不同参数组合网络质量的差异性。
现有技术通过基于现网反馈的数据(电子地图，天线文件，工参数据，测量报告(Measurement Report，MR)/道路测试(Drive Test，DT)数据构建仿真模型；其中，仿真模型会评估参数组合变化时，网络质量的变化情况；根据仿真模型对当前参数组合的评估结果，识别当前网络中的质差区域，确定需要调整参数的问题小区；通过对这些问题小区的参数组合利用遗传算法进行寻优，直至遗传算法收敛或者达到遗传算法的最大迭代次数，然后将优化得到的天线参数配置为上述问题小区的天线参数，并进行现网方案的下发。现有技术在获得新一轮测量数据时，需要重新建立仿真模型和初始化寻优算法，也就是说，每次获得的新一轮测量数据均需要重新建立仿真模型，然后重新初始化寻优算法，该手段使得端到端运行时间久，优化效率较低。
发明内容
本申请公开了一种天线配置参数优化方法、装置及存储介质，可以提高寻找最优天线配置参数的效率。
第一方面，本申请实施例提供一种天线配置参数优化方法，包括：根据优化区域对应的第一天线配置参数、所述第一天线配置参数的评分对第一预测模型进行增量式更新以得到第二预测模型；根据所述第二预测模型得到第二天线配置参数；确定是否将所述第二天线配置参数作为第一目标配置参数；若将所述第二天线配置参数作为第一目标配置参数，则将所述第二天线配置参数进行下发。
上述根据优化区域对应的第一天线配置参数、所述第一天线配置参数的评分对第一预测模型进行增量式更新以得到第二预测模型，可以理解为：根据优化区域对应的第一天线配置参数、所述第一天线配置参数的评分对第一预测模型的计算公式、参数或属性等进行更新，从而得到第二预测模型。而全量式更新则是根据第一天线配置参数、所述第一天线配置参数的评分以及生成第一预测模型之前所使用到的天线配置参数和评分来生成第二预测模型。也就是说，全量式更新未基于已经生成的第一预测模型来生成第二预测模型，其并未建立第一预测模型和第二预测模型之间的关联。增量式更新相较于全量式更新，其基于第一天线配置参数和第一天线配置参数的评分来对该已生成的第一预测模型进行更新进而得到新的预测模型。
通过本申请实施例，通过根据第一天线配置参数和第一天线配置参数的评分对第一预测模型进行增量式更新，进而根据更新后的预测模型得到新的天线配置参数。采用该手段，通过增量式更新该预测模型，基于更新后的预测模型可以快速收敛至评分较高的天线配置参数，进而提高了天线配置参数优化效率，相较于现有全量式更新的手段，本方案可有效提升交付质量和交付效率。
作为一种实现方式，通过确定是否达到预设条件来确定是否将所述第二天线配置参数作为所述第一目标配置参数。例如根据是否达到迭代次数N1，或者，根据是否连续N2次得到该第二天线配置参数(即N2次迭代获得相同的天线配置参数)，进而确定是否将该第二天线配置参数作为第一目标配置参数。其中，N1、N2均为正整数。如果当前达到迭代次数N1，或者，当前连续N2次得到该第二天线配置参数(此时，第一天线配置参数和该第二天线配置参数相同)，则确定将该第二天线配置参数作为第一目标配置参数。
作为一种实施方式，若不将所述第二天线配置参数作为所述第一目标配置参数，根据所述第二天线配置参数、所述第二天线配置参数的评分对所述第二预测模型进行增量式更新以得到第三预测模型；根据所述第三预测模型得到第三天线配置参数。
通过根据第二天线配置参数和该第二天线配置参数的评分对该第二预测模型进行增量式更新，得到第三预测模型。采用该手段，通过不断更新该预测模型，使得更新后的预测模型可以快速收敛至评分较高的天线配置参数，进而提高了天线配置参数优化效率，可有效提升交付质量和交付效率。
作为一种实施方式，所述方法还包括：获取测量数据和仿真数据，其中，所述测量数据是将第二目标配置参数下发后得到的数据，所述仿真数据是根据所述第二目标配置参数和第一仿真模型得到的数据；根据所述测量数据和仿真数据对所述第一仿真模型进行校正以得到第二仿真模型；根据所述第一天线配置参数和所述第二仿真模型得到所述第一天线配置参数 的评分。
本申请实施例通过基于第二目标配置参数下发后得到的测量数据，以及根据该第二目标配置参数进行仿真得到的仿真数据，来对仿真模型进行校正，使得该仿真模型的误差降低，贴近现网真实实施效果，进而提升仿真的准确度。
作为一种实施方式，上述根据所述测量数据和仿真数据对所述第一仿真模型进行校正以得到第二仿真模型，包括：根据所述测量数据和仿真数据对所述第一仿真模型的链路损耗进行校正以得到第二仿真模型。
通过对仿真模型的链路损耗进行校正，进而提升仿真的准确度。
作为一种实施方式，上述根据所述测量数据和仿真数据对所述第一仿真模型的链路损耗进行校正，包括：根据所述测量数据得到多个栅格中每个栅格的第一参考信号接收功率；根据所述第一仿真模型的仿真数据得到所述多个栅格中每个栅格的第二参考信号接收功率；根据所述每个栅格的第一参考信号接收功率和第二参考信号接收功率得到所述每个栅格的卡尔曼矩阵；根据所述每个栅格的第一参考信号接收功率、第二参考信号接收功率和卡尔曼矩阵对所述第一仿真模型的链路损耗进行校正。
本申请实施例基于卡尔曼滤波的覆盖评估校正技术，对仿真模型中的误差进行校正，使得仿真评估更加准确。
作为一种实施方式，所述方法还包括：获取存储的所述第一预测模型。
通过对第一预测模型进行存储，使得在对第一预测模型进行增量式更新时，可直接获取，进而提升了天线配置参数优化的效率。
作为一种实施方式，还包括存储校正后的链路损耗。采用该手段，使得在利用仿真模型进行仿真时，可直接获取校正后的链路损耗，提升了仿真效率，进而提升了天线配置参数优化的效率。
作为一种实施方式，还包括存储第二仿真模型。通过对第二仿真模型进行存储，使得后期进行仿真时，可直接获取第二仿真模型并利用该第二仿真模型，提升了仿真效率。
作为一种实施方式，所述方法还包括：根据所述优化区域对应的多组历史天线配置参数和每组所述历史天线配置参数的评分得到初始预测模型，以根据所述初始预测模型得到所述第一预测模型。
第二方面，本申请实施例提供一种天线配置参数优化装置，包括：第一模型生成模块，用于根据优化区域对应的第一天线配置参数、所述第一天线配置参数的评分对第一预测模型进行增量式更新以得到第二预测模型；第一参数生成模块，用于根据所述第二预测模型得到第二天线配置参数；判断模块，用于确定是否将所述第二天线配置参数作为第一目标配置参数；参数确定模块，用于若将所述第二天线配置参数作为第一目标配置参数，则将所述第二天线配置参数进行下发。
通过本申请实施例，通过根据第一天线配置参数和第一天线配置参数的评分对第一预测模型进行增量式更新，进而根据更新后的预测模型得到新的天线配置参数。采用该手段，通过更新该预测模型，可以提高预测模型的预测精准度，进而提高了天线配置参数优化效率，可有效提升交付质量和交付效率。
作为一种实施方式，所述装置还包括：第二模型生成模块，用于若不将所述第二天线配置参数作为所述第一目标配置参数，根据所述第二天线配置参数、所述第二天线配置参数的 评分对所述第二预测模型进行增量式更新以得到第三预测模型；第二参数生成模块，用于根据所述第三预测模型得到第三天线配置参数。
作为一种实施方式，所述装置还包括评分确定模块，用于：获取测量数据和仿真数据，其中，所述测量数据是将第二目标配置参数下发后得到的数据，所述仿真数据是根据所述第二目标配置参数和第一仿真模型得到的数据；根据所述测量数据和仿真数据对所述第一仿真模型进行校正以得到第二仿真模型；根据所述第一天线配置参数和所述第二仿真模型得到所述第一天线配置参数的评分。
作为一种实施方式，所述装置还包括获取模块，用于：获取存储的所述第一预测模型。
作为一种实施方式，所述装置还包括第三模型生成模块，用于：根据所述优化区域对应的多组历史天线配置参数和每组所述历史天线配置参数的评分得到初始预测模型，以根据所述初始预测模型得到所述第一预测模型。
第三方面，本申请提供了一种天线配置参数优化装置，包括处理器和存储器；其中，所述存储器用于存储程序代码，所述处理器用于调用所述程序代码，以执行上述方法。
第四方面，本申请提供了一种计算机存储介质，包括计算机指令，当所述计算机指令在电子设备上运行时，使得所述电子设备执行如第一方面任一种可能的实施方式提供的方法。
第五方面，本申请实施例提供一种计算机程序产品，当计算机程序产品在计算机上运行时，使得计算机执行如第一方面任一种可能的实施方式提供的方法。
第六方面，本申请实施例提供一种芯片系统，所述芯片系统应用于电子设备；所述芯片系统包括一个或多个接口电路，以及一个或多个处理器；所述接口电路和所述处理器通过线路互联；所述接口电路用于从所述电子设备的存储器接收信号，并向所述处理器发送所述信号，所述信号包括所述存储器中存储的计算机指令；当所述处理器执行所述计算机指令时，所述电子设备执行所述方法。
可以理解地，上述提供的第二方面所述的装置、第三方面所述的装置、第四方面所述的计算机存储介质、第五方面所述的计算机程序产品或者第六方面所述的芯片系统均用于执行第一方面中任一所提供的方法。因此，其所能达到的有益效果可参考对应方法中的有益效果，此处不再赘述。
附图说明
下面对本申请实施例用到的附图进行介绍。
图1a是本申请实施例提供的一种天线配置参数优化系统的场景示意图；
图1b是本申请实施例提供的一种天线配置参数优化方法的示意图；
图1c是本申请实施例提供的一种天线配置参数优化方法的示意图；
图2是本申请实施例提供的一种天线配置参数优化方法的流程示意图；
图3是本申请实施例提供的又一种天线配置参数优化方法的流程示意图；
图4是本申请实施例提供的另一种优化方法的流程示意图；
图5是本申请实施例提供的一种天线配置参数优化装置的结构示意图；
图6是本申请实施例提供的另一种天线配置参数优化装置的结构示意图。
具体实施方式
下面结合本申请实施例中的附图对本申请实施例进行描述。本申请实施例的实施方式部分使用的术语仅用于对本申请的具体实施例进行解释，而非旨在限定本申请。
本申请实施例可应用于5G或者未来的其他网络，如6G等。
参照图1a所示，为本申请实施例提供的一种天线配置参数优化系统的场景示意图。该天线配置参数优化系统包括计算服务器101、网络管理服务器102和基站103。该网络管理服务器102具体可以是操作维护中心(Operation and Maintenance Centre，OMC)，本申请以网络管理服务器102为OMC为例进行说明，应理解，网络管理服务器102还可以是操作支撑系统(Operations Support System,OSS)等等。其中，计算服务器101向OMC发送测量数据输入请求。OMC接收到该测量数据输入请求后，控制基站103开始收集数据。基站103开始对通讯网络中的天线进行测量数据的测量控制，并将测量数据实时上报到OMC。OMC收集完成测量数据之后，可将其导入计算服务器101，实现测量数据的上报。计算服务器101基于该测量数据进行天线配置参数优化处理，获得优化后的天线配置参数后，向OMC发送配置命令。该天线配置参数又可以叫做天线射频(Radio Frequency，RF)参数。OMC将优化后的RF参数转换为人机语言(Man Machine Language，MML)指令，并对基站天线的RF参数进行调整，实现调整参数的下发。
上述计算服务器101基于该测量数据进行天线配置参数优化处理，可参照图1b所示。首先，计算服务器101对输入数据(测量数据)基于电磁波传播特性建立仿真模型，当RF参数发生变化后，可以仿真评估出变化后的参考信号接收功率(Reference Signal Received Power，RSRP)值；然后根据RSRP门限以及仿真出来的RSRP值确定优化区域中的问题小区；然后利用寻优算法对上述问题小区的RF参数进行寻优；在寻优的过程中可通过进行仿真评估出不同天线配置参数的评分；根据RF参数寻优过程中尝试的天线配置参数和对应的评分训练预测模型，根据该预测模型可得到当前预测的最优的天线配置参数，通过将该最优的天线配置参数输入到上述仿真模型可得到该天线配置参数的评分，进而根据该天线配置参数和该天线配置参数的评分增量式更新该预测模型。计算服务器101还存储更新的该预测模型。通过对该预测模型不断更新，直到该预测模型对应的算法收敛或者达到迭代次数，将当前的预测模型对应的最优的天线配置参数作为目标配置参数，进而网络管理服务器102将该目标配置参数下发。其中，关于如何使用和更新预测模型的具体步骤可以参见图1c所示，具体如下：
计算服务器101确定是否存储有预测模型；若没有存储预测模型，计算服务器101读取存储的多组历史天线配置参数和每组天线配置参数的评分，并生成预测模型，该预测模型可以被称为初始预测模型。若存储有预测模型，则直接读取该存储的预测模型。
进一步地，计算服务器101根据该预测模型得到当前预测的最优天线配置参数，并基于仿真模型可得到该天线配置参数的评分；当该预测模型对应的算法未收敛或者未达到迭代次数时，计算服务器101根据该天线配置参数和该天线配置参数的评分增量式更新该预测模型，直到该预测模型对应的算法收敛或者达到迭代次数，计算服务器101将当前的预测模型对应的最优的天线配置参数作为目标配置参数，并存储该预测模型。
作为一种实现方式，计算服务器101还通过根据仿真得到的RSRP值和测量得到的RSRP值对上述仿真模型进行校正，得到校正后的仿真模型。
进一步地，计算服务器101还存储上述仿真模型，问题小区和上述RF参数寻优过程中 尝试的天线配置参数和对应的评分。
上述系统以计算服务器101进行天线配置参数优化、网络管理服务器102进行天线配置参数下发为例进行说明。作为另一种实现方式，上述计算服务器101和网络管理服务器102可集成为一体，进而直接进行天线配置参数优化以及下发。
应理解，图1a、图1b和图1c中涉及的天线配置参数优化方法具体内容在下文中将进行进一步的阐述。
下面对本申请实施例提供的天线配置参数优化方法的实现过程进行介绍。参照图2所示，为本申请实施例提供的一种天线配置参数优化方法的流程示意图。其中，本申请实施例以计算服务器执行该天线配置参数优化方法为例进行说明。该天线配置参数优化方法包括步骤201-204，具体如下：
201、根据优化区域对应的第一天线配置参数、所述第一天线配置参数的评分对第一预测模型进行增量式更新以得到第二预测模型；
其中，上述优化区域可以是定期优化的基站控制的多个通信小区对应的区域。
上述第一天线配置参数可以是该计算服务器在确定目标配置参数的过程中所获取的任一天线配置参数。该第一天线配置参数可包括以下一项或多项：物理方位角、物理下倾角、数字方位角、数字下倾角、水平波束宽度或垂直波束宽度等。
上述第一天线配置参数的评分用于表征该第一天线配置参数对应的网络质量(具体如信号质量)的优良程度。应理解，该评分具体也可以叫做标签等，本方案对此不做具体限定。
上述根据优化区域对应的第一天线配置参数、所述第一天线配置参数的评分对第一预测模型进行增量式更新以得到第二预测模型，可以理解为：根据优化区域对应的第一天线配置参数、所述第一天线配置参数的评分对第一预测模型的计算公式、参数或属性等进行更新，从而得到第二预测模型。而全量式更新则是根据第一天线配置参数、所述第一天线配置参数的评分以及生成第一预测模型之前所使用到的天线配置参数和评分来生成第二预测模型。也就是说，全量式更新未基于已经生成的第一预测模型来生成第二预测模型，其并未建立第一预测模型和第二预测模型之间的关联。增量式更新相较于全量式更新，其基于第一天线配置参数和第一天线配置参数的评分来对该已生成的第一预测模型进行更新进而得到新的预测模型。
具体地，根据天线配置参数集合X N和对应的天线配置参数的评分集合F N生成上述第一预测模型。上述天线配置参数集合X N＝{x i,i∈N}，表示多组天线配置参数x i的集合；评分集合F N＝{f(x i),i∈N}，表示该多组天线配置参数中每组参数对应的评分f(x i)的集合。下面对上述第一预测模型的增量式更新进行介绍。
作为一种实现方式，以第一预测模型为概率预测模型为例进行说明。
首先，定义任意两个天线配置参数x i,x j之间的协方差k(x i,x j)满足以下条件：
Figure PCTCN2021116534-appb-000001
其中，天线配置参数x i,x j均属于天线配置参数集合X N，α，θ均为非零系数。
然后，最大化在这两个超参α，θ下评分F N出现的概率，即通过最大化边缘对数似然来计算超参α和θ，边缘对数似然可表示为：
Figure PCTCN2021116534-appb-000002
其中，p(F N|α,θ)为在对应超参α和θ的情况下F N出现的概率。
根据上述最大化边缘对数似然计算得到超参α，θ，进而根据上述k(x i,x j)计算得到协方差矩阵K可表示为：
Figure PCTCN2021116534-appb-000003
增益矩阵F N可表示为：
Figure PCTCN2021116534-appb-000004
其中，第一预测模型的参数包括协方差矩阵K和增益矩阵F N。
根据第一天线配置参数x best和该第一天线配置参数的评分f(x best)更新上述第一预测模型。其中，该第一天线配置参数x best是根据该第一预测模型得到的，具体可参阅图4所述实施例中的描述，在此不再赘述。
具体地，上述第一预测模型更新后的协方差矩阵K可表示为：
Figure PCTCN2021116534-appb-000005
新的增益矩阵F N可表示为：F N＝F N+[f(x best)]   (4)
通过在原协方差矩阵如公式(1)所示的基础上新增一行一列得到协方差矩阵如公式(3)所示，在原增益矩阵F N如公式(2)所示的基础上新增一行得到增益矩阵F N如公式(4)所示。基于上述方式，可以实现上述第一预测模型的增量式更新。上述仅为一种示例，其还可以是其他形式的更新，本方案对此不做具体限定。
作为另一种实现方式，以第一预测模型为神经网络预测模型为例进行说明。根据天线配置参数集合X N和对应的天线配置参数的评分集合F N生成样本对。样本对以天线配置参数为输入数据，以天线配置参数的评分为预期输出，通过进行学习，进而训练一个神经网络。
神经网络满足如下条件：
W,B＝argmin W,B{WX N+B-F N}    (5)
相应地，该第一预测模型的参数为权重矩阵W和偏差矩阵B。
根据第一天线配置参数x best和该第一天线配置参数的评分f(x best)更新上述第一预测模型。该第一天线配置参数x best是根据该第一预测模型得到的，具体可参阅图4所述实施例中的描述，在此不再赘述。
上述第一预测模型更新后的样本对可表示为：
X N＝X N+[x best]
F N＝F N+[f(x best)]
上述第一预测模型更新后的权重矩阵W和偏差矩阵B可表示为：
W＝W
B＝argmin B{WX N+B-F N}   (6)
需要说明的是，对于第一预测模型来说，只更新偏差矩阵，不更新权重矩阵。基于上述方式，可以实现上述神经网络预测模型的增量式更新。
上述以两种不同的模型为例对模型的增量式更新进行说明。其中，其还可以是其他形式模型的增量式更新，本方案对此不做具体限定。
上述第一预测模型可以是通过以下方式得到的：
根据该优化区域对应的多组历史天线配置参数和每组该历史天线配置参数的评分得到初始预测模型，以根据该初始预测模型得到该第一预测模型。
作为一种实施方式，该初始预测模型可以是该第一预测模型。计算服务器通过将多组历史天线配置参数分别进行仿真得到各组历史天线配置参数的评分，通过根据该多组历史天线配置参数和每组该历史天线配置参数的评分得到该第一预测模型。其中，根据该第一预测模型得到上述多组历史天线配置参数中评分最高的历史天线配置参数，该评分最高的历史天线配置参数即为上述第一天线配置参数。应理解，上述多组历史天线配置参数可以是从预设天线配置参数集合中获取的。上述多组历史天线配置参数也可以是随机产生的。
作为另一种实施方式，该初始预测模型还可以是该优化区域对应的在时间顺序上的第一个预测模型。计算服务器对该初始预测模型进行增量式更新，通过不断迭代更新进而得到该第一预测模型。具体地，计算服务器通过将多组历史天线配置参数分别进行仿真得到各组历史天线配置参数的评分，通过根据该多组历史天线配置参数和每组该历史天线配置参数的评分得到该初始预测模型。通过根据该初始预测模型不断迭代更新得到该第一预测模型。其中，根据该初始预测模型不断迭代更新得到该第一预测模型的具体实现方式，可参阅上述实施例中第一预测模型迭代更新的描述，其更新的原理相同，在此不再赘述。
可选的，在步骤201之前，该方法还包括步骤200、获取存储的所述第一预测模型。通过对第一预测模型进行存储，使得在对第一预测模型进行增量式更新时，可直接获取，进而提升了天线配置参数优化的效率。
相应地，在步骤201之后，该方法还包括步骤2011、存储所述第二预测模型，以便后续对该第二预测模型进行增量式更新。其中，该步骤2011可以在步骤201之后、步骤202之前，其也可以在步骤202之后，本方案对此不做具体限定。
也就是说，计算服务器在得到任一预测模型时，还将该预测模型进行存储，以便后续进行增量式更新时使用。
可选的，计算服务器可通过基于测量数据和仿真数据对仿真模型进行校正进而得到上述步骤201中第一天线配置参数的评分。具体地，在步骤201之前，该方法还包括步骤3001-3003，可参阅图3所示，为本申请实施例提供的一种天线配置参数优化方法的流程示意图，具体如下：
3001、获取测量数据和仿真数据，其中，所述测量数据是将第二目标配置参数下发后得 到的数据，所述仿真数据是根据所述第二目标配置参数和第一仿真模型得到的数据；
上述测量数据包括以下一项或多项：电子地图、天线文件、代表当前射频参数的工参数据和测量到的MR数据、或DT数据等。其中，电子地图为利用计算机技术以数字方式存储和查阅的地图。天线文件为天线波瓣图文件。MR数据为用户设备(User Equipment，UE)上报的测量报告，包含主服小区标识、主服小区RSRP、邻区标识、邻区RSRP，以及主服和邻区各小区级广播波束的波束ID及对应RSRP。DT数据为道路测试的数据，其与最小化道路测试(Minimization of Drive-Test，MDT)数据类似。其中，MDT数据为是UE上报的带有经纬度信息的测量报告，可认为带有经纬度的MR数据。
需要说明的是，该第二目标配置参数相对于当前次寻找目标参数(即下文步骤204中的第一目标配置参数)是计算服务器上一次获得的目标配置参数。具体地，计算服务器通过网络管理服务器将第二目标配置参数进行下发后，通过网络管理服务器获取到上述测量数据。计算服务器还通过将第二目标配置参数输入至第一仿真模型得到上述仿真数据。
3002、根据所述测量数据和仿真数据对所述第一仿真模型进行校正以得到第二仿真模型；
作为一种实施方式，根据所述测量数据和仿真数据对所述第一仿真模型的链路损耗进行校正以得到第二仿真模型。
具体地，根据所述测量数据得到多个栅格中每个栅格的第一参考信号接收功率；根据所述第一仿真模型的仿真数据得到所述多个栅格中每个栅格的第二参考信号接收功率；根据所述每个栅格的第一参考信号接收功率和第二参考信号接收功率得到所述每个栅格的卡尔曼矩阵；根据所述每个栅格的第一参考信号接收功率、第二参考信号接收功率和卡尔曼矩阵对所述第一仿真模型的链路损耗进行校正，得到校正后的链路损耗，进而得到该第二仿真模型。
也就是说，通过对仿真模型的链路损耗进行校正，进而提升仿真的准确度。
具体地，当获取到第二目标配置参数下发后得到的测量数据时，计算服务器通过依据测量数据中每一条MR/DT数据中的经纬度信息，将MR/DT数据进行栅格化处理进而归属到多个栅格中。
针对每一个栅格，计算其最新测量的参考信号接收功率RSRP。具体地，通过去除测量的部分边缘数据，并对剩下的测量数据取平均。例如，针对每一个栅格，获得新测量的多时刻的多条测量信息MR＝{mr i},i∈N′；每个mr i信息包含测量获得的rsrp i，将多时刻的多条mr i汇总，按照每个mr i的rsrp i进行从大到小的排序，去除部分mr i，例如去除位于后面5％的mr i，进而形成新的MR new，该每个栅格的RSRP测量值可表示为：
Figure PCTCN2021116534-appb-000006
针对每一个栅格，计算其测量方差R：
Figure PCTCN2021116534-appb-000007
同时，针对每一个栅格，根据第二目标配置参数进行仿真进而得到上述每一个栅格的参考信号接收功率的仿真值RSRP’；
针对每一个栅格，获得其仿真方差。其中，通过获得该栅格周围如8个栅格的仿真值进而得到该栅格的仿真方差P -。
该栅格周围8个栅格的仿真值可表示为RSRP’ i,i∈[1-8]；
计算该栅格以及其周围8个栅格的仿真值的均值：
Figure PCTCN2021116534-appb-000008
计算该栅格的仿真方差：
Figure PCTCN2021116534-appb-000009
本方案仅以8个栅格为例进行说明，其还可以是其他设定数量，本方案对此不做具体限定。
针对每一个栅格，计算其卡尔曼矩阵：
Figure PCTCN2021116534-appb-000010
针对每一个栅格，根据其卡尔曼矩阵计算其校正后的参考信号接收功率RSRP”：
RSRP”＝RSRP’+Kal(RSRP-RSRP’)
根据该校正后的RSRP”，计算每个栅格校正后的链路损耗
Figure PCTCN2021116534-appb-000011
Figure PCTCN2021116534-appb-000012
该实施例通过根据仿真RSRP值和现网新反馈的测量RSRP值，采用卡尔曼滤波技术(多数据源融合)，综合考虑估计值和测量值的可信度，加权得出校正后的每个栅格的Linkloss，进而降低了仿真误差。通过将仿真RSRP值和最新的测量RSRP相融合，提升了仿真精度。
作为一种实施方式，针对每一个栅格，更新计算其仿真方差P -′＝(I-Kal)P -，I为单位矩阵。其中，计算服务器存储该仿真方差P -′，以便在下次进行仿真模型校正时进行使用。
基于上述校正后的链路损耗进而得到第二仿真模型。
3003、根据所述第一天线配置参数和所述第二仿真模型得到所述第一天线配置参数的评分。
通过将第一天线配置参数输入至上述校正后的仿真模型中计算得到每个栅格校正后的参考信号接收功率，进而得到该第一天线配置参数的评分。作为一种实施方式，本方案通过统计出整片优化区域中栅格的RSRP值大于RSRP门限的栅格占总栅格的比例，记为第一天线配置参数的评分。
本申请实施例，通过根据将第二目标配置参数下发得到的测量数据和将第二目标配置参数进行仿真得到的仿真数据来对仿真模型进行校正，进而得到仿真精度更高的仿真模型，以便在进行目标配置参数的寻优过程中提高天线配置参数的评分的准确度和精度，进一步提升天线配置参数优化效率。
202、根据所述第二预测模型得到第二天线配置参数；
具体地，该计算服务器通过将多个任意天线配置参数输入至该第二预测模型进而得到该第二天线配置参数。其中，该第二天线配置参数为该第二预测模型得到的较优的配置参数。例如，该第二天线配置参数为上述多个任意天线配置参数中评分最高的配置参数。
203、确定是否将所述第二天线配置参数作为第一目标配置参数；
具体地，该计算服务器通过确定是否达到预设条件来确定是否将所述第二天线配置参数作为所述第一目标配置参数。
例如根据是否达到迭代次数N1，或者，根据是否连续N2次得到该第二天线配置参数(即N2次迭代获得相同的天线配置参数)，进而确定是否将该第二天线配置参数作为第一目标配置参数。其中，N1、N2均为正整数。如果当前达到迭代次数N1，或者，当前连续N2次得到该第二天线配置参数(此时，第一天线配置参数和该第二天线配置参数相同)，则该计算服务器确定将该第二天线配置参数作为第一目标配置参数。
204、若将所述第二天线配置参数作为第一目标配置参数，则将所述第二天线配置参数进行下发。
当该计算服务器将该第二天线配置参数作为第一目标配置参数后，该计算服务器通过网络管理服务器将该第一目标配置参数(即第二天线配置参数)进行下发。即该计算服务器将该第一目标配置参数发送给网络管理服务器，以便该网络管理服务器将该第一目标配置参数进行下发。
作为一种实施方式，若不将所述第二天线配置参数作为第一目标配置参数，该计算服务器根据该第二天线配置参数、该第二天线配置参数的评分对该第二预测模型进行增量式更新以得到第三预测模型；并根据该第三预测模型得到第三天线配置参数。其中，不将第二天线配置参数作为第一目标配置参数，即未达到预设条件，例如当前未达到迭代次数N1，或者，当前未达到连续N2次得到该第二天线配置参数。
相应地，该计算服务器确定是否将该第三天线配置参数作为第一目标配置参数；若将该第三天线配置参数作为第一目标配置参数，则将该第三天线配置参数通过网络管理服务器进行下发。
该计算服务器通过重复执行上述步骤，直到得到第一目标配置参数，并通过网络管理服务器进行下发。
需要说明的是，本申请实施例以计算服务器为例进行说明。其中，网络管理服务器和计算服务器还可以集成为一体。相应地，若将上述第二天线配置参数作为第一目标配置参数，则计算服务器将该第二天线配置参数直接进行下发。
通过本申请实施例，通过根据第一天线配置参数和第一天线配置参数的评分对第一预测模型进行增量式更新，进而根据更新后的预测模型得到新的天线配置参数。采用该手段，通过增量式更新该预测模型，基于更新后的预测模型可以快速收敛至评分较高的天线配置参数，进而提高了天线配置参数优化效率，相较于现有全量式更新的手段，本方案可有效提升交付质量和交付效率。
下面对本申请实施例提供的天线配置参数优化方法的具体实现过程进行详细介绍。参照图4所示，为本申请实施例提供的一种天线配置参数优化方法的流程示意图。其包括步骤401-406，具体如下：
401、获取优化区域对应的第一测量数据；
其中，计算服务器可通过网络管理服务器获取优化区域对应的第一测量数据。网络管理服务器通过控制基站收集该第一测量数据。
该第一测量数据是该优化区域对应的初始测量数据，也就是说，该第一测量数据是该优化区域还未开始优化时对应的测量数据。
402、根据所述第一测量数据确定第三目标配置参数，并将所述第三目标配置参数通过网络管理服务器进行下发；
计算服务器根据第一测量数据确定需要进行天线配置参数优化的小区，进而确定第三目标配置参数。相应地，当前述实施例中的第一预测模型是初始预测模型时，该第三目标配置参数即为前述实施例中的第二目标配置参数。
其中，步骤402可包括步骤4021-4023，具体如下：
4021、根据所述第一测量数据构建栅格RSRP评估的仿真模型；
计算服务器利用上述第一测量数据，根据电磁波传播特性建立仿真模型。其中，该仿真模型包括链路损耗Linkloss。
其中，电磁波传播特性满足如下公式：
参考信号接收功率RSRP＝天线增益AntennaGain+功率Power-链路损耗Linkloss(7)；
上述链路损耗Linkloss为对上述第一测量数据进行栅格处理后得到的多个栅格中每个栅格的链路损耗Linkloss。
具体地，根据第一测量数据中每条MR/DT数据中的经纬度信息，将MR/DT数据进行栅格化处理得到多个栅格，进而得到该多个栅格中每个栅格的RSRP。根据第一测量数据中天线文件和工参数据，得到每个栅格的天线增益AntennaGain和功率Power。进而根据公式(7)，可得到每个栅格的链路损耗Linkloss。计算服务器存储该每个栅格的链路损耗Linkloss，以便后续使用该仿真模型。
4022、根据所述第一测量数据识别出需要调整的小区(问题小区)；
其中，计算服务器根据MR/DT数据，得到每个栅格的RSRP。通过基于每个栅格对应的参考信号接收功率RSRP，将参考信号接收功率小于预设门限的栅格，定义为弱覆盖栅格；对弱覆盖栅格进行聚类，获得弱覆盖区域；将小区的经纬度属于弱覆盖区域中的小区定义为问题小区。
4023、获取与所述问题小区对应的第三目标配置参数，并将所述第三目标配置参数通过网络管理服务器进行下发。
具体地，计算服务器随机生成一组天线配置参数x i,x i∈X all，X all表示全局天线配置参数，即所述优化区域对应的问题小区的可选天线配置参数的全集，也就是说所有可选的参数组成的集合。
针对天线配置参数x i，根据天线文件和工参数据，获得每个栅格对应的AntennaGain和Power。
根据天线配置参数x i以及上述仿真模型可得到该天线配置参数x i对应的每个栅格的RSRP值。
统计出整片优化区域中栅格的RSRP值大于RSRP门限的栅格占总栅格的比例，记为评分f(x i)。其中，计算服务器存储天线配置参数参数x i以及天线配置参数参数x i对应的评分。
针对多组不同的天线配置参数x i，重复执行上述步骤以获取各组天线配置参数分别对应的评分，直到满足尝试次数N3，进而输出尝试过程中评分最高的天线配置参数x *以及其评估得分f(x *)。
该评分最高的天线配置参数x *即为第三目标配置参数。该第三目标配置参数中包括上述问题小区中每个天线每个RF参数的配置。计算服务器将该第三目标配置参数通过网络管理服务器下发到现网，以调整现实通讯网络中天线的RF参数。
403、获取所述优化区域对应的第二测量数据，其中，所述第二测量数据是将所述第三目标配置参数下发后得到的数据；
计算服务器通过网络管理服务器获取到下发第三目标配置参数后对应的第二测量数据，即新的测量数据。其中，计算服务器可在间隔预设时间后向网络管理服务器发送获取下发上述第三目标配置参数后的第二测量数据。该预设时间可以是几个小时或者几天等，本方案对 此不做具体限定。当该第三目标配置参数是前述实施例中的第二目标配置参数时，相应地，该第二测量数据可以是前述实施例中的测量数据。
该第二测量数据可以仅包括MR/DT数据。其也可以包括电子地图、天线文件、代表当前射频参数的工参数据、测量到的MR数据和/或DT数据等，本方案对此不做具体限定。
404、根据所述第二测量数据确定是否需要继续优化；
其中，所述计算服务器根据新的测量数据，依据每一条MR/DT数据中的经纬度信息，将MR/DT数据栅格化处理并归属到步骤4021生成的多个栅格中，进而获得该多个栅格中每一个栅格的RSRP，再统计出整片优化区域中栅格的RSRP值大于RSRP门限的栅格占总栅格的比例。如果该比例满足预设要求，则进入步骤406结束优化；否则，若整片优化区域中栅格的RSRP值大于RSRP门限的栅格占总栅格的比例小于预设要求，则进入步骤405。
405、若需要继续优化，根据多组历史配置参数和每组历史配置参数的评分得到第一目标配置参数，并将所述第一目标配置参数进行下发。
作为一种实现方式，计算服务器根据第二预测模型得到第一目标配置参数；其中，计算服务器根据第一天线配置参数、所述第一天线配置参数的评分对第一预测模型进行增量式更新得到该第二预测模型。
具体地，1)、计算服务器根据第一次优化时得到上述第三目标配置参数的过程中尝试的多组历史天线配置参数和该多组历史天线配置参数中每组历史天线配置参数的评分得到初始预测模型，并存储该初始预测模型。具体过程可参阅前述实施例中步骤201中得到初始预测模型的描述，在此不再赘述。
然后计算服务器通过读取存储的该初始预测模型并不断迭代更新得到第一预测模型，并存储该第一预测模型。具体过程可参阅前述实施例中步骤201中得到第一预测模型的描述，在此不再赘述。
2)、计算服务器在后续更新时，读取存储的该第一预测模型，根据该第一预测模型得到第一天线配置参数，通过将第一天线配置参数进行仿真得到第一天线配置参数的评分，进而根据第一天线配置参数、所述第一天线配置参数的评分对第一预测模型进行增量式更新得到该第二预测模型。具体过程可参阅前述实施例中步骤201中对第一预测模型进行增量式更新得到第二预测模型的描述，在此不再赘述。
对于根据该第一预测模型得到第一天线配置参数，可以有多种实现方式。本申请实施例以两种实现方式为例进行说明。
作为第一种实现方式，以前述实施例中第一预测模型为概率预测模型为例进行说明。
针对任意一组天线配置参数x k，通过将该天线配置参数x k输入至该第一预测模型，计算其评分的预测值μ(x k)和该预测值的方差σ(x k)：
k＝[k(x k,x 1),k(x k,x 2)…,k(x k,x N)]
μ(x k)＝kK -1F N
σ 2(x k)＝k(x k,x k)-kK -1k T
根据该第一预测模型，针对全局所有天线配置参数中的每一个x k,x k∈X all，基于上述公式可计算各天线配置参数分别对应的评分的预测值μ(x k)和该预测值的方差σ(x k)。其中，也可以仅对全局所有天线配置参数中的部分天线配置参数进行采样计算，本方案对此不做具体限定。
针对全局所有天线配置参数中的每一个x k，根据累计概率密度公式，通过上述预测出来的μ(x k)和σ(x k)，可以计算出该天线配置参数对应的增益：
Figure PCTCN2021116534-appb-000013
该f(x *)是上一轮参数优化方案下发到现网的目标配置参数x *(即第三目标配置参数)所对应的评分。该增益EI(x k)的含义可理解为：天线配置参数x k的评分f(x k)比之前下发到现网的天线配置参数x *的评分f(x *)还要好的期望。
基于上述获得的各天线配置参数的EI(x k)，确定对应EI最大的天线配置参数x best，即上述第一天线配置参数x best：
Figure PCTCN2021116534-appb-000014
通过将x best输入至仿真模型得到其对应的评分f(x best)。
通过判断是否达到总的迭代次数N1，或者是否连续N2次得到该x best，来确定是否停止上述算法。若达到迭代次数N1，或者，当前连续N2次得到该x best，则输出该天线配置参数x best以及其评分f(x best)。上述N1例如可以是100等，上述N2例如可以是10等。本方案对此不做具体限定。
作为第二种实现方式，以前述实施例中第一预测模型为神经网络预测模型为例进行说明。
针对任意一组天线配置参数x k，通过将该天线配置参数x k输入至该第一预测模型，计算其评分的预测值μ(x k)和该预测值的方差σ(x k)：
μ(x k)＝Wx k+B
σ 2(x k)＝G
其中，G为非零常数。
根据该第一预测模型，针对全局所有天线配置参数中的每一个x k,x k∈X all，基于上述公式可计算各天线配置参数分别对应的评分的预测值μ(x k)和该预测值的方差σ(x k)。其中，也可以仅对部分天线配置参数进行采样计算，本方案对此不做具体限定。
针对全局所有天线配置参数中的每一个x k，根据累计概率密度公式，通过上述预测出来的μ(x k)和σ(x k)，可以计算出该天线配置参数对应的增益：
Figure PCTCN2021116534-appb-000015
其中，f(x *)是上一轮参数优化方案下发到现网的目标配置参数x *(即第三目标配置参数)所对应的评分。
基于上述获得的各天线配置参数的EI(x k)，确定对应EI最大的天线配置参数x best，即上述第一天线配置参数x best：
Figure PCTCN2021116534-appb-000016
通过将x best输入至仿真模型得到其对应的评分f(x best)。
通过判断是否达到总的迭代次数N1，或者是否连续N2次得到该x best，来确定是否停止上述算法。若达到迭代次数N1，或者，当前连续N2次得到该x best，则输出该天线配置参数x best以及其评分f(x best)。上述N1例如可以是100等，上述N2例如可以是10等。本方案对此不做具体限定。
若不满足上述条件，则根据该x best和评分f(x best)更新上述第一预测模型进而得到上述第二预测模型。该增量式更新可参阅前述实施例中步骤201的描述，在此不再赘述。
406、若不需要继续优化，结束该优化。
其中，上述仅以两次下发目标配置参数为例进行说明。其还可以是其他任意多次的迭代，例如，在上述步骤405将上述第一目标配置参数下发后，可重复执行步骤403-406，直到该优化区域对应的测量数据满足预设要求，进而停止优化。
通过本申请实施例，通过根据天线配置参数和该天线配置参数的评分对预测模型进行增量式更新，进而根据更新后的预测模型得到新的天线配置参数。采用该手段，通过不断增量式更新该预测模型，基于更新后的预测模型可以快速收敛至评分较高的天线配置参数，进而提高了天线配置参数优化效率，可有效提升交付质量和交付效率。
且，本方案提供的天线配置参数优化方法，当获取到新的测量数据时，基于上述存储的仿真模型，不需要再重新建立仿真模型，相较于现有技术，有效提升了整体优化效率。
参照图5所示，本申请实施例提供一种天线配置参数优化装置。该装置包括第一模型生成模块501、第一参数生成模块502、判断模块503和参数确定模块504，具体如下：
第一模型生成模块501，用于根据优化区域对应的第一天线配置参数、所述第一天线配置参数的评分对第一预测模型进行增量式更新以得到第二预测模型；
第一参数生成模块502，用于根据所述第二预测模型得到第二天线配置参数；
判断模块503，用于确定是否将所述第二天线配置参数作为第一目标配置参数；
参数确定模块504，用于若将所述第二天线配置参数作为第一目标配置参数，则将所述第二天线配置参数进行下发。
通过本申请实施例，通过根据第一天线配置参数和第一天线配置参数的评分对第一预测模型进行增量式更新，进而根据更新后的预测模型得到新的天线配置参数。采用该手段，通过更新该预测模型，可以提高预测模型的预测精准度，进而提高了天线配置参数优化效率，可有效提升交付质量和交付效率。
其中，所述装置还包括：第二模型生成模块，用于若所述第二天线配置参数不是第一目标配置参数，根据所述第二天线配置参数、所述第二天线配置参数的评分和所述第二预测模型得到第三预测模型；第二参数生成模块，用于根据所述第三预测模型得到第三天线配置参数。
其中，所述第一天线配置参数的评分是根据所述第一天线配置参数和第一仿真模型得到的，所述装置还包括仿真模型生成模块，用于：获取测量数据和仿真数据，其中，所述测量数据是将第二目标配置参数下发后得到的数据，所述仿真数据是根据所述第二目标配置参数和第二仿真模型得到的数据；根据所述测量数据和仿真数据对所述第二仿真模型进行校正以得到所述第一仿真模型。
其中，所述装置还包括获取模块，用于：获取存储的所述第一预测模型。
其中，所述装置还包括第三模型生成模块，用于：根据所述优化区域对应的多组历史天线配置参数和每组所述历史天线配置参数的评分得到初始预测模型，以根据所述初始预测模型得到所述第一预测模型。
上述各模块的具体实现方式可参阅前述实施例中的相应描述，在此不再赘述。
本申请还提供了一种天线配置参数优化装置，如图6所示，该天线配置参数优化装置包括至少一个处理器601，至少一个存储器602以及至少一个通信接口603。所述处理器601、所述存储器602和所述通信接口603通过所述通信总线连接并完成相互间的通信。
处理器601可以是通用中央处理器(CPU)，微处理器，特定应用集成电路(application-specific integrated circuit，ASIC)，或一个或多个用于控制以上方案程序执行的集成电路。
通信接口603，用于与其他设备或通信网络通信，如以太网，无线接入网(RAN)，无线局域网(Wireless Local Area Networks，WLAN)等。
存储器602可以是只读存储器(read-only memory，ROM)或可存储静态信息和指令的其他类型的静态存储设备，随机存取存储器(random access memory，RAM)或者可存储信息和指令的其他类型的动态存储设备，也可以是电可擦可编程只读存储器(Electrically Erasable Programmable Read-Only Memory，EEPROM)、只读光盘(Compact Disc Read-Only Memory，CD-ROM)或其他光盘存储、光碟存储(包括压缩光碟、激光碟、光碟、数字通用光碟、蓝光光碟等)、磁盘存储介质或者其他磁存储设备、或者能够用于携带或存储具有指令或数据结构形式的期望的程序代码并能够由计算机存取的任何其他介质，但不限于此。存储器可以是独立存在，通过总线与处理器相连接。存储器也可以和处理器集成在一起。
其中，所述存储器602用于存储执行以上方案的应用程序代码，并由处理器601来控制执行。所述处理器601用于执行所述存储器602中存储的应用程序代码。
存储器602存储的代码可执行以上提供的一种天线配置参数优化方法。
本申请实施例还提供一种芯片系统，所述芯片系统应用于电子设备；所述芯片系统包括一个或多个接口电路，以及一个或多个处理器；所述接口电路和所述处理器通过线路互联；所述接口电路用于从所述电子设备的存储器接收信号，并向所述处理器发送所述信号，所述信号包括所述存储器中存储的计算机指令；当所述处理器执行所述计算机指令时，所述电子设备执行所述方法。
本申请实施例还提供了一种计算机可读存储介质，该计算机可读存储介质中存储有指令，当其在计算机或处理器上运行时，使得计算机或处理器执行上述任一个方法中的一个或多个步骤。
本申请实施例还提供了一种包含指令的计算机程序产品。当该计算机程序产品在计算机或处理器上运行时，使得计算机或处理器执行上述任一个方法中的一个或多个步骤。
在上述实施例中，可以全部或部分地通过软件、硬件、固件或者其任意组合来实现。当使用软件实现时，可以全部或部分地以计算机程序产品的形式实现。所述计算机程序产品包括一个或多个计算机指令。在计算机上加载和执行所述计算机程序指令时，全部或部分地产生按照本申请实施例所述的流程或功能。所述计算机可以是通用计算机、专用计算机、计算机网络、或者其他可编程装置。所述计算机指令可以存储在计算机可读存储介质中，或者通 过所述计算机可读存储介质进行传输。所述计算机指令可以从一个网站站点、计算机、服务器或数据中心通过有线(例如同轴电缆、光纤、数字用户线)或无线(例如红外、无线、微波等)方式向另一个网站站点、计算机、服务器或数据中心进行传输。所述计算机可读存储介质可以是计算机能够存取的任何可用介质或者是包含一个或多个可用介质集成的服务器、数据中心等数据存储设备。所述可用介质可以是磁性介质，(例如，软盘、硬盘、磁带)、光介质(例如，DVD)、或者半导体介质(例如，固态硬盘(solid state disk，SSD))等。
本申请中，“至少一个”是指一个或者多个，“多个”是指两个或两个以上。“和/或”，描述关联对象的关联关系，表示可以存在三种关系，例如，A和/或B，可以表示：单独存在A，同时存在A和B，单独存在B的情况，其中A，B可以是单数或者复数。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，该流程可以由计算机程序来指令相关的硬件完成，该程序可存储于计算机可读取存储介质中，该程序在执行时，可包括如上述各方法实施例的流程。而前述的存储介质包括：ROM或随机存储记忆体RAM、磁碟或者光盘等各种可存储程序代码的介质。
以上所述，仅为本申请实施例的具体实施方式，但本申请实施例的保护范围并不局限于此，任何在本申请实施例揭露的技术范围内的变化或替换，都应涵盖在本申请实施例的保护范围之内。因此，本申请实施例的保护范围应以所述权利要求的保护范围为准。

Claims (13)

  1. 一种天线配置参数优化方法，其特征在于，包括：
    根据优化区域对应的第一天线配置参数、所述第一天线配置参数的评分对第一预测模型进行增量式更新以得到第二预测模型；
    根据所述第二预测模型得到第二天线配置参数；
    确定是否将所述第二天线配置参数作为第一目标配置参数；
    若将所述第二天线配置参数作为第一目标配置参数，则将所述第二天线配置参数进行下发。
  2. 根据权利要求1所述的方法，其特征在于，所述方法还包括：
    若不将所述第二天线配置参数作为所述第一目标配置参数，根据所述第二天线配置参数、所述第二天线配置参数的评分对所述第二预测模型进行增量式更新以得到第三预测模型；
    根据所述第三预测模型得到第三天线配置参数。
  3. 根据权利要求1或2所述的方法，其特征在于，所述方法还包括：
    获取测量数据和仿真数据，其中，所述测量数据是将第二目标配置参数下发后得到的数据，所述仿真数据是根据所述第二目标配置参数和第一仿真模型得到的数据；
    根据所述测量数据和仿真数据对所述第一仿真模型进行校正以得到第二仿真模型；
    根据所述第一天线配置参数和所述第二仿真模型得到所述第一天线配置参数的评分。
  4. 根据权利要求1至3任一项所述的方法，其特征在于，所述方法还包括：
    获取存储的所述第一预测模型。
  5. 根据权利要求1至4任一项所述的方法，其特征在于，所述方法还包括：
    根据所述优化区域对应的多组历史天线配置参数和每组所述历史天线配置参数的评分得到初始预测模型，以根据所述初始预测模型得到所述第一预测模型。
  6. 一种天线配置参数优化装置，其特征在于，包括：
    第一模型生成模块，用于根据优化区域对应的第一天线配置参数、所述第一天线配置参数的评分对第一预测模型进行增量式更新以得到第二预测模型；
    第一参数生成模块，用于根据所述第二预测模型得到第二天线配置参数；
    判断模块，用于确定是否将所述第二天线配置参数作为第一目标配置参数；
    参数确定模块，用于若将所述第二天线配置参数作为第一目标配置参数，则将所述第二天线配置参数进行下发。
  7. 根据权利要求6所述的装置，其特征在于，所述装置还包括：
    第二模型生成模块，用于若不将所述第二天线配置参数作为所述第一目标配置参数，根据所述第二天线配置参数、所述第二天线配置参数的评分对所述第二预测模型进行增量式更新以得到第三预测模型；
    第二参数生成模块，用于根据所述第三预测模型得到第三天线配置参数。
  8. 根据权利要求6或7所述的装置，其特征在于，所述装置还包括评分确定模块，用于：
    获取测量数据和仿真数据，其中，所述测量数据是将第二目标配置参数下发后得到的数据，所述仿真数据是根据所述第二目标配置参数和第一仿真模型得到的数据；根据所述测量数据和仿真数据对所述第一仿真模型进行校正以得到第二仿真模型；根据所述第一天线配置参数和所述第二仿真模型得到所述第一天线配置参数的评分。
  9. 根据权利要求6至8任一项所述的装置，其特征在于，所述装置还包括获取模块，用于：
    获取存储的所述第一预测模型。
  10. 根据权利要求6至9任一项所述的装置，其特征在于，所述装置还包括第三模型生成模块，用于：
    根据所述优化区域对应的多组历史天线配置参数和每组所述历史天线配置参数的评分得到初始预测模型，以根据所述初始预测模型得到所述第一预测模型。
  11. 一种天线配置参数优化装置，其特征在于，包括处理器和存储器；其中，所述存储器用于存储程序代码，所述处理器用于调用所述程序代码，以执行如权利要求1至5任一项所述的方法。
  12. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质存储有计算机程序，所述计算机程序被处理器执行以实现权利要求1至5任意一项所述的方法。
  13. 一种计算机程序产品，其特征在于，当计算机程序产品在计算机上运行时，使得所述计算机执行如权利要求1至5任意一项所述的方法。
PCT/CN2021/116534 2021-01-19 2021-09-03 天线配置参数优化方法、装置及存储介质 WO2022156226A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
EP21920605.9A EP4280660A4 (en) 2021-01-19 2021-09-03 METHOD AND DEVICE FOR OPTIMIZING ANTENNA CONFIGURATION PARAMETERS AND STORAGE MEDIUM

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202110068483.9A CN114818452A (zh) 2021-01-19 2021-01-19 天线配置参数优化方法、装置及存储介质
CN202110068483.9 2021-01-19

Publications (1)

Publication Number Publication Date
WO2022156226A1 true WO2022156226A1 (zh) 2022-07-28

Family

ID=82525059

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/116534 WO2022156226A1 (zh) 2021-01-19 2021-09-03 天线配置参数优化方法、装置及存储介质

Country Status (3)

Country Link
EP (1) EP4280660A4 (zh)
CN (1) CN114818452A (zh)
WO (1) WO2022156226A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115396908A (zh) * 2022-08-25 2022-11-25 中国联合网络通信集团有限公司 网络评估方法、装置及存储介质
CN117195740A (zh) * 2023-10-07 2023-12-08 香港中文大学（深圳） 一种基于机器学习的天线方向图预测方法
CN117939502A (zh) * 2024-03-20 2024-04-26 中铁四局集团有限公司 复杂环境条件下无线网络系统的远程控制方法及系统
CN118466255A (zh) * 2024-07-12 2024-08-09 航宇伟创科技(北京)有限公司 一种基于航电仿真器的仿真测试方法、系统及计算设备

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160165468A1 (en) * 2014-12-09 2016-06-09 Futurewei Technologies, Inc. Method and apparatus for optimizing cell specific antenna configuration parameters
CN106021813A (zh) * 2016-06-12 2016-10-12 中南大学 一种基于动态更新代理模型的快速多目标天线设计方法
CN110572835A (zh) * 2019-09-06 2019-12-13 中兴通讯股份有限公司 调节天线参数的方法及装置、电子设备、计算机可读介质
CN111372265A (zh) * 2018-12-26 2020-07-03 中国电信股份有限公司 天线参数配置方法、装置、系统及计算机可读存储介质

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112424998A (zh) * 2018-07-13 2021-02-26 诺基亚技术有限公司 控制辐射元件
CN111225391B (zh) * 2018-11-26 2021-09-21 华为技术有限公司 网络参数处理方法和设备

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160165468A1 (en) * 2014-12-09 2016-06-09 Futurewei Technologies, Inc. Method and apparatus for optimizing cell specific antenna configuration parameters
CN106021813A (zh) * 2016-06-12 2016-10-12 中南大学 一种基于动态更新代理模型的快速多目标天线设计方法
CN111372265A (zh) * 2018-12-26 2020-07-03 中国电信股份有限公司 天线参数配置方法、装置、系统及计算机可读存储介质
CN110572835A (zh) * 2019-09-06 2019-12-13 中兴通讯股份有限公司 调节天线参数的方法及装置、电子设备、计算机可读介质

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP4280660A4

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115396908A (zh) * 2022-08-25 2022-11-25 中国联合网络通信集团有限公司 网络评估方法、装置及存储介质
CN115396908B (zh) * 2022-08-25 2023-09-08 中国联合网络通信集团有限公司 网络评估方法、装置及存储介质
CN117195740A (zh) * 2023-10-07 2023-12-08 香港中文大学（深圳） 一种基于机器学习的天线方向图预测方法
CN117939502A (zh) * 2024-03-20 2024-04-26 中铁四局集团有限公司 复杂环境条件下无线网络系统的远程控制方法及系统
CN118466255A (zh) * 2024-07-12 2024-08-09 航宇伟创科技(北京)有限公司 一种基于航电仿真器的仿真测试方法、系统及计算设备

Also Published As

Publication number Publication date
EP4280660A4 (en) 2024-07-17
CN114818452A (zh) 2022-07-29
EP4280660A1 (en) 2023-11-22

Similar Documents

Publication Publication Date Title
WO2022156226A1 (zh) 天线配置参数优化方法、装置及存储介质
CN109379752B (zh) Massive MIMO的优化方法、装置、设备及介质
US9698897B2 (en) Method and system for auditing and correcting cellular antenna coverage patterns
CN108306699A (zh) 一种基于增益预估的天馈优化方法
WO2022017012A1 (zh) 一种网络配置方法及装置
CN113015192B (zh) 天线权值确定方法、装置、设备及存储介质
CN114501530B (zh) 基于深度强化学习的天线参数的确定方法和装置
US20130252653A1 (en) Methods and apparatus for channel selection within a wireless network
US20230224055A1 (en) Artificial intelligence based management of wireless communication network
CN114339913B (zh) 无线网络的邻小区更新方法、装置、介质及电子设备
CN101304595B (zh) 获取频率规划的方法、装置
CN111698695A (zh) 一种基于神经网络的lte指纹式定位方法
JP2018032939A (ja) 品質推定装置及び品質推定方法
CN114630348A (zh) 基站天线参数调整方法、装置、电子设备及存储介质
CN107113633A (zh) 优化通信网络的处理方法和装置
US11265731B1 (en) Small cell telecommunications network design
CN108282800B (zh) 无线蜂窝网络中的基站天线的参数优化方法
WO2012010283A1 (en) Telecommunications network node and methods
CN116896787A (zh) 终端定位方法、装置、设备及存储介质
CN112351449A (zh) 一种Massive MIMO单小区权值优化方法
CN112512115A (zh) 一种空中基站位置确定方法、装置及电子设备
CN114501338B (zh) 广播波束权值配置方法、装置及电子设备
US8055265B2 (en) Method and system for selecting radiation diagrams of antennas for mobile-radio communication networks
CN114363806B (zh) 一种高精度5g三角定位方法
JP7431415B2 (ja) 情報処理装置および情報処理方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21920605

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2021920605

Country of ref document: EP

Effective date: 20230818