A Bluetooth indoor positioning algorithm based on improved SSS

Abstract

This invention provides a Bluetooth indoor positioning algorithm based on an improved SSS, belonging to the field of indoor positioning technology. The algorithm's process is as follows: first, a relationship between the antenna received signal and the steering vector is constructed; a steering vector matrix is constructed to determine the direction vector of the signal source; a correlation matrix of the antenna received signal is calculated; the signal space is smoothly decomposed, and eigenvalue decomposition is performed using the correlation matrix to extract the signal subspace; the signal space is introduced into the SSS algorithm to obtain the incident angle from the antenna received signal; and finally, the least squares method is used to estimate the specific location of the transmitted signal. This invention has the advantages of low resource utilization and high positioning accuracy.

Description

Technical Field

[0001] The invention belongs to the technical field of indoor positioning and relates to a Bluetooth indoor positioning algorithm based on an improved SSS. Background Art

[0002] As indoor spaces become increasingly complex, demands for real-time and accurate indoor location tracking are increasing. For example, finding a car in a parking lot, searching for specific items indoors, and locating lost relatives are becoming increasingly difficult, leading to an unprecedented demand for indoor positioning. Satellite positioning is not practical indoors due to weak signals reaching the ground and their inability to penetrate buildings. Commonly used indoor positioning methods include Bluetooth, Wi-Fi, UWB, visible light, and inertial navigation. Bluetooth is particularly advantageous for its low cost, low power consumption, and ease of interaction with mobile phones, making it a clear advantage as an indoor positioning device. Numerous algorithms are currently used for indoor positioning, including the commonly used MUSIC and PDDA algorithms.

[0003] Indoor wireless positioning methods include those based on time of arrival (TOA), time difference of arrival (TDOA), angle of arrival (AOA), signal strength (RSS), and phase difference of arrival (PDOA). However, TOA and TDOA methods are difficult to implement in narrow-bandwidth UHF RFID systems. RSS-based methods are sensitive to multipath signal fading, so positioning systems based solely on RSS have limited accuracy. PDOA-based methods are also affected by multipath, resulting in inaccurate ranging results. Furthermore, AOA-based methods are affected by non-line-of-sight (NLOS) and multipath. MUSIC-based methods are computationally expensive and relatively slow. While Propagator Direct Data Acquisition (PDDA) has been proposed to estimate the direction of the received signal directly from the received data, eliminating the need to construct a covariance matrix or calculate its inverse, PDDA methods rely on calculating propagation vectors, which represent the cross-correlation between the measurement data from the first antenna element and the other antenna elements. However, their performance deteriorates in the presence of multipath. Currently, AoA estimation techniques based on signal subspace decomposition technology are widely used, such as multiple signal classification (MUSIC), Root-MUSIC, and signal parameter estimation via rotation invariance technology (ESPRIT). These techniques have been proposed as super-resolution AoA estimation techniques.

[0004] Sparse representation of space has been widely deployed in the field of AoA estimation. AoA estimation techniques based on sparsity minimization can provide accurate AoA estimation based on a single received sample (single base station). These techniques can also estimate the AoA of coherent signals. However, the disadvantage is that these techniques rely heavily on salient minimization techniques. But they require a lot of computing power. The present invention aims to reduce the computational burden and speed up the operation when accurately estimating the direction of the desired signal and the interference signal, and can apply a suitable beamforming algorithm to enhance the gain of the useful signal, suppress noise and interference, and can be effectively applied to the pseudo-spectrum and extract the correct peak therein, thereby determining the specific position coordinates. Summary of the Invention

[0005] The purpose of the present invention is to provide a Bluetooth indoor positioning algorithm based on an improved SSS to address the above-mentioned problems existing in the existing technology. The technical problem to be solved by the present invention is how to effectively eliminate the influence of coherent signals, make the estimated value close to the true value, and improve the positioning accuracy.

[0006] The object of the present invention can be achieved by the following technical solutions: A Bluetooth indoor positioning algorithm based on an improved SSS, characterized in that it includes the following steps:

[0007] 1) Constructing a relationship between the antenna received signal and the steering vector;

[0008] X(t)=A(θ)s(t)+n(t);

[0009] Where s(t) is the signal sent over the air, A(θ) is the steering vector of the antenna array, and n(t) represents the Gaussian noise signal;

[0010]

[0011] Where d is the distance between adjacent antennas; λ is the wavelength of the signal; M is the number of elements in the antenna, and θ represents the angle of arrival; the steering vector A(θ) describes the phase shift of the signal at each antenna as the distance from the transmitter changes.

[0012] If there are N transmitted signals, the steering vector can be expressed as:

[0013] A(θ)=[a(θ1),a(θ2),a(θ3),...,a(θ n )];

[0014] where a(θ k ) is the direction vector of the k-th signal source, which can be expressed as follows:

[0015]

[0016] 2) Construct a steering vector matrix and determine the direction vector;

[0017] For any array, assuming that the array elements are located in an arbitrary three-dimensional space, then when there are N transmitted signals, the three-dimensional space representation of the steering vector matrix is:

[0018] A(θ,φ)=[a(θ1,φ1),a(θ2,φ2),a(θ3,φ3),...,a(θ n ,φ n )]

[0019] Define the mth sensor in the antenna array as r m =(x m ,y m , z m ), then the steering vector is the direction vector of the nth signal source and can be expressed as follows:

[0020]

[0021] 3) Calculate the correlation matrix of the antenna received signal;

[0022] When there are N snapshots, the data sample covariance matrix can be calculated using the sampling covariance matrix approximation as follows:

[0023]

[0024] Under the premise that the signals are uncorrelated and the signals and noise are unrelated, since the received signal vector matrix is composed of M receiving antennas, coherent signals caused by multipath propagation may exist in a small indoor space. In this case, spatial smoothing technology is needed to de-correlate the signals.

[0025] When the antenna array is divided into p x and p y When the subarray is , the data covariance matrix of the smooth sampled forward space can be expressed as:

[0026]

[0027] in represents the antenna array (i x ,i y ) subarray covariance matrix;

[0028] The x(k) matrix model is a vector matrix of Mx1. Similarly, according to the characteristics of the transmitted signal, s(k) composed of D transmitted signals is a vector matrix of Dx1.

[0029] 4) Using the correlation matrix to perform eigenvalue decomposition and extract the signal subspace;

[0030]

[0031] Re-sort the signal's eigenvalues according to their eigenvalues, sorting them from large to small;

[0032] 5) According to the correlation matrix in step 3), the largest D eigenvalue is represented, and the corresponding eigenvector is:

[0033]

[0034] Among them, Q SS is the signal subspace;

[0035] 6) Based on the relationship between the received signal matrix and the steering vector in 1), the location information of the signal source can be obtained using the SSS algorithm:

[0036]

[0037] The signal direction angle θ and elevation angle can be obtained by searching the peak spectrum of the above

[0038] According to the obtained signal direction angle θ and elevation angle Enter the positioning formula

[0039] The specific location of the transmitted signal can be estimated.

[0040] In small indoor spaces, electromagnetic reflections can generate numerous coherent signals. When the incident signal strikes at a certain angle, the reflected coherent signal can be interpreted as the true signal, resulting in false signal estimation. The improved SSS algorithm effectively eliminates the influence of coherent signals, bringing the estimated value closer to the true value. This method is suitable for running on Bluetooth low energy processors with limited computing resources and is applicable to array antennas of various shapes. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is an overall flow chart of the present invention.

[0042] Figure 2 This is the multi-signal source incident principle of the present invention.

[0043] Figure 3 It is an explanatory diagram of the algorithm principle of the present invention.

[0044] Figure 4 This is the result of our actual elevation accuracy test.

[0045] Figure 5 This is the result of our actual azimuth accuracy test.

[0046] Figure 6 This is the actual size of the antenna board of the receiving device. DETAILED DESCRIPTION

[0047] The following are specific embodiments of the present invention and the accompanying drawings to further describe the technical solutions of the present invention, but the present invention is not limited to these embodiments.

[0048] The algorithm of the present invention is described in detail below using a uniform linear array.

[0049] 1) Constructing a relationship between the antenna received signal and the steering vector;

[0050] X(t)=A(θ)s(t)+n(t)

[0051] Where s(t) is the signal sent over the air, A(θ) is the steering vector of the antenna array, and n(t) represents the Gaussian noise signal.

[0052]

[0053] Where d is the distance between adjacent antennas, λ is the wavelength of the signal, M is the number of elements in the antenna, and θ represents the angle of arrival. The steering vector A(θ) describes how the signal at each antenna shifts in phase due to the varying distance from the transmitter.

[0054] If there are N transmitted signals, the steering vector can be expressed as:

[0055] A(θ)=[a(θ1),a(θ2),a(θ3),...,a(θ n )]

[0056] where a(θ k ) is the direction vector of the k-th signal source, which can be expressed as follows:

[0057]

[0058] 2) Construct a steering vector matrix and determine the direction vector;

[0059] For any array, assuming that the array elements are located in an arbitrary three-dimensional space, then when there are N transmitted signals, the three-dimensional space representation of the steering vector matrix is:

[0060] A(θ,φ)=[a(θ1,φ1),a(θ2,φ2),a(θ3,φ3),...,a(θ n ,φ n )]

[0061] Define the mth sensor in the antenna array as r m =(xm ,y m ,z m ), then the steering vector is the direction vector of the nth signal source and can be expressed as follows:

[0062]

[0063] 3) Calculate the correlation matrix of the antenna received signal;

[0064] When there are N snapshots, the data sample covariance matrix can be calculated using the sampling covariance matrix approximation as follows:

[0065]

[0066] In a small indoor space, there may be coherent signals generated by multipath propagation. In this case, spatial smoothing technology is needed to de-correlate the signals.

[0067] When the antenna array is divided into p x and p y When the subarray is , the data covariance matrix of the smooth sampled forward space can be expressed as:

[0068]

[0069] in represents the antenna array (i x ,i y ) subarrays.

[0070] Under the premise that the signals are uncorrelated with each other and the signals and noise are unrelated to each other, since the received signal vector matrix is composed of M receiving antennas,

[0071] The x(k) matrix model is an Mx1 vector matrix. Similarly, according to the characteristics of the transmitted signal, s(k) composed of D transmitted signals is a Dx1 vector matrix.

[0072] 4) Using the correlation matrix to perform eigenvalue decomposition and extract the signal subspace;

[0073]

[0074] Re-sort the signal's eigenvalues according to their eigenvalues, sorting them from large to small;

[0075] where ∑ SS It is a vector diagonal matrix of rank D, where the elements on the diagonal are: λ1,λ2,λ3,…,λ D ;

[0076] ∑ NSis a vector diagonal matrix of rank MD, where the elements on the diagonal are: D+1 ,λ D+2 ,λ D+3 ,…,λ M .

[0077] Obviously, SS The eigenvalues of NS The eigenvalue of Q SS It is R x The first D largest eigenvalues of correspond to the eigenvector matrix, which is an MxD matrix:

[0078] Q NS It is R x The eigenvector matrix corresponding to the last MD smaller eigenvalues is an M×(MD) matrix.

[0079] further:

[0080]

[0081] 5) According to the correlation matrix in step 3), the largest D eigenvalue is represented, and the corresponding eigenvector is Q ss =[q1, q2, q3, q4,..., q D ]

[0082] Among them, Q SS is the signal subspace;

[0083] 6) Based on the relationship between the received signal matrix and the steering vector in 1), the location information of the signal source can be obtained using the SSS algorithm:

[0084]

[0085] Where H represents the conjugate transpose, and the signal direction angle θ and elevation angle can be obtained by searching the peak spectrum above.

[0086] Assuming the coordinates of the base station are (0,0,0), the coordinates of the tag's location are:

[0087]

[0088] The specific position (x, y, z) of the transmitted signal can be estimated by the above formula, where the angle θ and It corresponds to the estimated azimuth and elevation angles of the tag to reach the base station that need to be measured, and h is the distance between the tag or receiver and the base station.

[0089] Figure 1This is the overall flow chart of the algorithm. First, the relationship between the antenna received signal and the steering vector is constructed, and the steering vector matrix is constructed to determine the direction vector of the signal source; the correlation matrix of the antenna received signal is calculated; then the signal space is smoothly decomposed, and the eigenvalue decomposition is performed using the correlation matrix to extract the signal subspace; the signal space is used to introduce the SSS algorithm to obtain the incident angle from the antenna received signal, and finally the least squares method is used to estimate the specific position of the transmitted signal.

[0090] Figure 2 This is the multi-signal source incident principle of the present invention.

[0091] Figure 3 This diagram explains the algorithm's principles. A single transmitting antenna at a fixed height serves as the signal source. When the transmitter is below the receiver antenna, AOA angle estimation can be used to determine the azimuth and elevation angles between the transmitter and receiver antennas, and the positioning formula can then be used to determine the specific coordinates. If you're in a small room, you can use the point type. If it's a long, narrow corridor, you can use the linear type. Multiple receivers can be arranged linearly.

[0092] Figure 4 This is the result of our actual elevation accuracy test;

[0093] Figure 5 The following figure shows the results of our actual azimuth accuracy test, conducted in an open, unobstructed 12m x 12m room. While maintaining the transmitter's azimuth constant, the elevation angle was varied from 0 to 80 degrees in 5-degree increments. Ten BLE packets were sampled for each angle, and the average was calculated. The angles from the three channels were then compared with the true angle. The final result shows minimal elevation error, but the azimuth error increases starting at 70 degrees for the 32-channel azimuth.

[0094] Figure 6 This is the actual size of the antenna board of the receiving device. On a 170mm*170mm black antenna board, a 4*4 antenna array is used, including an antenna array consisting of 16 patch antennas and 5 RF switches. The distance between the two patch antennas is 12.1mm, and the length and width of each patch antenna are 27.9mm.

[0095] The specific embodiments described herein are merely illustrative of the spirit of the present invention. Persons skilled in the art may make various modifications, additions, or substitutions to the described specific embodiments without departing from the spirit of the present invention or exceeding the scope of the appended claims.

Claims

1. A Bluetooth indoor positioning algorithm based on improved SSS, characterized in that: The process of this algorithm is as follows: 1) First, construct the relationship between the antenna receiving signal and the steering vector; 2) Construct a steering vector matrix to determine the direction vector of the signal source; For any array, assuming that the array elements are located in an arbitrary three-dimensional space, when there are N transmitted signals, the three-dimensional space representation of the steering vector matrix is: A(θ,φ)=[a(θ1,φ1),a(θ2,φ2),a(θ3,φ3),......,a(θ n ,f n )] Define the mth sensor in the antenna array as r m =(x m ,y m , z m ), then the steering vector is the direction vector of the nth signal source and can be expressed as follows: 3) Calculate the correlation matrix of the antenna received signal; 4) Smoothly decompose the signal space and use the correlation matrix to perform eigenvalue decomposition to extract the signal subspace; 5) Use the signal space to introduce the SSS algorithm and obtain the incident angle from the antenna received signal; Based on the relationship between the antenna received signal and the steering vector, the location information of the signal source can be obtained through the SSS algorithm: Where H represents the conjugate transpose, Q ss is the signal subspace; The signal direction angle θ and elevation angle can be obtained by searching the peak spectrum of the above According to the obtained signal direction angle θ and elevation angle Enter the positioning formula The specific location of the transmitted signal can be estimated, and h is the distance between the tag or receiver and the base station.

2. The improved SSS-based Bluetooth indoor positioning algorithm according to claim 1, characterized in that: The specific calculation steps are as follows: In step 1), a relationship between the antenna received signal and the steering vector is constructed; X(t)=A(θ)s(t)+n(t); Where s(t) is the signal sent over the air, A(θ) is the steering vector of the antenna array, and n(t) represents the Gaussian noise signal; Where d is the distance between adjacent antennas; λ is the wavelength of the signal; M is the number of elements in the antenna, and θ represents the angle of arrival; the steering vector A(θ) describes the phase shift of the signal at each antenna due to the change in distance from the transmitter; If there are N transmitted signals, the steering vector can be expressed as: A=[a(θ1),a(θ2),a(θ3),......,a(θ n )]; in: k∈n;a(θ k ) is the direction vector of the kth signal source; In step 3), the correlation matrix of the antenna received signal is calculated; When there are N snapshots, the data sample covariance matrix can be calculated using the sampling covariance matrix approximation as follows: Under the premise that the signals are uncorrelated and the signals and noise are unrelated, since the received signal vector matrix is composed of M receiving antennas, coherent signals caused by multipath propagation may exist in a small indoor space. In this case, spatial smoothing technology is needed to de-correlate the signals. When the antenna array is divided into p x and p y When the subarray is , the data covariance matrix of the smooth sampled forward space can be expressed as: in represents the antenna array (i x ,i y ) subarray covariance matrix; The x(k) matrix model is a vector matrix of Mx1. Similarly, according to the characteristics of the transmitted signal, s(k) composed of D transmitted signals is a vector matrix of Dx1. In step 4), the correlation matrix is used to perform eigenvalue decomposition and extract the signal subspace; where ∑ ss is a vector diagonal matrix of rank D; ∑ NS is a vector diagonal matrix of rank MD, Q ss It is R x The eigenvector matrix corresponding to the first D larger eigenvalues of is an MxD matrix: Q NS It is R x The eigenvector matrix corresponding to the last MD smaller eigenvalues is the matrix of Mx(MD); Re-sort the signal's eigenvalues according to their eigenvalues, sorting them from large to small; According to the correlation matrix in step 3), the largest D eigenvalue is represented, and the corresponding eigenvector is:

3. Apply the algorithm in claim 1 or claim 2 to a low-power Bluetooth processor with limited computing resources, or to array antennas of various shapes.

Images