Mobile path recognition method, recognition device, electronic equipment and readable storage medium

Abstract

The present disclosure provides a mobile path identification method, an identification device, an electronic device and a readable storage medium, and relates to the technical field of wireless communication. The user mobile path identification method comprises: performing an association operation on the collected MR data and the corresponding XDR data based on an association model to obtain MR association data, wherein the MR data is obtained based on the communication report information of the user terminal; performing a matching positioning operation on the MR association data and the fingerprints in the MR fingerprint library to obtain a specified size of the matched abnormal grid area; mapping the obtained real-time travel path data to the abnormal grid area to obtain a candidate path of the user terminal movement; and identifying the actual movement path of the user terminal from the candidate path based on an optimized hidden Markov model (HMM). Through the technical scheme of the present disclosure, the actual movement path of the user terminal can be reliably identified.

Description

Technical Field

[0001] This disclosure relates to the field of wireless communication technology, and in particular to a user mobility path identification method, a user mobility path identification device, an electronic device, and a computer-readable storage medium. Background Technology

[0002] Users often file complaints about communication network anomalies while moving on the road. Currently, the complaint handling process is almost entirely manual. Due to the difficulty in obtaining the user's accurate location and the rapid changes in the wireless signal propagation environment during user movement, resulting in significant errors, network optimization engineers handling complaints often struggle to obtain the user's precise movement path. They can only resort to blind testing (DT) or CQT (Call Quality Test) over a large area near the complaint location, leading to poor complaint handling effectiveness.

[0003] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0004] The purpose of this disclosure is to provide a user movement path identification method, a user movement path identification device, an electronic device, and a computer-readable storage medium, which at least to some extent overcome the problem of low detection accuracy of targets with circular trajectories in related technologies.

[0005] Other features and advantages of this disclosure will become apparent from the following detailed description, or may be learned in part from practice of this disclosure.

[0006] According to one aspect of this disclosure, a user movement path identification method is provided, comprising: performing an association operation on collected MR data and corresponding XDR data based on an association model to obtain MR association data, wherein the MR data is obtained based on communication reporting information of a user terminal; performing a matching and positioning operation on the MR association data and fingerprints in an MR fingerprint database to obtain a matched abnormal grid region of a specified size; mapping the acquired real-time travel path data to the abnormal grid region to obtain a candidate path for the user terminal's movement; and identifying the actual movement path of the user terminal from the candidate paths based on an optimized Hidden Markov Model (HMM).

[0007] In some embodiments, the step of performing an association operation between the collected MR data and the corresponding XDR data based on the association model to obtain MR associated data includes: collecting XDR data that is in the same time and area as the MR data; and performing an association operation between the MR data and the XDR data based on the association model to obtain the MR associated data.

[0008] In some embodiments, the acquisition of XDR data that is in the same time and area as the MR data includes: periodically monitoring the acquisition status information of the acquisition server to enable the acquisition server to acquire the MR data and the XDR data under normal conditions, wherein the acquisition status information includes configuration information and data transmission monitoring information during the data acquisition process.

[0009] In some embodiments, the method further includes: determining the user S1AP identifier on the base station side as an associated field, wherein the user S1AP identifier is configured as a non-short code on the base station side.

[0010] In some embodiments, the step of associating the MR data and XDR data based on the association model to obtain the MR-associated data includes: cleaning the MR data and XDR data to obtain cleaned MR data and cleaned XDR data; performing an initial association process on the cleaned MR data and cleaned XDR data based on the association field; recording the MR clean data that was not associated in the initial association process as unassociated data, and performing forward sliding window search and backward sliding window search on the unassociated data based on the acquisition duration to obtain search data; merging and deduplicating the search data to obtain processed search data; performing a secondary association process on the processed search data and cleaned XDR data; and writing the measurement information and location information in the MR data into the data table of the associated XDR data based on the processing results of the initial association process and the secondary association process to obtain the MR-associated data.

[0011] In some embodiments, the step of associating the MR data with the XDR data based on the association model to obtain the MR-associated data further includes: classifying the XDR data based on MME type, HTTP type, and COMMOM type to obtain MME data, HTTP data, and S1_COMMON data; performing an MME_MR association operation on the MR data and the MME data; performing an HTTP_MR association operation on the MR data and the HTTP data; and performing a COMMON_MR association operation on the MR data and the S1_COMMON data to obtain the MR-associated data.

[0012] In some embodiments, the step of performing a matching and positioning operation between the MR association data and fingerprints in the MR fingerprint database to obtain a matched abnormal grid region of a specified size includes: optimizing the K-nearest neighbor (KNN) model based on cross-validation to obtain an optimized KNN model; extracting the primary cell RSRP and neighboring cell RSRP of the user terminal from the MR association data as MR level data; and performing a matching and positioning operation between the MR level data and fingerprints in the MR fingerprint database based on the optimized KNN model to obtain the abnormal grid region of the specified size.

[0013] In some embodiments, optimizing the K-Nearest Neighbor (KNN) model based on cross-validation to obtain an optimized KNN model includes: dividing the obtained MR training set into K parts, designating one part of the K MR training set as the validation set, and designating the other K-1 parts as the training set to perform model training; repeating the model training K times to obtain K KNN models and corresponding K scores; and performing model optimization based on the K KNN models and corresponding K scores to obtain the optimized KNN model.

[0014] In some embodiments, the specified size is greater than or equal to 20m and less than or equal to 25m.

[0015] In some embodiments, the step of matching and locating the MR level data with fingerprints in the MR fingerprint database based on the KNN optimization model to obtain the abnormal grid region of the specified size specifically includes: matching and locating the MR level data with fingerprints in the MR fingerprint database based on the KNN optimization model to obtain candidate grid regions; obtaining a description location based on the identification of the communication reporting information; filtering out invalid grids from the candidate grid regions whose distance from the description location is greater than a distance threshold to obtain a first grid region; and filtering out invalid grids from the first grid region whose coverage is not provided by a primary coverage cell to obtain a second grid region, which serves as the abnormal grid region.

[0016] In some embodiments, before performing matching and localization between the MR level data and the fingerprints in the MR fingerprint database based on the KNN optimization model to obtain the abnormal grid region of the specified size, the method further includes acquiring CQT dialing test data, DT drive test data, MDT mobile device measurement data, and OTT data of multiple grid regions and using them as training data; training the model based on the training data to generate a propagation model; calculating the cell strength of each grid region based on the propagation model; and creating the MR fingerprint database based on the cell strength.

[0017] In some embodiments, mapping the acquired real-time travel path data to the abnormal grid area to obtain a candidate path for the user terminal's movement includes: acquiring terminal MR data of the user terminal at two points close to the reporting time of the communication reporting information; calculating the user's movement rate based on the terminal MR data; matching the user's movement rate with the real-time travel path data; and mapping the matching result to the abnormal grid area to obtain the candidate path.

[0018] In some embodiments, before performing the association operation between the collected MR data and the corresponding XDR data based on the association model, the method further includes: converting the acquired user complaint voice into complaint text to generate the communication reporting information based on the complaint text; and performing natural language processing on the complaint text to identify the descriptive location of the user complaint.

[0019] In some embodiments, the method of identifying the actual movement path of the user terminal from the candidate paths using the optimized Hidden Markov Model (HMM) includes: determining the described location as an observation point and the candidate path as a hidden state to calculate the observation probability of the described location for the candidate path; calculating the transition probability of each candidate path using a heap-optimized Dijkstra algorithm; inputting the observation probability and the transition probability into a Viterbi calculation model to calculate the corresponding Viterbi probability, and searching for the actual movement path from the candidate paths based on the Viterbi probability.

[0020] In some embodiments, calculating the observation probability of the described location for the candidate path includes: associating the MR association data with acquired network performance data and network alarm data based on cell coding to obtain initial sample data; analyzing the impact of the associated network performance data on the MR association data, weighting the MR association data based on the analysis results, and filtering performance sample data from the initial sample data based on the weighted labels; extracting the performance sample data corresponding to a specified alarm type from the performance sample data and using it as target performance data; obtaining the target performance data related to the candidate path, and calculating the observation probability based on the related target performance data.

[0021] In some embodiments, the heap-optimized Dijkstra algorithm for calculating the transition probability of each candidate path includes: calculating the path distance and great circle distance between different candidate paths based on the heap-optimized Dijkstra algorithm; and calculating the transition probability based on the path distance, the great circle distance, and the corresponding time interval between the candidate paths.

[0022] According to another aspect of this disclosure, a user movement path identification device is provided, comprising: an association module, configured to perform an association operation on collected MR data and corresponding XDR data based on an association model to obtain MR association data, wherein the MR data is obtained based on communication reporting information of a user terminal; a matching module, configured to perform a matching and positioning operation on the MR association data and fingerprints in an MR fingerprint database to obtain a matched abnormal grid region of a specified size; a mapping module, configured to map the acquired real-time travel path data to the abnormal grid region to obtain a candidate path for the user terminal's movement; and an identification module, configured to identify the actual movement path of the user terminal from the candidate paths based on an optimized Hidden Markov Model (HMM).

[0023] According to another aspect of this disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute the user movement path identification method of any one of the second aspect of the technical solutions described above by executing the executable instructions.

[0024] According to another aspect of this disclosure, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the user movement path identification method described above.

[0025] The user mobility path identification method and apparatus provided in the embodiments of this disclosure, when obtaining communication reporting information from a user terminal, acquires MR data and associates the MR data with the corresponding XDR data to obtain MR association data. Using the MR association data, a rough location of the user is achieved, and candidate paths for the user terminal's movement are obtained. Furthermore, the stability and efficiency of the HMM model's identification operation are improved during the process of identifying the user's actual mobility path based on the HMM model. Therefore, when obtaining communication reporting information, the actual mobility path of the user terminal can be reliably identified. While achieving reliable location of mobile users, in scenarios where communication reporting information is obtained based on user complaints, a rapid complaint response can be initiated based on the accurate location of the user, thereby improving the efficiency of user complaint handling and effectively saving the manual processing costs of complaint issues.

[0026] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0027] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure. It is obvious that the drawings described below are merely some embodiments of this disclosure, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0028] Figure 1 A flowchart illustrating a user movement path identification method according to an embodiment of this disclosure is shown;

[0029] Figure 2 A flowchart illustrating another user movement path identification method in an embodiment of this disclosure is shown;

[0030] Figure 3 This diagram illustrates an embodiment of the present disclosure of how MR data and XDR data are correlated.

[0031] Figure 4 A flowchart illustrating another user movement path identification method in an embodiment of this disclosure is shown;

[0032] Figure 5 This diagram illustrates a grid matching method based on an MR fingerprint database according to an embodiment of the present disclosure.

[0033] Figure 6 A flowchart illustrating yet another user movement path identification method in an embodiment of this disclosure is shown;

[0034] Figure 7 This diagram illustrates the GIS location of the complaint in the user movement path identification scheme of this embodiment.

[0035] Figure 8 This diagram illustrates a candidate grid region for a user movement path identification scheme in an embodiment of this disclosure.

[0036] Figure 9 This diagram illustrates the filtering of invalid grid regions in the user movement path identification scheme of this embodiment.

[0037] Figure 10 A schematic diagram of candidate paths for a user movement path identification scheme in an embodiment of this disclosure is shown.

[0038] Figure 11 This diagram illustrates the actual movement path of a user terminal in the user movement path identification scheme of this embodiment.

[0039] Figure 12 This diagram illustrates a user movement path identification device according to an embodiment of the present disclosure;

[0040] Figure 13A structural block diagram of an electronic device according to an embodiment of the present disclosure is shown. Detailed Implementation

[0041] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that this disclosure will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0042] Furthermore, the accompanying drawings are merely illustrative of this disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0043] Currently, positioning mobile communication users based on received signal strength has become a research hotspot. However, due to factors such as Rayleigh fading, frequent handover, and uplink / downlink interference, problems such as large positioning data errors, numerous invalid points, and difficulty in road network matching may occur.

[0044] The steps of the user movement path identification method in this example embodiment will now be described in more detail with reference to the accompanying drawings and embodiments.

[0045] like Figure 1 As shown, a user movement path identification method according to an embodiment of this disclosure includes:

[0046] Step S102: Based on the association model, perform an association operation between the collected MR data and the corresponding XDR data to obtain MR association data, wherein the MR data is obtained based on the communication reporting information of the user terminal.

[0047] The association model refers to, but is not limited to, association methods and association fields.

[0048] XDR data includes CDR (Call Detail Records) and TDR (Transaction Detail Records).

[0049] MR (Measurement Report) data consists of raw network data measured by user terminals. The measurement report carries relevant information about uplink and downlink radio links, including RSCP (Receive Signal Channel Power), ISCP (Interference Signal Code Power), BLER (Block Error Rate), and transmit power. MR data provides information about the signal level of a user terminal in a given cell, effectively providing an overview of the cell's coverage.

[0050] In addition, by correlating MR data and XDR data to obtain MR-correlated data, and using MR-correlated data to locate the user's position and detect the observation probability of the following candidate paths, it is beneficial to improve the accuracy of locating the user's movement path.

[0051] The communication reporting information can be complaint information proactively reported by users due to communication abnormalities, information reported by user terminals moving at high speeds outdoors, or communication information reported based on virtual road tests. Step S104: Perform a matching and positioning operation between the MR association data and the fingerprints in the MR fingerprint database to obtain the matched abnormal grid area of ​​a specified size.

[0052] The MR fingerprint database includes each known location point. The fingerprint feature information mainly includes the signal strength information of the surrounding cells to the location point. By matching the MR association data with the fingerprints in the MR fingerprint database, the abnormal grid area corresponding to the candidate communication reporting information can be obtained based on the matching result.

[0053] Step S106: Map the acquired real-time travel path data to the abnormal grid area to obtain the candidate path for the user terminal's movement.

[0054] The real-time travel route data includes internet road network information and real-time traffic information. The internet road network information includes road level information labels obtained and parsed from electronic maps. Real-time traffic information is obtained from internet travel data to further correct speed limits for road levels in electronic maps and to fully consider the impact of real-time road congestion on speed. Based on the interface of travel navigation apps, map road congestion levels are captured in 15-minute increments, including time, road type, and congestion status.

[0055] Step S108: Identify the actual movement path of the user terminal from the candidate paths based on the optimized Hidden Markov Model (HMM).

[0056] The HMM model can include an observation probability calculation model, a transition probability calculation model, and a Viterbi probability calculation model. Based on the obtained Viterbi probability, the candidate path with the highest probability is obtained, which is the actual movement path of the user terminal.

[0057] In this disclosure, the optimized HMM model refers to an HMM model obtained by optimizing one of the observation probability calculation model, transition probability calculation model, and Viterbi probability calculation model.

[0058] In this embodiment, when the communication reporting information of the user terminal is obtained, MR data is acquired and associated with the corresponding XDR data to obtain MR association data. The MR association data is used to achieve a rough location of the user and obtain candidate paths for the user terminal's movement. At the same time, it can further improve the stability and efficiency of the HMM model in the process of identifying the user's actual movement path based on the HMM model. Thus, when the communication reporting information is obtained, the actual movement path of the user terminal can be reliably identified. While achieving reliable location of mobile users, in scenarios where the communication reporting information is obtained based on user complaints, it can also quickly respond to complaints based on the accurate location of the user, thereby improving the efficiency of user complaint handling and effectively saving the manual processing costs of complaint issues.

[0059] In some embodiments, performing an association operation between the collected MR data and the corresponding XDR data based on the association model to obtain MR associated data includes: collecting XDR data that is in the same time and area as the MR data; and performing an association operation between the MR data and the XDR data based on the association model to obtain MR associated data.

[0060] In some embodiments, collecting XDR data that is in the same time and area as MR data includes: periodically monitoring the collection status information of the collection server to ensure that the collection server collects MR data and XDR data under normal conditions, wherein the collection status information includes configuration information and data transmission monitoring information during the data collection process.

[0061] In this embodiment, by strengthening the supervision of the acquisition and transmission process of MR data and XDR data, it is ensured that the acquisition configuration is correct and that the data is not lost, thereby improving the data quality and thus helping to improve the success rate of data association.

[0062] Specifically, to improve the success rate of data association, it is necessary to control the quality of the collected XDR and MR data, improve the data completeness and quality, and minimize the proportion of invalid data. This will improve the data association rate and accuracy. By monitoring data quality, such as regularly monitoring the status of the collection server to ensure that no files are lost during XDR and MR data collection, it is possible to investigate the causes on the MR / XDR collection side, check the base station configuration, and check for file loss during data transmission, thereby improving the data completeness. Investigating and resolving the causes on the MR data collection side can also improve the association accuracy of XDR and MR data.

[0063] In some embodiments, the method further includes: determining the user S1AP identifier on the base station side as an associated field, wherein the user S1AP identifier is configured as a non-short code on the base station side.

[0064] The S1AP ID is a content within the UE context and is the user index in the S1AP message.

[0065] In this embodiment, since the default short code MME (Mobility Management Entity, Network Node)_UES1APID is invalid during association, the short code data will be cleaned up. In order to improve the association accuracy of XDR data and MR data, the MME_UES1APID reported by the base station as an association field is restricted to not being a short code.

[0066] In addition, cleaning up invalid data, such as data with empty cell numbers, data with unreasonable start / end times, and MME_UES1APID, also helps to ensure the reliability of the remaining data association.

[0067] In some embodiments, such as Figure 2 As shown, step S102, which involves associating MR data with XDR data based on an association model to obtain MR-associated data, is a specific implementation method that includes:

[0068] Step S202: Perform invalid data cleaning on the MR data and XDR data to obtain cleaned MR data and cleaned XDR data.

[0069] Step S204: Perform initial association processing on the MR cleaning data and XDR cleaning data based on the association field.

[0070] Step S206: Record the MR cleaning data that was not associated in the first association process as unassociated data, and perform forward sliding window search and backward sliding window search on the unassociated data based on the acquisition time to obtain search data.

[0071] Step S208: Merge and deduplicate the search data to obtain the processed search data.

[0072] Step S210: Perform secondary correlation processing on the processed search data and XDR cleaned data.

[0073] Step S212: Based on the processing results of the first association processing and the second association processing, the measurement information and location information in the MR data are written into the data table of the associated XDR data to obtain MR association data.

[0074] This involves writing the measurement and location information from the MR data into a data table associated with the XDR data to achieve data backfilling. Data backfilling involves displaying the data through tables, forms, and lists, and then saving the data to the corresponding database table after performing operations on the data.

[0075] In addition, the XDR data and MR data of the same time, location and user are linked by the MME_UE_S1AP_ID field, and the signaling field and service field to be analyzed are backfilled to obtain the MR associated data.

[0076] In this embodiment, the association process is divided into two steps. The first step uses the association field MME_UE_S1AP_ID to associate XDR data and MR data. For parts that are not associated, a second association is performed. In the second association process, sliding window searches are performed 20 minutes before the start time and 20 minutes after the end time of the current MR data to obtain forward sliding window search results and backward sliding window search results. Merging, deduplication, and backfilling operations are performed on the search results. The measurement information and location information in the corresponding MR data are filled into the table storing XDR sensing data to complete the backfilling and obtain MR associated data. The above operations help improve the data association rate and the accuracy of the associated data.

[0077] In some embodiments, the association operation between MR data and XDR data based on the association model to obtain MR associated data further includes: classifying XDR data based on MME type, HTTP type, and COMMOM type to obtain MME data, HTTP data, and S1_COMMON data; performing an MME_MR association operation between MR data and MME data; performing an HTTP_MR association operation between MR data and HTTP data; and performing a COMMON_MR association operation between MR data and S1_COMMON data to obtain MR associated data.

[0078] Specifically, there are three types of association between XDR data and MR data: MME_MR association, HTTP_MR association, and COMMON_MR association. The difference between each type of association lies in the XDR data, which are MME data, HTTP data, and S1_COMMON data, respectively. Each type of association includes a preprocessing-association-backfilling process.

[0079] like Figure 3 As shown, XDR data and MR data are correlated, specifically the key parameters of the XDR data and the key parameters of the MR data.

[0080] Key parameters of XDR data include IMSI (International Mobile Subscriber Identity), ECGI (E-UTRAN Cell Global Identifier), MME Group ID, MME Code, and MME USS1AP ID.

[0081] Key parameters of MR data include MME Group ID, MME Code, MME US S1AP ID, AOA (Angle of Arrival) and TA (Time Advanced) values.

[0082] In addition, the associated parameters also include time, accuracy, and latitude.

[0083] like Figure 4 As shown, in some embodiments, step S104, which involves performing a matching and localization operation between the MR association data and the fingerprints in the MR fingerprint database to obtain a matched abnormal grid region of a specified size, includes the following specific implementation:

[0084] Step S402: Optimize the K-nearest neighbor (KNN) model based on cross-validation to obtain the optimized KNN model.

[0085] Step S404: Extract the primary cell RSRP and neighboring cell RSRP of the user terminal from the MR association data as MR level data.

[0086] Step S406: Based on the KNN optimization model, the MR level data is matched and located with the fingerprints in the MR fingerprint database to obtain an abnormal grid region of a specified size.

[0087] In some embodiments, the specified size is greater than or equal to 20m and less than or equal to 25m.

[0088] In this embodiment, by setting up a KNN optimization model, MR associated data is input into the KNN optimization model to output multiple grid-level (20m*20m) user complaint areas. Based on the association between XDR data and MR data and the application of KNN-based MR fingerprint positioning technology, 80% of the MR associated data can be located to the grid (20m*20m), thus improving the matching rate and accuracy.

[0089] In some embodiments, optimizing the K-Nearest Neighbor (KNN) model based on cross-validation to obtain an optimized KNN model includes: dividing the obtained MR training set into K parts, designating one part of the K parts of the MR training set as the validation set, and designating the other K-1 parts as the training set to perform model training; repeating the model training K times to obtain K KNN models and corresponding K scores; and performing model optimization based on the K KNN models and corresponding K scores to obtain the optimized KNN model.

[0090] In this embodiment, K-fold cross-validation is the most basic cross-validation method. It involves randomly dividing the training set into K parts, using one part as the validation set to evaluate the model, and using the remaining K-1 parts as the training set to train the model. This step is repeated K times, each time using a different subset as the validation set, ultimately resulting in K different models and K scores. The performance of these k models is then combined to optimize the KNN model, resulting in the optimized KNN model.

[0091] In some embodiments, matching and locating MR level data with fingerprints in the MR fingerprint database based on the KNN optimization model to obtain anomaly grid regions of a specified size specifically includes:

[0092] Based on the KNN optimization model, MR level data is matched and located with fingerprints in the MR fingerprint database to obtain candidate raster regions.

[0093] like Figure 5 As shown, there are multiple geographic grids, including geographic grid 1, geographic grid 2, geographic grid 3, and geographic grid 4, etc.

[0094] An MR fingerprint library is constructed based on MR fingerprints from multiple geographic grids, where MR fingerprints include grid signal strength.

[0095] After receiving the communication reporting information from the user terminal, data association is performed based on the reported information to obtain MR association data. Based on the MR level data in the MR association data, the difference is calculated with the grid signal strength of multiple geographic grids to obtain the corresponding signal strength difference. By further performing minimum Euclidean distance calculation, the geographic grid with the minimum Euclidean distance is obtained. If the ratio is grid 3, then grid 3 is determined as a candidate grid area.

[0096] Specifically, since most of the fields after associating XDR data and MR data do not contain location information, the MR data needs to be processed for location. In addition, since the terminal observes a set of signal strength sample values ​​of a cell at each location, the collection of a large number of sample value information forms the MR fingerprint database.

[0097] Furthermore, by utilizing the optimized KNN (k-nearest neighbor) model, the data in the fingerprint database is matched with the RSRP of the primary serving cell and the RSRP of the neighboring cells in the MR report, thus obtaining the location information at the time of MR reporting and realizing the positioning of MR data.

[0098] Obtain the location description based on the identification of communication reporting information.

[0099] The first grid region is obtained by filtering out invalid grids from the candidate grid region that are more than a distance threshold from the described location.

[0100] The invalid grids covered by cells without primary coverage are filtered out from the first grid area to obtain the second grid area, which is designated as the abnormal grid area.

[0101] In this embodiment, for the located grid area, invalid grids are first filtered out based on a distance algorithm, such as grids that are more than 1 km away from the described location. Then, based on the main coverage cell of the grid, invalid grids are filtered out again. The above two-layer filtering operation can filter out 30% of invalid grids, narrowing the range of user locations used for identification, thereby improving the reliability of user movement path identification.

[0102] In some embodiments, before performing matching and localization between MR level data and fingerprints in the MR fingerprint database based on the KNN optimization model to obtain an abnormal grid region of a specified size, the method further includes: acquiring CQT dialing test data, DT drive test data, MDT mobile device measurement data, and OTT data of multiple grid regions and using them as training data; training the model based on the training data to generate a propagation model; calculating the cell strength of each grid region based on the propagation model; and creating an MR fingerprint database based on the cell strength.

[0103] In this embodiment, the data sources in the fingerprint database include CQT dialing test data, DT drive test data, MDT mobile device measurement data, and OTT data. Due to the limitations of data collection, the data in the fingerprint database cannot cover all areas. Therefore, it is necessary to use this accurate data to train the propagation model and obtain the propagation model that most closely approximates the real wireless environment. This model is then used to calculate the strength of each cell in all grids and added to the fingerprint database, so that the data in the fingerprint database can cover the entire network area. By optimizing the MR fingerprint database, the data quality of the MR fingerprint database can be improved, the number of invalid samples can be reduced, and thus the positioning reliability based on the MR fingerprint database can be improved.

[0104] In some embodiments, the acquired real-time travel path data is mapped to anomaly grid areas to obtain candidate paths for the user terminal's movement, including:

[0105] Acquire terminal MR data of user terminals at two points close to the reporting time of communication reporting information.

[0106] Calculating user mobility rate based on terminal MR data specifically includes:

[0107] Fingerprint location data of suspected mobile users is selected, and further filtered based on the HOST link in DPI document S1_HTTP to extract the download_http_content and upload_http_content content of specific OTT and user identifier. The latitude and longitude of the user are extracted, and high-confidence latitude and longitude can be obtained through latitude and longitude conversion. Using the location latitude and longitude values ​​and timestamp, the speed value between any two points is obtained to further obtain the user's movement rate.

[0108] The process involves matching user movement speed with real-time travel route data.

[0109] Real-time travel route data includes information obtained from the Internet road network, and numerical values ​​of road level information labels in the road network obtained and parsed through electronic maps.

[0110] The matching results are mapped to the abnormal grid area to obtain candidate paths.

[0111] In this embodiment, by acquiring Internet road network information and real-time traffic information, the user's MR data is calculated based on any two points near the time of the complaint to match the user's movement speed with the Internet road network and traffic information, and candidate roads are selected by combining the rough location positioning results.

[0112] In some embodiments, before performing the association operation between the collected MR data and the corresponding XDR data based on the association model, the method further includes: converting the acquired user complaint voice into complaint text to generate communication reporting information based on the complaint text; and performing natural language processing on the complaint text to identify the descriptive location of the user complaint.

[0113] In this embodiment, for scenarios where users complain due to communication abnormalities, the user's voice is converted into complaint text, and the complaint text is further analyzed using natural language processing to identify location-related information, thereby identifying the location described in the user's complaint. Combined with candidate paths, the actual movement path of the user's terminal is identified.

[0114] In some embodiments, a specific implementation of step S104, which identifies the actual movement path of the user terminal from candidate paths based on an optimized Hidden Markov Model (HMM), includes:

[0115] The description location is determined as the observation point, and the candidate path is determined as the hidden state, so as to calculate the observation probability of the description location for the candidate path.

[0116] Specifically, in the HMM-based method, each location point in map matching is associated with all candidate road segments within a preset error radius (dynamically set: 50 meters in urban areas and 100 meters in rural areas). Each is considered an observation state, and each candidate road segment is represented as a hidden state. More specifically, a hidden state represents a candidate point, i.e., the point on the candidate road segment closest to the observation location. Each hidden state is assigned an observation probability, which represents the conditional probability of the observation state generated when the candidate point represented by each hidden state is the true location. The observation probability depends on its distance from the location point. Intuitively, candidate points farther from the location point have a lower observation probability. In the real state, the distance between the location point and the candidate points has measurement errors, generally assumed to be a zero-mean Gaussian distribution. For a location point O determined based on the description location... t Candidate point C in the candidate path t,i The observation probability is as shown in equation (1):

[0117]

[0118] Where σ refers to the standard deviation of the measured distance, g(O) t │C t,i )Specified site O t With candidate point C t,i The great circle distance, within a relatively short distance at low latitudes, can be approximated as the Euclidean distance on a two-dimensional plane.

[0119] Dijkstra's algorithm based on heap optimization calculates the transition probability of each candidate path.

[0120] Specifically, this invention incorporates a time distance method, the time interval between adjacent fingerprint location candidate points, and the difference in the great circle distance between candidate points. Based on the principle that the travel time between two points on a road network is the shortest, the Dijkstra algorithm is introduced.

[0121] The transition probability of each candidate path is calculated based on Dijkstra's algorithm, as shown in equations (2) and (3).

[0122] (C t,i |C t+1,j ) = e -βym (2)

[0123]

[0124] The evaluation process for β is shown in equation (4):

[0125]

[0126] Wherein d(C t,i C t+1,j ) refers to the path distance between candidate points, g(C t,i C t+1,j (i) refers to the great circle distance between candidate points. * and j * The road network refers to the roads, Δt refers to the time interval between candidate points, and β refers to the difference between the path distance and the great circle.

[0127] The observation probability and transition probability are input into the Viterbi calculation model to calculate the corresponding Viterbi probability, and the actual movement path is searched out from the candidate paths based on the Viterbi probability.

[0128] Specifically, actual road users usually prioritize routes that take the least amount of time based on their destination and origin. The trajectory with the highest overall probability P(O,C) is the actual movement path that is sought.

[0129] P(O,C)=P(c0|o0)Π i＝1,...,π [p(c i |o i )p(c i |c i-1 (5)

[0130] Where P(c0|o0) refers to the initial state probability.

[0131] In the Hidden Markov Model (HMM) analysis model, the hidden sequence with the highest probability, i.e. the optimal matching trajectory, is calculated by using the Viterbi computation model. At each step, each choice stores the minimum total cost or maximum value from each historical step to the current step, as well as the choice of each historical step under the current cost condition. Then, based on the observation probability and transition probability in the entire positioning data sequence, the Viterbi algorithm is used to solve the movement trajectory.

[0132] In this embodiment, by constructing and implementing a Hidden Markov Model (HMM), mobile user complaints can be resolved based on a fully automated solution system, thereby effectively saving manual processing costs and improving complaint handling efficiency.

[0133] In some embodiments, calculating the observation probability of a candidate path for a described location includes: associating MR-related data with acquired network performance data and network alarm data based on cell coding to obtain initial sample data; analyzing the impact of the associated network performance data on the MR-related data, weighting the MR-related data based on the analysis results, and selecting performance sample data from the initial sample data based on the weighted labels; extracting performance sample data corresponding to a specified alarm type from the performance sample data and using it as target performance data; acquiring target performance data related to the candidate path, and calculating the observation probability based on the related target performance data.

[0134] Specifically, when users complain about network anomalies, it indicates that there may be explicit or implicit problems in the network. The performance data and alarm data of these problematic networks can reflect the network's health status. Therefore, in the network optimization process, combining the analysis of mobile network wireless cell performance data and alarm data is also helpful in effectively locating problems. Furthermore, the OMC (Operation and Maintenance Center) network management and monitoring system can collect a large amount of performance and network alarm data from network devices.

[0135] Therefore, by applying mobile network wireless cell performance data and alarm data, and filtering out a large number of invalid samples, it is beneficial to improve the observation probability, thereby improving the accuracy of the HMM model output results.

[0136] The MR-related data, namely MR data and XDR data, are associated with performance data and alarm data through the cell code (cell_id).

[0137] Data association is performed within a specified data time period, and MR-related data is associated with alarm data, as shown in Table 1.

[0138] Table 1

[0139] MR_id / XDR_id cell_id Wireless alarm serial number ID Data time period 201-1123 380711-1 al2321019991 21:00～22:00

[0140] Data association is performed within a specified data time period, linking MR-related data with performance data, as shown in Table 2.

[0141] Table 2

[0142] MR_id / XDR_id cell_id Performance data encoding Data time period 201-1123 380711-1 01 Load Class 21:00～22:00

[0143] Furthermore, based on the impact of performance data analysis on MR-related data, the MR-related data are labeled with weights according to features to filter out performance sample data.

[0144] The performance data includes, but is not limited to: air interface performance indicators, equipment availability performance indicators, resource load performance indicators, call access performance indicators, and call hold performance indicators. Data with higher weight values ​​are selected and applied to the Viterbi calculation model.

[0145] Furthermore, based on the impact of alarm data analysis on MR-related data, we can further identify the feature number mr / xdr data, which is the target performance data.

[0146] Specifically, alarm data includes, but is not limited to, identifying RRU link failure alarms, DU cell outage alarms, network element connection interruption alarms, radio frequency unit maintenance link abnormality alarms, radio frequency unit maintenance link abnormality alarms, cell unavailability alarms, etc. Based on the above alarm data, further target performance data is identified.

[0147] Based on the performance data analysis process described above, 15% of invalid and inefficient sample data can be filtered out. Based on the alarm data analysis process described above, an additional 10% of invalid and inefficient sample data can be filtered out. The reduction in data samples helps to improve the hit rate of each path calculation in the Viterbi algorithm and reduce the calculation of invalid paths.

[0148] In this embodiment, by applying mobile network radio cell performance data and alarm data enhancement, the matching rate and accuracy of the observed data are improved, and the overall probability is increased by 5%. The application of performance data and alarm data improves the computational efficiency of big data processing by 50%, thereby reducing the distance g(O) between the hidden state and the observed state. t │C t,i ), to achieve the observation probability p(O) t │C t,i ) improvement.

[0149] In some embodiments, the heap-optimized Dijkstra algorithm for calculating the transition probability of each candidate path includes: calculating the path distance and great circle distance between different candidate paths using the heap-optimized Dijkstra algorithm; and calculating the transition probability based on the path distance, the great circle distance, and the corresponding time interval between the candidate paths.

[0150] Specifically, the commonly used transition probability model is constrained by the difference between the path distance between candidate points and the great circle distance between the location points, which affects the final accuracy and the stability of the model input results.

[0151] In the embodiments of this disclosure, by incorporating a time interval as a time distance and based on the shortest time distance between two points on the road network, a heap-optimized Dijkstra calculation is performed to improve the accuracy and stability of the model output.

[0152] Specifically, the Dijkstra process for heap optimization includes:

[0153] (1) Initialize the distance array dis, specifically by initializing the dis array to infinity and setting the distance of the initial point to 0.

[0154] (2) The loop is n-1 times because there are n points, only n-1 points need to be added to the set S. Each time the loop is performed, one point is added, so it is n-1 times.

[0155] In this context, the set of new points added to the collection is recorded using an array named st.

[0156] dis[i] represents the distance from the starting point s to point i. To update the shortest path from the starting point s to other points (such as point b) using the new point a, we compare the size of dis[a] + (the distance from a to b) and dis[b], and set dis[b] to the minimum of the two.

[0157] (3) After finding the point closest to the starting point, if we directly traverse all the remaining points, the time complexity is O(n). By using a heap for optimization, we can find the minimum value in a heap and reduce the time complexity from O(n) to O(logn). Thus, based on Dijkstra's heap optimization, the time complexity can be reduced from O(mn) to O(mlogn).

[0158] In this embodiment, the heap-optimized Dijkstra's algorithm optimizes the transition probabilities, effectively improving their values ​​and thus enhancing the effectiveness of the Hidden Markov Model (HMM). By combining the calculation of observation probabilities and Viterbi probabilities, the optimized HMM is applied to build a fully automated solution system. This system improves the stability and efficiency of the HMM by constructing and optimizing the HMM in three aspects: correcting the quality of observation data, increasing transition probabilities, referencing alarms, and optimizing the Viterbi algorithm for performance data. This HMM is then applied to a fully automated solution system to resolve mobile user complaints, effectively saving manual processing costs and improving complaint handling efficiency.

[0159] The user movement path recognition scheme in this disclosure will be described in detail below based on the user's voice complaint scenario.

[0160] like Figure 6 As shown, user movement path identification according to another embodiment of this disclosure includes:

[0161] Step S602: Apply a speech conversion model to convert the audio data of the user complaint call into text data.

[0162] For example, the text data converted from the complaint content includes: The 4G Internet signal near XXX Kindergarten in Town B, City A is poor, and sometimes there is no signal. Making calls and sending text messages is normal.

[0163] Step S604: Perform text parsing on the text data to extract the address of the user complaint data record.

[0164] Step S606: Use the GIS address database to output the seven-level pattern address corresponding to the recorded address, and perform GIS positioning of the complaining user.

[0165] Table 3 shows a schematic of the seven-level mode address.

[0166] Table 3

[0167] address Near XXX Kindergarten, Town B, City A Province C city A district D town B road XXX Detailed address ……

[0168] Figures 7 to 11 The image shows the user's location, identification location, and base station location.

[0169] Figure 7 A schematic diagram of the GIS positioning results is shown.

[0170] Step S608: Obtain MR data including user numbers based on text data, associate it with XDR data to obtain MR associated data, and use it as complaint MR data.

[0171] Step S610: Match the complaint MR data with the MR fingerprint database to obtain candidate grid regions.

[0172] Figure 8 A schematic diagram of the matched candidate grid regions is shown, as illustrated by the multiple squares in the figure.

[0173] Step S612: Filter out the wireless grids in the candidate grid area to obtain the abnormal grid area.

[0174] Specifically, 1) Location is determined by filtering out invalid grid cells based on the location described by the complaining user.

[0175] Based on a distance algorithm, invalid rasters (rasters whose distance from their described location is greater than 1km) are filtered out.

[0176] 2) Location is achieved by filtering out invalid grid cells based on information from surrounding base stations.

[0177] Then, based on the main coverage cell of the grid, invalid grids are filtered again (grids within the distance range, and the top 3 cells of the grid do not have the wireless cells described above).

[0178] Figure 9 This shows the abnormal raster regions that have had invalid raster regions filtered out.

[0179] Grids that are not within the range of the base station or within the valid range of the described location will be deleted as invalid grids. Figure 9 As shown in the black squares.

[0180] Step S614: Based on the abnormal grid area, the seven-level mode address, and the obtained Internet road network data and real-time traffic data, a candidate path for the user terminal to move is obtained.

[0181] Figure 10 A schematic diagram of candidate paths obtained based on raster mapping is shown, such as... Figure 10 As shown by the black dashed lines in the image.

[0182] Step S616: Input the MR association data on the candidate path into the HMM model to obtain the real location of the complaining user, and then obtain the actual movement path of the user terminal.

[0183] Figure 11 A schematic diagram of the actual movement path of the user terminal obtained based on the HMM model is shown, such as... Figure 11 As shown by the black dashed lines in the image.

[0184] In this embodiment, the road with the highest overall probability among the candidate roads is determined by transition probability evaluation and Viterbi algorithm, thus obtaining the true and effective location of the complaining user's outdoor road. The road range can be accurate to a 200m segment. Based on actual verification, the positioning accuracy exceeds 95%. In other words, applying this solution can effectively solve the problem of inaccurate location trajectory for mobile users, improve the user's experience, shorten the time for locating and resolving network problems, greatly reduce the workload of daily road testing and data analysis, and save network optimization costs.

[0185] The following reference Figure 12 This invention describes a user movement path identification device 1200 according to this embodiment of the invention. Figure 12 The user movement path identification device 1200 shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0186] The user movement path identification device 1200 is manifested in the form of a hardware module. The components of the user movement path identification device 1200 may include, but are not limited to: an association module 1202, used to perform an association operation between the collected MR data and the corresponding XDR data based on an association model to obtain MR association data, wherein the MR data is obtained based on the communication reporting information of the user terminal; a matching module 1204, used to perform a matching and positioning operation between the MR association data and fingerprints in the MR fingerprint database to obtain a matched abnormal grid region of a specified size; a mapping module 1206, used to map the acquired real-time travel path data to the abnormal grid region to obtain a candidate path for the user terminal's movement; and an identification module 1208, used to identify the actual movement path of the user terminal from the candidate paths based on an optimized Hidden Markov Model (HMM).

[0187] like Figure 13 As shown, the private network offloading device is an electronic device, specifically an electronic device, and the electronic device 1300 is manifested in the form of a general-purpose computing device. The components of the electronic device 1300 may include, but are not limited to: at least one processing unit 1310, at least one storage unit 1320, and a bus 1330 connecting different system components (including storage unit 1320 and processing unit 1310).

[0188] The storage unit stores program code that can be executed by the processing unit 1310, causing the processing unit 1310 to perform the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of the present invention. For example, the processing unit 1310 can perform actions such as... Figure 1 The schemes described in step S102 and step S108 are shown in the figure.

[0189] Storage unit 1320 may include readable media in the form of volatile storage units, such as random access memory (RAM) 13201 and / or cache memory 13202, and may further include read-only memory (ROM) 13203.

[0190] Storage unit 1320 may also include a program / utility 13204 having a set (at least one) of program modules 13205, such program modules 13205 including but not limited to: operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.

[0191] Bus 1330 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0192] Electronic device 1300 can also communicate with one or more external devices 1360 (e.g., keyboard, pointing device, Bluetooth device, etc.), one or more devices that enable a user to interact with electronic device 1300, and / or any device that enables electronic device 1300 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 1340. Furthermore, electronic device 1300 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 1350. As shown, network adapter 1350 communicates with other modules of electronic device 1300 via bus 1330. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 1300, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0193] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, terminal device, or electronic device, etc.) to execute the method according to the embodiments of this disclosure.

[0194] In exemplary embodiments of this disclosure, a computer-readable storage medium is also provided, on which a program product capable of implementing the methods described above is stored. In some possible embodiments, various aspects of the invention may also be implemented as a program product comprising program code that, when the program product is run on a terminal device, causes the terminal device to perform the steps of the various exemplary embodiments of the invention described in the "Exemplary Methods" section of this specification.

[0195] According to embodiments of the present invention, a program product for implementing the above-described method may employ a portable compact disc read-only memory (CD-ROM) and include program code, and may run on a terminal device, such as a personal computer. However, the program product of the present invention is not limited thereto. In this document, a readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0196] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0197] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of sending, propagating, or transmitting programs for use by or in conjunction with an instruction execution system, apparatus, or device.

[0198] The program code contained on the readable medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0199] Program code for performing the operations of this invention can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0200] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0201] Furthermore, although the steps of the method in this disclosure are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps may be omitted, multiple steps may be combined into one step, and / or a step may be broken down into multiple steps.

[0202] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, mobile terminal, or electronic device, etc.) to execute the methods according to the embodiments of this disclosure.

[0203] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the appended claims.

Claims

1. A method for identifying user movement paths, characterized in that, include: The collected MR data is correlated with the corresponding XDR data based on the correlation model to obtain MR correlated data, wherein the MR data is obtained based on the communication reporting information of the user terminal. The process of matching and locating the MR association data with fingerprints in the MR fingerprint database to obtain a matched abnormal grid region of a specified size includes: optimizing the K-nearest neighbor (KNN) model based on cross-validation to obtain an optimized KNN model; extracting the primary cell RSRP and neighboring cell RSRP of the user terminal from the MR association data as MR level data; and matching and locating the MR level data with fingerprints in the MR fingerprint database based on the optimized KNN model to obtain the abnormal grid region of the specified size. The acquired real-time travel path data is mapped to the abnormal grid area to obtain the candidate path for the user terminal's movement; The actual movement path of the user terminal is identified from the candidate paths based on an optimized Hidden Markov Model (HMM).

2. The user movement path identification method according to claim 1, characterized in that, The process of performing a correlation operation between the collected MR data and the corresponding XDR data based on the correlation model to obtain MR correlation data includes: Collect XDR data that is in the same time and area as the MR data; The MR data and the XDR data are correlated based on the correlation model to obtain the MR correlated data.

3. The user movement path identification method according to claim 2, characterized in that, The XDR data acquired at the same time and in the same region as the MR data includes: The acquisition status information of the acquisition server is monitored periodically to ensure that the acquisition server acquires the MR data and the XDR data under normal conditions. The data acquisition status information includes configuration information and data transmission monitoring information during the data acquisition process.

4. The user movement path identification method according to claim 2, characterized in that, Also includes: The user S1AP identifier on the base station side is identified as the associated field. Specifically, on the base station side, the user S1AP identifier is configured as non-invalid data for association between the MR data and the XDR data.

5. The user movement path identification method according to claim 4, characterized in that, The step of associating the MR data with the XDR data based on the association model to obtain the MR-associated data includes: The invalid data of the MR data and the XDR data are cleaned to obtain cleaned MR data and cleaned XDR data. Based on the association field, the MR cleaning data and the XDR cleaning data are initially associated. The MR cleaning data that was not associated in the initial association process is recorded as unassociated data. The unassociated data is then subjected to forward sliding window search and backward sliding window search based on the acquisition time to obtain search data. The search data is merged and deduplicated to obtain the processed search data; The processed search data and the XDR cleaned data are then subjected to a secondary correlation process. Based on the processing results of the first association processing and the second association processing, the measurement information and location information in the MR data are written into the data table of the associated XDR data to obtain the MR association data.

6. The user movement path identification method according to claim 2, characterized in that, The step of associating the MR data with the XDR data based on the association model to obtain the MR-associated data further includes: Based on the MME type, HTTP type, and COMMOM type, the XDR data is classified to obtain MME data, HTTP data, and S1_COMMON data. Perform an MME_MR association operation between the MR data and the MME data; Perform an HTTP_MR association operation between the MR data and the HTTP data; Perform a COMMON_MR association operation on the MR data and the S1_COMMON data to obtain the MR associated data.

7. The user movement path identification method according to claim 1, characterized in that, The optimization of the K-nearest neighbor (KNN) model based on cross-validation operations yields the following optimized KNN model: The obtained MR training set is divided into K parts. One part of the K parts of the MR training set is designated as the validation set, and the other K-1 parts are designated as the training set to perform model training. Repeat the model training process K times to obtain K K NNN models and K corresponding scores; The optimized KNN model is obtained by performing model optimization based on K KNN models and the corresponding K scoring methods.

8. The user movement path identification method according to claim 1, characterized in that, The specified size is greater than or equal to 20m and less than or equal to 25m.

9. The user movement path identification method according to claim 7, characterized in that, The step of matching and locating the MR level data with fingerprints in the MR fingerprint database based on the KNN optimization model to obtain the abnormal grid region of the specified size specifically includes: Based on the KNN optimization model, the MR level data is matched and located with the fingerprints in the MR fingerprint database to obtain candidate grid regions. Obtain the description location of a user complaint based on the communication reporting information, wherein the communication reporting information is generated based on the complaint text, and the description location is identified by performing natural language processing on the complaint text; Invalid grids that are more than a distance threshold from the described location are filtered out from the candidate grid region to obtain the first grid region; The invalid grids covered by cells without primary coverage are filtered out from the first grid area to obtain the second grid area, which is the abnormal grid area.

10. The user movement path identification method according to claim 1, characterized in that, Before performing matching and localization between the MR level data and fingerprints in the MR fingerprint database based on the KNN optimization model to obtain the abnormal grid region of the specified size, the method further includes: Acquire CQT call test data, DT road test data, MDT mobile device measurement data, and OTT data from multiple grid areas and use them as training data; The model is trained based on the training data to generate a propagation model; The cell strength of each grid region is calculated based on the propagation model. The MR fingerprint database is created based on the cell strength.

11. The user movement path identification method according to claim 1, characterized in that, The step of mapping the acquired real-time travel path data to the abnormal grid area to obtain the candidate path for the user terminal's movement includes: Acquire the terminal MR data of the user terminal at two points close to the reporting time of the communication reporting information; Calculate the user's mobility rate based on the terminal's MR data; The user's movement rate is matched with the real-time travel path data. The matching results are mapped to the abnormal grid area to obtain the candidate path.

12. The user movement path identification method according to any one of claims 1 to 11, characterized in that, Before performing the association operation between the collected MR data and the corresponding XDR data based on the association model, the following is also included: The acquired user complaint voice is converted into complaint text, and the communication reporting information is generated based on the complaint text.

13. The user movement path identification method according to claim 12, characterized in that, The optimized Hidden Markov Model (HMM) identifies the actual movement path of the user terminal from the candidate paths, including: The described location is determined as the observation point, and the candidate path is determined as the hidden state, so as to calculate the observation probability of the described location for the candidate path. The described location is obtained based on the identification of the communication reporting information. The Dijkstra algorithm based on heap optimization is used to calculate the transition probability of each candidate path. The observed probability and the transition probability are input into the Viterbi calculation model to calculate the corresponding Viterbi probability, and the actual movement path is searched out from the candidate paths based on the Viterbi probability.

14. The user movement path identification method according to claim 13, characterized in that, The calculation of the observation probability of the described location for the candidate path includes: Based on the cell code, the MR association data is associated with the acquired network performance data and network alarm data to obtain initial sample data; The impact of the associated network performance data on the MR associated data is analyzed, and the MR associated data is weighted based on the analysis results, so as to filter performance sample data from the initial sample data based on the weighted labels; Extract the performance sample data corresponding to the specified alarm type from the performance sample data and use it as the target performance data; Obtain the target performance data associated with the candidate path, and calculate the observation probability based on the associated target performance data.

15. The user movement path identification method according to claim 13, characterized in that, The Dijkstra algorithm based on heap optimization calculates the transition probability of each candidate path as follows: The Dijkstra algorithm with heap optimization is used to calculate the path distance and great circle distance between different candidate paths; The transition probability is calculated based on the path distance, the great circle distance, and the time interval between the candidate paths.

16. A user movement path identification device, characterized in that, include: The association module is used to perform an association operation between the collected MR data and the corresponding XDR data based on the association model to obtain MR associated data, wherein the MR data is obtained based on the communication reporting information of the user terminal. The matching module is used to perform a matching and positioning operation between the MR association data and fingerprints in the MR fingerprint database to obtain a matched abnormal grid region of a specified size. This includes: optimizing the K-nearest neighbor (KNN) model based on cross-validation to obtain an optimized KNN model; extracting the primary cell RSRP and neighboring cell RSRP of the user terminal from the MR association data as MR level data; and performing a matching and positioning operation between the MR level data and fingerprints in the MR fingerprint database based on the optimized KNN model to obtain the abnormal grid region of the specified size. The mapping module is used to map the acquired real-time travel path data to the abnormal grid area to obtain the candidate path for the user terminal's movement. The identification module is used to identify the actual movement path of the user terminal from the candidate paths based on an optimized Hidden Markov Model (HMM).

17. A network device, characterized in that, include: processor; as well as Memory for storing the executable instructions of the processor; The processor is configured to execute the user movement path identification method according to any one of claims 1 to 15 by executing the executable instructions.

18. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the user movement path identification method according to any one of claims 1 to 15.

Images