WO2017049914A1

WO2017049914A1 - Terminal positioning method, apparatus, and system

Info

Publication number: WO2017049914A1
Application number: PCT/CN2016/081894
Authority: WO
Inventors: 黄河; 张晓雷
Original assignee: 中兴通讯股份有限公司
Priority date: 2015-09-21
Filing date: 2016-05-12
Publication date: 2017-03-30
Also published as: CN106550447A; CN106550447B

Abstract

A terminal positioning method, apparatus, and system. The method comprises: generating a kernel mapping matrix according to a location-channel response database of a terminal; obtaining a principle eigenvector according to the kernel mapping matrix; generating a location-row vector function of the terminal according to the principle eigenvector and the kernel mapping matrix; obtaining a channel response vector of the current location of the terminal; and calculating location information corresponding to the channel response vector of the current location of the terminal according to the principle eigenvector and the location-row vector function of the terminal.

Description

Terminal positioning method, device and system

Technical field

This document relates to, but is not limited to, the field of positioning, and in particular, to a terminal positioning method, device and system.

Background technique

The terminal positioning method based on the RSSI (Received Signal Strength Indication) generally uses the channel energy between the AP (Access Point) and the terminal and the distance between the AP and the terminal to satisfy a certain statistical significance. Functional relationships, which in turn enable position estimation. Since the function relationship is unknown, this method needs to generate the terminal location and the receiving energy database of multiple APs in advance, and then learn or acquire the position-energy function relationship to realize wireless positioning. This method has a large database generating workload and a process. Computational complexity and other shortcomings.

Therefore, how to provide a terminal positioning method that can simplify the positioning calculation process is a technical problem to be solved by those skilled in the art.

Summary of the invention

The following is an overview of the topics detailed in this document. This Summary is not intended to limit the scope of the claims.

The embodiment of the invention provides a terminal positioning method, device and system, which can be easily realized.

The embodiment of the invention provides a terminal positioning method, including:

Generating a kernel mapping matrix according to the terminal location-channel response database; acquiring a main eigenvector according to the kernel mapping matrix; and generating a terminal position-row vector function according to the main eigenvector and the kernel mapping matrix;

Obtaining a channel response vector of the current location of the terminal;

According to the main feature vector and the terminal position-row vector function, the position information corresponding to the channel response vector of the current position of the terminal is calculated.

Optionally, the method further includes: acquiring a location-channel response vector of the plurality of known locations of the terminal, and establishing the terminal location-channel response database according to the location-channel response vector of the multiple known locations of the terminal. .

Optionally, generating the kernel mapping matrix according to the terminal location-channel response database comprises: calling a data matrix in the terminal location-channel response database, and calculating a kernel mapping matrix of the data matrix by using a kernel function.

Optionally, invoking the data matrix in the terminal location-channel response database comprises: calculating an initial range of the terminal by using a ray tracing algorithm, and calling a position-channel response vector of the terminal location-channel response database in an initial range to form a data matrix.

Optionally, the method further includes: correcting the initial range according to the location information, and continuing the subsequent steps.

Optionally, obtaining the main eigenvector according to the kernel mapping matrix comprises: normalizing the kernel mapping matrix to obtain a standardized kernel mapping matrix; performing eigenvalue decomposition on the normalized kernel mapping matrix to obtain a diagonal matrix composed of eigenvalues; A feature value contribution rate and a cumulative contribution rate are selected according to the cumulative contribution rate and the threshold value, and the feature vectors corresponding to all the main feature values are used as the main feature vector.

Optionally, generating the terminal position-row vector function according to the main feature vector and the kernel mapping matrix comprises: projecting the kernel mapping matrix to the main feature vector to obtain a new channel response feature vector matrix; and performing a new channel response feature vector matrix Dimensionality reduction processing; linear regression processing is performed on the new channel response feature vector matrix after dimensionality reduction according to the terminal position corresponding to the new channel response feature vector after each dimension reduction processing, and the terminal position-row vector function is obtained.

Optionally, the location information corresponding to the channel response vector of the current location of the terminal is calculated according to the main feature vector and the terminal location-row vector function, including: projecting the channel response vector of the current location of the terminal to the main feature vector to obtain a new feature vector. The new feature vector is substituted into the terminal position-row vector function to calculate the position information.

An embodiment of the present invention provides a terminal positioning apparatus, including:

a modeling module, configured to generate a kernel mapping matrix according to the terminal location-channel response database; obtain a main eigenvector according to the kernel mapping matrix; and generate a terminal position-row vector function according to the main eigenvector and the kernel mapping matrix;

Obtaining a module, configured to obtain a channel response vector of a current location of the terminal;

The calculation module is configured to calculate position information corresponding to the channel response vector of the current position of the terminal according to the main feature vector and the terminal position-row vector function.

Optionally, the modeling module is also set to:

Obtaining a location-channel response vector of a plurality of known locations of the terminal, and establishing the terminal location-channel response database according to a location-channel response vector of a plurality of known locations of the terminal.

Optionally, the modeling module is configured to generate a kernel mapping matrix according to the terminal location-channel response database in the following manner:

The data matrix in the terminal location-channel response database is called, and the kernel mapping matrix of the data matrix is calculated using the kernel function.

Optionally, the modeling module is configured to implement the calling of the data matrix in the terminal location-channel response database in the following manner:

The ray tracing algorithm is used to calculate the initial range of the terminal to be located, and the position-channel response vector of the terminal position-channel response database in the initial range is called to form a data matrix.

Optionally, the modeling module is further configured to: correct the initial range according to the location information.

Optionally, the modeling module is configured to obtain the main feature vector according to the kernel mapping matrix in the following manner:

Standardizing the kernel mapping matrix to obtain a standardized kernel mapping matrix; performing eigenvalue decomposition on the normalized kernel mapping matrix to obtain a diagonal matrix composed of eigenvalues; calculating each eigenvalue contribution rate and cumulative contribution rate, according to the cumulative contribution The main feature value is selected for the rate and the threshold value, and the feature vector corresponding to all the main feature values is used as the main feature vector.

Optionally, the modeling module is configured to generate a terminal position-row vector function according to the main feature vector and the kernel mapping matrix in the following manner:

The kernel mapping matrix is projected onto the main eigenvector to obtain a new channel response eigenvector matrix; the new channel response eigenvector matrix is subjected to dimensionality reduction processing; according to the terminal corresponding to each new channel response eigenvector after dimensionality reduction processing The position is subjected to linear regression processing on the new channel response feature vector matrix after the dimension reduction processing, and the terminal position-row vector function is obtained.

Optionally, the calculation module is set to:

The channel response vector of the current position of the terminal is projected to the main feature vector to obtain a new feature vector; the new feature vector is substituted into the terminal position-row vector function, and the position information is calculated.

The embodiment of the invention provides a terminal positioning system, which comprises the terminal positioning device provided by the embodiment of the invention.

Advantageous effects of embodiments of the present invention:

The embodiment of the invention provides a terminal positioning method, which generates a main feature vector according to a terminal position-channel response database, and then obtains a position-row vector function according to the data matrix and the main feature vector, and the channel response vector of the terminal to be tested is in the main feature. Vector projection, generating a new vector substituting position-row vector function to calculate the position; in this process, only the main feature vector needs to be generated according to the terminal position-channel response database, which has lower requirements on the terminal position-channel response database, only needs to be The data can be completed, which solves the problem that the generation of the relevant terminal location-channel response database is large. At the same time, the channel response vector of the terminal to be tested is projected in the main feature vector to generate a new vector substitution position-row vector function calculation position. The calculation process is simple and fast, and solves the complicated problem of the existing terminal positioning process.

Other aspects will be apparent upon reading and understanding the drawings and detailed description.

BRIEF abstract

1 is a schematic structural diagram of a terminal positioning apparatus according to a first embodiment of the present invention;

2 is a flowchart of a terminal positioning method according to a second embodiment of the present invention;

3 is a schematic diagram of networking of a terminal positioning system in a third embodiment of the present invention;

FIG. 4 is a flowchart of a terminal positioning method in a third embodiment of the present invention.

Embodiments of the invention

The invention will now be further illustrated by way of specific embodiments in conjunction with the accompanying drawings.

First embodiment:

FIG. 1 is a schematic structural diagram of a terminal locating device according to a first embodiment of the present invention. As shown in FIG. 1 , in the embodiment, the terminal locating device 1 provided by the present invention includes:

The modeling module 11 is configured to generate a kernel mapping matrix according to the terminal location-channel response database; obtain a main feature vector according to the kernel mapping matrix; and generate a terminal position-row vector function according to the main feature vector and the kernel mapping matrix;

The obtaining module 12 is configured to obtain a channel response vector of a current location of the terminal;

The calculating module 13 is configured to calculate position information corresponding to the channel response vector of the current position of the terminal according to the main feature vector and the terminal position-row vector function.

In some embodiments, the modeling module 11 in the above embodiment is further configured to:

In some embodiments, the modeling module 11 in the above embodiment is configured to generate a kernel mapping matrix according to the terminal location-channel response database in the following manner:

In some embodiments, the modeling module 11 in the above embodiment is configured to implement a call to the data matrix in the terminal location-channel response database in the following manner:

The initial range is corrected based on the location information. In this way, the embodiment of the present invention can implement terminal positioning with different precisions. When the terminal positioning accuracy requirement is low, the positioning can be achieved according to the position-channel response vector in the initial range. When the terminal positioning accuracy requirement is high, it is required. Corrected according to the initial range of positioning results until the user's needs are met.

In some embodiments, the modeling module 11 in the above embodiment is configured to implement obtaining a main feature vector according to the kernel mapping matrix in the following manner:

In some embodiments, the modeling module 11 in the above embodiment is configured to generate a terminal position-row vector function according to the main feature vector and the kernel mapping matrix in the following manner:

The kernel mapping matrix is projected onto the main eigenvector to obtain a new channel response eigenvector matrix; the new channel response eigenvector matrix is subjected to dimensionality reduction; according to each dimension reduction processing The terminal position corresponding to the new channel response feature vector performs linear regression processing on the channel response feature vector matrix after the dimensionality reduction process, and obtains the terminal position-row vector function.

In some embodiments, the computing module 13 in the above embodiment is configured to:

Correspondingly, the embodiment of the present invention provides a terminal positioning system, which includes the terminal positioning device 1 provided by the embodiment of the present invention.

The method in the following embodiments may be implemented in a server, and the above terminal locating device may be disposed in a server.

Second embodiment:

2 is a flowchart of a terminal locating method according to a second embodiment of the present invention. As shown in FIG. 2, in the embodiment, the terminal locating method provided by the embodiment of the present invention includes:

S201: Generate a kernel mapping matrix according to the terminal location-channel response database; obtain a primary eigenvector according to the kernel mapping matrix; and generate a terminal location-row vector function according to the primary eigenvector and the kernel mapping matrix;

S202: Obtain a channel response vector of a current location of the terminal.

S203: Calculate location information corresponding to a channel response vector of a current location of the terminal according to the main feature vector and the terminal position-row vector function.

In some embodiments, before the method S201 in the foregoing embodiment, the method further includes: acquiring a location-channel response vector of a plurality of known locations of the terminal, and establishing a terminal location according to the location-channel response vector of the plurality of known locations of the terminal- Channel response database.

In some embodiments, generating the kernel mapping matrix according to the terminal location-channel response database in the foregoing embodiment includes: calling a data matrix in the terminal location-channel response database, and calculating a kernel mapping matrix of the data matrix by using a kernel function.

In some embodiments, the data matrix in the call terminal location-channel response database in the above embodiment includes: calculating the initial range of the terminal by using a ray tracing algorithm, and calling the location-channel response of the terminal location-channel response database in the initial range. Vectors form the data matrix.

The initial range of the terminal may also be calculated by using a wireless positioning method (for example, a fingerprint positioning method based on Really Simple Syndication (RSS)).

In some embodiments, the method in the above embodiment further includes: correcting the initial range according to the location information, and continuing the subsequent process.

Optionally, the location information is added to the initial range, or the location information is added to the initial range, and one or more location information in the initial range that is the largest distance from the location information is deleted.

In some embodiments, obtaining the main feature vector according to the kernel mapping matrix in the foregoing embodiment includes: normalizing the kernel mapping matrix to obtain a normalized kernel mapping matrix; performing eigenvalue decomposition on the normalized kernel mapping matrix to obtain a feature value The diagonal matrix; calculating the contribution rate of each eigenvalue and the cumulative contribution rate, selecting the main eigenvalue according to the cumulative contribution rate and the threshold value, and using the feature vector corresponding to all the main eigenvalues as the main feature vector.

In some embodiments, generating the terminal position-row vector function according to the main feature vector and the kernel mapping matrix in the foregoing embodiment includes: projecting the kernel mapping matrix to the main feature vector to obtain a new channel response feature vector matrix; The channel response feature vector matrix is subjected to dimensionality reduction processing; linear regression processing is performed on the new channel response feature vector matrix after dimensionality reduction according to the terminal position corresponding to each new channel response feature vector after the dimensionality reduction processing, and the terminal position is obtained. - Line vector function.

In some embodiments, calculating the location information corresponding to the channel response vector of the current location of the terminal according to the primary feature vector and the terminal location-row vector function in the foregoing embodiment includes: performing channel response vector of the current location of the terminal to the primary feature vector. Projection, a new feature vector is obtained; the new feature vector is substituted into the terminal position-row vector function, and the position information is calculated.

The present invention will be further explained in conjunction with specific application scenarios.

Third embodiment:

This embodiment is described by taking an indoor positioning system as an example, and provides an algorithm that can effectively utilize multipath information to reduce positioning errors.

In practical applications, the energy of each radial component in the channel between the AP and the terminal also has a statistically significant functional relationship between the terminal locations. If the energy-terminal position relationship of each radial component can be fully utilized, not only Improve estimation accuracy while using KPCA (Kernel principal) Component analysis, kernel principal component analysis) can also reduce the database storage space between the terminal location and the AP receiving energy, and the corresponding workload required to establish the database. This embodiment proposes a radiolocation algorithm based on kernel-primary component analysis of terminal-AP channel response, because the multi-path energy-terminal position function relationship included in the channel response has nonlinear characteristics and geographical limitations. The eigenvectors suitable for the current terminal position estimation are extracted by gradually narrowing the geographical range and the kernel principal component analysis method, and the regression method is used to realize the terminal position estimation, which can fully utilize the energy-terminal position relationship of each radial component, which can not only improve the estimation. Accuracy, while also reducing the database storage space between the terminal location and the AP receiving energy, and the corresponding workload required to build the database.

In this embodiment, the kernel principal component analysis method is adopted for the channel response database, and the nonlinear feature vector in the multipath component can be extracted on the one hand, and the dimension of the feature vector can also be reduced, and the feature of the extracted smaller dimension is further The vector is subjected to regression analysis to obtain the position estimate. Firstly, all the data matrices H' in the given range Φ are taken out from the terminal position-channel response database, and then the kernel principal component analysis is used to reduce the dimensionality of H'.

Then, according to the data matrix H' and the corresponding known position, linear regression is performed to obtain the position-row vector linear function. Finally, the channel response vector of the current position of the terminal is projected onto the main eigenvector to obtain a new eigenvector, which is substituted into the position-row vector linearity. Function to estimate the terminal location.

Specifically, as shown in FIG. 3, it is assumed that there is a typical wireless local area network (WLAN) positioning system, in which there is a mobile terminal with a location to be determined and M APs (the AP location is unknown), so that the location of the mobile terminal is The two-dimensional vector x=[x _x ,x _y ] (corresponding to the two-dimensional plane), each AP can obtain the channel response between the AP and the terminal, wherein the discrete channel response between the mth AP and the terminal is L-dimensional The row vector h _m (x)=[h _m,1 (x) h _m,2 (x) ... h _m,L (x)], where h _m,l (x) is the mth time of the first sampling time The complex gain of the channel between the AP and the terminal, and the total vector of the discrete channel response between the M APs and the terminal can be expressed as an ML dimensional matrix h(x)=[h ₁ (x) h ₂ (x) .. h _M (x)] ^T .

It is assumed that the system has recorded the channel response vector between the terminal location and the AP in advance at N known locations, that is, the terminal location channel response "map", which may be expressed as a matrix form as follows:

H=[h(x(1)) ... h(x(n)) ... h(x(N))] ^T , where x(n) represents the nth known terminal position.

H is a three-dimensional matrix.

The wireless positioning problem is to give the current terminal total channel vector h(x)=[h ₁ (x) h ₂ (x) ... h _M (x)] ^T , how to use the existing terminal location channel response “map " Estimated to get the terminal location. Since the positioning method based on kernel principal component analysis has the same algorithm steps in the position estimation in each given range space, the following only takes one position estimation as an example:

Step 1: Take all the data matrices in the given range Φ from the terminal location-channel response database as follows: H'=[... h(x(n)) ...] ^T ; select a typical kernel function K(·,·), can obtain the N'×N' dimensional kernel mapping matrix K, N' of the corresponding kernel function and data matrix is the number of known positions in the given range F, K _i,j =K(h (x(i)), h(x(j))), performing eigenvalue decomposition on the kernel mapping matrix, wherein the diagonal matrix formed by the eigenvalues is D=Diag(λ ₁ , λ ₂ , . . . , λ _{N '),} the l th feature vectors a _l. Selecting L' main feature vectors, the corresponding main feature values and main feature vectors are respectively

with

Where L'<<N'.

The second step: projecting the kernel mapping matrix K to the main eigenvector to obtain a new channel response eigenvector matrix

among them,

among them,

The nth element of the lth feature vector.

The third step: performing a support vector machine regression to obtain a position-row vector linear function according to the new channel response feature vector matrix and the corresponding known position;

Where Wi is the weight coefficient of the regression, bx is the regression coefficient of the x-axis, and by is the regression of the y-axis coefficient.

The fourth step: projecting the channel response vector of the current position of the terminal to the main feature vector to obtain a new feature vector,

Substituting the position-line vector linear function of the linear regression after the third step in the third step, the terminal position is estimated.

The present invention will be further explained in conjunction with FIG. 4. As shown in FIG. 4, in the embodiment, the terminal positioning method provided by the present invention includes the following steps:

S401: Establish a terminal location-channel response database.

As shown in FIG. 3, the system model assumes that the number of APs is 3. The terminal positioning algorithm provided by the present invention includes a two-stage working mode: an offline phase, which is mainly used to establish a terminal location-channel response database, and an online phase, which is mainly used to implement Terminal positioning.

This step is offline, using the mobile device to collect the channel response information of each reference point and the AP, and correlating the location information at the time of acquisition to construct a location fingerprint database. The location fingerprint database (LFDB) is constructed in the offline phase. The location fingerprint data is composed of many database elements: DBE={L,R}, where L is the physical location and R represents the fingerprint collected at the location, expressed as :

Where N _r represents the number of APs in the range of device communication in the offline phase, r _i is the channel response data between the i-th AP and the terminal received by the sampling device, and id _i is the ID of the i-th AP.

A terminal position-channel response database is determined using a ray tracing algorithm. The ray in the ray tracing algorithm may be transmitted directly from the transmitter to the receiver, or it may be multiple reflections, diffraction, and transmission. When the receiver arrives at the receiver, this embodiment only considers the reflection condition, and the maximum number of reflections of the signal is 3 (because the energy of the signal has been greatly lost after 3 times of reflection, the influence can be ignored). Tracking calculates all losses in each ray propagation process. The tracking calculation is performed until the ray reaches the receiver, and the data of the reference point in the located area is counted to form a location fingerprint database, that is, a terminal location-channel response database.

In practical applications, each AP can obtain a channel response between the AP and the terminal, wherein the discrete channel response between the mth AP and the terminal is an L-dimensional row vector h' _m (x)=[h' _{m, 1} (x) h' _m,2 (x) ... h' _m,L (x)], where h' _m,l (x) is the channel between the mth AP and the terminal at the 1st sampling time Complex gain. The total vector of the discrete channel responses between the M APs and the terminal can be expressed as an ML dimensional matrix h'(x)=[h' ₁ (x) h' ₂ (x) ... h' _M (x)]. However, in the actual Orthogonal Frequency Division Multiplexing (OFDM) system, the channel response will be correspondingly dispersed at the sampling point. The discrete channel response between the mth AP and the terminal can be expressed as

Where 0 ≤ τ _m T _s ≤ T _G , t is time, T _s is the sampling interval, and τ _{m, l} is the normalized delay corresponding to the lth radial component of the channel between the mth AP and the terminal, T _G is the complex gain after the dispersion of the discrete channel response between the mth AP and the terminal in the sampling interval is expressed as:

In the OFDM system, the Fast Fourier Transformation (FFT) has a size of N. When τ _m,l is an integer, the complex gain h' _m,l does not diverge at other sampling times; but when τ _{m,l is a} non-integer, the complex gain h' _{m,l is} dispersed at other sampling times. Then the discrete channel response h _m (x)=[h _m,1 (x) h _m,2 (x) ... h _m,L (x)] between the diffused m-th AP and the terminal The response database is expressed as: H = [h(x(1)) ... h(x(n)) ... h(x(N))] ^T where x(n) represents the nth known terminal position.

S402: Calculate a main feature vector according to the terminal location-channel response database.

This step and all the following steps are in the online phase, mainly using the correspondence relationship between the positional relationship and the fingerprint database information, and determining the position by the measured fingerprint. First, the ray tracing algorithm is used to measure the channel response h of the target position. ), wherein the discrete channel response between the mth AP and the terminal is an L-dimensional row vector h' _m (x)=[h' _m,1 (x) h' _m,2 (x) ... h' _{m , L} (x)], where h' _m,l (x) is the complex gain of the channel between the mth AP and the terminal at the 1st sampling time. The dispersed discrete channel response h _m (x) = [h _m,1 (x) h _m,2 (x) ... h _m,L (x)]. In order to obtain position estimation with different precision, the position estimation range is gradually reduced (for example, the position range can be narrowed down according to the empirical value, and the updated position range can be centered on the previous estimated position, and the previous position range is multiplied by the reduction. The scale factor is obtained from the current position range), and the kernel principal component analysis and regression analysis method are used in each estimation range to obtain a position estimate with a small error range.

In this step, the data matrix H'=[... h(x(n)) ...] ^{T is first} selected, and the polynomial kernel function K(x _i , x _j )=(<x _i , x _j >+d) ^{p is selected.} , p∈N, d≥0, obtain the N×N-dimensional kernel mapping matrix K of the corresponding data matrix, where K _i,j =K(h(x(i)), h(x(j))), N is The number of samples, which is the number of known locations in the data matrix. among them,

Standardizing the above K matrix yields:

among them,

I is an N×N unit matrix, and a kernel mapping matrix

The eigenvalue decomposition is performed to obtain the eigenvalues and eigenvectors of the kernel mapping matrix, wherein the diagonal matrix composed of the eigenvalues is D=Diag(λ ₁ , λ ₂ , . . . , λ _N ), and the first eigenvector is a _l ; sequentially obtain each eigenvalue contribution rate, and the corresponding eigenvalue cumulative contribution rate

And compared with the cumulative threshold (such as 98%), when the cumulative contribution rate of the eigenvalue is greater than the cumulative threshold, the calculation is stopped, and the corresponding eigenvalues λ ₁ , λ ₂ ... λ _{L are selected} , according to the characteristic value Selecting the L main feature vectors, the corresponding main feature value and main feature vector are

with

S403: Projecting the data matrix to the main feature vector to obtain a new channel response feature vector matrix

This step is mainly to achieve the dimensionality reduction processing of the data matrix.

among them;

S404: A new channel response feature vector matrix according to the corresponding known position

Support vector machine regression is performed to obtain a position-row vector linear function.

specific,

α _i

Is a Lagrangian multiplier solved by SVR (Support Vector Machine);

N _NSV is the number of standard support vectors, and ε is the insensitivity function.

S405: Project a channel response vector of the current location of the terminal to the main feature vector to obtain a new feature vector.

The new feature vector obtained is:

S406: Calculate the terminal position according to the new feature vector and the position-row vector linear function.

In this step, the new feature vector calculated in step S405 is substituted into the position-row vector linear function obtained in step S404, and the terminal position (x ₀ , y ₀ ) is estimated.

S407: Correct the geographical range to obtain the most accurate positioning.

According to the initial estimated terminal position (x ₀ , y ₀ ), the geographical range is gradually reduced, that is, the range of the database is reduced; the coordinate adjustment precision is determined to be m, and the horizontal coordinate x is adjusted to x∈[x ₀ -m, x ₀ +m], The ordinate y is adjusted to y ∈ [y _{0 -} m, y ₀ + m], x, y ∈ N ^* ; all data matrices within a given range Φ are taken from the terminal position-channel response database as follows: H'= [h(x(1)) ... h(x(n)) ...] ^T , n ∈ Φ, and there is N'=|Φ|, and then re-estimate the terminal position according to the above steps S402-S406 ( x ₁ , y ₁ ). The above steps are iterated to obtain the terminal position (x ₂ , y ₂ ), (x ₃ , y ₃ ), and based on the empirical value, (x ₃ , y ₃ ) is taken as the final coordinate for estimating the position of the terminal.

In summary, through the implementation of the present invention, at least the following beneficial effects exist:

The embodiment of the invention provides a terminal positioning method, which generates a main feature vector according to a terminal position-channel response database, and then obtains a position-row vector function according to the data matrix and the main feature vector, and the channel response vector of the current position of the terminal is in the main feature. Vector projection, generating a new eigenvector substituting position-row vector function to calculate the position; in this process, only the main eigenvector needs to be generated according to the terminal position-channel response database, which has lower requirements on the terminal position-channel response database, only It takes several data to complete, and the terminal positioning is simply realized, which solves the problem of large database generation workload in related technologies. At the same time, the channel response vector of the current position of the terminal is projected in the main feature vector to generate a new feature vector. Substituting the position-row vector function to calculate the position, the calculation process is simple and fast, and solves the complicated problem of the related terminal positioning process.

Embodiments of the present invention also provide a computer readable storage medium storing computer executable instructions for performing any of the methods described above.

One of ordinary skill in the art will appreciate that all or a portion of the above steps may be performed by a program to instruct related hardware, such as a processor, which may be stored in a computer readable storage medium, such as a read only memory, disk or optical disk. Wait. Alternatively, all or part of the steps of the above embodiments may also be implemented using one or more integrated circuits. Correspondingly, each mode in the above embodiment The blocks/units may be implemented in the form of hardware, for example by means of an integrated circuit, or in the form of software function modules, for example by means of a processor executing the programs and instructions in the memory and memory. The invention is not limited to any specific form of combination of hardware and software.

The above is only a specific embodiment of the present invention, and is not intended to limit the present invention in any way. Any simple modification, equivalent change, combination or modification of the above embodiments in accordance with the technical spirit of the present invention is still in the present invention. The scope of protection of the technical solution of the invention.

Industrial applicability

The above technical solution simply realizes positioning.

Claims

A terminal positioning method includes:

Generating a kernel mapping matrix according to the terminal location-channel response database; acquiring a primary feature vector according to the kernel mapping matrix; and generating a terminal location-row vector function according to the primary feature vector and the kernel mapping matrix;

Obtaining a channel response vector of the current location of the terminal;

And calculating location information corresponding to the channel response vector of the current location of the terminal according to the primary feature vector and the terminal location-row vector function.
The method for locating a terminal according to claim 1, further comprising:

Obtaining a location-channel response vector of a plurality of known locations of the terminal, and establishing the terminal location-channel response database according to a location-channel response vector of a plurality of known locations of the terminal.
The terminal positioning method according to claim 2, wherein the generating a kernel mapping matrix according to the terminal location-channel response database comprises:

The data matrix in the terminal location-channel response database is invoked, and a kernel mapping matrix of the data matrix is calculated using a kernel function.
The terminal positioning method according to claim 3, wherein said invoking a data matrix in said terminal location-channel response database comprises:

The initial range of the terminal is calculated using a ray tracing algorithm, and the position-channel response vector of the terminal position-channel response database within the initial range is invoked to form the data matrix.
The terminal positioning method according to claim 4, further comprising: correcting the initial range according to the position information, and continuing the subsequent steps.
The method for locating a terminal according to any one of claims 1 to 5, wherein the obtaining a main feature vector according to the kernel mapping matrix comprises: normalizing the kernel mapping matrix to obtain a standardized kernel mapping matrix; The normalized kernel mapping matrix performs eigenvalue decomposition to obtain a diagonal matrix composed of eigenvalues; calculates each eigenvalue contribution rate and cumulative contribution rate, and selects a main eigenvalue according to the cumulative contribution rate and the threshold value, The feature vectors corresponding to all the main feature values are used as the main feature vector.
The terminal positioning method according to claim 6, wherein the generating the terminal position-row vector function according to the main feature vector and the kernel mapping matrix comprises:

Projecting the kernel mapping matrix to the main feature vector to obtain a new channel response feature vector matrix; performing dimensionality reduction processing on the new channel response feature vector matrix; according to each dimension reduction processed new channel The terminal position corresponding to the feature vector is subjected to linear regression processing on the new channel response feature vector matrix after the dimension reduction processing, and the terminal position-row vector function is obtained.
The terminal positioning method according to claim 7, wherein the calculating location information corresponding to the channel response vector of the current location of the terminal according to the main feature vector and the terminal position-row vector function comprises:

And mapping a channel response vector of the current location of the terminal to the main feature vector to obtain a new feature vector; and substituting the new feature vector into the terminal location-row vector function to calculate the location information.
A terminal positioning device includes:

a modeling module, configured to generate a kernel mapping matrix according to the terminal location-channel response database; obtain a primary feature vector according to the core mapping matrix; and generate a terminal location-row vector function according to the primary feature vector and the kernel mapping matrix;

Obtaining a module, configured to obtain a channel response vector of a current location of the terminal;

And a calculating module, configured to calculate location information corresponding to the channel response vector of the current location of the terminal according to the primary feature vector and the terminal location-row vector function.
The terminal locating device according to claim 9, wherein the modeling module is further configured to:

Obtaining a location-channel response vector of a plurality of known locations of the terminal, and establishing the terminal location-channel response database according to a location-channel response vector of a plurality of known locations of the terminal.
The terminal locating device of claim 10, wherein the modeling module is configured to generate a kernel mapping matrix according to the terminal location-channel response database in the following manner:

The data matrix in the terminal location-channel response database is invoked, and a kernel mapping matrix of the data matrix is calculated using a kernel function.
The terminal locating device of claim 11, wherein the modeling module is configured to invoke a data matrix in the terminal location-channel response database in the following manner:

Calculating an initial range of the terminal to be located by using a ray tracing algorithm, and calling a position-channel response vector of the terminal position-channel response database in the initial range to form the data matrix.
The terminal locating device according to claim 12, wherein the modeling module is further configured to:

The initial range is corrected based on the position information.
The terminal locating device according to any one of claims 9 to 13, wherein the modeling module is configured to implement acquisition of a main feature vector according to the kernel mapping matrix in the following manner:

Normalizing the kernel mapping matrix to obtain a standardized kernel mapping matrix; performing eigenvalue decomposition on the normalized kernel mapping matrix to obtain a diagonal matrix composed of eigenvalues; calculating each eigenvalue contribution rate and cumulative contribution rate And selecting a main feature value according to the cumulative contribution rate and the threshold value, and using the feature vector corresponding to all the main feature values as the main feature vector.
The terminal locating device according to claim 14, wherein the modeling module is configured to generate a terminal position-row vector function according to the main feature vector and the kernel mapping matrix in the following manner:

Projecting the kernel mapping matrix to the main feature vector to obtain a new channel response feature vector matrix; performing dimensionality reduction processing on the new channel response feature vector matrix; according to each dimension reduction processed new channel The terminal position corresponding to the feature vector is subjected to linear regression processing on the new channel response feature vector matrix after the dimension reduction processing, and the terminal position-row vector function is obtained.
The terminal positioning device according to claim 15, wherein said calculation module is configured to:

And mapping a channel response vector of the current location of the terminal to the main feature vector to obtain a new feature vector; and substituting the new feature vector into the terminal location-row vector function to calculate the location information.
A terminal positioning system comprising the terminal positioning device according to any one of claims 9 to 16.