Positioning method, equipment, device and storage medium based on measurement report

Abstract

The embodiment of the present application discloses a positioning method, device, apparatus and storage medium based on measurement reports, the method comprising: determining the cell category to which the MR data belongs according to the measurement report MR data to be positioned; inputting the MR data into a positioning prediction model corresponding to the cell category to obtain the positioning result of the MR data; wherein the positioning prediction model corresponding to the cell category refers to the one obtained after training using the Minimization of Drive Tests (MDT) historical data of the cell with the same cell category as the cell category as input samples, and using the positioning result data corresponding to the MDT historical data as output samples. The embodiment of the present application inputs the MR data to be positioned into the positioning prediction model trained by the MDT historical data, and can accurately obtain the positioning MR data with good real-time performance.

Description

Technical Field

[0001] The present application relates to the field of communication technology, and in particular to a positioning method, device, apparatus and storage medium based on measurement reports. Background Art

[0002] The wireless optimization method of measurement report MR (Measurement Report) is widely used in wireless networks. The location positioning of MR data is particularly important. The location information of MR can truly reflect the user's perception, facilitate understanding of the actual situation of the existing network, and quickly locate problems in the network and deal with them in time to achieve accurate network optimization and planning.

[0003] The existing MR position positioning adopts a fingerprint library positioning algorithm, that is, matching the MR information of the user's location with a sample data set of location fingerprints (fingerprint library); the fingerprint library of this algorithm is generally obtained through communication simulation, and the core of the algorithm lies in the matching design, generally using Euclidean distance, Mahalanobis distance and cosine similarity as grid matching thresholds.

[0004] However, the fingerprint library in actual applications is implemented through simulation, and then the data of Internet TV OTT is used to calibrate the fingerprint library. Since the simulation is calculated through propagation, although it is very comprehensive, it is a theoretical calculation value and there is still a big gap with the actual network. Among them, OTT calibration is only a small part of the data, so the accuracy of the MR position information in the fingerprint library is not high enough, which will lead to a relatively low positioning accuracy. In addition, the core algorithm of the fingerprint library is designed by matching through Euclidean distance, Mahalanobis distance and cosine similarity, and a single MR information is matched with the fingerprint library to calculate the result. When faced with the calculation of massive MR data in the existing network, the amount of calculation is large and the storage memory required is also large. The final result is low efficiency and resource consumption. Summary of the invention

[0005] Since the existing methods have the above-mentioned problems, the embodiments of the present application provide a positioning method, device, apparatus and storage medium based on measurement reports.

[0006] Specifically, the embodiments of the present application provide the following technical solutions:

[0007] In a first aspect, an embodiment of the present application provides a positioning method based on a measurement report, including:

[0008] Determine, according to the measurement report MR data to be positioned, the cell category to which the MR data belongs;

[0009] Inputting the MR data into a positioning prediction model corresponding to the cell category to obtain a positioning result of the MR data;

[0010] The positioning prediction model corresponding to the cell category is obtained by training using MDT historical data of cells of the same cell category as the input sample and positioning result data corresponding to the MDT historical data as the output sample.

[0011] Optionally, determining, according to the measurement report MR data to be positioned, the cell category to which the MR data belongs includes:

[0012] Determine the cell to which the MR data belongs according to the measurement result MR data to be positioned;

[0013] The cells are clustered using a Kmeans algorithm according to the characteristic information of the cells to obtain the cell category to which the cells belong.

[0014] Optionally, the characteristic information of the cell includes one or more of location, nature, room division, frequency, direction angle and hanging height.

[0015] Optionally, the training process of the positioning prediction model corresponding to the cell category includes:

[0016] Acquire MDT historical data of a cell of the same category as the cell, and positioning result data corresponding to the MDT historical data;

[0017] Performing one-hot encoding on the MDT historical data to obtain encoded data, and taking the cell where the MDT historical data is located as a unit, performing standardization processing on the encoded data according to the cell to obtain standardized input data;

[0018] The positioning result data is standardized according to the cells to obtain standardized output data;

[0019] The input data is used as an input sample, and the output data is used as an output sample. After model training is performed based on the XGBoost algorithm, a positioning prediction model corresponding to the cell category is obtained.

[0020] Optionally, the MDT historical data includes one or more of a cell identifier, a primary cell level, a primary cell physical identifier PCI, levels of the first four neighboring cells, PCIs of the first four neighboring cells, and a timing advance TADV of the first four neighboring cells;

[0021] The positioning result data corresponding to the MDT historical data includes one or more of longitude and latitude.

[0022] Optionally, before performing one-hot encoding on the MDT historical data, the method further includes:

[0023] The MDT historical data is cleaned; wherein the data cleaning includes: abnormal processing of missing values ​​and / or abnormal processing of data position drift.

[0024] Optionally, the exception handling of missing values ​​includes: filling different types of missing data with preset values ​​corresponding to the corresponding types;

[0025] and / or,

[0026] The abnormal processing of data position drift includes:

[0027] Determine the upper and lower quartiles of longitude and latitude of a single cell; the upper quartile is U, which means that only 1 / 4 of the values ​​of all samples are greater than U, and the lower quartile is L, which means that only 1 / 4 of the values ​​of all samples are less than L;

[0028] Determine the upper and lower limits; the difference between the upper quartile and the lower quartile is the interquartile range IQR, IQR = UL; the upper limit is U + 1.5 IQR, and the lower limit is L - 1.5 IQR;

[0029] The longitude and latitude are selected according to the upper bound and the lower bound and the intersection is taken, the part outside the intersection is removed as abnormal data, and the part inside the intersection is used as valid data for training.

[0030] Optionally, inputting the MR data into a positioning prediction model corresponding to the cell category to obtain a positioning result of the MR data includes:

[0031] One or more information including the cell identifier, main cell level, main cell PCI, levels of the first four neighboring cells, PCI of the first four neighboring cells, and TADV of the first four neighboring cells included in the MR data is input into the positioning prediction model corresponding to the cell category to obtain the longitude and / or latitude positioning results of the MR data.

[0032] In a second aspect, an embodiment of the present application further provides a data positioning device, including a memory, a transceiver, and a processor:

[0033] A memory for storing a computer program; a transceiver for sending and receiving data under the control of the processor; and a processor for reading the computer program in the memory and implementing the steps of the positioning method based on measurement reports as described in the first aspect when executing the computer program.

[0034] In a third aspect, the embodiment of the present application further provides a data locating device, including:

[0035] A first processing module, configured to determine the cell category to which the MR data belongs according to the measurement report MR data to be positioned;

[0036] A second processing module, used for inputting the MR data into a positioning prediction model corresponding to the cell category to obtain a positioning result of the MR data;

[0037] The positioning prediction model corresponding to the cell category is obtained by training using MDT historical data of cells of the same cell category as the input sample and positioning result data corresponding to the MDT historical data as the output sample.

[0038] In a fourth aspect, an embodiment of the present application further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the positioning method based on measurement reports as described in the first aspect.

[0039] As can be seen from the above technical solutions, the positioning method based on the measurement report in the embodiment of the present application, on the one hand, calls the positioning prediction model of the cell category to which the MR data to be positioned belongs to output the positioning result. Since the training stage of the model is completed in the early stage, when facing massive MR data, the positioning calculation can be performed faster, and the real-time performance is better. On the other hand, the accuracy of using the MDT historical data of the minimized drive test as the training set sample is higher, which can well guarantee the accuracy of the positioning prediction model and further improve the accuracy of MR data positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0041] Figure 1 This is one of the flow charts of the positioning method based on measurement report provided in the embodiment of the present application;

[0042] Figure 2 This is the second flowchart of the positioning method based on measurement report provided in the embodiment of the present application;

[0043] Figure 3 is a flow chart of model prediction provided by an embodiment of the present application;

[0044] Figure 4 is a structural schematic diagram of a data positioning device provided in an embodiment of the present application;

[0045] Figure 5 It is a structural diagram of a data positioning device according to an embodiment of the present application. DETAILED DESCRIPTION

[0046] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0047] Figure 1 A flowchart of a positioning method based on measurement report provided by an embodiment of the present application is shown. Figure 2 is a flowchart of another positioning method based on measurement report provided by an embodiment of the present application. Figure 3 This is a flow chart of a model prediction provided by an embodiment of the present application. Figure 1 , Figure 2 and Figure 3 The positioning method based on the measurement report provided in the embodiment of the present application is explained and illustrated in detail. Figure 1 As shown, a positioning method based on measurement report provided in an embodiment of the present application specifically includes:

[0048] Step 101: Determine the cell category to which the MR data belongs according to the measurement report MR data to be positioned;

[0049] In this step, it should be noted that, due to the large number of cells in the existing network, the accuracy of positioning with a single model in a single cell can be guaranteed, but it is difficult to implement a single model for each cell. A single model in a single cell is difficult to engineer in terms of model training, later model updates, and management. However, the richness and diversity of cells make it impossible to implement all cells with one model, which cannot guarantee the accuracy of the positioning model. Therefore, it is necessary to use Kmeans clustering for the cells, divide the cells with the measurement report MR (Measurement Report) data to be located into n categories, and train a positioning prediction model (which can be an XGBoost model) for each category. After obtaining the MR data to be located, first determine the cell category to which it belongs, and then call the positioning prediction model of the cell category to output the positioning result of the MR data.

[0050] Step 102: inputting the MR data into a positioning prediction model corresponding to the cell category to obtain a positioning result of the MR data;

[0051] The positioning prediction model corresponding to the cell category is obtained by training using Minimization of Drive-tests (MDT) historical data of cells of the same cell category as the cell category as input samples and using positioning result data corresponding to the MDT historical data as output samples.

[0052] In this step, it should be noted that after determining the cell category to which the MR data belongs, the positioning prediction model corresponding to the cell category is called, and the MR data is input into the positioning prediction model, thereby outputting the positioning result of the MR data. The positioning prediction model corresponding to each type of cell is obtained after training using the MDT historical data of the cell with the same cell category as the input sample and the positioning result data corresponding to the MDT historical data as the output sample.

[0053] In this step, it should be noted that MDT is the base station that sends relevant measurement configurations to the terminal according to the MDT measurement task configured by the network management. When the measurement conditions are met, the terminal performs measurements and reports the measurement information. MDT is similar to MR, and contains fields such as reference signal receiving power RSRP (Reference Signal Receiving Power), long-term evolution LTE (Long Term Evolution) reference signal receiving quality RSRQ (Reference Signal Receiving Quality), and contains the latitude and longitude information of the positioning system GPS (Global Positioning System), which can be used for big data analysis. Therefore, MDT data can be used as a training set for MR data to train the positioning prediction model. Among them, MDT data as a training set is also a constraint condition. The number of users in the cell must be sufficient. The more users there are, the better it can simulate the actual situation of MR.

[0054] It can be seen from the above technical scheme that the embodiment of the present application provides a positioning method based on measurement reports. First, according to the measurement report MR data to be positioned, the cell category to which the MR data belongs is determined. Then the MR data is input into the positioning prediction model corresponding to the cell category, so as to obtain the positioning result of the MR data. Among them, the positioning prediction model corresponding to the cell category is obtained after training in advance using the minimized drive test MDT historical data of the cell with the same cell category as the cell category as input samples, and using the positioning result data corresponding to the MDT historical data as output samples. On the one hand, according to the cell category to which the MR data to be positioned belongs, the embodiment of the present application calls the positioning prediction model of the cell category to output the positioning result. Since the training stage of the model is completed in the early stage, when facing massive MR data, the positioning calculation can be performed faster, and the real-time performance is better. On the other hand, the accuracy of using the minimized drive test MDT historical data as the training set sample is high, so that the accuracy of the positioning prediction model can be well guaranteed, and the accuracy of MR data positioning is further improved.

[0055] Based on the content of the above embodiment, in this embodiment, determining the cell category to which the MR data belongs according to the measurement result MR data to be positioned includes:

[0056] Determine the cell to which the MR data belongs according to the measurement result MR data to be positioned;

[0057] The cells are clustered using a Kmeans algorithm according to the characteristic information of the cells to obtain the cell category to which the cells belong.

[0058] In this embodiment, it should be noted that after obtaining the measurement report MR data to be positioned, the cell to which the MR data belongs is first determined, and then the Kmeans algorithm is used to cluster the cell according to the characteristic information of the cell, thereby outputting the cell category to which the cell belongs. The characteristic information of the cell includes: the location, nature, room division, frequency, direction angle, and hanging height of the cell. The Kmeans algorithm is unsupervised learning, and can perform clustering according to the characteristics of the clustered cells. The k value is generally not set very large. It can be combined with actual business by enumeration, such as setting k from 2 to a fixed value such as 10, repeatedly running Kmeans several times on each k value (to avoid local optimal solutions), and calculating the average contour coefficient of the current k, and finally selecting the k corresponding to the value with the largest contour coefficient.

[0059] Based on the contents of the above embodiments, in this embodiment, the characteristic information of the cell includes: one or more of location, nature, room division, frequency, direction angle and hanging height.

[0060] In this embodiment, it should be noted that the Kmeans algorithm is used to cluster the cell according to the characteristic information of the cell, so as to output the cell category to which the cell belongs. The characteristic information of the cell includes: the location, nature, room division, frequency, direction angle, and hanging height of the cell.

[0061] Based on the content of the above embodiment, in this embodiment, the training process of the positioning prediction model corresponding to the cell category includes:

[0062] Acquire MDT historical data of a cell of the same category as the cell, and positioning result data corresponding to the MDT historical data;

[0063] Performing one-hot encoding on the MDT historical data to obtain encoded data, and taking the cell where the MDT historical data is located as a unit, performing standardization processing on the encoded data according to the cell to obtain standardized input data;

[0064] The positioning result data is standardized according to the cells to obtain standardized output data;

[0065] The input data is used as an input sample, and the output data is used as an output sample. After model training is performed based on the XGBoost algorithm, a positioning prediction model corresponding to the cell category is obtained.

[0066] In this embodiment, it should be noted that in training the positioning prediction model corresponding to each cell category, the MDT historical data of the cell with the same cell category and the positioning result data corresponding to the MDT historical data are first obtained. Among them, the MDT historical data includes one or more of the cell identifier, the main cell level, the main cell physical identifier PCI (Peripheral Component Interconnect), the levels of the first 4 neighboring cells, the PCI of the first 4 neighboring cells, and the time advance TADV of the first 4 neighboring cells. Since the PCI value of the cell is a numerical value, there is no difference in size. For example, PCI "25" and "435" have no numerical difference in size. The two are just different codes, so it is necessary to perform unique hot encoding on the main cell and the neighboring cell PCI. Because the PCI is encoded, and the levels of the main neighboring cells are all large negative values, in order to unify, the levels need to be standardized according to the cell. The standardized processing is to transform the original data and convert the data into data with a mean of 0 and a standard deviation of 1. For single cell standardization, take the main cell level as an example, the standardization calculation method is as follows: the mean of all levels in the main cell is calculated as mean, the standard deviation is std, and each level value is x = (X-mean) / std, where x is the standardized level value and X is the original level value. Due to the particularity of the predicted longitude and latitude values, the difference in the data position of MR in the same cell mainly lies in the value after the decimal point, so it is also necessary to standardize the longitude and latitude by cell to obtain the output data after standardization.

[0067] In this step, the input data after standardization is used as the input sample, the output data is used as the output sample, and the positioning prediction model corresponding to the cell category is obtained after model training based on the XGBoost algorithm. Among them, XGBoost is a gradient boosting algorithm and a residual decision tree. Its basic idea is to gradually add one tree and one tree to the model. Every time a classification regression CRAT decision tree is added, the overall effect (the objective function decreases) should be improved. The XGBoost algorithm supports parallel computing and can prevent overfitting very well. Since the PCI coding of the main neighboring cell results in the feature being a coefficient matrix, XGBoost also specially designs an algorithm for sparse data. Therefore, the XGBoost algorithm is used to predict the longitude and latitude of MR. The model prediction process is as follows Figure 3 ; Input the feature information of MR to call the XGBoost model, and output the longitude and latitude of MR data. It can be seen that the embodiment of the present application uses MDT data as a training set, and uses the Kmeans algorithm combined with the XGBoost algorithm to locate MR data, improving the previous MR fingerprint positioning method, improving the accuracy and taking into account the efficiency.

[0068] Based on the content of the above embodiment, in this embodiment, the MDT historical data includes one or more of a cell identifier, a primary cell level, a primary cell physical identifier PCI, levels of the first four neighboring cells, PCIs of the first four neighboring cells, and timing advances TADV (Timing advance) of the first four neighboring cells;

[0069] The positioning result data corresponding to the MDT historical data includes one or more of longitude and latitude.

[0070] In this embodiment, optionally, the MDT historical data includes one or more of a cell identifier, a primary cell level, a primary cell physical identifier PCI, levels of the first four neighboring cells, PCIs of the first four neighboring cells, and time advance values ​​TADVs of the first four neighboring cells. After the positioning prediction model is trained using the MDT historical data as a training set, the output result includes one or more of longitude and latitude.

[0071] Based on the content of the above embodiment, in this embodiment, before performing one-hot encoding on the MDT historical data, the method further includes:

[0072] The MDT historical data is cleaned; wherein the data cleaning includes: abnormal processing of missing values ​​and / or abnormal processing of data position drift.

[0073] In this embodiment, it should be noted that since the data reported by the device has obvious position drift, analysis of the actual data will reveal that a small number of data locations are located at very far locations, and the actual data may also have data missing, so it is necessary to clean the MDT historical data before performing one-hot encoding on the MDT historical data. Data cleaning includes exception handling for missing values ​​and / or exception handling for data position drift. It can be seen that after first obtaining the historical MDT data of the cell of this category, the embodiment of the present application needs to clean the MDT data and perform feature processing on the cleaned data. The processed data is trained using the XGBoost model, so that the MR data of this category to be located calls the trained XGBoost model, that is, predicts the position result of the MR data.

[0074] Based on the content of the above embodiment, in this embodiment, the exception handling of missing values ​​includes: filling different types of missing data with preset values ​​corresponding to the corresponding types;

[0075] and / or,

[0076] The abnormal processing of data position drift includes:

[0077] Determine the upper and lower quartiles of longitude and latitude of a single cell; the upper quartile is U, which means that only 1 / 4 of the values ​​of all samples are greater than U, and the lower quartile is L, which means that only 1 / 4 of the values ​​of all samples are less than L;

[0078] Determine the upper and lower limits; the difference between the upper quartile and the lower quartile is the interquartile range IQR, IQR = UL; the upper limit is U + 1.5 IQR, and the lower limit is L - 1.5 IQR;

[0079] The longitude and latitude are selected according to the upper bound and the lower bound and the intersection is taken, the part outside the intersection is removed as abnormal data, and the part inside the intersection is used as valid data for training.

[0080] In this embodiment, it should be noted that the exception handling of missing values ​​includes filling different types of missing data with preset values ​​corresponding to the corresponding types. For example, PCI missing values ​​in MDT historical data are filled with -1, and level missing values ​​are filled with -999.

[0081] In this embodiment, it should be noted that for abnormal processing of data position drift, a single cell uses a box plot to remove abnormal values, and the specific implementation method is as follows:

[0082] First, define the upper and lower quartiles of the latitude and longitude of a single cell: the upper quartile is set to U, which means that only 1 / 4 of all samples have values ​​greater than U, that is, U is at 25% when sorted from large to small; similarly, the lower quartile is set to L, which means that only 1 / 4 of all samples have values ​​less than L, that is, L is at 75% when sorted from large to small. Then define the upper and lower bounds: set the difference between the upper and lower quartiles to IQR, that is: IQR = UL; the upper bound is set to U + 1.5 IQR, and the lower bound is set to L - 1.5 IQR; the longitude and latitude are selected according to the upper and lower bounds and the intersection is taken, and this part of the data is used for training. The box plot is more objective in selecting outliers and has certain advantages in identifying outliers.

[0083] Based on the content of the above embodiment, in this embodiment, the MR data is input into the positioning prediction model corresponding to the cell category to obtain the positioning result of the MR data, including:

[0084] One or more information including the cell identifier, main cell level, main cell PCI, levels of the first four neighboring cells, PCI of the first four neighboring cells, and TADV of the first four neighboring cells included in the MR data is input into the positioning prediction model corresponding to the cell category to obtain the longitude and / or latitude positioning results of the MR data.

[0085] In this embodiment, it should be noted that when the MR data is input into the positioning prediction model corresponding to the cell category, one or more information including the cell identifier, main cell level, main cell PCI, levels of the first four neighboring areas, PCI of the first four neighboring areas, and TADV of the first four neighboring areas included in the MR data can be input into the positioning prediction model corresponding to the cell category to obtain the longitude and / or latitude positioning results of the MR data.

[0086] Based on the same inventive concept, another embodiment of the present invention provides a data locating device, such as Figure 4 As shown, a data positioning device provided in an embodiment of the present application includes:

[0087] The first processing module 1 is used to determine the cell category to which the MR data belongs according to the measurement report MR data to be positioned;

[0088] A second processing module 2 is used to input the MR data into a positioning prediction model corresponding to the cell category to obtain a positioning result of the MR data;

[0089] The positioning prediction model corresponding to the cell category is obtained by training using MDT historical data of cells of the same cell category as the input sample and positioning result data corresponding to the MDT historical data as the output sample.

[0090] In this embodiment, it should be noted that, since there are many cells in the existing network, the accuracy of positioning with a single model in a single cell can be guaranteed, but it is difficult to implement a single model for each cell. A single model in a single cell is difficult to engineer in terms of model training, later model updates, and management. However, the richness and diversity of cells make it impossible to implement all cells with one model, which cannot guarantee the accuracy of the positioning model. Therefore, it is necessary to cluster the cells using Kmeans clustering, divide the cells of the MR data to be located into n categories, train a positioning prediction model for each category (which can be an XGBoost model), and after obtaining the MR data to be located, first determine the cell category to which it belongs, and then call the positioning prediction model of the cell category to output the positioning result of the MR data.

[0091] In this embodiment, it should be noted that after determining the cell category to which the MR data belongs, the positioning prediction model corresponding to the cell category is called, and the MR data is input into the positioning prediction model, thereby outputting the positioning result of the MR data. The positioning prediction model corresponding to each type of cell is obtained after training using the minimization of drive test (MDT) historical data of the cell of the same cell category as the cell category as input samples, and using the positioning result data corresponding to the MDT historical data as output samples.

[0092] In this embodiment, it should be noted that MDT is that the base station sends relevant measurement configurations to the terminal according to the MDT measurement task configured by the network management. When the measurement conditions are met, the terminal performs measurements and reports the measurement information. MDT is similar to MR, and includes fields such as reference signal receiving power RSRP (Reference Signal Receiving Power) and LTE reference signal receiving quality RSRQ (Reference Signal Receiving Quality). It contains the longitude and latitude information of the positioning system GPS (Global Positioning System) and can be used for big data analysis. Therefore, MDT data can be used as a training set for MR data to train the positioning prediction model. Among them, MDT data as a training set is also a constraint condition. The number of users in the cell must be sufficient. The more users there are, the better the real situation of MR can be simulated.

[0093] In this embodiment, it should be noted that the positioning data accuracy of MDT historical data is much higher than that of simulation, ensuring the accuracy of the training set samples; secondly, Kmeans and XGBoost model training, the model training stage is completed in the early stage, and the XGBoost model supports parallel computing, and the trained model is directly called by the MR data to be located, and the reasoning time is very fast, so it is efficient and has good real-time performance, which is much more efficient than the previous calculation of the results of each MR matching.

[0094] It can be seen from the above technical scheme that a data positioning device provided in an embodiment of the present application first determines the cell category to which the MR data belongs based on the measurement report MR data to be positioned. Then the MR data is input into the positioning prediction model corresponding to the cell category, thereby obtaining the positioning result of the MR data. Among them, the positioning prediction model corresponding to the cell category is obtained after training in advance using the minimized drive test MDT historical data of the cell with the same cell category as the cell category as input samples, and using the positioning result data corresponding to the MDT historical data as output samples. On the one hand, according to the cell category to which the MR data to be positioned belongs, the embodiment of the present application calls the positioning prediction model of the cell category to output the positioning result. Since the training stage of the model is completed in the early stage, when facing massive MR data, the positioning calculation can be performed faster, and the real-time performance is better. On the other hand, the accuracy of using the minimized drive test MDT historical data as the training set sample is high, so that the accuracy of the positioning prediction model can be well guaranteed, and the accuracy of MR data positioning is further improved.

[0095] The data locating device described in this embodiment can be used to execute the above method embodiment. Its principles and technical effects are similar and will not be described in detail here.

[0096] Based on the same inventive concept, another embodiment of the present invention provides a data positioning device, see Figure 5 The structural diagram of the data locating device is as follows: the network device 500 includes a memory 502, a transceiver 503, and a processor 501: wherein the processor 501 and the memory 502 may also be arranged physically separately.

[0097] The memory 502 is used to store computer programs; the transceiver 503 is used to send and receive data under the control of the processor 501.

[0098] Specifically, in Figure 5 In the embodiment, the bus system 504 may include any number of interconnected buses and bridges, specifically linking together various circuits of one or more processors represented by the processor 501 and the memory represented by the memory 502. The bus system 504 may also link together various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface. The transceiver 503 may be a plurality of components, namely, a transmitter and a receiver, providing a unit for communicating with various other devices on a transmission medium, which transmission medium includes a wireless channel, a wired channel, an optical cable, and other transmission media. The processor 501 is responsible for managing the bus architecture and general management, and the memory 502 may store data used by the processor 501 when performing operations.

[0099] The processor 501 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or a complex programmable logic device (CPLD). The processor may also adopt a multi-core architecture.

[0100] The processor 501 calls the computer program stored in the memory 502 to execute any of the methods provided in the embodiments of the present application according to the obtained executable instructions, for example:

[0101] Determine, according to the measurement report MR data to be positioned, the cell category to which the MR data belongs;

[0102] Inputting the MR data into a positioning prediction model corresponding to the cell category to obtain a positioning result of the MR data;

[0103] The positioning prediction model corresponding to the cell category is obtained by training using MDT historical data of cells of the same cell category as the input sample and positioning result data corresponding to the MDT historical data as the output sample.

[0104] Based on the content of the above embodiment, in this embodiment, determining the cell category to which the MR data belongs according to the measurement report MR data to be positioned includes:

[0105] Determine the cell to which the MR data belongs according to the measurement result MR data to be positioned;

[0106] The cells are clustered using a Kmeans algorithm according to the characteristic information of the cells to obtain the cell category to which the cells belong.

[0107] Based on the contents of the above embodiments, in this embodiment, the characteristic information of the cell includes: one or more of location, nature, room division, frequency, direction angle and hanging height.

[0108] Based on the content of the above embodiment, in this embodiment, the training process of the positioning prediction model corresponding to the cell category includes:

[0109] Acquire MDT historical data of a cell of the same category as the cell, and positioning result data corresponding to the MDT historical data;

[0110] Performing one-hot encoding on the MDT historical data to obtain encoded data, and taking the cell where the MDT historical data is located as a unit, performing standardization processing on the encoded data according to the cell to obtain standardized input data;

[0111] The positioning result data is standardized according to the cells to obtain standardized output data;

[0112] The input data is used as an input sample, and the output data is used as an output sample. After model training is performed based on the XGBoost algorithm, a positioning prediction model corresponding to the cell category is obtained.

[0113] Based on the content of the above embodiment, in this embodiment, the MDT historical data includes one or more of a cell identifier, a primary cell level, a primary cell physical identifier PCI, levels of the first four neighboring cells, PCIs of the first four neighboring cells, and timing advance values ​​TADVs of the first four neighboring cells;

[0114] The positioning result data corresponding to the MDT historical data includes one or more of longitude and latitude.

[0115] Based on the content of the above embodiment, in this embodiment, before performing one-hot encoding on the MDT historical data, the method further includes:

[0116] The MDT historical data is cleaned; wherein the data cleaning includes: abnormal processing of missing values ​​and / or abnormal processing of data position drift.

[0117] Based on the content of the above embodiment, in this embodiment, the exception handling of missing values ​​includes: filling different types of missing data with preset values ​​corresponding to the corresponding types;

[0118] and / or,

[0119] The abnormal processing of data position drift includes:

[0120] Determine the upper and lower quartiles of longitude and latitude of a single cell; the upper quartile is U, which means that only 1 / 4 of the values ​​of all samples are greater than U, and the lower quartile is L, which means that only 1 / 4 of the values ​​of all samples are less than L;

[0121] Determine the upper and lower limits; the difference between the upper quartile and the lower quartile is the interquartile range IQR, IQR = UL; the upper limit is U + 1.5 IQR, and the lower limit is L - 1.5 IQR;

[0122] The longitude and latitude are selected according to the upper bound and the lower bound and the intersection is taken, the part outside the intersection is removed as abnormal data, and the part inside the intersection is used as valid data for training.

[0123] Based on the content of the above embodiment, in this embodiment, the MR data is input into the positioning prediction model corresponding to the cell category to obtain the positioning result of the MR data, including:

[0124] One or more information including the cell identifier, main cell level, main cell PCI, levels of the first four neighboring cells, PCI of the first four neighboring cells, and TADV of the first four neighboring cells included in the MR data is input into the positioning prediction model corresponding to the cell category to obtain the longitude and / or latitude positioning results of the MR data.

[0125] Based on the same inventive concept, another embodiment of the present invention provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, all steps of the above-mentioned positioning method based on measurement reports are implemented, for example, according to the measurement report MR data to be positioned, the cell category to which the MR data belongs is determined; the MR data is input into a positioning prediction model corresponding to the cell category to obtain the positioning result of the MR data; wherein the positioning prediction model corresponding to the cell category refers to a model obtained after training using the minimization of drive tests (MDT) historical data of the cell with the same cell category as the cell category as input samples, and using the positioning result data corresponding to the MDT historical data as output samples.

[0126] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0127] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the embodiment scheme of the present application. Those of ordinary skill in the art may understand and implement it without creative labor.

[0128] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the positioning method based on the measurement report described in each embodiment or some parts of the embodiment.

[0129] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A positioning method based on measurement report, characterized in that: include: Determine, according to the measurement report MR data to be positioned, the cell category to which the MR data belongs; Inputting the MR data into a positioning prediction model corresponding to the cell category to obtain a positioning result of the MR data; The positioning prediction model corresponding to the cell category is obtained by training using MDT historical data of cells of the same cell category as the input sample and using positioning result data corresponding to the MDT historical data as the output sample; The determining, according to the measurement report MR data to be positioned, the cell category to which the MR data belongs, comprises: Determine the cell to which the MR data belongs according to the measurement result MR data to be positioned; Clustering the cells using a Kmeans algorithm according to the characteristic information of the cells to obtain the cell category to which the cells belong; The characteristic information of the cell includes: one or more of location, nature, room division, frequency, direction angle and hanging height; The training process of the positioning prediction model corresponding to the cell category includes: Acquire MDT historical data of a cell of the same category as the cell, and positioning result data corresponding to the MDT historical data; Performing one-hot encoding on the MDT historical data to obtain encoded data, and taking the cell where the MDT historical data is located as a unit, performing standardization processing on the encoded data according to the cell to obtain standardized input data; The positioning result data is standardized according to the cells to obtain standardized output data; The input data is used as an input sample, and the output data is used as an output sample. After model training is performed based on the XGBoost algorithm, a positioning prediction model corresponding to the cell category is obtained.

2. The positioning method based on measurement report according to claim 1, characterized in that: The MDT historical data includes one or more of a cell identifier, a primary cell level, a primary cell physical identifier PCI, levels of the first four neighboring cells, PCIs of the first four neighboring cells, and a timing advance value TADV of the first four neighboring cells; The positioning result data corresponding to the MDT historical data includes one or more of longitude and latitude.

3. The positioning method based on measurement report according to claim 1, characterized in that: Before performing one-hot encoding on the MDT historical data, the method further includes: The MDT historical data is cleaned; wherein the data cleaning includes: abnormal processing of missing values ​​and / or abnormal processing of data position drift.

4. The positioning method based on measurement report according to claim 3, characterized in that: The abnormal processing of missing values ​​includes: filling different types of missing data with preset values ​​corresponding to the corresponding types; and / or, The abnormal processing of data position drift includes: Determine the upper and lower quartiles of longitude and latitude of a single cell; the upper quartile is U, which means that only 1 / 4 of the values ​​of all samples are greater than U, and the lower quartile is L, which means that only 1 / 4 of the values ​​of all samples are less than L; Determine the upper and lower limits; the difference between the upper quartile and the lower quartile is the interquartile range IQR, IQR = UL; the upper limit is U + 1.5 IQR, and the lower limit is L - 1.5 IQR; The longitude and latitude are selected according to the upper bound and the lower bound and the intersection is taken, the part outside the intersection is removed as abnormal data, and the part inside the intersection is used as valid data for training.

5. The positioning method based on measurement report according to claim 2, characterized in that: Inputting the MR data into a positioning prediction model corresponding to the cell category to obtain a positioning result of the MR data includes: One or more information including the cell identifier, main cell level, main cell PCI, levels of the first four neighboring cells, PCI of the first four neighboring cells, and TADV of the first four neighboring cells included in the MR data is input into the positioning prediction model corresponding to the cell category to obtain the longitude and / or latitude positioning results of the MR data.

6. A data positioning device, characterized in that: Including memory, transceiver, processor: Memory for storing computer programs; a transceiver, for transmitting and receiving data under the control of the processor; A processor, configured to read the computer program in the memory and execute the steps of the positioning method based on measurement reports as described in any one of claims 1 to 5.

7. A data locating device, characterized in that: include: A first processing module, configured to determine the cell category to which the MR data belongs according to the measurement report MR data to be positioned; A second processing module, used for inputting the MR data into a positioning prediction model corresponding to the cell category to obtain a positioning result of the MR data; The positioning prediction model corresponding to the cell category is obtained by training using MDT historical data of cells of the same cell category as the input sample and using positioning result data corresponding to the MDT historical data as the output sample; The determining, according to the measurement report MR data to be positioned, the cell category to which the MR data belongs, comprises: Determine the cell to which the MR data belongs according to the measurement result MR data to be positioned; Clustering the cells using a Kmeans algorithm according to the characteristic information of the cells to obtain the cell category to which the cells belong; The characteristic information of the cell includes: one or more of location, nature, room division, frequency, direction angle and hanging height; The training process of the positioning prediction model corresponding to the cell category includes: Acquire MDT historical data of a cell of the same category as the cell, and positioning result data corresponding to the MDT historical data; Performing one-hot encoding on the MDT historical data to obtain encoded data, and taking the cell where the MDT historical data is located as a unit, performing standardization processing on the encoded data according to the cell to obtain standardized input data; The positioning result data is standardized according to the cells to obtain standardized output data; The input data is used as an input sample, and the output data is used as an output sample. After model training is performed based on the XGBoost algorithm, a positioning prediction model corresponding to the cell category is obtained.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the positioning method based on measurement reports as described in any one of claims 1 to 5 are implemented.

Images