An indoor positioning algorithm test method, device, equipment and storage medium

Abstract

The application discloses an indoor positioning algorithm test method, device, equipment and storage medium, and belongs to the technical field of indoor positioning technology test. The method comprises the following steps: acquiring real coordinate sequences of reference points, simulation signal characteristic parameters and measured signal characteristic parameters; calibrating a constructed three-dimensional environment simulation model based on the simulation signal characteristic parameters and the measured signal characteristic parameters to obtain a calibrated three-dimensional environment simulation model; obtaining a positioning result sequence by using a to-be-tested positioning algorithm based on the calibrated three-dimensional environment simulation model; and performing plane positioning error analysis based on the positioning result sequence and the real coordinate sequences to obtain a performance test evaluation report of the to-be-tested positioning algorithm. Through the application, the technical problems that the advantages of field measurement and pure simulation test cannot be fused and the environment dynamic change cannot be adapted in the prior test method are solved.

Description

Technical Field

[0001] This application belongs to the field of indoor positioning technology testing technology, and specifically relates to an indoor positioning algorithm testing method, device, equipment and storage medium. Background Technology

[0002] With the global networking of the BeiDou Navigation Satellite System, meter-level positioning can now be provided in open outdoor environments. However, satellite signals are severely attenuated in indoor spaces, underground parking lots, tunnels, and other sheltered environments. The propagation characteristics of indoor BeiDou signals after re-radiation by an indoor distribution system differ significantly from those of direct outdoor satellite signals.

[0003] In existing technologies, performance verification of indoor positioning algorithms using the "Indoor Distribution + BeiDou" method is mainly divided into two categories: field testing and pure simulation testing. However, field testing and pure simulation testing each use different evaluation indicators and data processing methods, lacking a unified performance evaluation framework. This makes it difficult to compare test results of different algorithms and in different scenarios, and lacks a method that can integrate the advantages of both methods to achieve efficient and accurate testing. As a result, the testing cost of indoor BeiDou positioning algorithms is high, the accuracy verification is inaccurate, and the scenario adaptability testing is incomplete, which seriously restricts the iterative optimization and large-scale engineering application of indoor BeiDou positioning technology.

[0004] Therefore, there is an urgent need for a technical solution that can meet the performance testing requirements of BeiDou positioning algorithms in complex indoor scenarios, including high precision, full-dimensionality, and reproducibility. Summary of the Invention

[0005] This application provides an indoor positioning algorithm testing method, apparatus, equipment, and storage medium, which can solve the technical problems that existing testing methods cannot integrate the advantages of on-site testing and pure simulation testing, and cannot adapt to dynamic environmental changes.

[0006] To achieve the above objectives, this application provides the following technical solution: An indoor positioning algorithm testing method, the method comprising: Obtain the true coordinate sequence, simulated signal characteristic parameters, and measured signal characteristic parameters for each reference point; The constructed three-dimensional environment simulation model is calibrated based on the simulated signal characteristic parameters and the measured signal characteristic parameters to obtain the calibrated three-dimensional environment simulation model. Based on the calibrated three-dimensional environment simulation model, a positioning result sequence is obtained through the positioning algorithm to be measured; Based on the positioning result sequence and the real coordinate sequence, a planar positioning error analysis is performed to obtain a performance test evaluation report of the positioning algorithm to be tested.

[0007] Based on the same inventive concept, this application also provides an indoor positioning algorithm testing device, the device comprising: The information acquisition module is configured to acquire the true coordinate sequence of each reference point, the characteristic parameters of the simulated signal, and the characteristic parameters of the measured signal. The model calibration module is configured to calibrate the constructed three-dimensional environment simulation model based on the simulation signal feature parameters and the measured signal feature parameters, so as to obtain the calibrated three-dimensional environment simulation model. The positioning algorithm testing module is configured to obtain a positioning result sequence based on the calibrated 3D environment simulation model using the positioning algorithm to be tested; and to perform planar positioning error analysis based on the positioning result sequence and the real coordinate sequence to obtain a performance test evaluation report of the positioning algorithm to be tested.

[0008] Based on the same inventive concept, this application also provides an electronic device, including: a memory and a processor; the processor is used to read and execute a computer program stored in the memory to implement the aforementioned indoor positioning algorithm testing method.

[0009] Based on the same inventive concept, this application also provides a computer storage medium storing computer-executable instructions, which, when executed, implement the aforementioned indoor positioning algorithm testing method.

[0010] Compared with the prior art, this application has the following advantages: By performing planar positioning error analysis on the positioning result sequence and the real coordinate sequence, we can intuitively distinguish the error distribution characteristics of the algorithm under static fixed point and dynamic motion trajectory, quantify the degree of error dispersion and extreme degradation, and form a unified and visualized performance evaluation standard. This standard can not only integrate the advantages of field measurement and pure simulation testing to achieve efficient and accurate testing, but also adapt to dynamic environmental changes. It can clearly present the shortcomings of the algorithm to the R&D personnel, and greatly shorten the overall R&D cycle of the positioning algorithm from development, verification to optimization and implementation.

[0011] This application solves the technical problems of existing testing methods that cannot integrate the advantages of field testing and pure simulation testing, and cannot adapt to dynamic environmental changes.

[0012] Other features and advantages of this application will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 A flowchart illustrating the method provided in the embodiments of this application; Figure 2 This is a schematic diagram of the overall architecture of an embodiment of this application; Figure 3 This is a schematic diagram of the cumulative distribution function curve according to an embodiment of this application; Figure 4 This is a schematic diagram showing the visualization output of system simulation results according to an embodiment of this application; Figure 5 This is a functional module diagram of an embodiment of the indoor positioning algorithm testing device of this application; Figure 6 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0016] To address the shortcomings of existing technologies, refer to Figure 1 This application discloses an indoor positioning algorithm testing method, the method comprising: Step S10: Obtain the true coordinate sequence, simulated signal characteristic parameters, and measured signal characteristic parameters for each reference point; In this embodiment, the real coordinates (preset during modeling) corresponding to each receiving point are recorded in the simulation scene as a sequence of real coordinates.

[0017] Based on the architectural drawings of the indoor environment to be tested (such as an underground parking lot), import information such as the 3D coordinates, material, and thickness of the walls to construct a 3D environment simulation model. Configure the parameters of the indoor distributed antenna (location, transmit power, type, azimuth angle, dual-band feeder length, etc.), call the ray tracing simulation engine, and calculate the simulation signal characteristic parameters at each reference point, including carrier-to-noise ratio (C / N0), dual-band pseudorange difference, and signal delay.

[0018] Establish a set of physical parameters to be inverted, P = [α_wall,1, α_wall,2, ..., Γ, n, L_diff], where α_wall is the attenuation coefficient (dB / m) of each wall material, Γ is the reflection coefficient, n is the path loss exponent, and L_diff is the diffraction loss factor. Initially, empirical values ​​are used.

[0019] Furthermore, representative reference points are selected in the scenario (usually directly below each antenna, at the edge of a wall, or at the boundary of signal coverage, with the number being 1 to 2 times the number of antennas). High-precision BeiDou reference stations or calibration equipment are used to collect measured signal characteristic parameters for each reference point, including: carrier-to-noise ratio C / N0_meas (dB-Hz), dual-frequency pseudorange difference Δρ_meas (m), and signal delay τ_meas (ns). The acquisition time for each reference point is no less than 1 minute, and the median value is taken as the measured signal characteristic parameter for that point.

[0020] There is no need to densely deploy data collection points throughout the test space. Only reference points such as directly below the antenna, the edge of the wall, and the signal boundary area are selected to complete the actual measurement sampling. The number of measurement points is only 1 to 2 times the total number of indoor antennas to complete the full-scene model calibration, which greatly reduces the workload of on-site data collection.

[0021] Step S20: The constructed three-dimensional environment simulation model is calibrated based on the simulated signal characteristic parameters and the measured signal characteristic parameters to obtain the calibrated three-dimensional environment simulation model. In some specific embodiments, step S20 includes: The total objective function value is obtained by calculating the weighted sum of squared errors based on the simulated signal characteristic parameters and the measured signal characteristic parameters. The objective function value is minimized using a Bayesian optimization algorithm. During the iteration process, the iteration converges to obtain the optimal parameter set. The optimal parameter set is then embedded into the constructed three-dimensional environment simulation model to obtain the calibrated three-dimensional environment simulation model. Among them, the deviation between the simulation observation data generated by the calibrated 3D environment simulation model and the actual observation data is within the preset range.

[0022] In this embodiment, for any reference point, the deviations between the simulated signal characteristic parameters and the measured signal characteristic parameters at that reference point are calculated in three dimensions: carrier-to-noise ratio, pseudorange difference, and time delay. Each deviation is divided by its corresponding standard deviation, normalized, squared, and then multiplied by the weight of that reference point to obtain the weighted sum of squared errors for that reference point. This process is repeated to obtain the weighted sum of squared errors for all reference points. The weighted sums of squared errors for all reference points are then summed to obtain the overall objective function value.

[0023] The objective function value is minimized using a Bayesian optimization algorithm (based on a Gaussian process surrogate model). During the iteration process, the Bayesian optimization algorithm calls ray tracing simulation to calculate predicted values, and intelligently searches the high-dimensional parameter space based on the predicted values ​​to efficiently approximate the globally optimal parameter set. Compared to grid search or genetic algorithms, Bayesian optimization can reduce the number of calls to time-consuming ray tracing simulations by more than 60%. It is the result of P after Bayesian optimization convergence, and the original parameter set P is an empirical value.

[0024] The obtained optimal parameter set Substitute the constructed three-dimensional environment simulation model into the data, and control the deviation between the simulation output and the measured value of each reference point within a preset threshold (e.g., carrier-to-noise ratio error less than 2dB, pseudorange error less than 0.5m, and time delay error less than 5 nanoseconds) to obtain the calibrated three-dimensional environment simulation model.

[0025] In this embodiment, a small amount of representative measured data is used in conjunction with a Bayesian optimization algorithm to retrieve core physical parameters such as wall attenuation and interface reflection from the ray tracing model. The simulation model is iteratively calibrated using a multi-dimensional weighted normalized objective function that considers carrier-to-noise ratio (CNR), pseudorange difference, and time delay until the root mean square error (RMSE) meets a preset accuracy threshold. Compared to the traditional modeling method that directly uses empirically fixed parameters, this method can control the RNR and pseudorange difference in the simulation output to within 2 dB and 0.5 m, respectively, improving the overall accuracy of the simulation. The Doppler, CNR, and pseudorange observation data generated based on the calibrated simulation model closely match the characteristics of real indoor signal propagation, eliminating the evaluation distortion caused by pure simulation scenarios and ensuring the authenticity and reliability of the positioning algorithm test results.

[0026] Meanwhile, Bayesian optimization based on the Gaussian process surrogate model can intelligently search high-dimensional parameter space. Compared with grid search and genetic algorithm, it can reduce the number of ray tracing simulation calls with huge computing power overhead, significantly shorten the model calibration iteration time, and reduce testing costs in hardware, computing power and manual field survey.

[0027] Step S30: Based on the calibrated three-dimensional environment simulation model, obtain the positioning result sequence through the positioning algorithm to be measured; In some specific embodiments, step S30 includes: Based on the calibrated three-dimensional environment simulation model, simulation observation data for diverse test scenarios are generated; The simulated observation data is input into the positioning algorithm to be measured, and the positioning result sequence output by the positioning algorithm is obtained.

[0028] In this embodiment, a calibrated three-dimensional environment simulation model is used to generate simulation observation data (including carrier-to-noise ratio, pseudorange, Doppler, etc.) for diverse test scenarios as needed. The generation frequency of the simulation observation data is matched with the requirements of the positioning algorithm to be measured.

[0029] The simulation observation data is input into the positioning algorithm to be measured (such as a cloud-based calculation engine or terminal SDK) to drive the algorithm to calculate the location and record the positioning result sequence output by the positioning algorithm to be measured (including latitude and longitude coordinates, floor, and timestamp).

[0030] Reference Figure 4 , Figure 4 This is a schematic diagram showing the visualization output of system simulation results according to an embodiment of this application. For example... Figure 4 As shown, this image displays an indoor signal coverage heatmap and antenna characteristic data generated using a ray tracing algorithm. This data is used to visually assess signal propagation characteristics and guide antenna deployment optimization. The left side of the image displays a 3D wireframe model of the building structure, with the signal strength distribution presented as a color-coded heatmap: red areas (-30 to -50 dBm) indicate strong signals, mainly concentrated around the antennas and in open passageways; blue areas (-80 to -100 dBm) indicate weak signals, typically appearing behind walls, in corners, and at signal boundaries. A color scale is attached to the upper right corner of the heatmap, clearly indicating the correspondence between signal strength and color. The right side contains an antenna characteristic data table, listing key parameters for each antenna, such as pseudorange difference, latency, carrier-to-noise ratio, and transmit power. This data can be directly exported for subsequent analysis or simulation calibration. Figure 4 The image also uses different colored ray trajectories to illustrate the propagation paths of direct (red), reflected (blue), and diffracted (green) rays, and marks the attenuation values ​​on the critical path. Figure 4 It fully presents the entire process of data from building model import, multipath parameter configuration, antenna deployment to simulation result output, providing intuitive and quantitative decision-making basis for the optimized deployment of indoor positioning systems. It is easy to understand that the visualized output of the system simulation results is the calibrated 3D environment simulation model, along with simulation observation data for diverse test scenarios generated on demand.

[0031] Step S40: Perform planar positioning error analysis based on the positioning result sequence and the real coordinate sequence to obtain a performance test evaluation report of the positioning algorithm to be tested.

[0032] In some specific embodiments, step S40 includes: The positioning result sequence and the real coordinate sequence are spatiotemporally aligned, and the planar positioning error value for each epoch is calculated using the three-dimensional Euclidean distance formula based on the longitude difference and latitude difference after conversion to planar distance. Sort the error values ​​of all epochs in ascending order and plot the cumulative distribution function curve; The quantile error of a preset value is extracted from the cumulative distribution function curve and used as a positioning accuracy indicator. Calculate key evaluation parameters, including average error, root mean square error, maximum error, and positioning delay; Based on the positioning accuracy indicators and key evaluation parameters corresponding to different test scenarios, different antenna configurations and different terminal types, a performance test evaluation report of the positioning algorithm to be tested is obtained.

[0033] In this embodiment, the positioning result sequence output by the positioning algorithm to be measured is spatiotemporally aligned with the real coordinate sequence (unifying the time reference and transforming the coordinates to the same projected coordinate system). Based on the longitude and latitude differences after conversion to planar distance, the planar positioning error value for each epoch is calculated using the three-dimensional Euclidean distance formula, which is as follows: ; In the formula, Indicates the difference in longitude. Indicates latitude difference, Indicates the height difference.

[0034] An epoch is a snapshot of a set of satellite observation data at a complete observation moment. Each frame of complete observation data output by the BeiDou / Global Navigation Satellite System receiver and simulation system is an epoch.

[0035] The planar positioning error values ​​of all epochs are sorted from smallest to largest, and the cumulative distribution function (CDF) curve is plotted. The 90th percentile error e_90 is extracted as the positioning accuracy indicator. (Referencing...) Figure 3 , Figure 3 This is a schematic diagram of the cumulative distribution function (CDF) curve according to an embodiment of this application. Figure 3 As shown in the figure, the horizontal axis represents the positioning error (meters), and the vertical axis represents the cumulative probability (0-100%). The CDF curves of the algorithm under test are plotted for three scenarios: static, walking dynamic, and vehicular dynamic. The error values ​​corresponding to the 50%, 75%, and 90% quantiles are marked, visually demonstrating the algorithm's accuracy performance in different scenarios. The CDF plot (Cumulative Distribution Function) shows the cumulative positioning error, reflecting the algorithm's positioning accuracy.

[0036] The raw data collected during testing includes the raw observations required for calculations. Different data can be collected depending on the test scenario and the terminal acquisition device. These data can all be used to calculate key evaluation parameters such as average error, root mean square error, maximum error, and positioning delay (end-to-end time from data input to result output). Furthermore, after calculating key evaluation parameters such as signal mean, standard deviation, and error distribution using Python data processing libraries, data tables and charts such as histograms and scatter plots are generated.

[0037] For different test scenarios (static locations, pedestrian dynamics, vehicle dynamics), different antenna configurations (omnidirectional / directional antennas, different spacing), and different terminal types (multi-brand mobile phones), the above process is repeated to obtain the positioning accuracy indicators and key evaluation parameters corresponding to different test scenarios, different antenna configurations, and different terminal types. Based on the positioning accuracy indicators and key evaluation parameters corresponding to different test scenarios, different antenna configurations, and different terminal types, a multi-dimensional performance test evaluation report of the positioning algorithm under test is obtained.

[0038] The performance evaluation report includes positioning accuracy indicators and key evaluation parameters, used to identify blind spots and weak coverage areas, and provides suggestions on the optimal installation location (three-dimensional coordinates), polarization, and power configuration of the output antenna, along with an estimated deployment cost.

[0039] Further, in one embodiment, after step S40, the following is included: The measured data stream reported in real time by the commercial terminal during the online simulation test of the calibrated 3D environment simulation model; When the average difference between the measured data of any local area and the predicted value of the current calibrated 3D environment simulation model is detected to be greater than the threshold, the fast back-calibration process is triggered to fine-tune the parameters of the local area. If the average difference between the measured data of the local area after fine-tuning and the predicted value of the current calibrated 3D environment simulation model is less than or equal to the threshold, then the parameters of the calibrated 3D environment simulation model are updated based on the parameters of the local area after fine-tuning.

[0040] In this embodiment, when deploying the calibrated 3D environment simulation model for online simulation testing, the system continuously monitors the measured data streams (such as C / N0 and RSSI) reported in real time by commercial terminals. When the average difference between a new batch of measured data in a local area and the current model prediction value is detected to be greater than the dynamic threshold θ (e.g., C / N0 difference > 3dB for 10 seconds), a fast back-calibration process is triggered.

[0041] The fast back-annotation process only fine-tunes the local parameters of the affected area (such as the attenuation coefficient of the nearby wall) using incremental learning or extended Kalman filtering. The fine-tuning range is limited to ±5%. After the adjustment is completed, short-term prediction verification is performed immediately (compared with 10 subsequent real-time data points).

[0042] The system checks whether the average difference between the measured data of the local area after fine-tuning and the predicted value of the current calibrated 3D environment simulation model is less than or equal to a threshold. If the average difference between the measured data of the local area after fine-tuning and the predicted value of the current calibrated 3D environment simulation model is less than or equal to the threshold, the short-term prediction verification is passed, and the parameters of the calibrated 3D environment simulation model are updated. Otherwise, the system rolls back, thus enabling the simulation model to quickly adapt to sudden environmental changes.

[0043] Reference Figure 2 , Figure 2 This is a schematic diagram of the overall architecture of an embodiment of this application. Figure 2 As shown, the test scenario modeling and simulation data generation involves importing architectural drawings and antenna configurations, calling the ray tracing engine to calculate signal characteristic parameters (carrier-to-noise ratio, pseudorange difference, and latency), and generating a simulation dataset. Architectural drawings include wall coordinates, building material type, and wall thickness, used to build a 3D building model and predict wall obstruction and signal reflection loss. Antenna configurations include antenna placement, transmit power, antenna model, azimuth angle, and feeder cable length, characterizing the signal transmission hardware attributes. These two datasets are used to build the physical simulation scenario. The test scenario modeling and simulation data generation uses architectural and antenna parameters to build a 3D scene, incorporating two simulation optimization techniques: adaptive mesh refinement and signal attenuation pruning. Adaptive mesh refinement refines the simulation mesh at wall edges and in areas with severe signal fluctuations, improving the accuracy of ray tracing electromagnetic calculations. Signal attenuation pruning removes ineffective rays, reducing the computational overhead of ray tracing simulation. This stage produces initial, uncalibrated simulation observation data (carrier-to-noise ratio, pseudorange, and Doppler).

[0044] The process of acquiring measured data and calibrating the simulation model involves setting up sampling points directly below the antenna, at wall corners, and in the transition zone between strong and weak signals. Using a high-precision BeiDou reference station and GNSS receiver, real carrier-to-noise ratio (CNR), pseudorange, and time delay observations are collected on-site as measured benchmark data. These are compared and analyzed with the simulation data to correct model parameters, and the calibrated simulation data is output. The process of acquiring measured data and calibrating the simulation model involves comparing the measured values ​​at the sampling points with the simulation output, performing three types of operations: comparing the deviations of CNR / pseudorange difference / time delay item by item, regression analysis to fit the error pattern, and iterative parameter correction. The Bayesian optimization method described earlier, which minimizes the weighted objective function, is used for iterative parameter tuning to reduce the system deviation between simulation and measurement.

[0045] Determine whether the accuracy of the simulation model meets the threshold (RMSE≤5dB). If the accuracy of the simulation model does not meet the threshold, continue to iteratively calibrate the simulation model parameters and repeat the error correction and optimization solution process. If the simulation model's accuracy meets the threshold, the process moves on to the practical evaluation stage of the positioning algorithm. An RMSE (Root Mean Square Error) of 5dB is the acceptable threshold for channel simulation, indicating that the overall deviation of the simulated signal strength is controllable.

[0046] The positioning algorithm testing and data acquisition process involves inputting calibrated and error-compliant simulation observation data (carrier-to-noise ratio, pseudorange, Doppler) into the positioning algorithm to drive its calculations. Simultaneously, the algorithm's output is recorded: calculated coordinates, identified floors, and timestamps for each sample. A fixed ground truth trajectory (a sequence of actual 3D coordinates of points) is pre-stored within the simulation scenario as a benchmark for accuracy comparison. This ground truth trajectory is pre-set during the simulation modeling phase and remains fixed throughout the process.

[0047] Test result analysis and performance evaluation, including systematic error quantification and statistics: Time alignment: The algorithm aligns the epoch timestamps with the ground truth points; 3D error calculation: The planar positioning error and 3D spatial positioning error are calculated epoch by epoch; Advanced statistics: The error samples are sorted, the CDF cumulative distribution curve is plotted, the 90% quantile error is extracted, the average positioning delay is calculated, and the maximum positioning error in a single test is calculated; These are used to quantify the positioning accuracy, delay, and error dispersion of the positioning algorithm in a single-point scenario.

[0048] Multi-scenario fusion testing and generalization capability assessment: Repeated testing was conducted by changing multiple sets of differentiated scenarios (Scenario 1, Scenario 2, Scenario 3, ... with modifications to building layout and antenna deployment schemes). Three types of special evaluations were added: calculation of positioning accuracy fluctuation during scenario switching, compatibility verification of different terminal models, quantification of the generalization adaptation level of the algorithm in unfamiliar scenarios, and judgment of whether the algorithm will experience a significant deterioration in accuracy after scenario migration.

[0049] In the report output phase, firstly, performance reports for each scenario are generated; then, all scenario data are summarized and integrated to form a final comprehensive performance evaluation report, which serves as the formal basis for the acceptance of the positioning algorithm and product iteration optimization.

[0050] A comprehensive evaluation framework covering positioning accuracy, computation latency, operational stability, scenario generalization capability, and terminal compatibility is established. By aligning the positioning results of each epoch with the true trajectory in time and space, the 3D positioning error is calculated. Combining error sequence sorting, CDF cumulative distribution curve, 90% quantile error, maximum error, average latency, and other multi-dimensional statistics, the algorithm performance is no longer judged solely by a single average error. It can intuitively distinguish the error distribution characteristics of the algorithm under static fixed-point and dynamic motion trajectory, quantify the degree of error dispersion and extreme degradation, and form a unified and visualized performance evaluation standard. This clearly presents the shortcomings of the algorithm to R&D personnel, significantly shortening the overall R&D cycle of positioning algorithms from development, verification to optimization and deployment.

[0051] In this embodiment, the true coordinate sequence, simulated signal characteristic parameters, and measured signal characteristic parameters of each reference point are obtained; the constructed three-dimensional environment simulation model is calibrated based on the simulated signal characteristic parameters and the measured signal characteristic parameters to obtain the calibrated three-dimensional environment simulation model; based on the calibrated three-dimensional environment simulation model, the positioning result sequence is obtained through the positioning algorithm to be tested; based on the positioning result sequence and the true coordinate sequence, planar positioning error analysis is performed to obtain the performance test evaluation report of the positioning algorithm to be tested.

[0052] By performing planar positioning error analysis on the positioning result sequence and the real coordinate sequence, we can intuitively distinguish the error distribution characteristics of the algorithm under static fixed point and dynamic motion trajectory, quantify the degree of error dispersion and extreme degradation, and form a unified and visualized performance evaluation standard. This standard can not only integrate the advantages of field measurement and pure simulation testing to achieve efficient and accurate testing, but also adapt to dynamic environmental changes. It can clearly present the shortcomings of the algorithm to the R&D personnel, and greatly shorten the overall R&D cycle of the positioning algorithm from development, verification to optimization and implementation.

[0053] This embodiment solves the technical problems of existing testing methods that cannot integrate the advantages of on-site testing and pure simulation testing, and cannot adapt to dynamic environmental changes.

[0054] Based on the same inventive concept, this application also provides an indoor positioning algorithm testing device.

[0055] In one embodiment, reference is made to Figure 5 , Figure 5 This is a functional module diagram of an embodiment of the indoor positioning algorithm testing device of this application. Figure 5 As shown, the indoor positioning algorithm testing device includes: Information acquisition module 10 is configured to acquire the true coordinate sequence of each reference point, the characteristic parameters of the simulated signal, and the characteristic parameters of the measured signal. The model calibration module 20 is configured to calibrate the constructed three-dimensional environment simulation model based on the simulation signal feature parameters and the measured signal feature parameters to obtain the calibrated three-dimensional environment simulation model. The positioning algorithm testing module 30 is configured to obtain a positioning result sequence based on the calibrated three-dimensional environment simulation model using the positioning algorithm to be tested; and to perform planar positioning error analysis based on the positioning result sequence and the real coordinate sequence to obtain a performance test evaluation report of the positioning algorithm to be tested.

[0056] Optionally, in one embodiment, the signal characteristic parameters include: carrier-to-noise ratio, dual-frequency pseudorange difference, and signal delay.

[0057] Optionally, in one embodiment, the model calibration module 20 is configured to: The total objective function value is obtained by calculating the weighted sum of squared errors based on the simulated signal characteristic parameters and the measured signal characteristic parameters. The objective function value is minimized using a Bayesian optimization algorithm. During the iteration process, the iteration converges to obtain the optimal parameter set. The optimal parameter set is then embedded into the constructed three-dimensional environment simulation model to obtain the calibrated three-dimensional environment simulation model. Among them, the deviation between the simulation observation data generated by the calibrated 3D environment simulation model and the actual observation data is within the preset range.

[0058] Optionally, in one embodiment, the model calibration module 20 is configured to: For any reference point, calculate the deviations between the simulated signal characteristic parameters and the measured signal characteristic parameters at that reference point in three dimensions: carrier-to-noise ratio, pseudorange difference, and time delay. Divide each deviation by its corresponding standard deviation, normalize it, square it, and then multiply it by the weight of the reference point to obtain the weighted sum of squared errors for the reference point. By analogy, the weighted sum of squared errors for all reference points is obtained; The total objective function value is obtained by summing the weighted squared errors of all reference points.

[0059] Optionally, in one embodiment, the positioning algorithm testing module 30 is configured to: Based on the calibrated three-dimensional environment simulation model, simulation observation data for diverse test scenarios are generated; The simulated observation data is input into the positioning algorithm to be measured, and the positioning result sequence output by the positioning algorithm is obtained.

[0060] Optionally, in one embodiment, the positioning algorithm testing module 30 is configured to: The positioning result sequence and the real coordinate sequence are spatiotemporally aligned, and the planar positioning error value for each epoch is calculated using the three-dimensional Euclidean distance formula based on the longitude difference and latitude difference after conversion to planar distance. Sort the error values ​​of all epochs in ascending order and plot the cumulative distribution function curve; The quantile error of a preset value is extracted from the cumulative distribution function curve and used as a positioning accuracy indicator. Calculate key evaluation parameters, including average error, root mean square error, maximum error, and positioning delay; Based on the positioning accuracy indicators and key evaluation parameters corresponding to different test scenarios, different antenna configurations and different terminal types, a performance test evaluation report of the positioning algorithm to be tested is obtained.

[0061] Optionally, in one embodiment, the indoor positioning algorithm testing device further includes a dynamic back-annotation and closed-loop verification module, configured to: The measured data stream reported in real time by the commercial terminal during the online simulation test of the calibrated 3D environment simulation model; When the average difference between the measured data of any local area and the predicted value of the current calibrated 3D environment simulation model is detected to be greater than the threshold, the fast back-calibration process is triggered to fine-tune the parameters of the local area. If the average difference between the measured data of the local area after fine-tuning and the predicted value of the current calibrated 3D environment simulation model is less than or equal to the threshold, then the parameters of the calibrated 3D environment simulation model are updated based on the parameters of the local area after fine-tuning.

[0062] The functions of each module in the above-mentioned indoor positioning algorithm testing device correspond to the steps in the above-mentioned indoor positioning algorithm testing method embodiment, and their functions and implementation processes will not be described in detail here.

[0063] Based on the same inventive concept, embodiments of this application also provide an electronic device, the structure of which is as follows: Figure 6 As shown, it includes: a memory and a processor, wherein the processor is used to read and execute the computer program stored in the memory to implement the aforementioned indoor positioning algorithm testing method.

[0064] Based on the same inventive concept, this application also provides a computer storage medium storing computer-executable instructions, which, when executed, implement the aforementioned indoor positioning algorithm testing method.

[0065] Finally, it should be noted that while some processes described in the embodiments of this application include multiple operations or steps that appear in a specific order, it should be understood that these operations or steps may not be executed in the order they appear in the embodiments of this application, or may be executed in parallel. The sequence number of the operation is only used to distinguish different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed sequentially or in parallel, and these operations or steps may be combined.

[0066] Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for testing indoor positioning algorithms, characterized in that, The method includes: Obtain the true coordinate sequence, simulated signal characteristic parameters, and measured signal characteristic parameters for each reference point; The constructed three-dimensional environment simulation model is calibrated based on the simulated signal characteristic parameters and the measured signal characteristic parameters to obtain the calibrated three-dimensional environment simulation model. Based on the calibrated three-dimensional environment simulation model, a positioning result sequence is obtained through the positioning algorithm to be measured; Based on the positioning result sequence and the real coordinate sequence, a planar positioning error analysis is performed to obtain a performance test evaluation report of the positioning algorithm to be tested.

2. The method according to claim 1, characterized in that, The signal characteristic parameters include: carrier-to-noise ratio, dual-frequency pseudorange difference, and signal delay.

3. The method according to claim 1, characterized in that, The calibration of the constructed three-dimensional environment simulation model based on the simulated signal feature parameters and the measured signal feature parameters to obtain the calibrated three-dimensional environment simulation model includes: The total objective function value is obtained by calculating the weighted sum of squared errors based on the simulated signal characteristic parameters and the measured signal characteristic parameters. The objective function value is minimized using a Bayesian optimization algorithm. During the iteration process, the iteration converges to obtain the optimal parameter set. The optimal parameter set is then embedded into the constructed three-dimensional environment simulation model to obtain the calibrated three-dimensional environment simulation model. Among them, the deviation between the simulation observation data generated by the calibrated 3D environment simulation model and the actual observation data is within the preset range.

4. The method according to claim 3, characterized in that, The step of calculating the weighted sum of squared errors based on the simulated signal feature parameters and the measured signal feature parameters to obtain the total objective function value includes: For any reference point, calculate the deviations between the simulated signal characteristic parameters and the measured signal characteristic parameters at that reference point in three dimensions: carrier-to-noise ratio, pseudorange difference, and time delay. Divide each deviation by its corresponding standard deviation, normalize it, square it, and then multiply it by the weight of the reference point to obtain the weighted sum of squared errors for the reference point. By analogy, the weighted sum of squared errors for all reference points is obtained; The total objective function value is obtained by summing the weighted squared errors of all reference points.

5. The method according to claim 1, characterized in that, The positioning result sequence obtained based on the calibrated three-dimensional environment simulation model and the positioning algorithm to be measured includes: Based on the calibrated three-dimensional environment simulation model, simulation observation data for diverse test scenarios are generated; The simulated observation data is input into the positioning algorithm to be measured, and the positioning result sequence output by the positioning algorithm is obtained.

6. The method according to claim 1, characterized in that, The planar positioning error analysis based on the positioning result sequence and the real coordinate sequence yields a performance test evaluation report for the positioning algorithm under test, including: The positioning result sequence and the real coordinate sequence are spatiotemporally aligned, and the planar positioning error value for each epoch is calculated using the three-dimensional Euclidean distance formula based on the longitude difference and latitude difference after conversion to planar distance. Sort the error values ​​of all epochs in ascending order and plot the cumulative distribution function curve; The quantile error of a preset value is extracted from the cumulative distribution function curve and used as a positioning accuracy indicator. Calculate key evaluation parameters, including average error, root mean square error, maximum error, and positioning delay; Based on the positioning accuracy indicators and key evaluation parameters corresponding to different test scenarios, different antenna configurations and different terminal types, a performance test evaluation report of the positioning algorithm to be tested is obtained.

7. The method according to claim 1, characterized in that, After performing planar positioning error analysis based on the positioning result sequence and the true coordinate sequence to obtain a performance test evaluation report for the positioning algorithm to be tested, the following steps are included: The measured data stream reported in real time by the commercial terminal during the online simulation test of the calibrated 3D environment simulation model; When the average difference between the measured data of any local area and the predicted value of the current calibrated 3D environment simulation model is detected to be greater than the threshold, the fast back-calibration process is triggered to fine-tune the parameters of the local area. If the average difference between the measured data of the local area after fine-tuning and the predicted value of the current calibrated 3D environment simulation model is less than or equal to the threshold, then the parameters of the calibrated 3D environment simulation model are updated based on the parameters of the local area after fine-tuning.

8. An indoor positioning algorithm testing device, characterized in that, The device includes: The information acquisition module is configured to acquire the true coordinate sequence of each reference point, the characteristic parameters of the simulated signal, and the characteristic parameters of the measured signal. The model calibration module is configured to calibrate the constructed three-dimensional environment simulation model based on the simulation signal feature parameters and the measured signal feature parameters, so as to obtain the calibrated three-dimensional environment simulation model. The positioning algorithm testing module is configured to obtain a positioning result sequence based on the calibrated 3D environment simulation model using the positioning algorithm to be tested; and to perform planar positioning error analysis based on the positioning result sequence and the real coordinate sequence to obtain a performance test evaluation report of the positioning algorithm to be tested.

9. An electronic device, characterized in that, include: Memory, processor; The processor is configured to read and execute the computer program stored in the memory to implement the steps of the indoor positioning algorithm testing method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed, implement the steps of the indoor positioning algorithm testing method according to any one of claims 1-7.

Images