Indoor positioning method for mobile terminals combining wireless base station field strength and sensor information

Abstract

A mobile terminal indoor positioning method combining wireless base station signal strength and sensor information is disclosed. The method involves the mobile terminal receiving base station signal fingerprint information and locally acquiring terminal sensor information from the current and previous moments. After filtering, the final positioning result for the current moment is calculated by traversing the positioning neighborhood from the previous moment. During movement, the terminal uses this method to iterate continuously, achieving continuous positioning of the terminal's trajectory. This invention improves data characteristic and enhances positioning accuracy by predicting the direction of travel using sensor information based on the base station power signal acquired by the mobile terminal.

Description

Technical Field

[0001] This invention relates to a technology in the field of wireless positioning, specifically to an indoor positioning method for mobile terminals that combines wireless base station field strength with sensor information. It is designed to address the unique characteristics of wireless base station power data, and improves data distinctiveness and positioning accuracy by predicting the direction of travel through sensor information. Background Technology

[0002] In wireless positioning methods, fingerprint-based positioning is widely used. However, mobile signals generated by 5G and other mobile base stations propagate based on remote radio frequency units (antennas). Within a given area, the signal characteristics are weak, making it difficult to estimate the distance to the base station using signal strength. This results in lower accuracy for positioning methods relying on the RSSI of 5G base station signals. Furthermore, the uncertainty of received signal power or path attenuation due to environmental factors and physical factors such as signal reflection and refraction during wireless signal propagation in space leads to similar or identical received signal power in certain adjacent areas or areas with geometric symmetry. This complicates practical wireless positioning problems and significantly impacts positioning accuracy. Summary of the Invention

[0003] To address the aforementioned shortcomings of existing technologies, this invention proposes an indoor positioning method for mobile terminals that combines wireless base station field strength with sensor information. By using the base station power signal acquired by the mobile terminal and sensor information to predict the direction of travel, the method improves data characteristics and enhances positioning accuracy.

[0004] This invention is achieved through the following technical solution:

[0005] This invention relates to an indoor positioning method for mobile terminals that combines wireless base station field strength with sensor information. The method involves the mobile terminal receiving base station signal fingerprint information and locally collecting terminal sensor information from the current and previous moments. After filtering, the final positioning result for the current moment is calculated by traversing the positioning neighborhood of the previous moment. During the movement of the terminal, the above method is used to perform cyclic traversal to achieve continuous positioning of the terminal trajectory.

[0006] The base station signal fingerprint information refers to the received power of the radio frequency signals of surrounding base stations received by the mobile terminal and maintained in real-time connection, preferably the RSSI value of the surrounding base station radio frequency signals; specifically, it is obtained as follows: at any specific location in a specific scenario, when the mobile terminal is at that location, it is assumed that it can receive signals transmitted by N surrounding base stations. The received RSSI values ​​of these N base stations are arranged into a two-dimensional data list. The first column of the data list is the base station ID number of these N base stations, which is arranged in ascending order according to the base station ID number; the second column of the data list is the received RSSI value corresponding to these N base stations.

[0007] The terminal sensor information refers to the information output in real time by the IMU sensor built into the mobile terminal, which includes the relative orientation and location information of the mobile terminal at the current moment. Specifically, it is obtained by combining map information, sensor information, base station signal fingerprint information, and the positioning result of the previous moment to obtain the positioning result of the mobile terminal at the current moment relative to the previous moment, so that the positioning result at the current moment and the positioning result at the previous moment maintain the continuity of the positioning trajectory.

[0008] The location neighborhood refers to the location grid where the location neighborhood of the mobile terminal at the previous moment is the base point of the location grid. The location neighborhood corresponds to a one-step reachable grid point in the geographic space that is centered on the base point of the location grid.

[0009] The aforementioned one-step reachable grid points refer to: in two-dimensional positioning space, corresponding to 8 surrounding grid points in two-dimensional space; and in three-dimensional positioning space, corresponding to 26 surrounding grid points in three-dimensional space.

[0010] The filtering process refers to: using the positioning grid point of the mobile terminal at the previous moment as the base point, and using the positioning grid point and its positioning neighborhood as candidate grid points for the positioning result at the current moment, the current positioning result is estimated using an optimized fusion positioning method. This estimation process is the filtering process for positioning grid points, and the final estimated grid points are the filtering results for positioning.

[0011] The traversal refers to: taking the positioning grid point of the mobile terminal at the previous moment as the base point, calculating the corresponding fusion positioning result of all possible neighboring grid points, and selecting the neighboring grid points that meet more fusion conditions as the final positioning grid result at the current moment through traversal comparison; when the terminal is in motion, the grid points after each step of positioning and filtering are traversed in a loop until the precise positioning motion trajectory is finally obtained.

[0012] The aforementioned loop traversal refers to: estimating the positioning result at the current moment based on base station signal fingerprint information and terminal sensor information using an optimized fusion positioning method, and looping until the motion trajectory positioning estimation ends.

[0013] The optimized fusion localization method refers to performing directional fusion, threshold fusion, and neighborhood fusion according to priority during the localization process. Specifically, the localization result after directional fusion is used as the initial value for threshold fusion, and the localization result after threshold fusion is used as the initial value for neighborhood fusion. The final result obtained is the optimized fusion localization result.

[0014] Technical effect

[0015] This invention effectively fuses received signals from the mobile communication network's RSSI (Receiving Signal Indicator) with sensor information from the mobile terminal. Particularly when the mobile terminal is in continuous motion, the localization and trajectory search for continuous motion are constrained within the neighborhood covered by the mobile terminal's sensors. This allows localization results based solely on the mobile communication network's RSSI information and those based solely on the mobile terminal's sensor information to compensate for each other's deficiencies. Furthermore, it utilizes only reference points within the neighborhood for calculations, enhancing the specificity of localizing reference points within the neighborhood in practical applications. Attached Figure Description

[0016] Figure 1 These are schematic diagrams illustrating three different signal enhancement methods of the present invention;

[0017] Figure 2 This is a comparison chart of the accuracy of different signal reinforcement methods used in actual tests;

[0018] In the figure: (a) represents the localization accuracy of KNN combined with different reinforcement methods, and (b) represents the localization accuracy of Bayes method combined with different reinforcement methods. Detailed Implementation

[0019] Example 1

[0020] like Figure 1 As shown in this embodiment, a mobile terminal indoor positioning method combining wireless base station field strength and sensor information includes the following steps:

[0021] Step 1) Fingerprint dataset collection: Assuming there are N positioning grid locations in the area to be located, at each location in the area to be located, the mobile terminal collects a large amount of average values ​​of the surrounding base station IDs (PCI numbers of the base stations) and the signal field strength (power) data (dBm) of the surrounding base stations received by the mobile terminal at a certain sampling rate as reference fingerprint data for positioning.

[0022] Step 2) Construction of the fingerprint dataset: Arrange the average values ​​of the surrounding base station IDs and signal strength data of the surrounding base stations collected at each location in the area to be located in Step 1) into a table, as shown in the table below, which represents the fingerprint information F of the nth positioning grid. nn = 1, 2, ..., N:

[0023]

[0024] In the table, ID k is the base station ID number, arranged in ascending order; It represents the average value of the field strength data of the k-th base station received in the n-th positioning grid, where n = 1, 2, ..., N; k = 1, 2, ..., K.

[0025] Step 3) Actual test data collection: During the actual test, the signals received by the mobile terminal are sampled at a certain frequency while the mobile terminal is moving, including: the ID number of the surrounding base stations, the instantaneous signal field strength corresponding to each surrounding base station, the sampling time interval of the mobile terminal, the instantaneous azimuth angle of the mobile terminal, the instantaneous velocity vector of the mobile terminal, and the instantaneous acceleration vector of the mobile terminal.

[0026] Step 4) Determining the initial location: At the initial location of the actual test, the distance between fingerprints based on the KNN algorithm or Bayesian algorithm is selected as the metric. All N positioning grids in the positioning area are traversed, and the grid with the minimum metric or the shortest distance is the initial location of the current positioning.

[0027] Step 5) Location Information Enhancement: In actual testing, after the mobile terminal leaves its initial location, it can obtain information including the ID numbers of surrounding base stations, the instantaneous signal strength of each surrounding base station, the sampling time interval Δt of the mobile terminal, and the instantaneous azimuth angle θ of the mobile terminal. i (t) Instantaneous velocity vector v of the mobile terminal i (t) Instantaneous acceleration vector a of the mobile terminal i (t), combined with the location information from the previous sampling time, can be used to fuse and enhance the location information, specifically including:

[0028] 5-1) Fusion of orientation information from mobile terminal sensor data: based on the instantaneous acceleration vector a of the mobile terminal i Multiplying (t) by the sampling time interval Δt yields an estimate of the instantaneous velocity vector v of the mobile terminal. i_est (t)=a i (t)×Δt;

[0029] 5-2) Compare this estimate with the instantaneous velocity vector v of the mobile terminal. i (t) is weighted and fused to obtain the estimated value v of the instantaneous velocity vector of the mobile terminal at the current moment. i_est_update (t)=a1×v i_est (t)+a2×v i(t), where a1 and a2 are two weighted values, a1+a2=1, and a1>0, a2>0;

[0030] 5-3) Multiply the updated velocity vector estimate by the sampling time interval Δt to obtain the displacement vector estimate d from the previous sampling time to the current sampling time. i_est (t)=v i_est_update (t)×Δt;

[0031] Step 6) Optimized fusion positioning method of mobile terminal sensor data and fingerprint information: When the fingerprint information received at the current moment is F i The corresponding signal field strength (power) is Classification algorithms (such as KNN or Bayesian algorithms) can be used to calculate the distance D between the fingerprint information and the fingerprints at various locations in the area to be located. in (n=1,2,...,N), and update the value d of the instantaneous displacement vector estimate of the mobile terminal sensor. i_est_update (t) is optimized and integrated with fingerprint information.

[0032] The optimized fusion refers to the following: during the process of determining the positioning grid from the previous time to the current time, the positioning grid result that can complete the most fusion processing conditions is selected as the preferred object. When only the above two fusion processing conditions are executed, the priority is: directional fusion processing + neighborhood fusion processing is better than directional fusion processing + threshold fusion processing is better than threshold fusion processing + neighborhood fusion processing. When only one of the above fusion processing conditions is executed, the priority is: neighborhood fusion processing is better than threshold fusion processing is better than directional fusion processing.

[0033] The aforementioned direction fusion processing refers to: comparing the estimated displacement vector value with the instantaneous azimuth angle θ of the mobile terminal. i (t) Perform direction information fusion to obtain the updated value of the mobile terminal's instantaneous displacement vector estimate at the current moment.

[0034] d i_est_update (t)=(a3×θ i (t)+a4×Arg(d i_est (t)))×||d i_est (t)||, where: ||.|| is the modulus operation, Arg(.) is the argument operation, a3 and a4 are the two weights, a3+a4=1, and a3>0, a4>0.

[0035] The threshold fusion processing mentioned above refers to: when the threshold d of the instantaneous displacement vector of any mobile terminal... th The modulus of the updated instantaneous displacement vector estimate after direction fusion is ||d i_est_update(t)|| and the threshold d th Comparison: When the threshold is exceeded, the magnitude of the instantaneous displacement vector is set to the threshold, but the argument remains unchanged. The result is vector d. i_est_update (t); If the threshold is not exceeded, the updated value is retained.

[0036] The neighborhood fusion processing refers to: fusing the estimated displacement vector d after threshold fusion processing. i_est_update (t) is superimposed with the previous positioning result to obtain the initial grid position for the current positioning. Then, a neighborhood grid of the previous positioning result is defined based on map information, typically selected as 8 grids adjacent to the previous positioning grid in two-dimensional space, or 26 grids adjacent to the previous positioning grid in three-dimensional space. If the initial grid position of the current positioning does not belong to the above-mentioned neighborhood grid, then the previous positioning grid is used as the starting point, and the value d is updated using the estimated instantaneous displacement vector of the mobile terminal at the current moment. i_est_update The directional phasor of (t) (a3×θ) i (t)+a4×Arg(d i_est (t))) is the direction of mobile terminal movement from the previous positioning grid to the current positioning grid. The neighborhood of the center point of each of the above neighborhood grids that is closest to the phasor of this direction is taken as the neighborhood fusion grid point of the current positioning.

[0037] Step 7) Iterate and dynamically optimize the localization: In the next sampling and localization process, take the result of the previous localization as the starting localization position, and then iterate and dynamically repeat the above steps 5 and 6 to obtain the next localization result until the localization process ends.

[0038] Example 2

[0039] This embodiment uses a Redmi K30u mobile terminal for actual testing, and the test area is an indoor hall scene.

[0040] Step 1) Collect fingerprint data. The test scenario consists of 30 grids, each with a side length of 3 meters × 3 meters. Data is collected 5 times per second per grid, i.e., the sampling frequency is set to 5Hz, and the sampling interval is 0.2 seconds (i.e., Δt = 0.2 seconds). The data collection time for each grid lasts for 60 seconds, meaning that 300 data points are collected per grid. The IDs (PCI numbers) of 10 surrounding base stations and the signal strength (power) data (dBm) of the surrounding base stations received by the mobile terminal are recorded to create the fingerprint dataset.

[0041] Step 2) Analyze the data collected in Step 1) for the nth (n = 1, 2, ..., 30) grid according to the following table:

[0042]

[0043] Among them, 135 (ID 1), 167 (ID 2), ..., 378 (ID 10) are the ID numbers of the 10 surrounding base stations received by the mobile terminal. These are the average values ​​of the 300 base station signal strength (power) data points corresponding to these 10 base station IDs. This constructs the base station signal strength dataset for all 30 grid points.

[0044] Step 3) Based on the above, begin testing in a real-world scenario. First, data collection needs to be performed on the mobile terminal within the test scenario. This involves the tester holding the test mobile terminal and walking normally within the test scenario. During this movement, the received signal data of the mobile terminal is collected at the same 5Hz sampling frequency. This data includes: the ID numbers of surrounding base stations, the instantaneous signal strength corresponding to each surrounding base station, and the instantaneous azimuth angle θ of the mobile terminal. i (t) Instantaneous velocity vector v of the mobile terminal i (t) Instantaneous acceleration vector a of the mobile terminal i (t).

[0045] Step 4) First, determine the initial location. In this step, only two sets of information are used: the ID numbers of the surrounding base stations receiving the current signal and the instantaneous signal strength corresponding to each surrounding base station. Arrange the ID numbers of the surrounding base stations receiving the signal in the order given in Step 2), and let the instantaneous signal strength of each corresponding base station ID be... Traditional fingerprint localization methods are used for positioning. Here, the distance between fingerprints based on the KNN algorithm or Bayesian algorithm is selected as the metric. In the KNN algorithm, the parameter is selected as K=5. All 30 positioning grids in the positioning area are traversed, and the grid with the smallest metric or the shortest distance is selected as the initial position of the current positioning.

[0046] Step 5) Strengthen the location information of the mobile terminal.

[0047] 5.1) Fusion of directional information from mobile terminal sensor data: based on the instantaneous acceleration vector a of the mobile terminal i Multiplying (t) by the sampling time interval Δt (in this embodiment, Δt = 0.2 seconds) yields the estimated value v of the instantaneous velocity vector of the mobile terminal. i_est (t)=0.2×a i (t).

[0048] 5.2) Perform velocity vector fusion of mobile terminal sensor data: fuse the estimated instantaneous velocity vector v of the mobile terminal. i_est (t) and the instantaneous velocity vector v of the mobile terminal i(t) Perform weighted fusion, and set both weights a1 and a2 to 0.5 to obtain the estimated value v of the instantaneous velocity vector of the mobile terminal at the current moment. i_est_update (t)=0.5×v i_est (t)+0.5×v i (t).

[0049] 5.3) Perform displacement vector fusion of mobile terminal sensor data: fuse the estimated instantaneous velocity vector v of the mobile terminal. i_est_update Multiplying (t) by the sampling time interval Δt (in this embodiment, Δt = 0.2 seconds) yields the estimated displacement vector d from the previous sampling time to the current sampling time. i_est (t)=0.2×v i_est_update (t).

[0050] Step 6) Optimize, fuse, and enhance the mobile terminal positioning information accordingly. Specifically, in the process of determining the positioning grid from the previous moment to the current moment, the positioning grid result that can complete the most fusion processing conditions is selected as the preferred option. When only the above two fusion processing conditions are executed, the priority is: directional fusion processing + neighborhood fusion processing is better than directional fusion processing + threshold fusion processing is better than threshold fusion processing + neighborhood fusion processing. When only one of the above fusion processing conditions is executed, the priority is: neighborhood fusion processing is better than threshold fusion processing is better than directional fusion processing.

[0051] The direction fusion process refers to: converting the displacement vector estimate d obtained in step 5) into a tangible vector fusion value. i_est (t) and the instantaneous azimuth angle θ of the mobile terminal i (t) Perform directional information fusion, and set two weighting values ​​a3 and a4 to 0.5 to obtain the updated value d of the estimated instantaneous displacement vector of the mobile terminal at the current time. i_est_update (t)=(0.5×θ i (t)+0.5×Arg(d i_est (t)))×||d i_est (t)||, where:

[0052] ||.|| is the modulo operation, and Arg(.) is the argument operation.

[0053] The threshold fusion processing mentioned above refers to: selecting the threshold d of the instantaneous displacement vector of the mobile terminal. th = 3 meters, the magnitude of the updated instantaneous displacement vector estimate after direction fusion ||d i_est_update (t)|| and the threshold d th Comparison: When

[0054] ||d i_est_updateIf (t)||>3 meters, then the magnitude of the instantaneous displacement vector ||d i_est_update (t)||Reset to 3 meters, while keeping its argument unchanged, the result is still expressed using vector d. i_est_update (t) is used to represent; when ||d i_est_update If (t)||≤3 meters, then retain the updated value d of the original instantaneous displacement vector estimate. i_est_update (t) remains unchanged.

[0055] The neighborhood fusion processing refers to: fusing the estimated displacement vector d after threshold fusion processing. i_est_update (t) is superimposed with the previous positioning result to obtain the initial grid position of the current positioning. The neighborhood grid of the previous positioning result is then defined based on map information. In this embodiment, eight grids adjacent to the previous positioning grid in two-dimensional space are selected (with the forward direction pointed to by the mobile terminal as the forward direction, i.e., forward, backward, left, right, left-front, left-back, right-front, and right-back, a total of eight directions). If the initial grid position of the current positioning does not belong to the above-mentioned neighborhood grids, the grid of the previous positioning is used as the starting point, and two weighting values ​​a3 and a4 are both set to 0.5. The value d is updated with the estimated instantaneous displacement vector of the mobile terminal at the current moment. i_est_update The directional phasor of (t) (0.5×θ) i (t)+0.5×Arg(d i_est (t))) is the direction of mobile terminal movement from the previous positioning grid to the current positioning grid. The neighborhood of the center point of each of the above-mentioned neighborhood grids that is closest to the phasor of this direction is taken as the neighborhood fusion grid point of the current positioning, that is... Where: j = 1, 2, ..., 8, (x j ,y j ) represents the coordinates of the center point of the 8 neighboring grids, d j The value d represents the position of the center point of the 8 neighboring grids, updated with the estimated instantaneous displacement vector of the mobile terminal at the current moment, starting from the previous positioning result. i_est_update (t) represents the perpendicular distance of the ray in the direction (x, y). opt This is the current location grid optimization result after neighborhood fusion.

[0056] Step 7) Iterate through the dynamic optimization localization process. In the sampling and localization process of the next moment, take the localization result of the previous moment as the starting localization position, and then iterate through and dynamically repeat the above steps 5 and 6 to obtain the next localization result until the localization process ends.

[0057] After obtaining the sampled test data of the above actual scenario, as the object of performance comparison, the traditional KNN algorithm was first used, and the KNN parameter was optimized and set to K=5, and the localization was performed without adding any constraints. Figure 2 (a) shows the localization accuracy of KNN combined with different reinforcement methods. Due to poor signal characteristics, the overall localization accuracy is low, only 65.52%. When neighborhood reinforcement is adjusted and threshold reinforcement is added, the localization accuracy reaches 68.01%. Further adding neighborhood constraints increases the localization accuracy to 70.15%. When using the newly added directional information from the sensor, the localization accuracy can reach 72.03% when the direction is poor and 76.94% when the direction information is good. It can be seen that the performance is significantly improved compared to the original localization accuracy of 65.52% under the unconstrained condition.

[0058] Using the same reinforcement method, practical tests were conducted under the Bayes algorithm. Figure 2 (b) shows the localization accuracy of the Bayes method combined with different reinforcement methods. It can be seen that the accuracy is 52.95% before using the reinforcement methods. After fusing the Bayes method with neighborhood and reinforcement methods, the accuracy reaches 58.98%. Furthermore, after fusing the Bayes method with orientation and neighborhood information, the accuracy reaches 63.66%, demonstrating a significant performance improvement.

[0059] A comprehensive test was conducted on a large number of test and training sets. The obtained localization results were compared with the actual localization values ​​to calculate the average localization accuracy. Under the KNN algorithm, the average localization accuracy before and after augmentation was 6.1m and 4.7m, respectively; while under the Bayes algorithm, the average localization accuracy before and after augmentation was 7.3m and 4.9m, respectively. Both algorithms showed a significant improvement in accuracy.

[0060] Compared with existing technologies, this method combines the received base station signal power with the sensor information of the receiving terminal to generate neighborhood constraints in the positioning scenario. This removes positioning interference caused by flying points in non-adjacent areas on the map during continuous positioning, thereby significantly improving the accuracy and continuity of positioning. This invention not only relies on the received signal strength data itself but also expands the information and dimensions of the reference signal, predicting and constraining factors such as travel direction, threshold, and neighborhood, thereby improving signal characteristics and the overall positioning accuracy of the mobile terminal.

[0061] The above-described specific implementations can be partially adjusted by those skilled in the art in different ways without departing from the principles and purpose of the present invention. The scope of protection of the present invention is defined by the claims and is not limited to the above-described specific implementations. All implementation schemes within the scope of the claims are bound by the present invention.

Claims

1. A method for indoor positioning of a mobile terminal combining wireless base station field strength and sensor information, characterized in that, The mobile terminal receives base station signal fingerprint information and locally collects terminal sensor information at the current and previous times. After filtering, it calculates and obtains the final positioning result at the current time by traversing the positioning neighborhood of the previous time. The terminal uses the above method to traverse continuously during the movement to achieve continuous positioning of the terminal trajectory. The base station signal fingerprint information refers to the received power of the radio frequency signal of the surrounding base station that is received by the mobile terminal and is in a real-time connection state. The aforementioned location neighborhood refers to: if the location neighborhood of the mobile terminal at the previous moment is taken as the base point of the location grid, then the location neighborhood corresponds to a one-step reachable grid point in geographic space centered on the base point of the location grid; The filtering process refers to: using the positioning grid point of the mobile terminal at the previous moment as the base point, and using the positioning grid point and its positioning neighborhood as the candidate grid points for the positioning result at the current moment, the positioning result at the current moment is estimated using an optimized fusion positioning method. This estimation process is the filtering process for positioning grid points, and the grid points obtained by the final estimation are the filtering results for positioning. The optimized fusion localization method refers to the following: during the localization process, directional fusion, threshold fusion, and neighborhood fusion are performed according to priority. Specifically, the localization result after directional fusion is used as the initial value for threshold fusion, and the localization result after threshold fusion is used as the initial value for neighborhood fusion. The final result obtained is the optimized fusion localization result. The optimized fusion positioning method is as follows: during the process of determining the positioning grid from the previous time to the current time, the positioning grid result that can complete the most fusion processing conditions is selected as the preferred object. When only the above two fusion processing conditions are executed, the priority is: directional fusion processing + neighborhood fusion processing is better than directional fusion processing + threshold fusion processing is better than threshold fusion processing + neighborhood fusion processing. When only one of the above fusion processing conditions is executed, the priority is: neighborhood fusion processing is better than threshold fusion processing is better than directional fusion processing.

2. The indoor positioning method for mobile terminals according to claim 1, characterized in that, The base station signal fingerprint information is the RSSI value of the radio frequency signals of surrounding base stations; specifically, it is obtained in the following way: at any specific location in a specific scenario, when the mobile terminal is at that location, assuming that it can receive signals transmitted by N surrounding base stations, the RSSI values ​​of these N base stations are arranged into a two-dimensional data list. The first column of the data list is the base station ID number of these N base stations, and their order is arranged in ascending order according to the size of the base station ID number; the second column of the data list is the received RSSI value corresponding to these N base stations.

3. The indoor positioning method for mobile terminals according to claim 1, characterized in that, The terminal sensor information refers to the information output in real time by the IMU sensor built into the mobile terminal, which includes the relative orientation and location information of the mobile terminal at the current moment. Specifically, it is obtained by combining map information, sensor information, base station signal fingerprint information, and the positioning result of the previous moment to obtain the positioning result of the mobile terminal at the current moment relative to the previous moment, so that the positioning result at the current moment and the positioning result at the previous moment maintain the continuity of the positioning trajectory.

4. The indoor positioning method for mobile terminals according to claim 1, characterized in that, The traversal refers to: taking the positioning grid point of the mobile terminal in the previous moment as the base point, calculating the corresponding fused positioning result of all possible neighboring grid points, and selecting the neighboring grid point that meets more fusion conditions as the final positioning grid result of the current moment through traversal comparison. When the terminal is in motion, the grid points after each step of positioning and filtering are traversed in a loop until the precise positioning motion trajectory is finally obtained. The aforementioned loop traversal refers to: estimating the positioning result at the current moment based on base station signal fingerprint information and terminal sensor information using an optimized fusion positioning method, and looping until the motion trajectory positioning estimation ends.

5. The indoor positioning method for mobile terminals according to claim 1, characterized in that, The aforementioned direction fusion processing refers to: combining the displacement vector estimate with the instantaneous azimuth angle θ of the mobile terminal. i (t) Perform direction information fusion to obtain the updated value d of the mobile terminal's instantaneous displacement vector estimate at the current moment. i_est_update (t)=(a3×θ i (t)+a4×Arg(d i_est (t)))×||d i_est (t)||, where: ||.|| is the modulus operation, Arg(.) is the argument operation, a3 and a4 are the two weights, a3+a4=1, and a3>0, a4>0; The threshold fusion processing mentioned above refers to: when the threshold d of the instantaneous displacement vector of any mobile terminal... th The modulus of the updated instantaneous displacement vector estimate after direction fusion is ||d i_est_update (t)|| and the threshold d th Comparison: When the threshold is exceeded, the magnitude of the instantaneous displacement vector is set to the threshold, but the argument remains unchanged. The result is vector d. i_est_update (t); If the threshold is not exceeded, the updated value is retained; The neighborhood fusion processing refers to: fusing the estimated displacement vector d after threshold fusion processing. i_est_update (t) The initial grid position of the current location is obtained by superimposing it with the previous location result; and the neighborhood grid of the previous location result is delineated according to the map information, which is selected as 8 grids adjacent to the previous location grid in two-dimensional space, or 26 grids adjacent to the previous location grid in three-dimensional space; if the initial grid position of the current location does not belong to the above neighborhood grid, then the grid of the previous location is used as the starting point, and the value d is updated with the estimated value of the instantaneous displacement vector of the mobile terminal at the current time. i_est_update The directional phasor of (t) (a3×θ) i (t)+a4×Arg(d i_est (t))) is the direction of mobile terminal movement from the previous positioning grid to the current positioning grid. The neighborhood of the center point of each of the above neighborhood grids that is closest to the phasor of this direction is taken as the neighborhood fusion grid point of the current positioning.

Images