A single station terminal fingerprint positioning method based on square weighted K nearest neighbor algorithm

Abstract

The application discloses a single station terminal fingerprint positioning method based on a square weighted K nearest neighbor algorithm and belongs to the technical field of fingerprint positioning. The method comprises the following steps: based on a channel frequency response (CFR) matrix, a position fingerprint matrix (FRCPM) with higher positioning precision is extracted; a fingerprint compression is performed on the fingerprint matrix (FRCPM) to obtain a compressed fingerprint (FCPV) with smaller data scale and being more beneficial to improving positioning speed; and a square weighted K nearest neighbor matching method with higher positioning precision is obtained by improving the weighted K nearest neighbor method.

Description

Technical Field

[0001] This invention relates to a single-station terminal fingerprint localization method based on the squared weighted K-nearest neighbor algorithm, belonging to the field of signal and information processing technology. Background Technology

[0002] Today, location-based services (LBS) are widely used in many fields such as express delivery and logistics, smart homes, and emergency rescue, and LBS has become a basic need for people's daily production and life. Currently, one of the most mature LBS technologies is the Global Positioning System (GPS), which originated for military applications. Traditional positioning methods are generally based on triangulation. The biggest problem with these methods is that in non-line-of-sight (NLOS) scenarios where electromagnetic wave scattering and diffraction phenomena occur, they are affected by multipath effects, resulting in a significant decrease in positioning accuracy.

[0003] In recent years, fingerprint positioning technology has received widespread attention and research due to its ability to overcome the problems of traditional wireless network positioning technologies and to effectively locate targets by fully utilizing multipath information from the channel. Fingerprint positioning technology is mainly divided into two types: one based on Received Signal Strength (RSS) and the other based on Channel State Information (CSI). RSS fingerprints are coarse and sensitive to environmental changes, thus limiting the accuracy of RSS-based fingerprint positioning methods, while CSI-based fingerprint positioning performs relatively better. In CSI-based fingerprint positioning methods, fingerprint matrix extraction, fingerprint matrix compression, and the matching algorithm during online positioning have a significant impact on positioning accuracy and speed. Therefore, it is of great significance to invent a fingerprint extraction, compression, and matching algorithm with higher accuracy and faster speed. Summary of the Invention

[0004] To address the issue of low positioning performance in traditional fingerprint extraction, compression, and matching methods, this invention proposes a single-station terminal fingerprint positioning method based on the squared weighted K-nearest neighbor algorithm. The fingerprint matrix FRCPM is extracted based on the channel frequency response (CFR) and compressed into FCPV. During online positioning, the squared weighted K-nearest neighbor matching algorithm is used, which significantly improves the performance of fingerprint positioning.

[0005] The present invention provides a single-station terminal fingerprint localization method based on the squared weighted K-nearest neighbor algorithm, comprising the following steps:

[0006] Step 1: Obtain the fingerprint matrix FRCPM data of all reference points in the area to be located;

[0007] Step 2: Compress the fingerprint matrix FRCPM of all reference points to obtain the compressed fingerprint FCPV. Store the position coordinates of all reference points in the area to be located and the compressed fingerprint FCPV in a one-to-one correspondence to obtain the fingerprint database.

[0008] The reference point refers to the area to be located being gridded with a fixed side length, and all grid nodes are used as fingerprint positioning reference points.

[0009] Step 3: Extract the fingerprint matrix FRCPM of the point to be located, and compress the fingerprint matrix FRCPM of the point to be located to obtain the compressed fingerprint FCPV of the point to be located; match the compressed fingerprint FCPV of the point to be located with the compressed fingerprint FCPV in the fingerprint database to estimate the coordinates of the point to be located.

[0010] Furthermore, the fingerprint matrix FRCPM of the reference points is obtained in step 1, and the specific method is as follows:

[0011] Record the coordinates of each reference point and collect the uplink signal of the terminal device at each reference point; the base station directly obtains the channel frequency response (CFR) matrix or the channel impulse response (CIR) matrix by performing channel estimation on the uplink signal of the terminal device;

[0012] If the channel estimation result is the channel impulse response (CIR) matrix, the channel frequency response (CFR) matrix can be obtained by performing simple processing such as fast Fourier transform or interpolation on the channel impulse response (CIR) matrix. The Hadamard product of the channel frequency response (CFR) matrix and its conjugate matrix is ​​defined as the frequency response channel power matrix (FRCPM), and the FRCPM is used as the location fingerprint.

[0013] Furthermore, in step 2, the fingerprint matrix FRCPM of the reference points is compressed, and the specific method is as follows:

[0014] Adding the elements of each column of the fingerprint matrix FRCPM yields an element with N elements. l The row vector is called the Frequency Channel Power Vector (FCPV), which is also known as the Compressed Fingerprint FCPV.

[0015] The compressed fingerprint FCPV of the k-th reference point is denoted as FC. k FC k The j-th element FC kj Represented as:

[0016]

[0017] Furthermore, in step 3, the compressed fingerprint FCPV of the point to be located is matched with the compressed fingerprint FCPV in the fingerprint database to estimate the coordinates of the point to be located. The specific steps are as follows:

[0018] Step 3.1: Determine the optimal K value for the squared-weighted K-nearest neighbor algorithm; specifically, the optimal K value for the squared-weighted K-nearest neighbor algorithm is determined by cross-validation.

[0019] Step 3.2: Match the compressed fingerprint FCPV of the point to be located with the data in the fingerprint database to estimate the point to be located. The coordinates are represented as:

[0020]

[0021]

[0022] In the formula, Represents the coordinates of the point to be located. w represents the coordinates of the k-th reference point. k The weights corresponding to the k-th reference point satisfy ∑w k =1,d k It is the distance of the fingerprint between the k-th reference point and the point to be located; specifically, d k It is the Euclidean distance of the fingerprint between the k-th reference point and the point to be located. △d is to prevent a very small positive number with a denominator of 0.

[0023] To address the problems of traditional fingerprint extraction, compression, and matching methods for fingerprint localization, this invention adopts the above technical solutions to achieve the following technical effects: the fingerprint matrix FRCPM has higher positional recognition; the compressed fingerprint FCPV has a smaller data size and is composed of real numbers, which facilitates processing and improves localization speed; and the squared weighted K-nearest neighbor matching algorithm has higher localization accuracy than the traditional weighted K-nearest neighbor matching algorithm. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the workflow of the method of the present invention;

[0025] Figure 2 This is a comparison chart of the positioning accuracy of the method of this invention and several existing methods; Detailed Implementation

[0026] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings:

[0027] like Figure 1The overall process of the technical solution shown is as follows: fingerprint localization is divided into offline and online stages. The offline stage involves fingerprint extraction based on the CSI obtained from channel estimation, acquiring the fingerprint matrix FRCPM data of all reference points in the area to be located, then compressing it into a compressed fingerprint FCPV, and finally storing the FCPV and coordinate data of each reference point in the fingerprint database. The online stage involves extracting the fingerprint matrix FRCPM of the point to be located, compressing it into an FCPV, and finally using the squared-weighted K-nearest neighbor algorithm for matching to estimate the coordinates of the point to be located. The specific operations of each step are described in detail below.

[0028] This invention discloses a single-site terminal fingerprint localization method based on the squared-weighted K-nearest neighbor algorithm. The method is based on a MIMO-OFDM single-site system, where the base station is equipped with N base stations in a uniform linear array (ULA) configuration. t With one antenna, the total number of OFDM subcarriers is N. l Numerous scatterers are randomly distributed within the area to be located; the method includes the following steps:

[0029] Step 1: Obtain the fingerprint matrix FRCPM data of all reference points in the area to be located;

[0030] The reference point refers to the area to be located being gridded with a fixed side length, and all grid nodes are used as fingerprint positioning reference points.

[0031] The coordinates of each reference point are recorded and the uplink signal of the terminal device at each reference point is collected; the base station directly obtains the channel frequency response (CFR) matrix or the channel impulse response (CIR) matrix by performing channel estimation on the uplink signal of the terminal device.

[0032] If the channel estimation result is the channel impulse response (CIR) matrix, the channel frequency response (CFR) matrix can be obtained by performing simple processing such as fast Fourier transform or interpolation on the channel impulse response (CIR) matrix; the Hadamard product of the channel frequency response (CFR) matrix and its conjugate matrix is ​​defined as the frequency response channel power matrix (FRCPM).

[0033] The channel frequency response (CFR) vector of the nth link at the kth reference point is represented as h. k,n The entire channel frequency response (CFR) matrix at the k-th reference point can be expressed as: H k It is an N t ×Nl The complex matrix, representing the FRCPM of the k-th reference point as F k Then F k Expressed as a formula FRCPM is an N t ×N l A real matrix, [F k ] i,j F represents k The element in the i-th row and j-th column, then [F k ] i,j The physical meaning represented is the channel strength of the i-th communication link to the j-th subcarrier, and FRCPM is used as the location fingerprint;

[0034] Step 2: Compress the fingerprint matrix FRCPM of all reference points to construct a fingerprint database. The specific steps are as follows:

[0035] Step 2.1: Sum the elements of each column of the fingerprint matrix FRCPM to obtain a matrix with N elements. l The row vector is called the Frequency Channel Power Vector (FCPV), which is also known as the compressed fingerprint FCPV.

[0036] The compressed fingerprint FCPV of the k-th reference point is denoted as FC. k FC k The j-th element FC kj Represented as:

[0037]

[0038] Step 2.2: Store the position coordinates of all reference points in the area to be located and the compressed fingerprint FCPV one by one to obtain the fingerprint database;

[0039] Step 3: Extract the fingerprint matrix FRCPM of the point to be located, and compress the fingerprint matrix FRCPM of the point to be located to obtain the compressed fingerprint FCPV of the point to be located; the method of extracting the fingerprint matrix FRCPM of the point to be located is the same as the method of obtaining the fingerprint matrix FRCPM of the reference point in Step 1, and the method of compressing the fingerprint matrix FRCPM of the point to be located is the same as the method of compressing the fingerprint matrix FRCPM of the reference point in Step 2.

[0040] Then, the squared-weighted K-nearest neighbor matching algorithm is used to match the compressed fingerprint FCPV of the point to be located with the compressed fingerprint FCPV in the fingerprint database to estimate the coordinates of the point to be located; the specific steps include:

[0041] Step 3.1: Determine the optimal K value for the squared weighted K-nearest neighbor algorithm using cross-validation.

[0042] For example, using the 4-fold cross-validation method, the reference point is first divided into 4 equal parts. The data from the first part is used as the test set, and the other 3 parts are used as the fingerprint database. Squared weighted K-nearest neighbor matching with a K value of 1 is used to obtain the average positioning error Err1. Then, the second, third, and fourth parts of the data are used as the test set to perform positioning, and the average positioning errors Err2, Err3, and Err4 are obtained respectively. Then, the average of these 4 average positioning errors is calculated. The above test positioning process is repeated by changing the K value until all possible values ​​of K value have been traversed. The average positioning error corresponding to all K values ​​is compared. The K value with the smallest average error is the optimal K value.

[0043] Step 3.2: Match the compressed fingerprint FCPV of the point to be located with the data in the fingerprint database to estimate the point to be located. The coordinates are represented as:

[0044]

[0045]

[0046] In the formula, Represents the coordinates of the point to be located. w represents the coordinates of the k-th reference point. k The weights corresponding to the k-th reference point satisfy ∑w k =1,d k It is the Euclidean distance of the fingerprint between the k-th reference point and the point to be located. Δd is used to prevent a very small positive number with a denominator of 0, and can be taken as 10. -6 .

[0047] Compared to the weighted K-nearest neighbor algorithm, the squared-weighted K-nearest neighbor algorithm appropriately increases the weight of the nearest neighbor and reduces the weight of points that are slightly farther away; the value of K is determined by cross-validation.

[0048] like Figure 2 As shown, under the same experimental conditions, the compressed fingerprint FCPV method of the present invention combined with the square-weighted K-nearest neighbor matching method is compared with the existing fingerprint angle-delay domain channel power matrix (ADCPM), fingerprint similarity criterion Manhattan distance, and weighted K-nearest neighbor and Gaussian-weighted K-nearest neighbor matching methods. It can be seen that the compressed fingerprint FCPV method of the present invention has higher positioning accuracy than the existing fingerprint ADCPM method, and the square-weighted K-nearest neighbor matching method of the present invention has higher positioning accuracy than the existing weighted K-nearest neighbor and Gaussian-weighted K-nearest neighbor matching methods.

Claims

1. A single-station terminal fingerprint localization method based on the squared-weighted K-nearest neighbor algorithm, characterized in that, Includes the following steps: Step 1: Obtain the fingerprint matrix FRCPM data of all reference points in the area to be located. The specific method is as follows: The coordinates of each reference point are recorded and the uplink signal of the terminal device at each reference point is collected; the base station directly obtains the channel frequency response (CFR) matrix or the channel impulse response (CIR) matrix by performing channel estimation on the uplink signal of the terminal device. If the channel estimation result is the channel impulse response (CIR) matrix, the channel frequency response (CFR) matrix can be obtained by performing a fast Fourier transform or simple interpolation on the channel impulse response (CIR) matrix. The Hadamard product of the channel frequency response (CFR) matrix and its conjugate matrix is ​​defined as the frequency response channel power matrix (FRCPM), and the FRCPM is used as the location fingerprint. Step 2: Compress the fingerprint matrix FRCPM of all reference points to obtain the compressed fingerprint FCPV. The specific method is as follows: The number of elements is obtained by summing the elements in each column of the fingerprint matrix FRCPM. The row vector is called the Frequency Channel Power Vector (FCPV), which is also known as the Compressed Fingerprint FCPV. The first The compressed fingerprint FCPV of each reference point is represented as follows: ,but The element Represented as: ； The position coordinates of all reference points in the area to be located are stored one-to-one with the compressed fingerprint FCPV to obtain the fingerprint database; Step 3: Extract the fingerprint matrix FRCPM of the point to be located, and compress the FRCPM to obtain the compressed fingerprint FCPV of the point to be located; match the compressed fingerprint FCPV of the point to be located with the compressed fingerprint FCPV in the fingerprint database to estimate the coordinates of the point to be located. The specific steps are as follows: Step 3.1: Determine the optimal K value for the squared weighted K-nearest neighbor algorithm; Step 3.2: Match the compressed fingerprint FCPV of the point to be located with the data in the fingerprint database to estimate the point to be located. The coordinates are represented as: In the formula, Represents the coordinates of the point to be located. Representing the The coordinates of the reference points Representing the The weights corresponding to each reference point satisfy the following condition: , It is the first The distance of the fingerprint between the reference point and the point to be located. This is to prevent a very small positive number with a denominator of 0.

2. The single-station terminal fingerprint localization method based on the squared-weighted K-nearest neighbor algorithm according to claim 1, characterized in that, The reference point refers to the area to be located being gridded with a fixed side length, and all grid nodes are used as fingerprint positioning reference points.

3. The single-station terminal fingerprint localization method based on the squared-weighted K-nearest neighbor algorithm according to claim 1, characterized in that, It is the first The Euclidean distance of the fingerprint between a reference point and the point to be located.

4. The single-station terminal fingerprint localization method based on the squared weighted K-nearest neighbor algorithm according to claim 1, characterized in that, In step 3.1, cross-validation is used to determine the optimal K value for the squared weighted K-nearest neighbor algorithm.

Images