A three-dimensional positioning method for unknown interference in NLOS environment

Abstract

The present invention belongs to the field of indoor positioning technology, and specifically relates to a three-dimensional positioning method for unknown interference in an NLOS environment. In the NLOS environment, the present invention establishes a three-dimensional constraint optimization positioning model based on known ranging information, establishes constraints through the positioning results of the linear least squares method, and improves the search range of the particle swarm optimization, chicken swarm optimization, and ant lion optimization algorithm accordingly to solve the proposed positioning method. The following steps are included: 4 fixed base stations and a mobile tag point to be found are set indoors, and a spatial three-dimensional coordinate system is established with a point indoors as the origin; the three-dimensional coordinate values ​​of the four base stations and the distance values ​​of each base station and the tag node are obtained in an ultra-wideband communication mode; the three-dimensional coordinates of the tag node are obtained by the least squares method, the two-dimensional coordinates except the height coordinates are retained, and the height value of the tag node is recalculated by constraint optimization.

Description

Technical Field

[0001] The invention belongs to the technical field of indoor positioning, and in particular relates to a three-dimensional positioning method for unknown interference in a NLOS environment. Background Art

[0002] Indoor positioning technology is a key factor driving many of the current innovative applications. Compared with positioning technologies based on received signal strength (RSS) (such as Bluetooth, WiFi and radio frequency identification), ultra-wide band (UWB) positioning technology has higher positioning accuracy. UWB technology has extremely high temporal resolution, so positioning is often achieved through ranging. However, in a non-line-of-sight (NLOS) environment, the positioning target is very likely to overlap with various objects. Although UWB technology has strong penetration capabilities, it still produces certain ranging errors, and the performance of various positioning algorithms is also affected.

[0003] In order to improve the positioning accuracy under NLOS interference, the current implementation methods are mainly divided into four categories: accurate modeling of NLOS errors, error correction based on data learning, NLOS path identification, and integration of other sensor data to achieve positioning. These methods have effectively improved the positioning accuracy under specific conditions, but they also have certain shortcomings. For example, machine learning can extract features from ranging data in depth, but it needs to rely on a large amount of sample data to train parameter networks in order to achieve high-performance learning efficiency. By pre-calibrating and calibrating the distribution characteristics of NLOS errors, the literature "Shi X, Mao G, Yang Z, et al. MLE-based localization and performance analysis in probabilistic LOS / NLOS environment [J]. Neurocomputing, 2017, 270: 101-109" can achieve the best positioning effect. However, the actual environment is more complex and changeable, and a time-consuming and expensive testing process is required to fit the accurate prior distribution of errors, which leads to excessively high algorithm costs and limited applications. Summary of the invention

[0004] In view of the above problems, the present invention proposes a three-dimensional positioning method for unknown interference in a NLOS environment.

[0005] The technical solution of the present invention is:

[0006] A three-dimensional positioning method for unknown interference in a NLOS environment comprises the following steps:

[0007] S1. Set up L fixed base stations in the positioning area, L ≥ 3;

[0008] S2. Select a point in the positioning area as the origin and establish a three-dimensional spatial coordinate system;

[0009] S3, through the ultra-wideband communication mode, obtain the three-dimensional coordinate values ​​of all base stations and the distance data from each base station to the target r = [r1, r2, ..., r L ],r l is the ranging data of the lth base station, l = 1, 2, ..., L, and the position of the lth base station is x l =[x l y l z l ] T ;

[0010] S4, the target two-dimensional coordinates are estimated by LLS method. The cost function for establishing the target height is:

[0011]

[0012] The target height estimation result is obtained by using the swarm intelligence optimization algorithm:

[0013]

[0014] The positioning result is

[0015] S5. Establish constraint conditions based on the positioning result obtained in step S4:

[0016]

[0017] Among them, △ x ,△ y ,△ z To surround Deviation of setting;

[0018] Define the distance difference between different base stations and the target as {△r ij |1≤i≤L,1≤j≤L,i≠j}, the theoretical distance difference between different base stations and the target is △r ij and It is expressed as:

[0019] △r ij =r i -r j

[0020]

[0021] Where (x, y, z) is the coordinates of the target to be located;

[0022] Select the base station with serial number j as the reference base station, j = 1, 2, ..., L, reduce the deviation between the measured data difference of the remaining base stations to the reference base station j and the theoretical data difference, and establish the cost function

[0023]

[0024] in, The data of all base stations are fully traversed to reduce the deviation through the cost function, and the optimization result is:

[0025]

[0026] in, It represents the target positioning result solved by the cost function under the jth reference base station. It means that the final target position coordinates are obtained by averaging the positioning results of all groups.

[0027] The beneficial effect of the present invention is that the present invention establishes two types of optimization positioning models, which do not rely on prior information of errors and can achieve three-dimensional positioning in LOS and NLOS environments. The measured results show that: 1) compared with other positioning methods, the method of the present invention has higher positioning accuracy and is insensitive to base station deployment; 2) compared with the positioning results of particle swarm optimization, chicken swarm optimization, and ant lion optimization algorithms with improved search range, the positioning method of the present invention can achieve more stable positioning performance with less time consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a CDF comparison of the absolute positioning error between the method of the present invention and other methods.

[0029] Figure 2 This is the RMSE comparison between the method of the present invention and other methods.

[0030] Figure 3 It is the CDF comparison between the method of the present invention and other methods.

[0031] Figure 4 Different parameter values z positioning performance. DETAILED DESCRIPTION

[0032] The present invention is described in detail below with reference to the accompanying drawings and embodiments:

[0033] Example

[0034] In this example, the positioning environment is set as a 7.6m×9.2m×3m laboratory, in which four UWB base stations are placed with coordinates of (0, 0, 1.3)m, (5, 0, 1.7)m, (0, 5, 1.7)m, and (5, 5, 1.3)m, and a movable tag is placed for positioning.

[0035] In the LOS environment, the tags were placed at 324 different locations, and several sets of distance measurement data at the same location formed a test file, with a total of 77,046 sets of distance data collected. In the NLOS environment, 37,956 sets of distance data were measured at 231 test points, and NLOS interference was completely unknown. At the above test points, the tags were placed at different heights of 0.88m, 1.3m, 1.7m, and 2m. The distance measurement data was pre-processed by program cleaning to exclude abnormal and sudden distance measurement values.

[0036] Model building

[0037] L (L ≥ 3) UWB base stations are deployed in space, and the measurement model can be expressed as

[0038] r=f(x)+n (1)

[0039] Where r = [r1, r2, …, r L ] is the measurement vector, f is a cluster of known nonlinear functions about the target position, x = [xyz] T is the position coordinate of the target, n=[n1,n2,…,n L ] is the measurement error, the position of the lth base station is x l =[x l y l z l ] T , the distance between the base station and the target is calculated using a two-way ranging technology based on time of flight (TOF). Then formula (2) is further expressed as

[0040]

[0041] Where d l is the distance between the base station and the target.

[0042] The LLS method is used as the basic positioning method and extended to three-dimensional positioning:

[0043] First, transform equation (3) into a set of linear equations with zero mean perturbations:

[0044]

[0045] At the same time, the error component m is introduced l and the dummy variable R

[0046]

[0047] R = x 2 +y 2 +z 2 (5)

[0048] Substituting equation (5) (6) into equation (4), we get

[0049]

[0050] Write equation (7) in matrix form,

[0051] θ=[xyz R] T (7)

[0052] q=[m1 m2…m L ] T (8)

[0053]

[0054]

[0055] Then formula (7) can be written as

[0056] Aθ+q=b (11)

[0057] Ignoring the quadratic term, we get

[0058]

[0059] Assuming that the measurement noise has zero mean, the cost function of the linear least squares method is obtained

[0060]

[0061] Formula (14) is about The quadratic function of Take the derivative and set the result to 0 to get the LLS estimate

[0062]

[0063] Therefore, the positioning result is expressed as

[0064]

[0065] When the target is in the same position, the UWB system can collect multiple sets of distance data. If there is no real-time positioning output requirement, cluster analysis can be performed on the multiple sets of position coordinates calculated to reduce the impact of abnormal solutions on the results.

[0066] The positioning method of the present invention is:

[0067] 1) Positioning algorithm design under LOS environment

[0068] First, the x and y coordinates of the target are solved by the LLS algorithm. Then the target height is solved by establishing and optimizing the cost function.

[0069] The cost function is constructed as

[0070]

[0071] In the formula, is the LLS estimate of the target two-dimensional coordinates. The result is expressed as

[0072]

[0073] The positioning result is

[0074] 2) Design of anti-interference positioning algorithm in NLOS environment

[0075] a) Design constraints

[0076] In the NLOS environment, a set of rough three-dimensional coordinates of the target is calculated using the method in the LOS environment, expressed as [x LLS ,y LLS ,z opt ] T , establish constraints

[0077]

[0078] Formulas (19), (20), and (21) can be transformed into six inequality constraints, △ x ,△ y ,△ z For around x LLS ,y LLS ,z opt Set deviation.

[0079] b) Constructing the cost function

[0080] Let the difference in measured distance from different base stations to the target be {△r ij |1≤i≤L,1≤j≤L,i≠j}, let the theoretical distance difference from different base stations to the target be Then △r ij and It can be expressed as

[0081] △r ij =r i -r j (twenty one)

[0082]

[0083] Select the base station with serial number j (j = 1, 2, ..., L) as the reference base station, reduce the deviation between the measured data difference of the remaining base stations to the reference base station j and the theoretical data difference, and establish the cost function

[0084]

[0085] In the formula

[0086] Formula (17) and (24) are used as the fitness functions of the three methods, and the optimization type is to find the minimum value. In the three groups, the fitness value of the individual is calculated and each possible solution is regarded as the position of the individual. Through three different heuristic mechanisms, the optimization is iterated to finally obtain the individual with the best fitness value, and its position is the target coordinate. The data of all base stations is fully traversed to reduce the deviation through formula (24), and the optimization result is expressed as

[0087]

[0088] In the formula It represents the target positioning result obtained by formula (24) under the jth reference base station, and formula (26) represents the final target position coordinates obtained by averaging the positioning results of all groups.

[0089] Positioning method solution

[0090] 1) Input the ranging data r between all base stations and targets in the space and the location coordinates of the base stations.

[0091] 2) Calculate the target two-dimensional coordinates [x LLS ,y LLS ] T , the swarm intelligence optimization method is used to solve equation (17) and obtain the three-dimensional coordinates of the target in the LOS environment.

[0092] 3) In the NLOS environment, through steps 1) and 2) and equation (18), the target rough coordinates [x LLS ,y LLS ,z opt ] T , initialize △ x ,△ y ,△ z Construct constraints and number reference base stations.

[0093] 4) Construct the cost function formula (24) of the reference base station, use the swarm intelligence optimization method to solve it, and obtain a set of positioning results.

[0094] 5) Determine whether the positioning results of all reference base stations have been solved, otherwise repeat 4).

[0095] 6) Take the average of the L groups of positioning results to obtain the three-dimensional coordinates of the target in the NLOS environment.

[0096] In order to fully compare the solving performance of different intelligent optimization algorithms, the population size of the three algorithms is set to 50, and the maximum number of iterations is set to 300. For two types of test files, the above method is used to obtain the positioning results of the method of the present invention. At the same time, the positioning results of the LLS algorithm and the improved least squares algorithm are compared. x ,△ y is 0.2m, z The absolute positioning error, cumulative distribution function (CDF) and root mean square error (RMSE) of the calculated coordinates are used as performance evaluation criteria.

[0097]

[0098] Where (x, y, z) is the actual position of the tag to be located. is the estimated position of the positioning algorithm. The comparison results are shown in Table 1 and Table 2.

[0099] Table 1 Average experimental results of 324 test points / m

[0100]

[0101] Table 2 Average experimental results of 231 test points / m

[0102]

[0103] The actual positioning results show that: Figure 1-Figure 4 As shown, 1) in the LOS environment, when the UWB signal is not blocked, in terms of 3D positioning RMSE, the method of the present invention is about 0.26m and 0.4m lower than LLS and the algorithm in the literature "Wang Guijie, Jiao Liangbao, Cao Xuehong. Research on indoor 3D positioning algorithm based on least squares method [J]. Computer Technology and Development, 2020, 30(04): 69-73" respectively, and has the best positioning performance; 2) in the NLOS environment, the result of the method of the present invention is about 0.07m higher than the traditional LLS and the algorithm in the literature "Wang Guijie, Jiao Liangbao, Cao Xuehong. Research on indoor 3D positioning algorithm based on least squares method [J]. Computer Technology and Development, 2020, 30(04): 69-73" in 2D positioning RMSE, and about 1.66m and 0.30m higher in 3D positioning RMSE respectively; 3) compared with PSO and ALO, the combination of the method of the present invention and CSO for optimization has more practical application value, and the performance comparison is shown in Table 3.

[0104] Table 3 Optimization performance comparison

[0105]

Claims

1. A three-dimensional positioning method for unknown interference in a NLOS environment, characterized in that: The following steps are involved: S1. Set up L fixed base stations in the positioning area, L ≥ 3; S2. Select a point in the positioning area as the origin and establish a three-dimensional spatial coordinate system; S3, through the ultra-wideband communication mode, obtain the three-dimensional coordinate values ​​of all base stations and the distance data from each base station to the target r = [r1, r2, ..., r L ],r l is the ranging data of the lth base station, l = 1, 2, ..., L, and the position of the lth base station is x l =[x l y l z l ] T ; S4, the target two-dimensional coordinates are estimated by LLS method. The cost function for establishing the target height is: The target height estimation result is obtained by using the swarm intelligence optimization algorithm: The positioning result is S5. Establish constraint conditions based on the positioning result obtained in step S4: Among them, △ x ,△ y ,△ z To surround Deviation of setting; Define the distance difference between different base stations and the target as {△r ij |1≤i≤L,1≤j≤L,i≠j}, the theoretical distance difference between different base stations and the target is △r ij and It is expressed as: △r ij ＝r i -r j Where (x, y, z) is the coordinates of the target to be located; Select the base station with serial number j as the reference base station, j = 1, 2, ..., L, reduce the deviation between the measured data difference of the remaining base stations to the reference base station j and the theoretical data difference, and establish the cost function in, The data of all base stations are fully traversed to reduce the deviation through the cost function, and the optimization result is: in, It represents the target positioning result solved by the cost function under the jth reference base station. It means that the final target position coordinates are obtained by averaging the positioning results of all groups.

Images