5G high-precision positioning method and system based on inter-system measurement

Abstract

The invention provides a 5G high-precision positioning method and system based on inter-system measurement, electronic equipment and a storage medium, and aims to solve the problem that 5G application positioning precision is insufficient, the method comprises the following steps: obtaining position information and a sample of 5G neighbor cell measurement information from 4G MR data, the measurement information comprising a frequency point and a PCI; based on the engineering parameters, establishing a mapping relation from the ECI, the 5G frequency point and the PCI of the 4G cell to the NCI of the 5G cell; for each sample, identifying the NCI corresponding to the 5G neighbor cell measurement information by using the mapping relationship, and forming 5G sampling points of the position information and the complete signal characteristics of the master server and the neighbor cell 5G cell based on the signal strength; training a model by using all the sampling points, taking the position information as a label and taking the 5G cell signal characteristics as input, and constructing a 5G positioning fingerprint database; and in response to a 5G measurement report of the terminal, matching the terminal position from the fingerprint database. According to the invention, the 5G positioning precision can be obviously improved.

Description

Technical Field

[0001] This disclosure relates to the field of communication technology, specifically to a 5G high-precision positioning method based on heterogeneous system measurement, a 5G high-precision positioning system based on heterogeneous system measurement, an electronic device, and a computer-readable storage medium. Background Technology

[0002] In today's digital age, Location-Based Services (LBS) have become deeply integrated into people's daily lives. From navigation for daily commutes to location-based personalized recommendations, LBS is essential. Reliable LBS relies on accurate location information, which in turn depends on high-precision positioning technology. Positioning technologies are diverse. Based on the source of the positioning reference signal, they can be categorized into 4G / 5G positioning, GPS / BeiDou positioning, UWB (Ultra-Wideband) positioning, Bluetooth positioning, Wi-Fi positioning, and RFID (Radio Frequency Identification) positioning. From the perspective of how the positioning request is initiated, they can be divided into active positioning and passive positioning. Active positioning, such as in navigation applications, requires the terminal to actively call the positioning service to determine its own location and navigate to the destination. Technologies supporting active positioning include 5G positioning and GPS / BeiDou positioning. Passive positioning, on the other hand, uses the network to locate the terminal, for emergency rescue and equipment monitoring, etc. The terminal being located is unaware or only passively responds. Technologies supporting passive positioning include 4G / 5G positioning.

[0003] However, the effective application of positioning technology relies heavily on the support of corresponding positioning infrastructure. For example, Bluetooth positioning requires the construction of Bluetooth positioning base stations and positioning terminals, and the availability of such hardware significantly impacts the scope and effectiveness of positioning technology applications. Currently, only 4G and 5G networks have achieved full coverage, allowing positioning technologies based on these networks to reach almost all mobile phone users and possess the broadest application prospects. Passive positioning technology utilizes communication signaling and MR (Measurement Report) data in the form of big data to locate terminals, and is widely used in many fields such as social governance and emergency disaster relief. In 4G networks, location information is collected using MDT (Minimization of Drive Tests) data and combined with MR data to train a positioning fingerprint database, achieving a positioning accuracy of up to 50m. However, for common 5G applications, the lack of MDT data means that the industry's main positioning methods, such as simulated fingerprint positioning and triangulation, only achieve an accuracy of 100-150m, which is insufficient to meet the refined optimization requirements of 5G networks. Summary of the Invention

[0004] To address the issue of insufficient positioning accuracy in existing 5G applications due to the lack of MDT (Multi-Level Data) data, this disclosure provides a 5G high-precision positioning method based on inter-system measurement, a 5G high-precision positioning system based on inter-system measurement, an electronic device, and a computer-readable storage medium. This method effectively converts 4G MR (Matching Location Data) data into 5G positioning fingerprint database data, solving both the problem of 5G positioning data source and the conversion from 4G data to 5G high-precision sampling points, significantly improving 5G positioning accuracy to a practical level.

[0005] In a first aspect, this disclosure provides a 5G high-precision positioning method based on heterogeneous system measurements, the method comprising:

[0006] Sample data containing high-precision location information and 5G neighbor cell measurement information is obtained from MR data of 4G network. The 5G neighbor cell measurement information includes at least the frequency point and PCI (Physical Cell Identifier) ​​of the 5G neighbor cell.

[0007] Based on 4G and 5G engineering parameters, establish a mapping relationship between 4G cell ECI (E-UTRAN Cell Identifier), 5G frequency points, and 5G PCI to 5G cell NCI (NR Cell Identifier).

[0008] For each sample data, the mapping relationship is used to identify the 5G cell NCI corresponding to the 5G neighbor cell measurement information, and the 5G neighbor cells are sorted based on signal strength to form a 5G sampling point containing the high-precision location information, the primary 5G cell, and the neighboring 5G cells with complete signal characteristics.

[0009] Using all 5G sampling points, the high-precision location information is used as the location label, and the corresponding 5G cell signal features are used as the input features to train a machine learning positioning model in order to build a 5G high-precision positioning fingerprint database.

[0010] In response to receiving a 5G measurement report from the terminal to be located, the location of the terminal to be located is matched or predicted from the 5G high-precision positioning fingerprint database.

[0011] Furthermore, the acquisition of sample data containing high-precision location information and 5G neighbor cell measurement information includes:

[0012] Parse the raw XML format data of 4G MR;

[0013] Extract timestamps, latitude and longitude, 4G cell ECI, and 5G neighbor cell measurement information from the data;

[0014] Data that simultaneously contains valid MDT high-precision latitude and longitude coordinates and at least one 5G neighbor cell measurement information is filtered out and used as the sample data.

[0015] Furthermore, establishing the mapping relationship between 4G cell ECI, 5G frequency points, and 5G PCI to 5G cell NCI includes:

[0016] Based on the 4G engineering parameters, obtain the ECI of the 4G cell and the latitude and longitude of its installation location;

[0017] Based on the 5G engineering parameters, obtain the NCI, 5G frequency point, PCI, and latitude and longitude of the installation location of the 5G cell;

[0018] For a specific 4G cell, within a preset geographical distance range around it, search for 5G candidate cells in the 5G engineering parameters that have the same 5G frequency and PCI as the 5G neighbor cell measurement information contained in a certain MR data from that specific 4G cell.

[0019] From the 5G candidate cells, select the 5G cell that is closest to the specific 4G cell, and establish a mapping relationship from the ECI of the specific 4G cell, the 5G frequency point, and the PCI to the NCI of the closest 5G cell.

[0020] Furthermore, the ranking of 5G neighboring cells based on signal strength includes:

[0021] All 5G neighboring cells measured in the sample data are sorted in descending order according to their RSRP (Reference Signal Received Power) values;

[0022] The 5G neighbor cell with the highest RSRP is designated as the primary 5G cell, and the remaining 5G neighbor cells are designated as neighboring 5G cells.

[0023] Furthermore, the 5G sampling point includes the following fields: timestamp, longitude, latitude, primary serving cell NCI, primary serving cell frequency, primary serving cell PCI, primary serving cell RSRP, primary serving cell RSRQ (Reference Signal Received Quality), and at least one neighboring cell NCI, neighboring cell frequency, neighboring cell PCI, neighboring cell RSRP, and neighboring cell RSRQ.

[0024] Furthermore, the machine learning localization model is the LightGBM model.

[0025] Secondly, this public provides a 5G high-precision positioning system based on heterogeneous system measurement, the system comprising:

[0026] The data acquisition module is configured to acquire sample data containing high-precision location information and 5G neighbor cell measurement information from MR data of the 4G network. The 5G neighbor cell measurement information includes at least the frequency point and PCI of the 5G neighbor cell.

[0027] The cell identifier mapping module is configured to establish a mapping relationship between 4G cell identifier ECI, 5G frequency point and 5G PCI to 5G cell NCI based on 4G engineering parameters and 5G engineering parameters.

[0028] The sampling point construction module is configured to identify the 5G cell NCI corresponding to the 5G neighbor cell measurement information in each sample data using the mapping relationship, and sort the 5G neighbor cells based on signal strength to form a 5G sampling point containing the high-precision location information, the primary 5G cell and the neighboring 5G cells with complete signal characteristics.

[0029] The model training module is configured to use all 5G sampling points, with the high-precision location information as the location label and the corresponding 5G cell signal features as the input features, to train a machine learning positioning model in order to build a 5G high-precision positioning fingerprint database.

[0030] The positioning module is configured to, in response to receiving a 5G measurement report from the terminal to be positioned, match or predict the location of the terminal to be positioned from the 5G high-precision positioning fingerprint database.

[0031] Furthermore, the data acquisition module is specifically configured as follows:

[0032] Parse the raw XML format data of 4G MR;

[0033] Extract timestamps, latitude and longitude, 4G cell ECI, and 5G neighbor cell measurement information from the data;

[0034] Data that simultaneously contains valid MDT high-precision latitude and longitude coordinates and at least one 5G neighbor cell measurement information is filtered out and used as the sample data.

[0035] Thirdly, this disclosure provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the processor runs the computer program stored in the memory, the processor executes a 5G high-precision positioning method based on heterogeneous system measurement as described in any of the first aspects.

[0036] Fourthly, this disclosure provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the 5G high-precision positioning method based on heterogeneous system measurement as described in any of the first aspects above.

[0037] Beneficial effects:

[0038] This disclosure provides a 5G high-precision positioning method, system, electronic device, and storage medium based on inter-system measurement for 5G. Utilizing existing MDT measurement report data from 4G networks, and through a series of data processing and correlation methods, it indirectly assigns high-precision location tags to 5G signal characteristics, thereby successfully constructing a 5G high-precision positioning fingerprint database. This improves the passive positioning accuracy of 5G networks from the hundreds of meters (100-150 meters) to the tens of meters level, effectively solving the high-precision positioning problem caused by missing MDT data in 5G networks and significantly improving positioning accuracy. Moreover, the solution requires no modification to existing 5G network hardware and does not require waiting for the widespread adoption of 5G terminal MDT functionality; instead, it fully utilizes the data resources of the already extensively covered and maturely operating 4G networks. By mining underutilized 5G neighbor cell measurement information in 4G MR, the data value of the 4G network is extended to the 5G network, achieving a low-cost, high-efficiency technological upgrade. Meanwhile, the NCI identification method based on engineering parameters and geographical distance constraints, according to network planning principles, can accurately map measurement information to a unique 5G cell through association query and distance calculation, fundamentally avoiding positioning errors caused by cell identifier confusion, laying a solid data foundation for training a high-precision positioning model, and ensuring the accuracy and reliability of the constructed fingerprint database. Attached Figure Description

[0039] Figure 1 This is a flowchart illustrating a 5G high-precision positioning method based on heterogeneous system measurement, provided in Embodiment 1 of this disclosure.

[0040] Figure 2 A schematic diagram illustrating an example of an RRC Connection Reconfiguration message provided in an embodiment of this disclosure;

[0041] Figure 3 A partial screenshot of the 4G MR data acquisition specification (draft for comments) provided in this embodiment of the disclosure;

[0042] Figure 4 A schematic diagram of a 4G measurement of a 5G MR sample provided in an embodiment of this disclosure;

[0043] Figure 5 This is a schematic diagram of the overall construction process of 5G high-precision positioning based on heterogeneous system measurement, provided in an embodiment of this disclosure.

[0044] Figure 6 This is an architecture diagram of a 5G high-precision positioning system based on heterogeneous system measurement, provided in Embodiment 2 of this disclosure;

[0045] Figure 7 This is an architectural diagram of an electronic device provided in Embodiment 3 of this disclosure. Detailed Implementation

[0046] To enable those skilled in the art to better understand the technical solutions of this disclosure, the disclosure will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments and drawings described herein are merely for explaining the invention and are not intended to limit the invention.

[0047] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence; furthermore, in the absence of conflict, the embodiments and features in the embodiments of this disclosure can be arbitrarily combined with each other.

[0048] The terminology used in the embodiments of this disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of this disclosure. The singular forms “a,” “the,” and “the” as used in the embodiments of this disclosure and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0049] In the following description, the use of suffixes such as “module,” “part,” or “unit” to denote elements is solely for the purpose of illustrative purposes and has no specific meaning in itself. Therefore, “module,” “part,” or “unit” may be used interchangeably.

[0050] Current common passive positioning methods for 5G, such as simulated fingerprint positioning and triangulation, suffer from the following problems: Insufficient hardware infrastructure: The effective application of positioning technology relies on the support of corresponding basic positioning hardware infrastructure. For example, Bluetooth positioning requires the construction of Bluetooth positioning base stations and positioning terminals. The availability of hardware infrastructure greatly affects the application scope and effectiveness of positioning technology. Weak cross-network feature adaptability: 5G and 4G signal characteristics differ significantly (such as frequency bands, beamforming, and modulation methods). Existing methods have not been optimized for these differences, leading to deviations in cross-network signal feature matching and reducing positioning accuracy. Missing 5G MDT data: 5G terminals do not yet support MDT (Minimum Drive Test) functionality, making it impossible to directly obtain accurate location data. Existing technologies rely on simulated fingerprint or triangulation positioning, with positioning accuracy generally below the 100-meter level, making it difficult to meet the refined optimization requirements of 5G networks.

[0051] This disclosure focuses on addressing the issue of insufficient 5G positioning accuracy by proposing a solution based on periodic 4G MR measurements of 5G neighboring cells to construct a high-precision 5G positioning fingerprint database. Through in-depth research and practice, it aims to improve 5G positioning accuracy, providing solid technical support for the precise application of 5G networks in various fields, facilitating more scientific and rational 5G network planning and construction, and promoting the effective application and innovative development of 5G technology in more scenarios. This will play a significant role in enhancing the level of digital services in society, strengthening public safety capabilities, and promoting the upgrading of intelligent industries.

[0052] The following detailed embodiments illustrate the technical solutions of this disclosure and how they solve the technical problems existing in the prior art. It is understood that in the embodiments of this disclosure, the executing entity may perform some or all of the steps in the embodiments of this disclosure. These steps or operations are merely examples, and the embodiments of this disclosure may also perform other operations or variations thereof. Furthermore, the steps may be performed in different orders as presented in the embodiments of this disclosure, and it is not necessary to perform all the operations in the embodiments of this disclosure. Moreover, the following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0053] Example 1

[0054] Figure 1 This is a flowchart illustrating a 5G high-precision positioning method based on heterogeneous system measurement, as provided in Embodiment 1 of this disclosure. Figure 1 As shown, the method includes:

[0055] Step S101: Obtain sample data containing high-precision location information and 5G neighbor cell measurement information from the MR data of the 4G network. The 5G neighbor cell measurement information includes at least the frequency point and PCI of the 5G neighbor cell.

[0056] Step S102: Based on 4G engineering parameters and 5G engineering parameters, establish the mapping relationship between 4G cell ECI, 5G frequency point and 5G PCI to 5G cell NCI;

[0057] Step S103: For each sample data, the 5G cell NCI corresponding to the 5G neighbor cell measurement information is identified using the mapping relationship, and the 5G neighbor cells are sorted based on signal strength to form a 5G sampling point containing the high-precision location information, the primary 5G cell, and the neighboring 5G cells with complete signal characteristics.

[0058] Step S104: Using all 5G sampling points, with the high-precision location information as the location label and the corresponding 5G cell signal features as the input features, train a machine learning positioning model to build a 5G high-precision positioning fingerprint database.

[0059] Step S105: In response to receiving a 5G measurement report from the terminal to be located, match or predict the location of the terminal to be located from the 5G high-precision positioning fingerprint database.

[0060] This disclosed embodiment utilizes 4G and 5G engineering parameter files, based on the uniqueness of 5G frequency points and PCI combinations, and combined with geographical distance constraints, to establish cross-system associations between 4G and 5G cells, identifying the unique identifier (NCI) of the 5G cell. Furthermore, it arranges 5G neighboring cells in reverse RSRP order to determine the primary server and neighboring cells, forming a complete 5G sampling point containing multi-dimensional signal indicators. This scheme, from data extraction to sampling point construction, achieves an effective conversion from 4G MR data to 5G positioning fingerprint database data, solving both the 5G positioning data source problem and the conversion from 4G data to high-precision 5G sampling points.

[0061] First, the principle of inter-system measurement is explained. The principle of 4G inter-system measurement is as follows:

[0062] 1) Interoperability between 4G and 5G.

[0063] Interoperability between 4G and 5G ensures that the sample can freely switch between 4G and 5G networks to achieve different communication services. For example, when the sample is performing small packet data services (not sensitive to download speed) or voice services, the terminal switches to 4G, and when it needs to perform large packet data services, it switches back to 5G.

[0064] To ensure smooth interoperability between 4G and 5G, it is essential to support inter-system measurements between 4G and 5G. LTE terminals have only one transmitter and receiver; when a UE is providing services on a specific frequency, it cannot measure other frequencies. If the UE needs to perform inter-frequency or inter-system measurements, it must suspend services related to its current service frequency and instead begin measurement from a new subframe, allocating a time slot for inter-frequency measurement. This is the measurement gap (GAP) mechanism in LTE.

[0065] According to the 3GPP specification, the configuration for measuring GAPs includes the gap mode and the gapOffset parameter. There are two main gap modes: gp0 and gp1. The gap period in gp0 mode is 40ms, and in gp1 mode it is 80ms. In both modes, the duration of each GAP is 6ms. The gapOffset parameter affects the start time of the GAP. In gp0 mode, the gapOffset ranges from 0 to 39; in gp1 mode, it ranges from 0 to 791. The measurement GAP length (MGL) is fixed at 6ms. The measurement GAP repetition period (MGRP) depends on the selected gap mode, i.e., 40ms or 80ms. The gapOffset is used to determine the first subframe of each GAP. These parameters are configured in the MeasConfig field of the RRC (Radio Resource Control) Connection Reconfiguration message. An example of an RRC Connection Reconfiguration message is... Figure 2 As shown, this is used to guide the UE in performing effective inter-frequency and inter-system measurements.

[0066] Typical inter-system measurements between 4G and 5G networks are based on the need for 4G / 5G interoperability. These measurement results are event-based and are ultimately collected and processed by the base station, outputting to the MRE file of the OMC (Operation and Maintenance Center). Due to the needs of network maintenance and management, the 3GPP specification supports enabling periodic inter-system measurements between 4G and 5G networks, with the periodic inter-system measurement results ultimately output to the MRO of the OMC.

[0067] 2) 4G MR data acquisition specifications.

[0068] The detailed specifications for 4G MR data acquisition include key fields for 5G inter-system measurement results, such as the SSB (Synchronization Signal and PBCH Block) reference signal received power (RSRP), SSB reference signal received quality (RSRQ), cell carrier number (Arfcn), and physical cell identifier (PCI). These key fields are crucial for building a high-precision 5G positioning fingerprint database.

[0069] like Figure 3 The 4G MR data acquisition specification (draft for comments) shown not only includes the aforementioned key measurement information, but also the high-precision latitude and longitude location information of the MDT reported by the samples. The fusion of location information and 5G inter-system measurement information lays a solid foundation for training a pure 5G high-precision fingerprint database.

[0070] Figure 4 This presentation showcases sample examples of MR measurements taken from 4G data for 5G, providing a clear visual representation of the various measurement data formats. These sample data serve as crucial input for the subsequent processing of the 5G positioning fingerprint database.

[0071] Based on heterogeneous system measurement, this embodiment of the disclosure achieves 5G high-precision positioning, and the overall process is as follows: Figure 5 Specifically:

[0072] In step S101, the 4G / 5G interoperability and periodic inter-system measurement mechanism defined by the 3GPP standard is utilized. When the terminal is in a 4G connection state, by configuring the measurement gap, the signals of surrounding 5G neighboring cells (such as SS-RSRP, SS-RSRQ) are periodically measured, and these measurement results, along with the high-precision latitude and longitude provided by the 4G MDT, are reported to the 4G MR. 4G MR data files are periodically collected from the operation and maintenance center through the backend server. The files are parsed to extract key fields, including: timestamp, latitude and longitude provided by the MDT function, ECI of the serving 4G cell, and 5G neighboring cell measurement information. Subsequently, the data is cleaned, retaining only data entries that simultaneously contain valid high-precision latitude and longitude and at least one 5G neighboring cell measurement record, forming the initial sample dataset.

[0073] Generally, 4G MR only contains the carrier frequency and PCI of the 5G cell, but not the unique identifier (NCI) of the 5G cell. To solve this problem, engineering parameter files for both 4G and 5G can be used. The 4G engineering parameters include the 4G cell identifier (ECI, equal to enbid×256+cellid) and key information such as the cell's installation location (latitude and longitude). The 5G engineering parameters include the 5G cell identifier (NCI), the cell's installation location (latitude and longitude), the cell carrier frequency, and the PCI.

[0074] By loading the latest 4G and 5G engineering parameter tables (usually databases or spreadsheet files), and analyzing each piece of sample data based on the engineering parameter file, and adhering to the network planning principle that "within the same 4G cell coverage area, the same set of 5G frequency points and PCIs should correspond to a unique 5G cell," the candidate 5G cell closest to the 4G cell is selected. This establishes a correlation between 4G and 5G cells, namely, establishing a mapping relationship from 4G cell ECI, 5G frequency point, and 5G PCI to 5G cell NCI, thereby identifying the 5G cell's NCI. This solves the 5G cell identification problem. It accurately maps non-unique (5G frequency point, PCI) combinations to unique NCIs, laying the foundation for constructing accurate 5G signal fingerprints.

[0075] For each sample data point, the generated mapping table is used to convert the (Arfcn (Absolute Radio Frequency Channel Number), PCI) of each record in its 5G neighbor cell measurement list to the corresponding NCI. Next, all 5G neighbor cells with identified NCIs are sorted in descending order based on their RSRP values, with the cell with the highest RSRP becoming the primary serving cell and the rest becoming neighbor cells. Finally, the MDT latitude and longitude coordinates from this sample data are used as location tags and combined with the sorted 5G cell signal characteristics (NCI, RSRP, RSRQ, etc.) to form a standard 5G sampling point. The format of this sampling point is equivalent to an ideal 5G MDT data point.

[0076] All generated 5G sampling point sets were divided into training and testing sets. A machine learning framework was used, with 5G cell signal features from the sampling points (such as the primary serving cell's NCI, primary serving cell's RSRP, and the NCI and RSRP of each neighboring cell) as input features X, and the corresponding latitude and longitude as output labels Y, to train a regression model. Model parameters were adjusted through cross-validation until the model's positioning error on the test set reached the expected target (e.g., 50 meters), thus completing the construction of the 5G high-precision positioning fingerprint database.

[0077] When positioning of a terminal is required, the network side obtains the 5G MR report reported by the terminal. The 5G signal features in the report are extracted, and the machine learning positioning model trained in step S104 is used to make predictions, thereby obtaining the estimated latitude and longitude coordinates of the terminal and completing high-precision positioning.

[0078] This disclosure utilizes existing MDT measurement report data from 4G networks and, through a series of data processing and correlation methods, indirectly assigns high-precision location tags to 5G signal characteristics, thereby successfully constructing a 5G high-precision positioning fingerprint database. This improves the passive positioning accuracy of 5G networks from the hundreds of meters (100-150 meters) to the tens of meters level, effectively solving the high-precision positioning problem caused by the lack of MDT data in 5G networks and significantly improving positioning accuracy. Moreover, the solution requires no modification to existing 5G network hardware and does not require waiting for the widespread adoption of 5G terminal MDT functionality; instead, it fully utilizes the data resources of the already extensively covered and maturely operating 4G networks. By mining underutilized 5G neighbor cell measurement information in 4G MR, the data value of 4G networks is extended to 5G networks, achieving a low-cost, high-efficiency technological upgrade. Meanwhile, the NCI identification method based on engineering parameters and geographical distance constraints, according to network planning principles, can accurately map measurement information to a unique 5G cell through association query and distance calculation, fundamentally avoiding positioning errors caused by cell identifier confusion, laying a solid data foundation for training a high-precision positioning model, and ensuring the accuracy and reliability of the constructed fingerprint database.

[0079] Furthermore, the acquisition of sample data containing high-precision location information and 5G neighbor cell measurement information includes:

[0080] Parse the raw XML format data of 4G MR;

[0081] Extract timestamps, latitude and longitude, 4G cell ECI, and 5G neighbor cell measurement information from the data;

[0082] Data that simultaneously contains valid MDT high-precision latitude and longitude coordinates and at least one 5G neighbor cell measurement information is filtered out and used as the sample data.

[0083] The original 4G MR format is XML, which needs to be parsed before the required information can be extracted. The parsing process is as follows: First, a Document Builder Factory is established, then a Document Builder is created, followed by the creation of a Document and the reading of the XML file through the file path to be parsed. After creating the Node List, the XML information is read, and the timestamp, precision, latitude, 4G network element identifiers enbid and cellid, 4G primary serving cell measurement information, 4G neighbor cell measurement information, and 5G neighbor cell measurement information are extracted and saved.

[0084] Only data containing MDT latitude and longitude coordinates and 5G neighbor cell measurement information is effective for training a 5G positioning fingerprint database. This data includes high-precision location information and 5G neighbor cell measurement results, laying the foundation for building an accurate 5G positioning fingerprint database. By filtering out this data, interference from irrelevant data can be reduced, improving the efficiency and accuracy of fingerprint database training.

[0085] The process of acquiring sample data containing high-precision location information and 5G neighbor cell measurement information is explained below:

[0086] Parsing the raw XML format data of 4G MR: The Operations Support System (OSS) or network management system generates 4G MR data files periodically (e.g., every 15 minutes). These files typically follow the format defined by the 3GPP standard and are stored and transmitted in XML format. Parsing technology: On the server backend, mature XML parsing technologies are used for processing. For example, in a Java environment, Document Builder Factory and Document Builder can be used to create a DOM (Document Object Model) tree; in a Python environment, the xml.etree.ElementTree library can be used. The purpose of the parsing process is to transform structured XML text data into in-memory objects that are easy for computer programs to process and access.

[0087] Extracting Key Data Fields: After successful parsing, the program will traverse the XML document structure, locate and extract the values ​​of the following key fields: Timestamp: Records the time when the measurement report was generated, used for data traceability and timeliness management. Latitude and Longitude: This is the core source of high-precision location information. This latitude and longitude data is obtained by the terminal's MDT function through the built-in GPS or Beidou module when collecting measurement reports. Its accuracy is usually at the meter level, far exceeding the positioning accuracy based on cell identifiers. 4G Serving Cell ECI: This field is the "anchor point" for data association. It uniquely identifies the 4G cell that the terminal is connected to when reporting this MR. 5G Neighbor Cell Measurement Information: The program needs to extract a list of 5G neighbor cells. For each neighbor cell in the list, at least the following needs to be extracted: Frequency Point: Indicates the carrier frequency used by the 5G neighbor cell. PCI: Physical Cell Identifier, used to distinguish different cells within a local area. RSRP / RSRQ: Signal strength and quality measurements of the 5G neighbor cell.

[0088] Filtering and selecting valid sample data: Not all parsed MR data is useful for building a 5G fingerprint database, so strict filtering is required: First condition: It must contain valid high-precision latitude and longitude coordinates (MDT). Invalid data with empty latitude and longitude fields, values ​​of zero, or values ​​clearly outside a reasonable range (e.g., not within the country's administrative region) must be excluded. This condition ensures that each sample has an accurate "location label." Second condition: It must contain at least one valid 5G neighbor cell measurement information. Records with empty 5G neighbor cell lists or invalid key measurements such as RSRP must be excluded. This condition ensures that the sample contains learnable "5G signal characteristics." Only data entries that meet both of these conditions will be retained as "sample data" for subsequent processes. This data cleaning step is crucial, as it directly determines the quality of the subsequently constructed 5G sampling point set, thus affecting the accuracy of the final positioning model.

[0089] The steps described in this embodiment standardize the data source processing: by parsing unstructured XML data into structured data objects, the foundation for subsequent automated processing is laid. Precise field extraction and rigorous data filtering ensure that the data used to train the model possesses both the key attributes of "high-precision location" and "effective 5G signal," thereby improving the reliability of the fingerprint database from the source. Filtering out a large amount of irrelevant data significantly reduces the amount of data that needs to be processed in subsequent steps, improving the overall efficiency of the solution and making it more suitable for processing massive amounts of carrier-grade data, thus enhancing processing efficiency.

[0090] Furthermore, establishing the mapping relationship between 4G cell ECI, 5G frequency points, and 5G PCI to 5G cell NCI includes:

[0091] Based on the 4G engineering parameters, obtain the ECI of the 4G cell and the latitude and longitude of its installation location;

[0092] Based on the 5G engineering parameters, obtain the NCI, 5G frequency point, PCI, and latitude and longitude of the installation location of the 5G cell;

[0093] For a specific 4G cell, within a preset geographical distance range around it, search for 5G candidate cells in the 5G engineering parameters that have the same 5G frequency and PCI as the 5G neighbor cell measurement information contained in a certain MR data from that specific 4G cell.

[0094] From the 5G candidate cells, select the 5G cell that is closest to the specific 4G cell, and establish a mapping relationship from the ECI of the specific 4G cell, the 5G frequency point, and the PCI to the NCI of the closest 5G cell.

[0095] The steps to establish the mapping relationship aim to solve the key technical challenge of "how to accurately map the non-unique 5G cell identifier (frequency point + PCI) in 4G MR to the globally unique 5G cell identifier (NCI)".

[0096] In constructing 5G sampling points, the first step is to use engineering parameter files to identify the unique identifier (NCI) of each 5G cell. Generally, there are only a few frequency points for 5G, and the PCI value range is [0, 1023], resulting in very limited combinations of frequency points and PCIs. During the planning phase, the PCI reuse distance must be strictly controlled (generally greater than 3000 meters) to prevent confusion and interference between cells.

[0097] Generally, 4G MR only contains the carrier frequency and PCI of the 5G cell, but not the cell's unique identifier (NCI). To solve this problem, we can utilize the engineering parameter files for both 4G and 5G. The 4G engineering parameters contain key information such as the 4G cell identifier (ECI, equal to enbid×256+cellid) and the cell's installation location (latitude and longitude). The 5G engineering parameters contain key information such as the 5G cell identifier (NCI), cell installation location (latitude and longitude), cell carrier frequency, and PCI. By analyzing these engineering parameter files, we can establish a correlation between 4G and 5G cells, thereby identifying the 5G cell's NCI.

[0098] Under normal circumstances, for the same 4G cell, the same set of 5G frequency points and PCI measured within its coverage area must belong to the same 5G cell (otherwise, confusion and interference between 5G cells would occur). Therefore, for a 4G cell, it is possible to query the 5G cells within a 3-kilometer radius of it.

[0099] Calculate the distance to the 5G cell, and group the cells according to 5G frequency and PCI. Find the 5G cell closest to the 4G cell in each group, and designate it as the 5G neighbor cell for that 4G cell, 5G frequency, and PCI. Following this method, a quick reference table of 4G and 5G neighbor cell relationships can be established for the entire network. The 5G cell identifier (NCI) can be quickly found using ECI, 5G frequency, and 5G PCI to complete the identification of 5G cells.

[0100] The specific implementation process is explained below:

[0101] 1. Data Preparation: Loading Project Parameters

[0102] 4G Engineering Parameters: This file typically contains key information for each 4G cell, such as ECI, base station name, latitude and longitude, antenna height, azimuth, and downtilt angle. In this embodiment, the core focus is on utilizing the ECI and latitude / longitude fields. 5G Engineering Parameters: This file contains key information for each 5G cell, such as NCI, base station name, latitude and longitude, frequency (Arfcn), and PCI. In this embodiment, the core focus is on utilizing the NCI, latitude / longitude, frequency, and PCI fields. These engineering parameters are the foundation of network planning and optimization, maintained by the operator's network department, and possess high authority and accuracy.

[0103] 2. Determine the search center and scope.

[0104] Specific 4G cell: The “specific 4G cell” here refers to the serving cell corresponding to the sample data extracted in step S101, whose ECI is known.

[0105] Obtain center coordinates: By querying the 4G engineering parameters and using the ECI as an index, the precise installation location latitude and longitude (lat4G, lon4G) of this 4G cell can be obtained. These coordinates will serve as the center point for subsequent searches.

[0106] Preset geographical distance range: A circular search area is defined with the center point of the precise installation location of the 4G cell as the center. The range setting needs to balance accuracy and coverage, with 3 kilometers being the preferred option. This range is based on the typical radius of cellular network coverage and the network planning principle of avoiding PCI confusion, ensuring coverage of all potential 5G neighboring cells that the terminal may measure under this 4G cell.

[0107] 3. Searching and matching candidate cells

[0108] At this point, it's necessary to address a specific MR data entry from a particular 4G cell that reports a 5G neighbor cell, identified as (ArfcnX, PCIY). Initial screening: Using 5G engineering parameters, first filter out all 5G cells whose geographical locations fall within the aforementioned preset distance range. Precise matching: From the geographical screening results, further search for 5G cells that simultaneously meet the following two conditions: frequency equal to ArfcnX, and PCI equal to PCIY. All cells meeting these conditions constitute a "5G candidate cell set." Under ideal network planning, this set should contain only one cell. However, in complex real-world environments, multiple candidates may exist (e.g., due to planning flaws or newly added base stations not being optimized in a timely manner).

[0109] 4. Adjudication and Mapping

[0110] Decision-making principle: Based on the fundamental network planning principle that "within the coverage area of ​​a single 4G cell, the same pair of (frequency points, PCI) should uniquely correspond to one 5G cell." If multiple candidates exist, the one with the strongest signal (i.e., the closest distance) is the correct option. Decision execution: Calculate the distance between the specific 4G cell location (lat4G, lon4G) and the location of every 5G cell in the candidate set. Precise calculations are performed using spherical distance formulas (such as the Haversine formula).

[0111] Establish the mapping: Select the nearest 5G cell and use its globally unique identifier (NCI) as the answer. At this point, a robust mapping relationship is established from 4G cell ECI, 5G frequency and 5G PCI to 5G cell NCI.

[0112] Generate mapping table: Perform this operation on all 4G cells and all combinations of (5G frequency points, 5G PCI) appearing in their MR data across the entire network, and finally generate a quick reference table of 4G-5G neighbor cell relationship mapping covering the entire network.

[0113] By creatively introducing geographical distance constraints as a decision criterion through the above steps, non-unique (frequency points, PCI) are mapped to unique NCIs, overcoming the core technical obstacle of converting 4G data to 5G data and solving the ambiguity problem of identifier mapping. Furthermore, the mapping process is based on solid network planning principles, maximizing the accuracy of the mapping relationship and providing a reliable data foundation for building a high-precision 5G positioning fingerprint database. This fundamentally avoids positioning deviations caused by cell identification errors, ensuring data accuracy. The entire process relies on standard engineering parameters maintained by the operator's existing network, ensuring reliable data sources, clear algorithm logic, and automated execution. It is highly suitable for processing large-scale network data and possesses significant engineering application value.

[0114] Furthermore, the ranking of 5G neighboring cells based on signal strength includes:

[0115] All 5G neighboring cells measured in the sample data are sorted in descending order according to their RSRP values;

[0116] The 5G neighbor cell with the highest RSRP is designated as the primary 5G cell, and the remaining 5G neighbor cells are designated as neighboring 5G cells.

[0117] Sort 5G neighboring cells as a key preprocessing step in building high-quality 5G sampling points. Its purpose is to simulate the actual connection status of the terminal in the 5G network and to build a "signal fingerprint" that can most effectively represent the current location of the terminal.

[0118] In constructing a complete 5G sampling point, the 5G neighboring cells are first arranged in reverse order according to their Reference Signal Received Power (RSRP). This step aims to sort the neighboring cells based on signal strength to determine the primary serving cell and other neighboring cells. For example, in the detailed explanation of 5G NSA handover air interface signaling, the handover process for 5G terminals in different scenarios is mentioned, which involves judging and selecting cell signal strength. By arranging the RSRPs of 5G neighboring cells in reverse order, the cell with the strongest signal can be quickly identified as the primary serving cell, while other cells are designated as neighboring cells.

[0119] This sorting method ensures that cells with higher signal strength are prioritized during subsequent location processes, improving the accuracy and reliability of positioning. Simultaneously, it better reflects the coverage and signal quality distribution of the 5G network, providing a reference for emergency response and disaster relief.

[0120] The following is a detailed explanation of this process:

[0121] 1. Sorting criteria: Received Reference Signal Power (RSRP)

[0122] RSRP Definition: RSRP is a key measurement characterizing signal strength in LTE and 5G networks. It is defined as the linear average of the received power of resource elements (REs) carrying a cell-specific reference signal within the measurement bandwidth. Simply put, the RSRP value directly reflects how strong the signal received by the terminal from a particular cell is, and the unit is usually dBm. RSRP is the most stable and direct indicator of downlink signal strength. Signal strength is closely related to distance and path loss; therefore, the cell with the strongest RSRP is usually the geographically closest cell to the terminal and the most likely to establish a primary connection. Ranking by RSRP most accurately reflects the wireless environment in which the terminal is located.

[0123] 2. Sorting operation: Descending order

[0124] Operation Process: In the data processing program, all 5G neighboring cells that have completed NCI identification (usually a list) in a sample dataset are processed. A sorting algorithm (such as quicksort) is used, with the RSRP value of each neighboring cell as the sorting key, to sort them in descending order. Result: After sorting, the first element in the list is the 5G neighboring cell with the largest RSRP value, the second element is the one with the second largest RSRP value, and so on.

[0125] 3. Role Determination: Main Server and Neighboring Servers

[0126] Determining the primary 5G cell: The first 5G neighbor cell in the sorted list, i.e., the cell with the highest RSRP, is officially determined as the primary 5G cell for this sampling point. In the subsequently constructed feature vector, the various parameters of this cell (NCI, RSRP, RSRQ, etc.) will play a core role.

[0127] Determine the list of neighboring 5G cells: All 5G neighboring cells starting from the second element in the sorted list are collectively referred to as neighboring 5G cells. They constitute the auxiliary signal environment characteristics of the terminal's current location. During modeling, the few neighboring cells with the strongest signals are usually retained (e.g., the first 3-6) to balance feature dimensionality and information content.

[0128] By sorting by signal strength, the system accurately reproduces the 5G network status perceived by the terminal at that location, ensuring that the constructed "5G sampling points" are truly and uniquely mapped to their geographical locations. This lays a reliable foundation for machine learning models to learn the "signal-location" relationship. Distinguishing between the "dominant serving cell" and "neighboring cells" helps the machine learning model identify the signal sources that contribute most to positioning. The model can prioritize learning the strong correlation between strong signal cells and location, thereby improving learning efficiency and positioning accuracy. After sorting, the cell signal features in each 5G sampling point have a fixed order (dominant serving cell first, neighboring cells arranged in order of strength). This ensures that the feature vector structure input to the machine learning model is uniform and standardized, avoiding model confusion or training difficulties caused by random input order, and forming standardized feature vectors. Even if the signal of the dominant serving cell fluctuates due to temporary obstruction, its powerful neighboring cell list signal patterns can still provide the model with sufficient auxiliary information for location determination, thereby improving the stability and reliability of the entire positioning system in complex wireless environments and enhancing positioning robustness.

[0129] Furthermore, the 5G sampling point includes the following fields: timestamp, longitude, latitude, primary serving cell NCI, primary serving cell frequency, primary serving cell PCI, primary serving cell RSRP, primary serving cell RSRQ, and at least one neighboring cell NCI, neighboring cell frequency, neighboring cell PCI, neighboring cell RSRP, and neighboring cell RSRQ.

[0130] Each 5G sampling point is essentially a standardized training sample that can be directly used by machine learning models.

[0131] In 5G sampling points, 1. Longitude and latitude are the core value of the sampling points, directly derived from the high-precision GPS / BeiDou coordinates provided by 4G MDT data. In machine learning, this will serve as the label for supervised learning, i.e., the target variable that the model needs to predict. Its accuracy directly determines the upper limit of the model. Timestamp: Records the collection time of the sample data. It can be used for data management, such as dividing the training set and test set by time, or troubleshooting model failure caused by network changes (such as base station relocation). 2. The primary serving cell feature fields are the "core anchor points" of positioning. Primary Serving Cell NCI: Globally unique 5G cell identifier. This is the most important category feature, allowing the model to directly learn the strong correlation between a specific cell and its coverage area. Primary Serving Cell Frequency: Indicates the carrier frequency used by the primary serving cell. Different frequency bands (such as 2.6GHz, 3.5GHz) have different propagation characteristics (coverage range, penetration), which are key features for the model to distinguish signal propagation behavior. Primary Serving Cell PCI: Physical Cell Identifier, as a supplement to NCI, provides additional category information. Primary Serving Cell RSRP: Signal strength of the primary serving cell. The most crucial continuous numerical feature for determining the distance between the terminal and the base station is the RSRP. A larger value generally indicates a closer distance or less obstruction. The primary serving cell's RSRQ (Representative Signal Quality Q) integrates signal strength and interference levels. It provides supplementary information beyond RSRP, helping to determine the purity of the signal environment. The neighbor cell list feature fields form the "context" of the location, including at least one neighbor cell's NCI, frequency, PCI, RSRP, and RSRQ. This part defines the characteristics of the neighbor cell list. Each neighbor cell contains information of the same dimensions as the primary serving cell. A single cell's signal can only provide limited location information (possibly anywhere within the cell's coverage area). However, the combination of signals from multiple neighbor cells forms a unique "signal fingerprint," greatly improving the uniqueness of the location. Implementation: Typically, the strongest N (e.g., N=6) neighbor cell information is retained in descending order of RSRP. For samples with fewer than N neighbor cells, preset values ​​(such as 0 or very small values) can be used to fill the gaps, keeping the feature vector length fixed.

[0132] By clearly defining the data structure of 5G sampling points, data standardization was achieved: 4G MR data from diverse sources and with complex formats were transformed into standardized 5G sampling points with a unified structure and clear fields, laying the foundation for large-scale, automated model training. Furthermore, a rich feature vector was constructed: this structure includes identifier features (NCI, PCI), numerical features (RSRP, RSRQ), configuration features (frequency points), and spatial relationship features (neighbor cell list and relative signal strength), providing machine learning models with comprehensive information describing the wireless signal environment, enabling the models to learn extremely complex "signal-location" mapping relationships.

[0133] Furthermore, the machine learning localization model is the LightGBM model.

[0134] The LightGBM model was specifically chosen for the 5G high-precision positioning task disclosed herein due to its comprehensive advantages in high-dimensional data processing, computational efficiency, prediction accuracy, and friendliness to class features. LightGBM can efficiently process high-dimensional feature vectors composed of signals from multiple neighboring cells. It significantly improves processing efficiency through feature parallelization and unique one-sided gradient sampling and mutually exclusive feature binding techniques.

[0135] The LightGBM machine learning algorithm is used to train a localization model, employing latitude and longitude as location labels and 5G cell network signal indicators as features. During training, LightGBM continuously adjusts the model parameters to minimize the error between the model's predicted location and the actual location. Through learning from a large number of samples, the model gradually masters the relationship between 5G cell network signal indicators and location, thereby achieving localization of unknown locations.

[0136] The specific training process of the LightGBM model includes:

[0137] Data preparation: The entire 5G sampling point set is randomly divided into training, validation, and test sets. Numerical features (such as RSRP) are standardized, and categorical features (such as NCI) are one-hot encoded or subjected to efficient categorical feature histogram statistics.

[0138] Model training: The model is trained using a LightGBM regressor. The training process involves iteratively building multiple decision trees, each of which learns to correct the prediction error (residual) of the previous tree. The final model is a weighted combination of these weak learners.

[0139] Loss function: The mean squared error or mean absolute error is usually used as the loss function to minimize the error between the latitude and longitude predicted by the model and the actual location.

[0140] Hyperparameter tuning: Key hyperparameters of LightGBM, such as learning rate, maximum tree depth, and number of leaf nodes, are optimized using techniques such as cross-validation to prevent overfitting and achieve optimal performance.

[0141] After training is complete, the final LightGBM model is saved, which is the constructed "5G high-precision positioning fingerprint database".

[0142] After the model is trained, for the MR that needs to be located, the corresponding location label is located based on the signal indicators of the primary serving cell and neighboring cells, thus completing the final location.

[0143] When a new 5G MR needs to be located, it is first converted into a feature vector according to the same data processing flow (NCI recognition, feature sorting, etc.), and then input into the trained LightGBM model.

[0144] The model will directly output a two-dimensional coordinate (longitude, latitude) as the estimated location of the terminal.

[0145] This disclosed embodiment innovatively utilizes MDT (Minimum Drive Test) data from existing 4G networks to successfully solve the industry problem of low passive positioning accuracy in 5G networks due to the lack of high-precision location tag data. Its overall technical effect is reflected in the following: without changing the existing 5G network infrastructure and terminals, it achieves an order-of-magnitude improvement in passive positioning accuracy on the 5G network side, from the hundreds of meters (100-150 meters) to the tens of meters level. The solution effectively combines 5G neighbor cell signal information contained in 4G measurement reports with high-precision latitude and longitude tags through a complete data conversion process, constructing a high-quality 5G positioning fingerprint database that can be used for machine learning. This technical approach is not only low-cost and highly implementable, but also breaks through the data source bottleneck for 5G high-precision positioning, providing reliable core location capability support for the deep application of 5G technology in key areas such as emergency communication, precise navigation, network optimization, and smart cities, significantly enhancing the service capabilities and industrial value of 5G networks.

[0146] Example 2

[0147] Embodiment 2 of this disclosure also provides a 5G high-precision positioning system based on heterogeneous system measurement, such as... Figure 6 As shown, the system includes:

[0148] The data acquisition module 11 is configured to acquire sample data containing high-precision location information and 5G neighbor cell measurement information from the MR measurement report data of the 4G network. The 5G neighbor cell measurement information includes at least the frequency point and physical cell identifier (PCI) of the 5G neighbor cell.

[0149] The cell identifier mapping module 12 is configured to establish a mapping relationship between the 4G cell identifier ECI, the 5G frequency point and the 5G PCI to the 5G cell globally unique identifier NCI based on the 4G engineering parameters and 5G engineering parameters.

[0150] The sampling point construction module 13 is configured to identify the 5G cell NCI corresponding to the 5G neighbor cell measurement information in each sample data using the mapping relationship, and sort the 5G neighbor cells based on the signal strength to form a 5G sampling point containing the high-precision location information, the primary 5G cell and the neighboring 5G cells with complete signal characteristics.

[0151] The model training module 14 is configured to use all 5G sampling points, with the high-precision location information as the location label and the corresponding 5G cell signal features as the input features, to train a machine learning positioning model in order to build a 5G high-precision positioning fingerprint database.

[0152] The positioning module 15 is configured to, in response to receiving a 5G measurement report from the terminal to be positioned, match or predict the location of the terminal to be positioned from the 5G high-precision positioning fingerprint database.

[0153] Furthermore, the data acquisition module 11 is specifically configured as follows:

[0154] Parse the raw XML format data of 4G MR;

[0155] Extract timestamps, latitude and longitude, 4G cell ECI, and 5G neighbor cell measurement information from the data;

[0156] Data that simultaneously contains valid MDT high-precision latitude and longitude coordinates and at least one 5G neighbor cell measurement information is filtered out and used as the sample data.

[0157] Furthermore, the cell identifier mapping module 12 is specifically configured as follows:

[0158] Based on the 4G engineering parameters, obtain the ECI of the 4G cell and the latitude and longitude of its installation location;

[0159] Based on the 5G engineering parameters, obtain the NCI, 5G frequency point, PCI, and latitude and longitude of the installation location of the 5G cell;

[0160] For a specific 4G cell, within a preset geographical distance range around it, search for 5G candidate cells in the 5G engineering parameters that have the same 5G frequency and PCI as the 5G neighbor cell measurement information contained in a certain MR data from that specific 4G cell.

[0161] From the 5G candidate cells, select the 5G cell that is closest to the specific 4G cell, and establish a mapping relationship from the ECI of the specific 4G cell, the 5G frequency point, and the PCI to the NCI of the closest 5G cell.

[0162] Furthermore, the sampling point construction module 13 is specifically configured as follows:

[0163] All 5G neighboring cells measured in the sample data are sorted in descending order according to the value of the Reference Signal Received Power (RSRP).

[0164] The 5G neighbor cell with the highest RSRP is designated as the primary 5G cell, and the remaining 5G neighbor cells are designated as neighboring 5G cells.

[0165] Furthermore, the 5G sampling point includes the following fields: timestamp, longitude, latitude, primary serving cell NCI, primary serving cell frequency, primary serving cell PCI, primary serving cell RSRP, primary serving cell reference signal reception quality RSRQ, at least one neighboring cell NCI, neighboring cell frequency, neighboring cell PCI, neighboring cell RSRP, and neighboring cell RSRQ.

[0166] Furthermore, the machine learning localization model is the LightGBM model.

[0167] The 5G high-precision positioning system based on heterogeneous system measurement in this embodiment is used to implement the 5G high-precision positioning method based on heterogeneous system measurement in the first embodiment of the method. Therefore, the description is relatively simple. For details, please refer to the relevant descriptions in the previous method embodiments, which will not be repeated here.

[0168] In addition, such as Figure 7 As shown, Embodiment 3 of this disclosure also provides an electronic device, including a memory 100 and a processor 200. The memory 100 stores a computer program. When the processor 200 runs the computer program stored in the memory 100, the processor 200 executes the various possible methods described above.

[0169] The memory 100 is connected to the processor 200. The memory 100 can be a flash memory, a read-only memory, or another type of memory. The processor 200 can be a central processing unit or a microcontroller.

[0170] Furthermore, embodiments of this disclosure also provide a computer-readable storage medium storing a computer program, which is executed by a processor using the various possible methods described above.

[0171] The computer-readable storage medium includes volatile or non-volatile, removable or non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, computer program modules or other data). Computer-readable storage media include, but are not limited to, RAM (Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), flash memory or other memory technologies, CD-ROM (Compact Disc Read-Only Memory), Digital Video Disc (DVD) or other optical disc storage, cartridges, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer.

[0172] It is understood that the above embodiments are merely exemplary embodiments used to illustrate the principles of this disclosure, and this disclosure is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and substance of this disclosure, and these modifications and improvements are also considered to be within the scope of protection of this disclosure.

Claims

1. A 5G high-precision positioning method based on inter-system measurement, characterized in that, The method comprises: obtaining sample data containing high-precision position information and 5G adjacent area measurement information from 4G network measurement report (MR) data, wherein the 5G adjacent area measurement information at least includes the frequency point and physical cell identifier (PCI) of the 5G adjacent area; establishing a mapping relationship from the E-UTRAN cell identifier (ECI) of a 4G cell, the frequency point and the PCI of a 5G cell to the NR cell identifier (NCI) of the 5G cell based on 4G engineering parameters and 5G engineering parameters; for each piece of sample data, identifying the NCI of the 5G cell corresponding to the 5G adjacent area measurement information in the sample data by using the mapping relationship, and sorting the 5G adjacent areas based on signal strength to form a 5G sampling point containing the high-precision position information, the main 5G cell and the adjacent 5G cells and the complete signal features of the 5G cells; training a machine learning positioning model by using all the 5G sampling points, taking the high-precision position information as a position label and taking the corresponding 5G cell signal features as input features, to construct a 5G high-precision positioning fingerprint library; in response to receiving a 5G measurement report of a terminal to be positioned, matching or predicting the position of the terminal to be positioned from the 5G high-precision positioning fingerprint library.

2. The method of claim 1, wherein, The obtaining of the sample data containing high-precision position information and 5G adjacent area measurement information comprises: parsing the original XML format data of the 4G MR; extracting the time stamp, latitude and longitude, 4G cell ECI and 5G adjacent area measurement information in the data; filtering and screening out data containing valid minimum drive test (MDT) high-precision latitude and longitude and at least one piece of 5G adjacent area measurement information as the sample data.

3. The method of claim 1, wherein, The establishing of the mapping relationship from the ECI of a 4G cell, the frequency point and the PCI of a 5G cell to the NCI of the 5G cell comprises: obtaining the ECI of a 4G cell and the latitude and longitude of the installation position of the 4G cell according to the 4G engineering parameters; obtaining the NCI of a 5G cell, the frequency point and the PCI of the 5G cell and the latitude and longitude of the installation position of the 5G cell according to the 5G engineering parameters; for a specific 4G cell, searching for 5G candidate cells having the same frequency point and PCI as the 5G adjacent area measurement information contained in a piece of MR data from the specific 4G cell within a preset geographical distance range around the specific 4G cell; from the 5G candidate cells, selecting the 5G cell closest to the specific 4G cell to establish a mapping relationship from the ECI of the specific 4G cell, the frequency point and the PCI to the NCI of the 5G cell closest to the specific 4G cell.

4. The method of claim 1, wherein, The sorting of the 5G adjacent areas based on signal strength comprises: arranging all the 5G adjacent areas measured in the sample data in descending order according to the value of reference signal received power (RSRP); determining the 5G adjacent area with the strongest RSRP as the main 5G cell and the remaining 5G adjacent areas as the adjacent 5G cells.

5. The method of claim 4, wherein, The 5G sampling point comprises the following fields: timestamp, longitude, latitude, main service cell NCI, main service cell frequency point, main service cell PCI, main service cell RSRP, main service cell reference signal received quality RSRQ, NCI of at least one adjacent cell, adjacent cell frequency point, adjacent cell PCI, adjacent cell RSRP, and adjacent cell RSRQ.

6. The method of claim 1, wherein, The machine learning positioning model is a LightGBM model. 7.A 5G high-precision positioning system based on inter-system measurement, characterized in that, The system comprises: a data acquisition module configured to acquire sample data comprising high-precision position information and 5G adjacent cell measurement information from MR data of a 4G network, wherein the 5G adjacent cell measurement information at least comprises frequency points and PCIs of 5G adjacent cells; a cell identification mapping module configured to establish mapping relationships from 4G cell identifications ECI, 5G frequency points, and 5G PCIs to 5G cell NCIs based on 4G engineering parameters and 5G engineering parameters; a sampling point construction module configured to, for each piece of sample data, identify 5G cell NCIs corresponding to 5G adjacent cell measurement information in the sample data by using the mapping relationships, and sort 5G adjacent cells based on signal strengths to form 5G sampling points comprising complete signal features of the high-precision position information, main service 5G cell, and adjacent 5G cells; a model training module configured to train a machine learning positioning model by using all 5G sampling points, taking the high-precision position information as position labels and corresponding 5G cell signal features as input features, to construct a 5G high-precision positioning fingerprint library; a positioning module configured to match or predict a position of a terminal to be positioned from the 5G high-precision positioning fingerprint library in response to receiving a 5G measurement report of the terminal to be positioned.

8. The system of claim 7, wherein, The data acquisition module is specifically configured to: analyze original XML format data of 4G MR; extract timestamps, longitude and latitude, 4G cell ECI, and 5G adjacent cell measurement information from the data; filter and screen out data comprising valid MDT high-precision longitude and latitude and at least one 5G adjacent cell measurement information as the sample data.

9. An electronic device, comprising: A computer program is stored in a memory and a processor, and when the processor executes the computer program stored in the memory, the processor executes the 5G high-precision positioning method based on heterogeneous system measurement according to any one of claims 1-6.

10. A computer-readable storage medium, characterized in that, A computer program is stored on a computer readable storage medium, and when the computer program is executed by a processor, the 5G high-precision positioning method based on heterogeneous system measurement according to any one of claims 1-6 is implemented.

Images