Anti-interference indoor positioning method and system

Abstract

The application discloses an anti-interference indoor positioning method and system, and relates to the technical field of indoor positioning. The method comprises the following steps: constructing a space model; selecting a near neighbor fingerprint reference point of a signal domain in a test point sub-region by using a multi-resolution method to obtain a signal domain near neighbor reference point set; selecting K near neighbor reference points from the signal domain near neighbor reference point set; obtaining a physical domain near neighbor reference point set by using a line clustering method to further screen the physical domain; and obtaining a position estimation of a test point in a conjugate space by using a WKNN algorithm based on the near neighbor reference point set under the double screening of the signal domain and the physical domain. The conjugate space is introduced to realize multi-level constraints on signal similarity and distribution structure robustness of the near neighbor points, to allow complementary information to be obtained from two dimensions of energy and physics, to provide near neighbor positioning information which is effective in both signal and physical distances, and to balance the signal similarity and the distribution structure robustness.

Description

Technical Field

[0001] This invention relates to the field of indoor positioning technology, and in particular to an anti-interference indoor positioning method and system. Background Technology

[0002] With the increasing intelligence and societal development, the demand for high-precision positioning is becoming increasingly prominent. Although the Global Navigation Satellite System (GNSS) provides accurate real-time navigation services outdoors, the ability to receive satellite signals indoors is limited by the shielding effect of buildings. Therefore, there is an urgent need for an efficient and accurate indoor positioning solution.

[0003] In diverse indoor environments, various access points (APs) exist, and indoor positioning can be achieved using information such as Time of Arrival (TOA), Time Difference of Arrival (TDOA), Angle of Arrival (AOA), Channel State Information (CSI), and Received Signal Strength (RSS). In particular, due to its versatility and independence from dedicated hardware modules, RSS-based Bluetooth or Wi-Fi signals have become the most commonly used signal sources in indoor positioning applications. Fingerprint matching methods using RSS as data have also become a widely used approach among researchers.

[0004] The fingerprint matching process consists of two stages: offline and online. A common fingerprint localization method involves similarity matching between reference points (RPs) in the fingerprint database and test points (TPs) to be located. However, this involves a large amount of computation. To reduce the complexity and time consumption of online localization, clustering is performed offline to simplify the process, thereby reducing online computation and improving localization accuracy. Commonly used clustering methods include K-means, DBSCAN, and DPC, which divide the target region into subspaces and describe each space as a class. Since fingerprint information differs from simple digital quantities due to noise interference, even if signals are similar in the energy domain, they may differ significantly in the physical domain.

[0005] Besides offline clustering, online matching methods also affect the localization results. WKNN utilizes the matching degree between the TP (Technical Neighbor) and the RPs (Representative Points) in the fingerprint database, selecting the K RPs with the highest matching degree for weighted location calculation. However, it lacks consideration for scenarios such as multipath effects, AP aging, and receiver device movement; the selection of fixed nearest neighbor RPs inevitably introduces redundant information or loses important information. Therefore, establishing a fingerprint localization system with strong reliability and high accuracy is particularly important.

[0006] To address the issue of spatial ambiguity, the WKNN method, based on signal similarity and spatial location, matches nearest neighbors in the signal domain. However, this approach neglects the consideration of AP volatility. To resolve test point misjudgments caused by unstable APs, stability and discriminant factors are introduced to assist AP selection. However, since it only considers the ubiquitous signal space and ignores the joint consideration of physical space, the positioning results are significantly biased. Considering the physical trajectory characteristics of signal propagation, the method divides the region into circular boundaries to constrain spatial ambiguity and utilizes the relative RSS of APs to reduce the cost of fingerprint database construction, making it more adaptable to the differences in transmission power caused by AP heterogeneity. However, its positioning accuracy is still easily affected by RSS fluctuations and environmental interference, limiting the generalization ability of the positioning model. Summary of the Invention

[0007] In view of the above-mentioned problems, the present invention is proposed.

[0008] Therefore, the problem that this invention aims to solve is that the positioning accuracy of existing technologies is still easily affected by RSS fluctuations and environmental interference, which limits the generalization ability of the positioning model.

[0009] To solve the above-mentioned technical problems, the present invention provides the following technical solution: an anti-interference indoor positioning method, comprising: determining ground wireless access points and airborne wireless access points within the positioning area; deploying ground fingerprint reference points and airborne fingerprint reference points; assigning a unique physical address (MAC) to each wireless access point; extracting the coexisting wireless access points (APs) between the ground and airborne wireless access points using the MACs; obtaining S MAC addresses through their intersection to determine a ground-plane AP database; obtaining a ground-plane original fingerprint database by combining coordinates from the AP database; collecting the signal strength of a test point TP; coarsely locating a sub-region of TP based on the Euclidean distance of the signal strength between the reference point and TP; obtaining a set of nearest neighbor reference points for the sub-region; selecting nearest neighbor fingerprint reference points in the signal domain within the sub-region using a multi-resolution method; selecting K nearest neighbor reference points from the signal domain nearest neighbor reference point set; further filtering from the physical domain using a linear clustering method to obtain a set of physical domain nearest neighbor reference points; and using the WKNN algorithm to obtain a position estimate of the test point TP in the conjugate space based on the nearest neighbor reference point set filtered by both the signal and physical domains.

[0010] As a preferred embodiment of the anti-interference indoor positioning method of the present invention, the method for determining the ground wireless access points includes detecting N ground wireless access points (APs) on the ground plane during the offline phase within the positioning area Ω. n The deployment method of the ground fingerprint reference points includes deploying M fingerprint reference points RP at equal intervals on the ground plane. m The deployment method of the aerial fingerprint reference points includes deploying reference points in the same manner as on the ground plane, on an empty plane with the same two-dimensional coordinates but at a height of Z. The method for determining the airborne wireless access points includes detecting J wireless access points (APs) in the airspace. j .

[0011] As a preferred embodiment of the anti-interference indoor positioning method of the present invention, the step of selecting the nearest neighbor fingerprint reference point in the signal domain within the sub-region using a multi-resolution method includes calculating the sub-region Ω based on the Pearson coefficient. u All nearest neighbor reference points (RP) k Correlation coefficient ξ with test point TP k Represented as,

[0012]

[0013] Among them, E TP This represents the RSS measurement value of the ground-plane wireless access point included in the test point TP. RSS represents the received signal strength. k Cov(G) represents the RSS measurement of ground-plane wireless access points included in the nearest reference points. k E TP ) represents the nearest neighbor reference point RP k Covariance with the signal strength at the test point TP Represented as ground plane RP k AP collected at the location n Signal variance This represents the AP data collected at test point TP. n Signal variance; based on correlation coefficient ξ k The multi-resolution model is configured to divide the range layer by layer, as expressed by the formula:

[0014]

[0015] Where, ε k Represented as the number of layers; and Let ε represent the lower and upper bounds of each layer, respectively, and g represent the ascending gradient with g > 0; define the ε-th... k Layer resolution is expressed as,

[0016]

[0017] Thus, the resolution is obtained. With the number of layers ε k The mapping relationship is represented as follows:

[0018]

[0019] Here, ε represents a very small non-negative number to prevent the denominator from being 0. Represented as a floor function, based on resolution. Calculate the level ε k Calculate the Euclidean distance between the nearest neighbor reference point and the test point TP in this layer. Select Minimum σ t Using each reference point RP as a confidence neighbor, the final candidate reference point RP set in the signal space is obtained.

[0020] As a preferred embodiment of the anti-interference indoor positioning method of the present invention, wherein: selecting K nearest neighbor reference points from the set of nearest neighbor reference points in the signal domain includes, for the set of reference points in the signal domain Based on the signal Euclidean scale, the reference points RP are sorted in ascending order. The reference point RP with the smallest Euclidean distance is taken as the initial reference point RP. initial From RP initial The groups are initially determined based on the Euclidean distance, with each group selecting three nearest reference points as a set of reference points {RP}. k ,RP k+1 ,RP k+2 Each reference point RP participates in the selection only once, thus resulting in multiple sets of reference points.

[0021] As a preferred embodiment of the anti-interference indoor positioning method of the present invention, the step of using the linear clustering method to further filter the physical domain to obtain the set of physical domain nearest neighbor reference points also includes, for the reference point set {RP} k ,RP k+1 ,RP k+2}, calculate RP respectively k With RP k+1 RP k With RP k+2 RP k+1 With RP k+2 distance The stability of each angle can be determined using the triangle cosine theorem, expressed by the formula:

[0022]

[0023] Where, χ k , χ k+1 , χ k+2 Represented as {RP k ,RP k+1 ,RP k+2 Let the cosine of the included angle of the triangle formed by {RP} be the reference point set for the current group. k ,RP k+1 ,RP k+2 The linearity of} is,

[0024]

[0025] The calculated linear convergence is the linear convergence of the current group of reference point sets. That is, each group of reference point sets shares a unified linear convergence, denoted as .

[0026] Traversal If the number of RPs within the remaining reference point combination cannot form a triangle (i.e., the number of remaining RPs is 2 or 1), then the linear convergence degree is set to 0. or

[0027] Calculate the mean L of the set of all reference points. th Represented as,

[0028]

[0029] like Then discard the current reference point set {RP} k}, thus obtaining the final set of physical domain reference points.

[0030] As a preferred embodiment of the anti-interference indoor positioning method of the present invention, the method of obtaining the position estimate of the test point TP in the conjugate space using the WKNN algorithm includes, based on the nearest neighbor set filtered by the signal and physical domains. The conjugate space value is calculated as follows:

[0031] C sk =ξ k +iχ k

[0032] Among them, C sk Represented as reference point RP k The degree of proximity of the test point TP to the signal domain and the rationality of the physical structure, where i represents the imaginary unit; for the nearest reference point RP k The set, with its conjugate weights configured as follows:

[0033]

[0034] Where ε is a very small non-negative number to prevent the denominator from being 0.

[0035] As a preferred embodiment of the anti-interference indoor positioning method of the present invention, the step of obtaining the position estimate of the test point TP in the conjugate space using the WKNN algorithm further includes introducing a negative exponential function. Utilizing smoothing properties to limit nearest neighbor reference points (RP) k The range of conjugate weight fluctuations; nearest neighbor set selected by dual domains of signal and physical domains. As the coordinates of the test point TP (x) t ,y t The ideal reference set for location estimation is obtained by exponentially quantifying the nearest neighbor reference points (RP). k The localization weights, based on WKNN, yield the position estimate in the conjugate space as follows:

[0036]

[0037] in, The weights of the reference point RPk are represented as (x k ,y k ) represents the physical coordinates of the reference point RPk.

[0038] Another objective of this invention is to provide an anti-interference indoor positioning system that can accurately locate the indoor location to be measured.

[0039] To address the aforementioned technical problems, this invention provides the following technical solution: a system for an anti-interference indoor positioning method, comprising: a spatial model establishment module, a signal domain nearest neighbor selection module, a physical domain nearest neighbor selection module, and a conjugate spatial positioning module; the spatial model establishment module is used to detect wireless access points within the positioning area, deploy reference points, establish a ground plane original fingerprint database, and determine the sub-region of the test point TP; the signal domain nearest neighbor selection module uses a distance measurement method to select signal domain nearest neighbor fingerprint reference points within the region to which the test point TP belongs, obtaining a set of signal domain nearest neighbor reference points; the physical domain nearest neighbor selection module uses the reasonableness of the physical geometry of the reference points as the basis for judging whether the signal domain nearest neighbor reference points are reliable, thereby obtaining a set of highly reliable physical domain nearest neighbor reference points; the conjugate spatial positioning module selects K nearest neighbor reference points from the set of nearest neighbor reference points and uses the WKNN method to estimate the location of the point to be located TP.

[0040] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the anti-interference indoor positioning method described above.

[0041] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the anti-interference indoor positioning method described above.

[0042] The beneficial effects of this invention are: the invention uses a multi-resolution method to make the reference point RP k During the signal space search process, multiple sets of recursive resolutions are used. Dominant, used to regulate the reference point RP k To address the issue of balancing timeliness and accuracy in search, this method is used to improve the accuracy of reference points (RP). k The precise analysis of the correlation between the test point TP and the signal domain simplifies complex retrieval into direct filtering.

[0043] This invention uses a dual-scale measurement method to provide a more comprehensive reference point RP. k The screening criteria improved positioning accuracy. This was achieved through the linear convergence threshold L. th Its adaptive characteristics and dynamic adaptation to environmental changes solve the problem of the lack of flexibility of fixed thresholds, thus enabling adaptive selection of the number of nearest neighbors.

[0044] This invention achieves multi-level constraints on the signal similarity and distribution structure robustness of nearest neighbors by introducing conjugate space. This strategy allows for the acquisition of complementary information from both energy and physical dimensions, providing nearest neighbor location information that is effective for both signal and physical distances, taking into account both physical and signal distance performance characteristics, and achieving a balance between signal similarity and distribution structure robustness. Attached Figure Description

[0045] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is a flowchart of an anti-interference indoor positioning method in Example 1.

[0047] Figure 2 This is a cumulative probability distribution diagram of positioning errors for five algorithms in an anti-interference indoor positioning method in Example 2.

[0048] Figure 3 This is a box plot of the positioning errors of five algorithms in an anti-interference indoor positioning method in Example 2.

[0049] Figure 4 This is a module structure diagram of an anti-interference indoor positioning system in Example 3. Detailed Implementation

[0050] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0051] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0052] Example 1, referring to Figure 1 This is the first embodiment of the present invention, which provides an anti-interference indoor positioning method including, as follows: Figure 1 As shown:

[0053] Step 1: Offline Phase. Within the positioning area, determine ground-based and airborne wireless access points, and deploy ground-based and airborne fingerprint reference points. Each wireless access point is assigned a unique physical address (MAC address). Wireless access can be achieved via Bluetooth or Wi-Fi.

[0054] The method for determining terrestrial wireless access points includes detecting N terrestrial wireless access points (APs) on the ground plane during the offline phase within the positioning area Ω. n .

[0055] The deployment methods for ground fingerprint reference points include deploying M fingerprint reference points RP at equal intervals on the ground plane. m .

[0056] The deployment methods for aerial fingerprint reference points include deploying reference points in the same manner as on the ground plane, on an empty plane with the same two-dimensional coordinates but at a height of Z.

[0057] The methods for determining airborne wireless access points include detecting J wireless access points (APs) in the airspace. j .

[0058] Step 2: Extract the coexisting wireless access points (APs) of terrestrial and airborne wireless access points through MAC addresses, and obtain S MAC addresses by intersection to determine the terrestrial AP database. Based on the AP database, obtain the original terrestrial fingerprint database by combining coordinates.

[0059] Step 3: Collect the signal strength of the test point TP, and roughly locate the sub-region of TP based on the Euclidean distance of the signal strength between the reference point and TP, and obtain the set of nearest reference points of the sub-region.

[0060] Step 4: Use a multi-resolution method to select the nearest fingerprint reference points in the signal domain within the sub-region to obtain the set of nearest reference points in the signal domain.

[0061] Calculate the subregion Ω based on the Pearson coefficient. u All nearest neighbor reference points (RP) k Correlation coefficient ξ with test point TP k Represented as,

[0062]

[0063] Among them, E TP This represents the RSS measurement value of the ground-plane wireless access point included in the test point TP. RSS represents the received signal strength. k Cov(G) represents the RSS measurement of ground-plane wireless access points included in the nearest reference points. k E TP ) represents the nearest neighbor reference point RP k Covariance with the signal strength at the test point TP Represented as ground plane RP k AP collected at the location n Signal variance This represents the AP data collected at test point TP. n Signal variance.

[0064] Based on the correlation coefficient ξ k The multi-resolution model is configured to divide the range layer by layer, as expressed by the formula:

[0065]

[0066] Where, ε k Represented as the number of layers; and Let g represent the lower and upper bounds of each layer, respectively, and g represent the ascending gradient with g > 0.

[0067] Define the εth k Layer resolution is expressed as,

[0068]

[0069] Thus, the resolution is obtained. With the number of layers ε k The mapping relationship is represented as follows:

[0070]

[0071] Here, ε represents a very small non-negative number to prevent the denominator from being 0. Represented as a floor function, based on resolution. Calculate the level ε k.

[0072] Calculate the Euclidean distance between the nearest neighbor reference point and the test point TP in this layer. Select Minimum σ t Using each reference point RP as a confidence neighbor, the final candidate reference point RP set in the signal space is obtained.

[0073] Thus, this invention makes the reference point RP k During the signal space search process, multiple sets of recursive resolutions are used. Dominant, used to regulate the reference point RP k To address the issue of balancing timeliness and accuracy in search, this method is used to improve the accuracy of reference points (RP). k Accurate analysis of the correlation between the test point TP and the signal domain simplifies complex retrieval into direct filtering; correlation coefficient ξ k Preliminary screening of highly correlated nearest neighbor reference points (RPs) k Euclidean criterion It delves into the differences between signal features. The dual-scale measurement method provides a more comprehensive reference point RP. k The screening criteria improved the positioning accuracy.

[0074] Step 5: Select K nearest neighbor reference points from the signal domain nearest neighbor reference point set, and use the linear clustering method to filter again from the physical domain to obtain the physical domain nearest neighbor reference point set.

[0075] For the set of reference points in the signal domain Based on the signal Euclidean scale, the reference points RP are sorted in ascending order. The reference point RP with the smallest Euclidean distance is taken as the initial reference point RP. initial .

[0076] From RP initial The groups are initially determined based on the Euclidean distance, with each group selecting three nearest reference points as a set of reference points {RP}. k ,RP k+1 ,RP k+2 Each reference point RP participates in the selection only once, thus resulting in multiple sets of reference points.

[0077] For the reference point set {RP k ,RP k+1 ,RP k+2}, calculate RP respectively k With RP k+1 RP k With RP k+2 RP k+1 With RP k+2 distance The stability of each angle can be determined using the triangle cosine theorem, expressed by the formula:

[0078]

[0079] Where, χ k , χ k+1 , χ k+2 Represented as {RP k ,RP k+1 ,RP k+2 Let the cosine of the included angle of the triangle formed by {RP} be the reference point set for the current group. k ,RP k+1 ,RP k+2 The linearity of} is,

[0080]

[0081] The calculated linear convergence is the linear convergence of the current group of reference point sets. That is, each group of reference point sets shares a unified linear convergence, denoted as .

[0082] Traversal If the number of RPs within the remaining reference point combination cannot form a triangle (i.e., the number of remaining RPs is 2 or 1), then the linear convergence degree is set to 0. or

[0083] Calculate the mean L of the set of all reference points. th Represented as,

[0084]

[0085] like Then discard the current reference point set {RP} k}, thus obtaining the final set of physical domain reference points.

[0086] Thus, by using the linear clustering threshold L th Its adaptive characteristics, dynamically adapting to environmental changes, solve the problem of inflexibility caused by fixed thresholds, thus allowing for adaptive selection of the number of nearest neighbors. Furthermore, it provides a physical domain reference point set under linear clustering. For the set of reference points in the signal domain By considering the geometric structure, the accuracy of the nearest neighbor information in the energy domain is verified through the robustness of the physical domain structure, avoiding the problem of over-reliance on the signal space. The positioning accuracy is improved by using high-confidence nearest neighbor points under dual constraints.

[0087] Step 6: Based on the set of nearest neighbor reference points selected by both the signal domain and the physical domain, the WKNN algorithm is used to obtain the position estimate of the test point TP in the conjugate space.

[0088] Nearest neighbor set based on signal and physical domain filtering The conjugate space value is calculated as follows:

[0089] C sk =ξ k +iχ k

[0090] Among them, C sk Represented as reference point RP k The degree of proximity of the test point TP signal domain to the rationality of the physical structure, where i is represented by the imaginary unit.

[0091] For the nearest reference point RP k The set, with its conjugate weights configured as follows:

[0092]

[0093] Where ε is a very small non-negative number to prevent the denominator from being 0.

[0094] Introducing the negative exponential function Utilizing smoothing properties to limit nearest neighbor reference points (RP) k The range of conjugate weight fluctuations.

[0095] Nearest neighbor set selected by both signal and physical domains As the coordinates of the test point TP (x) t ,y t The ideal reference set for location estimation is obtained by exponentially quantifying the nearest neighbor reference points (RP). k The localization weights, based on WKNN, yield the position estimate in the conjugate space as follows:

[0096]

[0097] in, The weights of the reference point RPk are represented as (x k ,y k ) represents the physical coordinates of the reference point RPk.

[0098] Thus, by introducing conjugate space, multi-level constraints on the signal similarity and distribution structure robustness of nearest neighbors are achieved. This strategy allows for the acquisition of complementary information from both energy and physical dimensions, providing nearest neighbor location information that is effective for both signal and physical distances, taking into account both physical and signal distance performance characteristics, and achieving a balance between signal similarity and distribution structure robustness.

[0099] Example 2, refer to Figure 2 and Figure 3This is the second embodiment of the present invention, which differs from the first embodiment in that: an anti-interference indoor positioning method further includes, in order to verify and explain the technical effect of the method, a comparative test is conducted between the traditional technical solution and the method of the present invention, and the test results are compared by means of scientific demonstration to verify the real effect of the method.

[0100] A T-shaped connecting corridor on a floor of a college within a university was selected as the real-world test site. This site experiences frequent pedestrian traffic and signal overlap and interference. Choosing this scenario allows for the examination of the impact of environmental variability and dynamic signal characteristics on positioning, ensuring the test results have high representativeness in real-world measurements. To fully achieve the fingerprint database aging effect, one year after the completion of the offline ground-level database collection, further offline air-level and online ground-level database collections were conducted in the same area.

[0101] The reference points (RPs) are arranged in a uniform grid, with an interval of τ = 1m between adjacent reference points, and a total of M = 160 RPs are deployed. At height Z in the global positioning area... 地 =0.7m ground plane and height Z 空 N=313 and J=362 AP signals were detected in an empty plane with a diameter of 2m. After filtering, S=161 APs were ultimately retained for the experiment. To avoid dimensionality differences during computation and without loss of generality, the AP signal intensities not detected at the reference points were padded with -100dBm. Q=50 samples were collected at each reference point, with a sampling time interval of t=2.3s. TP samples were collected in the middle of a T-shaped corridor, with an interval of d=0.5m between adjacent test points, for a total of 160 test points. To eliminate the potential impact of equipment differences on algorithm performance evaluation, the same mobile terminal was used as the data acquisition tool in both offline and online processes. All data processing operations were performed on the Matlab 2022b platform to ensure consistency and comparability.

[0102] To verify the overall positioning performance of the anti-jamming indoor positioning method (AJIP) proposed in this invention, this paper selects four positioning algorithms—AAS, WKNN, EWKNN, and VWKNN—as comparison algorithms and uses error descriptive statistics as a metric to analyze the effectiveness of the AJIP method in this embodiment.

[0103] For the AAS method, please refer to Tao Y, ZHAO L. FingerprintLocalizationWithAdaptive AreaSearch[J]. IEEE Communications Letters, 2020, 24(7): 1446-1450.

[0104] For the WKNN method, please refer to SiyangL IU, de Lacerda R, FiorinaJ. WKNN indoor Wi-Fi localization method using k-means clusteringbased radio mapping [C] / / 2021IEEE93rd Vehicular Technology Conference (VTC2021-Spring). IEEE, 2021:1-5.

[0105] For the EWKNN method, please refer to Wang Peizhong, Zheng Nanshan, Zhang Yanzhe. Indoor positioning algorithm based on dynamic K value and AP MAC address filtering [J]. Computer Science, 2016, 43(01):163-165.

[0106] For the VWKNN method, please refer to Xu Tian, ​​He Jingsha, Zhu Nafei, et al. Improved VWKNN location fingerprinting algorithm based on discrete coefficients [J]. Journal of Beijing University of Aeronautics and Astronautics, 2022, 48(7):1242-1251.

[0107] Furthermore, to ensure the comparability of the experiments, the parameters of the comparison algorithms have all been adjusted to the optimal configuration for this experimental scenario.

[0108] Figure 2 The document presents the cumulative error probability distributions of five algorithms. Overall, the upward trends of the five algorithms are relatively consistent. However, the AJIP algorithm proposed in this embodiment begins to show its advantage after the positioning error exceeds 0.8m, with a cumulative error probability reaching 30.63%. Furthermore, the cumulative error probability of the AJIP algorithm proposed in this embodiment reaches 73.13% within 2m and 85.63% within 3m, outperforming the other four comparison algorithms, with an optimal optimization of 10%. Although the AJIP algorithm proposed in this embodiment is not always optimal between 0.4m and 0.8m, it can be seen that the AJIP proposed in this embodiment does not show a significant gap compared to the comparison algorithms at this point. Within this range, the maximum error of the AJIP proposed in this embodiment is only slightly worse than the optimal algorithm VWKNN by 3.13%.

[0109] The VWKNN algorithm improves positioning accuracy by pre-removing highly volatile access points (APs), but some highly volatile APs with high single-point discriminativeness are also removed. The AJIP algorithm proposed in this embodiment, through dual-plane screening, retains APs with high volatility but high discriminativeness, resulting in only a slight loss of accuracy at some test points (TPs) and maintaining an advantage in most areas. The EWKNN algorithm finds nearest neighbors through a threshold, but it relies too heavily on the number of nearest neighbors; when the number of nearest neighbors is small, the threshold lacks representativeness, leading to poor positioning performance. The WKNN algorithm is a classic positioning algorithm that can adapt to the current environment, but fixing the K value will cause distant reference points (RPs) to be excluded. k It also incorporates calculations, thus achieving a high degree of fit with the VWKNN and EWKNN algorithms; the AAS algorithm uses the difference in AP signal intensity to determine the reference point RP. k While sorting is performed to find the best nearest neighbor, drastic fluctuations in the AP signal at test point TP can reduce positioning stability, resulting in a long-term lower cumulative error distribution compared to the AJIP algorithm proposed in this embodiment. Experiments show that the AJIP algorithm proposed in this embodiment has strong dynamic adaptability and can effectively filter out outliers and find the best nearest neighbor in complex environments, thereby improving positioning accuracy.

[0110] The box plot of the positioning error distribution of the proposed AJIP algorithm in this embodiment compared with four other algorithms is shown below. Figure 3 As shown, compared to the other four algorithms, the AJIP positioning error proposed in this embodiment is largely concentrated in a low-error region of less than 2.147m, which is significantly smaller than the other four algorithms. The AJIP positioning error of the middle 50% data interval proposed in this embodiment is the narrowest, at 1.420m, and the median error is the smallest at 1.286m. Although the number of abnormal test points with positioning errors exceeding 1.5 times the middle 50% data interval is greater than that of WKNN and VWKNN, the number of abnormal points is controlled between 1 and 2, and the error value of the abnormal points is smaller than that of the other two algorithms. Therefore, the AJIP algorithm proposed in this embodiment has strong anti-interference capabilities because it utilizes multi-resolution... Hierarchical screening and linear clustering threshold L th The constraints can largely eliminate non-nearest neighbor reference points RP. k This improved positioning accuracy.

[0111] Table 1 shows the performance comparison results of the five algorithms. The positioning algorithm AJIP proposed in this embodiment significantly outperforms the other four algorithms in terms of both average error and error variance, reaching as high as 14.8% and 23.8% respectively. Furthermore, the maximum error of AJIP proposed in this embodiment is also better than the other four algorithms by 3.3% compared to their best values. Although its minimum error is not optimal, it is only 0.0034m worse than the optimal WKNN algorithm. The average positioning error of AJIP proposed in this embodiment is 1.6840m, representing a 14.8% to 19.5% accuracy improvement compared to the comparison algorithms. Experimental results show that AJIP proposed in this embodiment has significant advantages in solving the problems of access point aging and movement, as well as signal interference in the environment. This is because the dual-plane fingerprint database construction strategy can largely screen out unstable and low-contribution access points (APs), and the use of conjugate space (CS) during online matching can maximize the relationship between the test point (TP) and the reference point (RP). k Based on the similar characteristics and the rationality of the physical layout of the reference points, the reference point RP is increased. k The rationality of the selection.

[0112] Table 1 Performance Comparison of 5 Positioning Algorithms

[0113]

[0114] Example 3, referring to Figure 4 This is the third embodiment of the present invention, which differs from the previous two embodiments in that: a system for an anti-interference indoor positioning method is characterized by comprising a spatial model establishment module 100, a signal domain nearest neighbor selection module 200, a physical domain nearest neighbor selection module 300, and a conjugate spatial positioning module 400; the spatial model establishment module 100 is used to detect wireless access points within the positioning area, deploy reference points, establish a ground plane original fingerprint database, and determine the sub-region of the test point TP; the signal domain nearest neighbor selection module 200 uses a distance measurement method to select signal domain nearest neighbor fingerprint reference points within the area to which the test point TP belongs, thereby obtaining a set of signal domain nearest neighbor reference points; the physical domain nearest neighbor selection module 300 uses the reasonableness of the physical geometry of the reference points as the basis for judging whether the signal domain nearest neighbor reference points are reliable, thereby obtaining a set of highly reliable physical domain nearest neighbor reference points; the conjugate spatial positioning module 400 selects K nearest neighbor reference points from the set of nearest neighbor reference points and uses the WKNN method to estimate the location of the point to be located TP.

[0115] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0116] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0117] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0118] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0119] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An anti-interference indoor positioning method, characterized in that: include, Within the positioning area, determine the ground wireless access points and air wireless access points, deploy ground fingerprint reference points and air fingerprint reference points, and assign a unique physical address (MAC) to each wireless access point; By extracting the MAC addresses of the wireless access points (APs) that coexist with the ground wireless access points and the air wireless access points, the intersection of which yields S MAC addresses to determine the ground plane AP database. Based on the AP database, the coordinates are combined to obtain the original ground plane fingerprint database. The signal strength of the test point TP is collected, and the sub-region of TP is roughly located based on the Euclidean distance between the reference point and TP, so as to obtain the set of nearest reference points of the sub-region. A multi-resolution method is used to select nearest-neighbor fingerprint reference points in the signal domain within the sub-region to obtain a set of nearest-neighbor reference points in the signal domain. Select K nearest neighbor reference points from the set of nearest neighbor reference points in the signal domain, and use the linear clustering method to filter them again from the physical domain to obtain the set of nearest neighbor reference points in the physical domain. Based on the nearest neighbor reference point set under dual screening in the signal domain and physical domain, the WKNN algorithm is used to obtain the position estimate of the test point TP in the conjugate space. The linear convergence method includes: grouping based on Euclidean distance; selecting three nearest reference points for each group as a set of reference points; each reference point participating in the selection only once, thus obtaining multiple sets of reference points; determining the stability of each angle using the triangle cosine theorem to obtain the cosine value of the included angle of the triangle formed by the set of reference points; determining the linear convergence of the current group of reference points based on the cosine value; each set of reference points sharing a unified linear convergence; using the rationality of the physical geometry of the reference points as the basis for judging whether the nearest neighbor reference points in the signal domain are reliable; calculating the mean of the linear convergence of all reference point sets; if the linear convergence of the current reference point set is less than the mean, discarding the current reference point set to obtain the final physical domain reference point set. In the conjugate space, the conjugate space value represents the proximity of the reference point and the test point in the signal domain and the rationality of the physical geometry of the reference point, and is connected in imaginary units; conjugate weights are configured for each set of neighboring reference points to limit the fluctuation range of the conjugate weights of the neighboring reference points, and the position estimation in the conjugate space is obtained based on WKNN.

2. The anti-interference indoor positioning method as described in claim 1, characterized in that: The method for determining the terrestrial wireless access point includes, within the positioning area Inside, during the offline phase, it was detected at the ground plane. terrestrial wireless access points ; The deployment method of the ground fingerprint reference points includes deploying them at equal intervals on the ground plane. fingerprint reference points ; The deployment method of the aerial fingerprint reference points includes: having the same two-dimensional coordinates on the ground plane but with a height of [missing information]. On the empty plane, reference points are deployed in the same manner as on the ground plane. ; The method for determining the airborne wireless access point includes detecting it on the air plane. wireless access points .

3. The anti-interference indoor positioning method as described in claim 2, characterized in that: The method of selecting nearest-neighbor fingerprint reference points in the signal domain within the sub-region using a multi-resolution approach includes calculating the sub-region based on the Pearson coefficient. All neighboring reference points Correlation coefficient with test point TP Represented as, in, This represents the RSS measurement value of the ground-plane wireless access points included in the test point TP. RSS indicates the received signal strength. Represented as the RSS measurement value of ground-plane wireless access points included in the nearest reference points. Represented as the nearest reference point Covariance with the signal strength at the test point TP Represented as ground plane Collected from Signal variance This is represented as data collected at test point TP. Signal variance; Based on correlation coefficient The multi-resolution model is configured to divide the range layer by layer, as expressed by the formula: in, Represented as the number of layers; and These represent the lower and upper limits of each layer, respectively. Represented as an ascending gradient and ; Definition of the first Layer resolution is expressed as, Thus, the resolution is obtained. With number of layers The mapping relationship is represented as follows: in, Represent it as a very small non-negative number to prevent the denominator from being 0. Represented as a floor function, based on resolution. Computational hierarchy ; Calculate the Euclidean distance between the nearest neighbor reference point and the test point TP in this layer. Select smallest Using each reference point RP as a confidence neighbor, the final candidate reference point RP set in the signal space is obtained. .

4. The anti-interference indoor positioning method as described in claim 3, characterized in that: The step of selecting K nearest neighbor reference points from the set of nearest neighbor reference points in the signal domain includes, for the set of reference points in the signal domain Based on the Euclidean distance of the signal, the reference points RP are sorted in ascending order. The reference point RP with the smallest Euclidean distance is taken as the initial reference point. ; from The groups are initially determined based on the Euclidean distance, with each group selecting three nearest reference points as a set of reference points. Each reference point RP participates in the selection only once, thus resulting in multiple sets of reference points.

5. The anti-interference indoor positioning method as described in claim 4, characterized in that: The method of using linear clustering to further filter the physical domain to obtain the set of nearest neighbor reference points also includes, for the reference point set Calculate separately and , and , and distance , , The stability of each angle can be determined using the triangle cosine theorem, expressed by the formula: in, , , Represented as The cosine of the included angle of the triangle formed is given by the set of reference points for the current group. The linear convergence is, The calculated linear convergence is the linear convergence of the current group of reference point sets. That is, each group of reference point sets shares a unified linear convergence, denoted as . ; Traversal If the number of RPs within the remaining reference point combination cannot form a triangle (i.e., the number of remaining RPs is 2 or 1), then the linear convergence degree is set to 0. or ; Calculate the mean of the set of all reference points. Represented as, like Then discard the current reference point set. To obtain the final set of physical domain reference points .

6. The anti-interference indoor positioning method as described in claim 5, characterized in that: The location estimation of the test point TP in the conjugate space obtained by using the WKNN algorithm includes a nearest neighbor set based on signal and physical domain filtering. The conjugate space value is calculated as follows: in, Represented as a reference point The proximity of the test point to the TP signal domain and the reasonableness of the physical geometry of the reference point, Represented as the imaginary unit; For neighboring reference points The set, with its conjugate weights configured as follows: in, It is a very small non-negative number to prevent the denominator from being 0.

7. The anti-interference indoor positioning method as described in claim 6, characterized in that: The method of using the WKNN algorithm to obtain the position estimate of the test point TP in the conjugate space also includes introducing a negative exponential function. Using smoothing properties to limit nearest neighbor reference points The range of conjugate weight fluctuations; Nearest neighbor set selected by both signal and physical domains As the coordinates of the test point TP An ideal reference set for location estimation is obtained by exponentially quantifying nearest neighbor reference points. The localization weights, based on WKNN, yield the position estimate in the conjugate space as follows: in, Represented as a reference point The weight, Represented as a reference point The physical coordinates.

8. A system employing an anti-interference indoor positioning method as described in any one of claims 1 to 7, characterized in that: It includes a spatial model building module (100), a signal domain nearest neighbor filtering module (200), a physical domain nearest neighbor filtering module (300), and a conjugate spatial positioning module (400). The spatial model building module (100) is used to detect wireless access points in the positioning area, deploy reference points, establish a ground plane original fingerprint database, and determine the sub-region of the test point TP. The signal domain nearest neighbor filtering module (200) uses a distance measurement method to filter signal domain nearest neighbor fingerprint reference points within the area to which the test point TP belongs, and obtains a set of signal domain nearest neighbor reference points; The physical domain nearest neighbor filtering module (300) uses the reasonableness of the physical geometry of the reference point as the basis for judging whether the signal domain nearest neighbor reference point is reliable, thereby obtaining a set of highly reliable nearest neighbor reference points in the physical domain. The conjugate spatial positioning module (400) selects K nearest neighbor reference points from the set of nearest neighbor reference points and uses the WKNN method to estimate the location of the point TP to be located.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the anti-interference indoor positioning method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the anti-interference indoor positioning method according to any one of claims 1 to 7.

Images