5G wireless communication network drive test system based on active learning

Abstract

The present invention discloses a 5G wireless communication network drive test system based on active learning, belonging to the field of mobile communication network testing and optimization technology. The system includes obtaining a terminal QoE indicator set and mapping it into a three-dimensional spatial matrix; generating a dynamic drive test path based on the QoE gradient change of the spatial matrix; evaluating the drive test path measurement data to obtain a joint confidence map; screening an active annotation sample set based on the confidence map; and constructing an incremental diagnostic model based on the sample set to output an optimization decision instruction set. The present invention adopts an active learning framework, integrating QoE spatial gradient analysis, two-dimensional confidence assessment, and incremental diagnostic modeling technology. It can achieve dynamic and intelligent planning of 5G network drive test paths, efficient screening of test data, and accurate diagnosis of network problems, significantly improving the efficiency of network drive testing, the accuracy of problem diagnosis, and the response speed of network optimization.

Description

Technical Field

[0001] The present invention relates to the technical field of mobile communication network testing and optimization, and in particular to a 5G wireless communication network drive testing system based on active learning. Background Art

[0002] The rapid deployment and application of 5G (fifth-generation mobile communication technology) networks are driving growing demand for diverse services, such as ultra-high-definition video, virtual reality, and connected vehicles. To ensure high-quality user experience (QoE) for these services and continuously optimize network performance, accurate and efficient testing and evaluation of deployed 5G networks is crucial. Traditional network drive testing is a common method for obtaining real-world wireless signal coverage, interference, and service performance data.

[0003] However, existing 5G network drive testing methods often have limitations. Traditional drive testing typically employs pre-set fixed routes or a grid-based, comprehensive coverage approach. This can lead to inefficient testing, generate large amounts of redundant data in areas with good network quality, and underinvest in areas experiencing frequent problems or dynamic changes. Furthermore, large-scale drive testing generates enormous amounts of data, making subsequent analysis and processing complex. The long lead time for test results to be fed back to network optimization makes it difficult to quickly respond to real-time changes in network status and meet the demands of refined operations and maintenance. Summary of the Invention

[0004] To address the above issues, the present invention provides a 5G wireless communication network drive test system based on active learning. This system adopts an active learning framework and integrates QoE spatial gradient analysis, two-dimensional confidence assessment, and incremental diagnostic modeling technology. It can achieve dynamic and intelligent planning of 5G network drive test paths, efficient screening of test data, and accurate diagnosis of network problems, significantly improving the efficiency of network drive testing, the accuracy of problem diagnosis, and the response speed of network optimization.

[0005] The above objectives can be achieved through the following solutions:

[0006] The 5G wireless communication network drive test system based on active learning includes a data acquisition and mapping module for acquiring a set of terminal service quality experience (QoE) indicators in a preset area and mapping the QoE indicator set into a three-dimensional rasterized spatial matrix; a path generation module for generating a dynamic drive test path based on the QoE attenuation gradient change of the three-dimensional rasterized spatial matrix; a data evaluation module for performing a two-dimensional confidence evaluation on the measurement data in the dynamic drive test path to obtain a space-time joint confidence distribution map; a sample screening module for screening an actively labeled sample set based on the fuzzy interval distribution of the joint confidence distribution map; and a model construction and decision module for constructing an incremental diagnostic model based on the actively labeled sample set, and generating and outputting a network optimization decision instruction set based on the incremental diagnostic model.

[0007] Optionally, the data acquisition and mapping module includes: a data extraction unit for extracting end-to-end delay probe data from a preset base station interface; a data parsing unit for parsing the video buffering event duration distribution in the delay probe data; an indicator generation unit for generating a QoE indicator set based on the video buffering event duration distribution and associating it with the switching request message density in the preset user movement trajectory; and a spatial matrix mapping unit for mapping the QoE indicator set into a three-dimensional rasterized spatial matrix.

[0008] Optionally, the path generation module includes: a variation coefficient calculation unit, used to calculate the signal strength variation coefficient per unit area in the QoE attenuation gradient change; a path node generation unit, used to generate a multi-level optimized path node when the difference in the signal strength variation coefficient of adjacent grids exceeds a preset threshold; and a path construction unit, used to construct a topological structure containing spatial weights based on the multi-level optimized path nodes to form a dynamic path test path.

[0009] Optionally, the generation of multi-level optimized path nodes includes: using a preset coverage hole prediction algorithm to analyze the building penetration loss pattern; marking potential signal shielding areas in the three-dimensional rasterized space matrix according to the loss pattern; and using the boundary points of the signal shielding areas as mandatory constraints for generating priority path nodes.

[0010] Optionally, the data evaluation module includes: a spatial dimension analysis unit, used to analyze the measurement data in the dynamic road test path in the spatial dimension, and to statistically calculate the signal quality fluctuation parameters reflecting the overlapping coverage area status of the micro base station; a time dimension analysis unit, used to analyze the measurement data in the dynamic road test path in the time dimension, and to extract the burst frequency change rate of the tracking area update TAU event; a confidence fusion unit, used to weightedly fuse the signal quality fluctuation parameters and the burst frequency change rate to obtain a joint confidence distribution map.

[0011] Optionally, the statistical signal quality fluctuation parameters reflecting the status of the overlapping coverage area of ​​the micro base station include: obtaining a reference signal reception quality RSRQ measurement report set within a preset time period; establishing an RSRQ variation coefficient matrix of the multi-cell joint coverage area based on the RSRQ measurement report set; identifying a set of interference-sensitive areas that meet the preset collaborative coverage conditions in the RSRQ variation coefficient matrix, and using the interference-sensitive area set as the signal quality fluctuation parameter.

[0012] Optionally, the model construction and decision module includes: a feature extraction unit for extracting PRACH preamble detection failure sequence features from the actively labeled sample set; a training sample generation unit for associating the PRACH preamble detection failure sequence features with a preset TCP retransmission response rate timing map to generate a composite training sample set; a model construction unit for taking network status features as input and network performance indicators as output, and using the composite training sample set to establish and train a neural network model to obtain an incremental diagnostic model; a model updating unit for using a preset network optimization objective function as a constraint condition, and using a preset incremental learning algorithm and the composite training sample set to update the incremental diagnostic model; a decision generation and output unit for analyzing the network status based on the incremental diagnostic model, and generating a network optimization decision instruction set for output.

[0013] Optionally, the extracting of PRACH preamble detection failure sequence features from the actively labeled sample set includes: constructing a time domain-frequency domain joint distribution template based on a preset preamble sequence; comparing the cyclic prefix correlation value of the measurement data in the dynamic road test path with the matching degree distribution of the time domain-frequency domain joint distribution template; generating a network topology correlation parameter of the preamble detection failure event according to the attenuation slope of the matching degree distribution, and using the network topology correlation parameter as the PRACH preamble detection failure sequence feature.

[0014] Optionally, the use of a preset incremental learning algorithm and the composite training sample set to update the incremental diagnostic model includes: monitoring the feature weight correlation coefficient matrix of adjacent diagnostic cycles; when the eigenvalue divergence of the correlation coefficient matrix exceeds a preset range, processing the measurement data of the current cycle, identifying the fuzzy interval measurement data corresponding to the preset confidence fuzzy interval, and using the fuzzy interval measurement data to preferentially update the incremental diagnostic model.

[0015] Optionally, the system also includes a feedback and adaptive control module, including: a feedback processing unit, used to generate model correction parameters based on the execution feedback of the network optimization decision instruction set; an adjustment instruction unit, used to generate adjustment instructions based on the model correction parameters, and use the adjustment instructions to dynamically adjust the sampling density gradient parameters of the three-dimensional rasterized spatial matrix; a reconstruction trigger unit, used to reconstruct the dynamic road test path based on the corrected three-dimensional rasterized spatial matrix.

[0016] Compared with the prior art, the present invention has the following advantages:

[0017] 1. Improved drive test efficiency and targeting: By constructing a QoE spatial matrix and dynamically generating drive test paths based on its attenuation gradient, test resources can be precisely focused on areas where network quality is degrading or potential problems exist. This avoids the redundancy and inefficiency associated with traditional fixed paths or blind full-coverage testing, significantly improving drive test efficiency and resource utilization.

[0018] 2. Optimized data value and analysis load: Using a two-dimensional confidence assessment to measure the value of data, combined with active learning ideas to screen out the most information-rich key samples for analysis and modeling, it can effectively compress the scale of data to be processed, significantly reduce the storage, transmission and analysis complexity of massive drive test data, and improve data utilization efficiency.

[0019] 3. Enhanced diagnostic accuracy and adaptability: By building and incrementally updating diagnostic models based on actively screened high-value samples and introducing execution feedback for adaptive correction and adjustment of model parameters, the system can more accurately and quickly diagnose dynamically changing network problems, improving diagnostic accuracy and the system's intelligent adaptability, and shortening the "test-analysis-optimization" closed-loop cycle.

[0020] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0022] Figure 1 This is a framework diagram of a 5G wireless communication network drive test system based on active learning according to an embodiment of the present invention.

[0023] Figure 2 3. It is a structural diagram of a 5G wireless communication network drive test system based on active learning according to an embodiment of the present invention.

[0024] Figure 3 Schematic diagram of the spatial distribution of QoE indicators in an embodiment of the present invention.

[0025] Figure 4 4 is a comparison curve of active learning effects of an embodiment of the present invention.

[0026] Figure 5 Schematic diagram of dynamic drive test paths and nodes according to an embodiment of the present invention. DETAILED DESCRIPTION

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0028] Reference Figure 1 One embodiment of the present invention proposes a 5G wireless communication network drive test system based on active learning. This system adopts an active learning framework and integrates QoE spatial gradient analysis, two-dimensional confidence assessment, and incremental diagnostic modeling technology. It can achieve dynamic and intelligent planning of 5G network drive test paths, efficient screening of test data, and accurate diagnosis of network problems, significantly improving the efficiency of network drive testing, the accuracy of problem diagnosis, and the response speed of network optimization.

[0029] like Figure 2 As shown, the system of this embodiment specifically includes:

[0030] A data acquisition and mapping module is used to obtain a terminal service quality experience (QoE) indicator set in a preset area and map the QoE indicator set into a three-dimensional rasterized spatial matrix;

[0031] Specifically, the data acquisition and mapping module is responsible for collecting QoE indicator data that reflects the actual user experience in the preset area to form an initial indicator set. To facilitate spatial analysis and processing, the module further converts and stores these indicator data, which may be geographically discrete, into a structured three-dimensional gridded spatial matrix through mapping or gridding technology. This matrix can summarize the basic distribution of QoE within a specified spatial range, such as Figure 3As shown in the figure, the spatial distribution of QoE is intuitively displayed. The darker areas in the figure are potential problem areas with low QoE, which provides an intuitive basis for subsequent dynamic path planning.

[0032] A path generation module, configured to generate a dynamic drive test path according to a QoE attenuation gradient change of the three-dimensional rasterized spatial matrix;

[0033] Specifically, the Path Generation Module receives and analyzes the three-dimensional spatial matrix generated by the Data Acquisition and Mapping Module. It calculates the rate of change of the QoE metric in the matrix across different spatial directions, known as the attenuation gradient, to identify areas where QoE values ​​are rapidly declining or falling below a preset quality level. These areas are identified as key locations where network coverage issues or performance bottlenecks may exist. Based on this identification, the module plans and generates dynamic, non-fixed drive test routes, aiming to guide subsequent drive test equipment to prioritize or focus on these identified key areas, thereby improving the targetedness and efficiency of the testing process.

[0034] A data evaluation module, configured to perform a two-dimensional confidence evaluation on the measurement data in the dynamic drive test path to obtain a space-time joint confidence distribution map;

[0035] Specifically, the data evaluation module receives measurement data collected during the drive test. Because data collection can be affected by various factors, such as the environment and equipment status, resulting in variations in data quality, this module assesses the credibility of this data. It comprehensively considers both spatial and temporal dimensions to quantitatively assess the confidence of the measurement data. The evaluation results are ultimately organized into a joint spatial-temporal confidence distribution map. This map characterizes the reliability or uncertainty of measurement data obtained at different geographic locations and time points, providing a key basis for screening valuable samples.

[0036] A sample screening module, configured to screen an actively labeled sample set according to the fuzzy interval distribution of the joint confidence distribution map;

[0037] Specifically, the sample screening module receives the spatial-temporal joint confidence distribution map from the data evaluation module. It aims to apply active learning (AL) strategies to optimize sample selection. It analyzes this distribution map and focuses on identifying areas with low confidence or high uncertainty, known as fuzzy intervals. Prioritizing data samples from these fuzzy intervals is generally more effective than random sampling in improving model performance or uncovering hidden issues, such as Figure 4 As shown in the performance comparison chart, the active learning of the present invention can achieve higher diagnostic accuracy with fewer training samples compared to random sampling, thereby improving the model learning efficiency.

[0038] The model building and decision module is used to build an incremental diagnosis model based on the actively labeled sample set, and generate and output a network optimization decision instruction set based on the incremental diagnosis model.

[0039] Specifically, the model building and decision-making module receives a set of actively annotated samples with high information value selected by the sample screening module. The core task of this module is to use these key samples to train or update a diagnostic model that can diagnose network performance problems. This model has incremental learning capabilities, which means that it can use new samples to continuously optimize its own performance and adapt to changes in the network environment without the need for complete retraining every time. After using the model to complete the diagnosis of the current network status or problem, the module will combine the preset network optimization goals or policy rules to convert the diagnosis results into a specific, executable network optimization decision instruction set, such as generating instructions for adjusting base station transmission power, modifying switching thresholds, or recommending coverage enhancement in specific areas, for reference by network operation and maintenance personnel or directly input into the automated execution system.

[0040] Through an active learning framework, integrating QoE spatial gradient analysis, two-dimensional confidence assessment, and incremental diagnostic modeling technology, it is possible to achieve dynamic and intelligent planning of 5G network drive test paths, efficient screening of test data, and accurate diagnosis of network problems, significantly improving the efficiency of network drive tests, the accuracy of problem diagnosis, and the response speed of network optimization.

[0041] Optionally, the data acquisition and mapping module includes:

[0042] A data extraction unit, configured to extract end-to-end delay probe data from a preset base station interface;

[0043] Specifically, the data extraction unit serves as an important way for the data acquisition and mapping module to obtain raw data. Its core task is to obtain probe measurement data that can reflect the end-to-end transmission delay performance of the service from specific interfaces of the mobile communication network. The data extraction unit can actively pull or passively receive end-to-end delay probe data from these interfaces through configured protocols and authentication methods, using mechanisms such as timed polling, event subscription notifications, or reading log files in specific formats. This type of probe data is usually generated by active detection programs deployed in the network or on user terminals. The obtained raw delay probe data provides the basic measurement basis for the subsequent data parsing unit to analyze specific service events.

[0044] A data parsing unit, configured to parse the duration distribution of video buffering events in the delay probe data;

[0045] Specifically, the data analysis unit receives raw latency probe data from the data extraction unit. Its primary task is to deeply interpret this latency data to identify and quantify key events that impact the user's video viewing experience, namely, video buffering events. After identifying the onset and end times of each video buffering event, the data analysis unit calculates the duration of that buffering event. This is done by compiling statistical data on the duration of all buffering events occurring within a specific time window or geographic area.

[0046] An indicator generating unit, configured to generate a QoE indicator set based on the distribution of duration of the video buffering event and in association with the handover request message density in a preset user movement trajectory;

[0047] Specifically, the indicator generation unit integrates the video buffering distribution characteristics from the data analysis unit and incorporates contextual information from other dimensions for comprehensive evaluation, generating a set of indicators that more comprehensively reflect QoE. The contextual information this unit needs to obtain primarily includes pre-defined user movement trajectories and handover request message density. User movement trajectory data can typically be obtained by processing Global Positioning System (GPS) logs recorded by the drive test equipment itself, reflecting the test terminal's movement path in geographic space. Handover request message density is typically obtained from the network side, for example, by analyzing base station or core network signaling logs to extract the frequency or density of handover request events occurring within a specific geographic area or time period. The key to the indicator generation unit lies in performing correlation analysis, which involves temporally and spatially matching and correlating the video buffering experienced by a user at a specific location or time period along their movement trajectory with the handover request density encountered at that location or in adjacent areas. This correlation aims to identify scenarios where network mobility management issues may lead to degraded video service QoE. For example, the conditional probability of a long buffering event can be calculated in areas where the handover density exceeds a preset threshold. Based on the correlation analysis results and the statistical characteristics of the buffering event distribution itself, such as the average buffering time, the probability of long buffering events, and the handover request message density, the unit finally generates a set of quantitative QoE indicators. This set can be expressed as a comprehensive QoE score for each geographic grid unit or drive test track segment. , the score can be calculated using the following exemplary weighted model:

[0048] ,

[0049] Where: Represents the preset QoE full score value, 、 、 These are the preset weight coefficients for the impact of average buffering time, long buffering events, and switching density. Represents the average buffering duration calculated based on the duration distribution of video buffering events. Represents the probability of a long buffering event where the buffering duration exceeds the preset threshold, calculated based on the distribution. represents the handover request message density associated with the current location, This model intuitively reflects that the longer the average buffering time, the higher the probability of long buffering, and the greater the handover density, the greater the penalty points and the lower the final comprehensive QoE score.

[0050] The space matrix mapping unit is used to map the QoE indicator set into a three-dimensional rasterized space matrix.

[0051] Specifically, the spatial matrix mapping unit receives the set of QoE metrics output by the metric generation unit, which contains the quantitative scores corresponding to each geographic location or drive test segment. The core task of this unit is to accurately organize and populate these QoE score data, which may still contain discrete geographic coordinates or regional identifiers, into the corresponding grid cells of a pre-defined three-dimensional rasterized spatial matrix covering the target test area. This mapping process ensures that QoE metric values ​​from different sources and formats are unified onto a standardized spatial grid, forming structured QoE spatial distribution data, providing a regular input for all subsequent modules that rely on this spatial matrix for analysis.

[0052] Optionally, the path generation module includes:

[0053] a coefficient of variation calculation unit, configured to calculate a coefficient of variation of signal strength per unit area in the QoE attenuation gradient change;

[0054] Specifically, the coefficient of variation calculation unit receives and processes the QoE attenuation gradient change information analyzed by the path generation module. This information is derived from the three-dimensional gridded spatial matrix generated by the data acquisition and mapping module. The QoE attenuation gradient reflects the speed and direction of QoE spatial variation. This unit's task is to further quantify the instability or volatility of this gradient change. Using a basic grid cell or a preset neighborhood within the three-dimensional spatial matrix as a unit area, it calculates the coefficient of variation (CV) of the QoE attenuation gradient within that unit area. The CV is a standardized statistic, typically defined as the ratio of the standard deviation to the mean. It measures the dispersion of gradient data relative to its average level and is unaffected by the mean itself. By calculating the CV of the gradient for each unit area, this unit can identify areas where QoE changes are not only rapid but also extremely erratic. The calculated signal strength CV results serve as a key input for the subsequent path node generation unit to determine whether to set optimized path nodes.

[0055] A path node generating unit, configured to generate a multi-level optimized path node when the difference in the signal strength variation coefficient of adjacent grids exceeds a preset threshold;

[0056] Specifically, the path node generation unit receives the signal strength coefficient of variation values ​​corresponding to each unit area in the three-dimensional gridded spatial matrix output by the coefficient of variation calculation unit. The core function of this unit is to identify the geographical boundaries or areas where the instability of network QoE changes has changed significantly based on the spatial distribution differences of these coefficients of variation, and to generate key nodes for path planning at these locations. It traverses the grid cell i in the spatial matrix and compares its coefficient of variation value with each adjacent grid cell j. In order to quantify this difference and determine whether to generate a node, an indicator representing the boundary strength between adjacent grids can be calculated. :

[0057] ,

[0058] Where: Represents adjacent grids The boundary strength indicator between 、 Grid and grid The signal intensity variation coefficient value, It is a preset threshold used to determine whether the difference in coefficient of variation is significant. When it is greater than zero, it indicates that this is the boundary where the QoE gradient volatility changes significantly, which has a high detection value. The path node generation unit will generate an optimized path node here. In addition, in order to achieve multi-level optimization, the generated path nodes can be given different priorities or attributes. The level of the priority or attribute can be adjusted according to the calculated boundary strength. The specific value of is determined according to the preset classification rules. Generally, the larger the boundary strength value, the higher the optimization priority of the corresponding node. Finally, the unit outputs a set of path nodes with priority or attribute information for use by subsequent path construction units.

[0059] The path construction unit is used to construct a topological structure including spatial weights according to the multi-level optimized path nodes to form a dynamic drive test path.

[0060] Specifically, the path construction unit receives a multi-level optimized set of path nodes with varying priorities or attributes, output by the path node generation unit. These nodes represent key locations of interest on the map. Its primary task is to organize these discrete nodes into a topological structure that reflects their spatial relationships and connectivity, typically represented by a weighted graph. In this graph, path nodes are vertices, while potentially feasible path segments connecting two nodes constitute edges. The key lies in assigning spatial weights to these edges. This weight can be a comprehensive value that incorporates not only cost factors such as the physical distance between nodes and estimated travel time, but also information related to test value, such as the average QoE gradient and coefficient of variation of the area traversed by the path segment, or directly utilizes node priority information. After constructing this spatially weighted topological structure, the path construction unit employs one or more path planning and optimization algorithms, such as heuristic search algorithms like A* search (A-Star Search), or customized algorithms tailored to specific objectives, to find an optimal or suboptimal path sequence within this weighted topological graph. Finally, the dynamic drive test path output by the unit is a specific, ordered sequence of instructions, such as a series of geographical coordinate points or road segment identifiers, which can be directly used by the drive test equipment to navigate and perform test tasks, such as Figure 5 The diagram below illustrates how dynamically generated drive test paths connect key test nodes. Nodes of varying shapes represent different priorities, with high-priority nodes representing areas requiring the most attention. This provides clear guidance for intelligent and efficient allocation of test resources.

[0061] Optionally, generating a multi-level optimization path node includes:

[0062] Analyze building penetration loss patterns using a pre-set coverage hole prediction algorithm;

[0063] Specifically, in order to accurately identify and quantify the obstruction effect of buildings on 5G signal propagation, and then predict potential coverage holes, a preset coverage hole prediction algorithm is used to carefully analyze the building penetration loss pattern in the target area. The prediction algorithm can be based on a ray tracing model based on the principles of computational electromagnetics, which requires input of detailed three-dimensional building geometry model information and electromagnetic parameters of building materials. Typical three-dimensional model information can come from building information models or high-precision geographic information system data, while material electromagnetic parameters include dielectric constants and conductivity of walls or glass of different materials. More computationally efficient engineering empirical models can also be used, such as the logarithmic distance path loss model combined with additional loss factors for specific scenarios, where building penetration loss is a key consideration. Penetration loss describes the energy attenuation of the signal when it penetrates an obstacle. For example, a simplified multi-wall penetration loss model It can be expressed as:

[0064] ,

[0065] Where: represents the total predicted penetration loss of the signal through the building, Represents the basic loss value through the first exterior wall, Represents the average loss value through each interior wall, is the number of equivalent interior walls or floors that the signal path passes through, By applying this type of algorithm, a predicted penetration loss value can be calculated for each grid affected by a building in the three-dimensional space matrix, forming a detailed building penetration loss pattern distribution.

[0066] marking potential signal masking areas in the three-dimensional gridded spatial matrix according to the loss pattern;

[0067] Specifically, based on the predicted building penetration loss pattern, areas where the signal may be too weak to meet basic business needs, namely potential signal shielding areas, are clearly marked in the three-dimensional gridded space matrix. The marking is based on comparing the predicted signal reception strength with the minimum signal strength threshold required to maintain the QoE of a specific business. The predicted signal reception strength can be estimated by subtracting the total loss including path loss and penetration loss from the transmission power of the transmitter. If the predicted reception strength of a grid is lower than the preset service availability signal threshold, the grid is determined to be in a potential signal shielding area. The marking process can be achieved through an indicator function accomplish:

[0068] ,

[0069] Where: For grid The masking mark, is the signal transmission power, Is the arrival grid The total path loss, is the preset minimum available signal strength threshold. By performing this judgment and marking on all relevant grids, the spatial scope and shape of the potential signal obstruction area are clearly defined on the three-dimensional space matrix.

[0070] The boundary points of the signal shielding area are used as mandatory constraints for generating priority path nodes.

[0071] Specifically, after clarifying the spatial scope of the potential signal shielding area, it is necessary to identify the boundary points of these areas and convert these points into mandatory constraints when generating path nodes. The identification of boundary points can be done by applying the edge detection algorithm in image processing to the two-dimensional or three-dimensional space of the marking result matrix. Common algorithms include, for example, the Canny operator or the Sobel operator, which are used to find the transition position where the marking value jumps. The identified boundary point sets are of great testing significance, as they represent the critical transition zone for the quality of network coverage. These boundary point information is then used as mandatory constraints to influence the process of the path node generation unit generating multi-level optimized path nodes. This physical shielding constraint based on prediction makes the generation of path nodes more physically meaningful and of practical testing value.

[0072] For example, consider planning a drive test route for a city block dominated by tall buildings. First, a ray tracing-based coverage hole prediction algorithm, combined with a precise 3D building model of the block, analyzes and calculates the signal penetration loss pattern through a specific high-rise building C. It predicts that there is a large area of ​​signal strength below the acceptable threshold on the shadow side of the building C. Based on this prediction, the corresponding grid cells behind building C in the 3D gridded spatial matrix are marked as potential signal shadowing areas. Next, a set of boundary points is identified where this shadowing area intersects with the surrounding unshadowed areas. Finally, when generating multi-level optimized path nodes, this boundary point information is applied as mandatory constraints. For example, high-priority path nodes must be generated within a specific range outside these boundary points, while avoiding generating any mandatory nodes within the marked shadowed areas. This ensures that the node set output by the path node generation unit ensures that the subsequently generated dynamic drive test route focuses on exploring the coverage edges of building C's shadow, rather than ineffectively exploring coverage holes or wandering through areas with good signal strength.

[0073] Optionally, the data evaluation module includes:

[0074] A spatial dimension analysis unit, configured to analyze the measurement data in the dynamic drive test path in the spatial dimension and to generate statistics on signal quality fluctuation parameters reflecting the state of the overlapping coverage area of ​​the micro base station;

[0075] Specifically, the spatial dimension analysis unit focuses on assessing the reliability of measurement data from a geographic perspective. It receives measurement data collected along dynamic drive test paths, which typically contains a wealth of radio environment information, such as received signal strength indicator (RSSI), reference signal received power (RSRP), reference signal received quality (RSRQ), and identifiers of the current serving cell and neighboring cells. A key task of this unit is to identify areas of overlapping micro base station coverage. This can be achieved by analyzing the cell identifiers of multiple micro base stations with strong signals in the measurement data, or by combining the geographic location and coverage information of the micro base stations recorded in a pre-set base station engineering parameter database. Within these identified overlapping coverage areas, signal quality is often unstable due to potential issues such as pilot contamination, increased interference, or frequent handovers. Therefore, this unit further calculates signal quality fluctuation parameters to quantify this instability. For example, it can calculate the standard deviation or coefficient of variation of the RSRP value of the primary serving cell over a short distance, or it can count the number or frequency of handovers within the serving cell. The calculated values ​​of these fluctuation parameters will serve as one of the bases for evaluating the confidence of the data space dimension.

[0076] A time dimension analysis unit, configured to analyze the measurement data in the dynamic drive test path in the time dimension and extract a burst frequency change rate of a tracking area update (TAU) event;

[0077] Specifically, the temporal analysis unit focuses on the dynamic characteristics of measurement data over time to assess its timeliness and stability. It also analyzes received measurement data, particularly when it includes terminal Layer 3 signaling message records. This unit identifies and extracts specific mobility management events, such as Tracking Area Update (TA) events. TAUs are signaling procedures initiated by a terminal to maintain network reachability when it crosses the boundary of a pre-defined Tracking Area (TA) during mobility. This unit monitors the number of TAU events within a continuous time window, calculates the TAU frequency, and further analyzes the rate of change of this frequency over time, such as by calculating the first-order difference of the frequency time series or the amplitude of short-term fluctuations, to determine the burst frequency change rate (BFR). A high BFR may indicate a sudden change in user mobility, network boundary configuration issues, or anomalies in the paging or location update signaling process. These factors may affect the representativeness and stability of the measurement data at that time point and serve as a crucial reference for assessing the confidence level of the data's temporal dimension.

[0078] The confidence fusion unit is used to perform weighted fusion on the signal quality fluctuation parameter and the burst frequency change rate to obtain a joint confidence distribution map.

[0079] Specifically, the confidence fusion unit brings together the evaluation results of data quality or stability from the two dimensions of space and time. It receives the quantitative signal quality fluctuation parameters output by the spatial dimension analysis unit and the quantitative TAU event burst frequency change rate parameters output by the time dimension analysis unit. The goal of this unit is to comprehensively calculate a final joint confidence score based on information from these two or more dimensions to represent the overall reliability of the measurement data obtained at a specific time and space point. The score can be designed so that the greater the fluctuation and the higher the rate of change, the lower the confidence. This can be achieved through a preset weighted fusion model, as shown below:

[0080] ,

[0081] Where: represents the calculated joint confidence score, is the preset upper limit of the confidence score, 、 are the preset weight coefficients assigned to the spatial dimension parameters and the time dimension parameters, is the input quantization signal quality fluctuation parameter value, is the input quantified TAU event burst frequency change rate value, represents a normalization function. The more dramatic the spatial fluctuations or temporal changes, the larger the normalized value, resulting in a greater deduction from the full score and a lower final joint confidence score. By performing this fusion calculation on the data from all sampling points along the drive test path, a spatial-temporal joint confidence distribution map covering the entire test area and time range can be generated.

[0082] For example, when the road test vehicle drives to an area B where micro base stations are densely deployed and happens to be the junction of multiple tracking areas, the spatial dimension analysis unit analyzes the measurement data and finds that the service cell changes frequently and the standard deviation of the received signal power is large, and calculates a higher signal quality fluctuation parameter. At the same time, the time dimension analysis unit analyzes the signaling data of this period and detects that the terminal has initiated multiple tracking area update events, and the short-term change rate of the frequency of occurrence is also very high. After receiving these two parameters that indicate high data instability, the confidence fusion unit calculates according to the preset weights and fusion formula to obtain a significantly low joint confidence score. This low score value is then filled in the position corresponding to area B and the test time in the space-time joint confidence distribution map, clearly indicating that the data confidence of this time-space point is low and may need to be given priority by the subsequent sample screening module.

[0083] Optionally, the statistically reflecting the signal quality fluctuation parameter of the overlapping coverage area state of the micro base station includes:

[0084] Obtain a reference signal received quality (RSRQ) measurement report set within a preset time period;

[0085] Specifically, to quantitatively assess the signal quality stability in areas with overlapping micro base station coverage, it is first necessary to obtain basic data reflecting the signal quality status in that area. The spatial dimension analysis unit collects and combines Reference Signal Received Quality (RSRQ) measurement reports reported by all terminals located within the identified micro base station overlapping coverage area within a preset time period from designated sources, such as detailed measurement logs recorded by drive test terminal equipment, user measurement reports collected by the network management system, or data obtained through mobility-driven testing mechanisms. These reports constitute the RSRQ measurement report set used for subsequent analysis. RSRQ, as a key metric, comprehensively reflects the strength of the received signal as well as the level of co-channel interference and noise, and is a common basis for evaluating signal quality at the cell edge and in overlapping coverage areas.

[0086] Establishing an RSRQ coefficient of variation matrix for a multi-cell joint coverage area based on the RSRQ measurement report set;

[0087] Specifically, after collecting enough RSRQ measurement report samples, it is necessary to quantify the degree of signal quality fluctuation based on these measurement report sets. For the multi-cell joint coverage area within the analysis range, calculate the coefficient of variation of all RSRQ sample values ​​in the area. The calculation process is to first calculate the arithmetic mean and standard deviation of all RSRQ samples in the area, and then obtain the coefficient of variation of RSRQ. By summarizing the RSRQ coefficient of variation values ​​calculated for each analysis area, an RSRQ coefficient of variation matrix or mapping diagram can be established. The values ​​in this matrix reflect the relative volatility of signal quality at different overlapping coverage locations. Higher values ​​indicate more severe fluctuations.

[0088] An interference-sensitive area set that meets a preset cooperative coverage condition is identified in the RSRQ variation coefficient matrix, and the interference-sensitive area set is used as a signal quality fluctuation parameter.

[0089] Specifically, in the established RSRQ variation coefficient matrix, specific areas that meet the preset collaborative coverage conditions are identified to form an interference-sensitive area set. The collaborative coverage conditions here usually refer to the criteria for signal quality fluctuations exceeding the normal or acceptable range. For example, an RSRQ variation coefficient threshold can be preset. When the RSRQ variation coefficient value corresponding to a certain area in the matrix is ​​greater than the RSRQ variation coefficient threshold, the area is judged to meet the conditions and is classified as an interference-sensitive area. These areas may have significant co-channel interference, unreasonable switching parameter configuration or other problems that cause drastic fluctuations in signal quality. Ultimately, the location information or identifiers of all identified interference-sensitive areas are collected to form a set. This interference-sensitive area set itself, or features further extracted based on this set, is used as the signal quality fluctuation parameter output by the spatial dimension analysis unit to reflect the state of the overlapping coverage area.

[0090] For example, assume that the spatial dimension analysis unit is analyzing the known overlapping coverage area C of the micro base stations BS1 and BS2. It first obtains the RSRQ measurement reports reported by all terminals entering or in area C in the past 5 minutes. Based on these reports, the variation matrix establishment step calculates that the overall RSRQ variation coefficient in area C is 0.4. Subsequently, the area identification step compares this calculated value of 0.4 with the preset, and the collaborative coverage condition may be set to an RSRQ variation coefficient greater than 0.35, which is considered interference sensitive. Because the calculated value is greater than the threshold and meets the conditions, area C is identified and marked as an interference-sensitive area. This identification result constitutes part of the output signal quality fluctuation parameter to indicate that the spatial signal stability of the area is poor, and its information will be used by the subsequent confidence fusion unit.

[0091] Optionally, the model building and decision-making module includes:

[0092] a feature extraction unit, configured to extract PRACH preamble detection failure sequence features from the actively marked sample set;

[0093] Specifically, the feature extraction unit processes the received actively annotated sample sets, focusing on analyzing the underlying information reflecting the terminal's initial network access. By parsing the sample data, it identifies failure events during the Physical Random Access Channel (PRACH) process and extracts sequence features that quantify the access difficulty, such as the frequency or pattern of failures, providing input for subsequent correlation analysis.

[0094] A training sample generating unit, configured to associate the PRACH preamble detection failure sequence characteristics with a preset TCP retransmission response rate time series graph to generate a composite training sample set;

[0095] Specifically, the training sample generation unit constructs training data that reflects the correlation between lower-layer access and upper-layer transmission performance. It performs spatiotemporal correlation matching between the PRACH failure features output by the feature extraction unit and a pre-defined time series graph that records the performance of the network's Transmission Control Protocol (TCP) layer. By combining the matched features from different layers, a composite training sample set containing cross-layer information is generated for training the diagnostic model.

[0096] A model building unit is used to establish and train a neural network model using the composite training sample set with network status characteristics as input and network performance indicators as output to obtain an incremental diagnostic model;

[0097] Specifically, the model building unit is responsible for the initial, from-scratch creation of the incremental diagnostic model. It receives the composite training sample set provided by the training sample generation unit and uses the network status characteristics within it as model inputs, such as the PRACH preamble detection failure sequence characteristics, and associated network performance indicators as desired outputs or labels, such as the TCP retransmission acknowledgment rate. By executing a standard supervised learning training process, it builds and trains an initial neural network diagnostic model. This initial model forms the basis for subsequent incremental updates, providing a baseline model for optimization by the model update unit.

[0098] A model updating unit, configured to update the incremental diagnostic model using a preset incremental learning algorithm and the composite training sample set, taking a preset network optimization objective function as a constraint condition;

[0099] Specifically, the model update unit uses the composite training sample set provided by the training sample generation unit to adjust and optimize the incremental diagnostic model by executing a preset incremental learning algorithm. This learning method enables the model to continuously adapt to network changes based on new samples. During the update process, the model's adjustment direction is also constrained by the preset network optimization objective function to ensure that its evolution conforms to the overall network optimization strategy. This unit ultimately outputs an updated incremental diagnostic model with improved performance.

[0100] The decision generation and output unit is used to analyze the network status according to the incremental diagnosis model and generate a network optimization decision instruction set for output.

[0101] Specifically, the decision generation and output unit utilizes the latest incremental diagnostic model provided by the model update unit. Based on this model, it performs diagnostic analysis on the current network status to identify potential network performance issues or areas for optimization. After completing the diagnosis, it combines the model's diagnostic conclusions with pre-defined network optimization strategies or expert rules to translate them into specific, executable network optimization decision instructions. These instructions are then formatted and output to guide subsequent network adjustments and optimization efforts.

[0102] For example, low-confidence samples from region C are screened. The feature extraction unit identifies a high frequency of PRACH access failures. The training sample generation unit finds that the preset TCP profile for the corresponding time period in this region also shows a high retransmission rate. It then associates the access failure signature with the high retransmission rate to generate composite training samples. The model update unit uses these samples, using an incremental learning algorithm, to update the diagnostic model within the constraints of network stability. The updated model can better identify TCP performance degradation caused by such access issues and guide subsequent decision-making.

[0103] Optionally, extracting PRACH preamble detection failure sequence features from the actively marked sample set includes:

[0104] Constructing a time domain-frequency domain joint distribution template based on a preset leading sequence;

[0105] Specifically, to deeply analyze the causes of access failures in the physical random access channel from a signal processing perspective, a reference template must be constructed based on a preamble sequence. This preamble sequence typically refers to a specific waveform sequence used for random access defined in 5G or other cellular network standards. The constructed joint time-frequency distribution template describes the energy or phase distribution characteristics of this ideal preamble sequence in both time and frequency dimensions, serving as a reference for subsequent comparisons.

[0106] comparing a cyclic prefix correlation value of measurement data in the dynamic drive test path with a matching degree distribution of the time-domain-frequency-domain joint distribution template;

[0107] Specifically, the actual signal collected is compared with this ideal template, which is the measurement data from the dynamic drive test path corresponding to the actively annotated sample set. The cyclic prefix (CP) correlation value of this actual measurement data is calculated. The CP is a technique used to combat multipath in OFDM systems, and calculating its correlation value helps assess the signal synchronization status and channel quality. By comparing the similarity between the CP correlation values ​​of the measured data and the ideal template, a matching degree distribution is generated, which reflects the extent to which the received signal conforms to the expected preamble sequence characteristics.

[0108] A network topology correlation parameter of a preamble detection failure event is generated according to the attenuation slope of the matching degree distribution, and the network topology correlation parameter is used as the PRACH preamble detection failure sequence feature.

[0109] Specifically, quantized feature parameters are generated based on the matching degree distribution. The focus is on the attenuation slope of the matching degree as it changes with certain factors. If the matching degree decays rapidly, it usually indicates poor received signal quality, synchronization difficulties, or strong interference, which is likely the cause of the preamble detection failure. By calculating and quantifying this attenuation slope, a network topology correlation parameter for the preamble detection failure event can be generated. This parameter is intended to characterize the strength of the correlation between this access failure and a specific network topology location or channel state. Ultimately, the generated network topology correlation parameter is used as the PRACH preamble detection failure sequence feature output by the feature extraction unit for subsequent analysis and model training.

[0110] Optionally, the updating of the incremental diagnostic model by using a preset incremental learning algorithm and the composite training sample set includes:

[0111] Monitor the feature weight correlation coefficient matrix of adjacent diagnostic cycles;

[0112] Specifically, to ensure the stability and adaptability of the incremental diagnostic model during continuous learning, changes in the model's internal state are monitored. The model's feature weight correlation coefficient matrix is ​​calculated and compared between consecutive diagnostic cycles. This matrix reflects the model's current understanding of the correlations between input features. By comparing this matrix between consecutive update cycles, it is possible to determine whether the model's internal structure or understanding of the relationships between features has undergone significant changes.

[0113] When the eigenvalue divergence of the correlation coefficient matrix exceeds a preset range, the measurement data of the current cycle is processed, the fuzzy interval measurement data corresponding to the preset confidence fuzzy interval is identified, and the incremental diagnostic model is preferentially updated using the fuzzy interval measurement data.

[0114] Specifically, the degree of difference between the feature weight correlation coefficient matrices of adjacent periods is calculated, for example, by calculating the eigenvalue divergence or other distance measurement between them. This quantified difference value is compared with a preset range or threshold. If the calculated divergence exceeds this preset range, it indicates that the model may have experienced significant concept drift or learned an unstable state. According to the preset confidence fuzzy interval standard, the data points in the current measurement data that fall into these high uncertainty intervals are identified, namely fuzzy interval measurement data. These data are considered to have high learning value because they contain information that the model is not yet sure about. These identified fuzzy interval measurement data are implanted into a priority queue associated with the preset historical training set. This means that these data will be selected first in subsequent model updates to help the model learn faster and improve its prediction or diagnosis capabilities in fuzzy and uncertain areas.

[0115] Optionally, the system further includes a feedback and adaptive control module, including:

[0116] A feedback processing unit, configured to generate model correction parameters based on execution feedback of the network optimization decision instruction set;

[0117] Specifically, the feedback processing unit collects actual execution feedback information of the network optimization instructions and generates model correction parameters for guiding the improvement of the diagnosis model by analyzing the deviation between the feedback information and the expected effect.

[0118] An adjustment instruction unit, configured to generate an adjustment instruction based on the model correction parameter, and dynamically adjust a sampling density gradient parameter of the three-dimensional rasterized spatial matrix using the adjustment instruction;

[0119] Specifically, the adjustment instruction unit receives the model correction parameters and determines whether the construction parameters of the three-dimensional space matrix need to be adjusted. If necessary, it generates corresponding adjustment instructions to dynamically update the sampling density gradient parameters and other settings of the space matrix.

[0120] A reconstruction triggering unit is used to reconstruct the dynamic drive test path based on the modified three-dimensional rasterized spatial matrix.

[0121] Specifically, when the three-dimensional rasterized spatial matrix is ​​updated to a modified state according to the adjustment instruction, the reconstruction trigger unit detects the update and triggers the path generation module to replan based on the modified spatial matrix and generate a new dynamic drive test path.

[0122] For example, after the decision instruction is executed, the feedback processing unit analyzes the effect feedback and generates model correction parameters indicating that monitoring of a specific area should be strengthened. The adjustment instruction unit generates instructions based on this information, increasing the sampling density parameter of the area in the three-dimensional spatial matrix. After the matrix is ​​updated, the reconstruction trigger unit triggers the path generation module to regenerate the dynamic drive test path based on the corrected matrix. The new path naturally increases coverage of the key area.

[0123] It should be noted that the electrical connections between the above-mentioned units do not necessarily mean direct connections of lines. Indirect connections are applicable to the embodiments of the present invention as long as the purpose of the present invention is achieved. The above description is only an exemplary embodiment of the present invention and is not intended to limit the scope of the present invention.

[0124] That is, any equivalent changes and modifications made according to the teachings of the present invention are still within the scope of the present invention. Those skilled in the art will readily conceive of other embodiments of the present invention after considering the disclosure of the specification and practical truths. This application is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary technical means in the art not described herein.

Claims

1. A 5G wireless communication network drive test system based on active learning, characterized in that: The system comprises: A data acquisition and mapping module is used to obtain a terminal service quality experience (QoE) indicator set in a preset area and map the QoE indicator set into a three-dimensional rasterized spatial matrix; A path generation module, configured to generate a dynamic drive test path based on the QoE attenuation gradient variation of the three-dimensional gridded spatial matrix; wherein the path generation module includes: a variation coefficient calculation unit, configured to calculate the coefficient of variation of signal strength per unit area in the QoE attenuation gradient variation; a path node generation unit, configured to generate a multi-level optimized path node when the difference in the signal strength variation coefficients of adjacent grids exceeds a preset threshold; and a path construction unit, configured to construct a topological structure containing spatial weights based on the multi-level optimized path nodes to form a dynamic drive test path; A data evaluation module, configured to perform a two-dimensional confidence evaluation on the measurement data in the dynamic drive test path to obtain a space-time joint confidence distribution map; A sample screening module, configured to screen an actively labeled sample set according to the fuzzy interval distribution of the joint confidence distribution map; The model building and decision module is used to build an incremental diagnosis model based on the actively labeled sample set, and generate and output a network optimization decision instruction set based on the incremental diagnosis model.

2. The 5G wireless communication network drive test system based on active learning according to claim 1, characterized in that The data acquisition and mapping module includes: A data extraction unit, configured to extract end-to-end delay probe data from a preset base station interface; A data parsing unit, configured to parse the duration distribution of video buffering events in the delay probe data; An indicator generating unit, configured to generate a QoE indicator set based on the distribution of duration of the video buffering event and in association with the handover request message density in a preset user movement trajectory; The space matrix mapping unit is used to map the QoE indicator set into a three-dimensional rasterized space matrix.

3. The 5G wireless communication network drive test system based on active learning according to claim 1, characterized in that Generating a multi-level optimization path node includes: Analyze building penetration loss patterns using a pre-set coverage hole prediction algorithm; marking potential signal masking areas in the three-dimensional gridded spatial matrix according to the loss pattern; The boundary points of the signal shielding area are used as mandatory constraints for generating priority path nodes.

4. The 5G wireless communication network drive test system based on active learning according to claim 1, characterized in that The data evaluation module includes: A spatial dimension analysis unit, configured to analyze the measurement data in the dynamic drive test path in the spatial dimension and to generate statistics on signal quality fluctuation parameters reflecting the state of the overlapping coverage area of ​​the micro base station; A time dimension analysis unit, configured to analyze the measurement data in the dynamic drive test path in the time dimension and extract a burst frequency change rate of a tracking area update (TAU) event; The confidence fusion unit is used to perform weighted fusion on the signal quality fluctuation parameter and the burst frequency change rate to obtain a joint confidence distribution map.

5. The 5G wireless communication network drive test system based on active learning according to claim 4, characterized in that The signal quality fluctuation parameters that reflect the state of the overlapping coverage area of ​​the micro base station include: Obtain a reference signal received quality (RSRQ) measurement report set within a preset time period; Establishing an RSRQ coefficient of variation matrix for a multi-cell joint coverage area based on the RSRQ measurement report set; An interference-sensitive area set that meets a preset cooperative coverage condition is identified in the RSRQ variation coefficient matrix, and the interference-sensitive area set is used as a signal quality fluctuation parameter.

6. The 5G wireless communication network drive test system based on active learning according to claim 1, characterized in that The model building and decision-making module includes: a feature extraction unit, configured to extract PRACH preamble detection failure sequence features from the actively marked sample set; A training sample generating unit, configured to associate the PRACH preamble detection failure sequence characteristics with a preset TCP retransmission response rate time series graph to generate a composite training sample set; A model building unit is configured to use the network state characteristics as input and the network performance indicators as output, and to establish and train a neural network model using the composite training sample set to obtain an incremental diagnostic model; a model updating unit is configured to use a preset network optimization objective function as a constraint condition, and to update the incremental diagnostic model using a preset incremental learning algorithm and the composite training sample set; The decision generation and output unit is used to analyze the network status according to the incremental diagnosis model and generate a network optimization decision instruction set for output.

7. The 5G wireless communication network drive test system based on active learning according to claim 6, characterized in that The extracting the PRACH preamble detection failure sequence feature from the active marked sample set includes: Constructing a time domain-frequency domain joint distribution template based on a preset leading sequence; comparing a cyclic prefix correlation value of measurement data in the dynamic drive test path with a matching degree distribution of the time-domain-frequency-domain joint distribution template; A network topology correlation parameter of a preamble detection failure event is generated according to the attenuation slope of the matching degree distribution, and the network topology correlation parameter is used as the PRACH preamble detection failure sequence feature.

8. The 5G wireless communication network drive test system based on active learning according to claim 6, characterized in that The updating of the incremental diagnostic model using the preset incremental learning algorithm and the composite training sample set includes: Monitor the feature weight correlation coefficient matrix of adjacent diagnostic cycles; When the eigenvalue divergence of the correlation coefficient matrix exceeds a preset range, the measurement data of the current cycle is processed, the fuzzy interval measurement data corresponding to the preset confidence fuzzy interval is identified, and the incremental diagnostic model is preferentially updated using the fuzzy interval measurement data.

9. The 5G wireless communication network drive test system based on active learning according to claim 1, characterized in that The system also includes a feedback and adaptive control module, including: A feedback processing unit, configured to generate model correction parameters based on execution feedback of the network optimization decision instruction set; An adjustment instruction unit, configured to generate an adjustment instruction based on the model correction parameter, and dynamically adjust a sampling density gradient parameter of the three-dimensional rasterized spatial matrix using the adjustment instruction; A reconstruction triggering unit is used to reconstruct the dynamic drive test path based on the modified three-dimensional rasterized spatial matrix.

Images