A passive sensing device and method based on intelligent reflective surface

Abstract

The present invention discloses a passive sensing device and method based on an intelligent reflective surface, wherein the device comprises an intelligent reflective surface and a sensor. The method comprises the following steps: the electromagnetic wave emitted by the wireless communication access point is reflected by the target after propagating through the free space, and the reflected signal is reflected twice when passing through the intelligent reflective surface. The control of the secondary reflection process is achieved by adjusting the relevant parameters of the intelligent reflective surface, and the sensor receives the secondary reflection signal, and uses the signal to achieve the perception of the space target; through the interference removal method based on the atomic norm, the interference signal can be suppressed, and the effective target echo signal can be extracted; finally, the receiving signal matrix is ​​constructed based on the Hankel matrix, and the perception of the space target is achieved based on the MUSIC method. The present invention improves the target perception performance of the communication perception integrated system, while reducing the overall system cost.

Description

Technical Field

[0001] The present invention belongs to the technical field of communication perception and wireless positioning, and in particular to a passive perception device and method based on an intelligent reflective surface. Background Art

[0002] The integrated communication and perception system is similar to the communication radar perception system and has become a research hotspot this year. The integrated communication and perception system can realize the functions of target perception and wireless communication at the same time. In the integrated communication and perception system, signals can be actively transmitted to detect targets, or passive detection can be achieved by receiving signals through sensors. However, in the integrated communication and perception system, it is difficult for sensors to distinguish the signals reflected from the targets in the received signals, so it is very important to study the denoising method for the integrated communication and perception system. At the same time, high-resolution spatial spectrum estimation methods also play an important role in achieving accurate perception. The traditional fast Fourier transform is limited by the Rayleigh limit and has limited resolution; the super-resolution spatial spectrum estimation method represented by the multiple signal classification and rotation invariant subspace method has certain requirements on the signal-to-noise ratio and the number of snapshots; the sparse reconstruction method represented by compressed sensing divides the grid in the spatial domain to achieve super-resolution estimation of parameters, but if the target is not on the grid, off-grid errors will occur. Therefore, it is of great significance to achieve high-resolution DOA estimation with fewer snapshots.

[0003] Patent number 202111224815.4, it uses a programmable metasurface to generate multiple sets of random dual-beam surface codes, each set of dual-beam pointing angles are different, and the orthogonal matching pursuit algorithm is used to obtain information such as the angle of the incoming wave. This method can only detect a limited number of directions and does not have communication capabilities.

[0004] Patent number 202111416744.8 estimates the channel state information matrix of the received signal through the pilot signal carried by the 5G NR signal actively transmitted by the user to the base station, and then uses the MUSIC method to estimate the DOA of the signal. This method requires active cooperation from the user end, and the MUSIC method requires multiple snapshots to accumulate to achieve better results. Summary of the invention

[0005] The purpose of the present invention is to provide a passive sensing device and method based on an intelligent reflective surface to solve the technical problems in the prior art that interference signals are not considered, multiple snapshots need to be accumulated to estimate the target, and there is an error platform in the target parameter estimation.

[0006] In order to solve the above technical problems, the specific technical solutions of the present invention are as follows:

[0007] A passive sensing device based on an intelligent reflecting surface includes an intelligent reflecting surface and a sensor, wherein the intelligent reflecting surface controls the amplitude and phase of the electromagnetic wave signal reflected by it through a controller, and the sensor includes an antenna and a data processing device, and the data processing device is composed of a radio frequency module and a signal processing unit, wherein the radio frequency module has only a single radio frequency channel.

[0008] The present invention also proposes a sensing method of a passive sensing device based on an intelligent reflective surface, comprising the following steps:

[0009] Step 1: Create a receiving signal model;

[0010] Step 2: The received signal model mainly includes the target signal, the interference signal and the environmental noise. The interference signal and the environmental noise in the received signal are suppressed by the interference removal method based on the atomic norm and the measurement matrix optimization, so as to complete the extraction of the target signal.

[0011] Step 3: The target signal reconstruction method based on Hankel matrix calculates the Hankel matrix of the target signal, takes the Hankel matrix of the target signal as the reconstructed signal, uses the MUSIC method to decompose the reconstructed signal to obtain the signal subspace and the noise subspace, and uses the noise subspace to estimate the spatial spectrum, thereby realizing the perception of the target space.

[0012] Furthermore, the received signal model in step 1 includes a target signal from the target and reflected by the smart reflective surface, an interference signal from a wireless communication access point directly propagated through the smart reflective surface, and environmental noise.

[0013] Furthermore, the interference removal method in step 2 includes an interference removal method based on the atomic norm and an interference removal method based on measurement matrix optimization, wherein the interference removal method based on the atomic norm calculates the atomic norm of the target signal to obtain a denoised vector after removing the interference signal, and the interference removal method based on measurement matrix optimization realizes the removal of interference signals in a specific direction by adjusting the measurement matrix; the interference removal method based on the atomic norm specifically includes the following steps: constructing a sparse reconstruction optimization model including L2 norm and atomic norm; for the sparse reconstruction optimization model, constructing a convex optimization problem of semi-positive definite programming based on the Toeplitz matrix of the model; using an iterative method to solve the above-mentioned convex optimization problem to obtain a denoised vector; the sparse reconstruction optimization model including L2 norm and atomic norm is essentially to minimize the sum of the square of L2 norm and atomic norm, wherein the square of L2 norm is the square of L2 norm of the difference between the received signal minus the target signal and then minus the interference signal, and the atomic norm is the atomic norm of the target signal.

[0014] Furthermore, a convex optimization problem of semi-positive definite programming is constructed based on the Toeplitz matrix. This process is a method for solving the atomic norm of the target signal. In this process, the atomic norm of the target signal is expressed as the sum of the trace of the Toeplitz matrix composed of a vector u to be solved and a variable v to be solved, wherein the constraints that the vector u to be solved and the variable v to be solved need to satisfy are that the Toeplitz matrix composed of the vector u to be solved, the matrix jointly composed of the variable v to be solved and the target signal needs to be a semi-positive definite matrix, and the vector u and the variable v that satisfy the above constraints are iteratively solved through semi-positive definite programming.

[0015] Furthermore, an interference removal method based on measurement matrix optimization suppresses interference signals in a specific direction, which includes the following steps: modeling the measurement matrix optimization problem as an optimization problem of semi-positive definite programming; solving the aforementioned semi-positive definite programming problem to obtain a matrix, and performing eigendecomposition on the matrix; using the eigenvector obtained by the eigendecomposition to represent a row of the measurement matrix.

[0016] Furthermore, the signal reconstruction method based on the Hankel matrix in step 3 only requires a snapshot of the received signal.

[0017] Furthermore, in step 3, the signal reconstruction method based on the Hankel matrix takes out multiple groups of ML elements from the denoised M*1-dimensional vector and reconstructs the matrix. The dimension of the reorganized matrix is ​​(ML)*L, where M is the number of array elements of the smart metasurface, and L is the number of rows of the reconstructed signal matrix.

[0018] Furthermore, in step 3, the reconstructed matrix is ​​subjected to singular value decomposition using the MUSIC method to obtain several singular values, among which some singular values ​​are several orders of magnitude larger than the remaining singular values. The singular vectors corresponding to these singular values ​​constitute the signal subspace, and the number of these singular values ​​depends on the number of targets. The singular vectors corresponding to other singular values ​​that are several orders of magnitude smaller than these results constitute the noise subspace.

[0019] Furthermore, in step 3, the spatial spectrum is composed of the ratio of the L2 norm square of the steering vector of the smart reflection surface to the L2 norm square of the aforementioned noise subspace, wherein the steering vector is used to describe the spatial phase difference of the smart reflection surface, and the spatial phase difference indicates that there is a certain difference in phase between signals received by array elements with different positions in space. The smart reflection surface is composed of multiple array elements, and the structure of the steering vector is determined by the relative positions of the array elements constituting the smart reflection surface. The specific value of the steering vector is determined by the angle between the target and the smart reflection surface.

[0020] The invention provides a passive sensing device and method based on a smart reflective surface, which has the following advantages:

[0021] 1. The present invention uses a sensor with only a single radio frequency channel to achieve the perception of the target.

[0022] 2. The interference removal method provided by step 2 of the present invention can effectively filter out interference signals and environmental noise, thereby improving the performance of target perception.

[0023] 3. The signal reconstruction method and MUSIC method provided in step 3 of the present invention have a higher performance in estimating the direction of the target than methods such as orthogonal matching pursuit and fast Fourier transform. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a scene schematic diagram of the present invention;

[0025] Figure 2 This is a simulation result diagram of the spatial spectrum estimation of the present invention when the signal-to-noise ratio is 20dB;

[0026] Figure 3 The estimation performance of the present invention under different signal-to-noise ratios (without measurement matrix optimization);

[0027] Figure 4 The estimation performance of the present invention under different signal-to-noise ratios (with measurement matrix optimization);

[0028] Figure 5 Schematic diagram of root mean square error (RMSE) of wireless communication access points and sensors at different positions in the present invention;

[0029] Figure 6 The figure is a schematic diagram of a passive sensing device based on an intelligent reflective surface according to the present invention. DETAILED DESCRIPTION

[0030] In order to better understand the purpose, structure and function of the present invention, the passive sensing device and method based on the intelligent reflective surface of the present invention are further described in detail below in conjunction with the accompanying drawings.

[0031] The present invention provides a passive sensing device based on an intelligent reflecting surface, comprising an intelligent reflecting surface and a sensor, wherein the intelligent reflecting surface controls the amplitude and phase of an electromagnetic wave signal reflected by the intelligent reflecting surface through a controller, and the sensor comprises an antenna and a data processing device, and the data processing device is composed of a single radio frequency channel and a signal processing unit.

[0032] The present invention also proposes a sensing method of a passive sensing device based on an intelligent reflective surface, comprising the following steps:

[0033] Step 1: Create a receiving signal model;

[0034] The received signal model includes a target signal from a target and reflected by a smart reflective surface, an interference signal from a wireless communication access point and directly transmitted through the smart reflective surface, and environmental noise.

[0035] Step 2: The received signal model mainly includes the target signal, the interference signal and the environmental noise. The interference signal and the environmental noise in the received signal are suppressed by the interference removal method based on the atomic norm and the measurement matrix optimization, so as to complete the extraction of the target signal.

[0036] The interference removal method includes an interference removal method based on an atomic norm and an interference removal method based on measurement matrix optimization, wherein the interference removal method based on the atomic norm calculates the atomic norm of the target signal to obtain a denoised vector after removing the interference signal, and the interference removal method based on the measurement matrix optimization realizes the removal of interference signals in a specific direction by adjusting the measurement matrix; the interference removal method based on the atomic norm specifically includes the following steps: constructing a sparse reconstruction optimization model including an L2 norm and an atomic norm; for the sparse reconstruction optimization model, constructing a convex optimization problem of semi-positive definite programming based on the Toeplitz matrix of the model; solving the above convex optimization problem by an iterative method to obtain a denoised vector; the sparse reconstruction optimization model including the L2 norm and the atomic norm is essentially to minimize the sum of the square of the L2 norm and the atomic norm, wherein the square of the L2 norm is the square of the L2 norm of the difference between the received signal minus the target signal and then minus the interference signal, and the atomic norm is the atomic norm of the target signal.

[0037] A convex optimization problem of semi-positive definite programming is constructed based on the Toeplitz matrix. This process is a method for solving the atomic norm of the target signal. In this process, the atomic norm of the target signal is represented as the sum of the trace of the Toeplitz matrix composed of a vector u to be solved and a variable v to be solved, wherein the constraints that the vector u to be solved and the variable v to be solved need to satisfy are that the Toeplitz matrix composed of the vector u to be solved, the matrix jointly composed of the variable v to be solved and the target signal needs to be a semi-positive definite matrix, and the vector u and the variable v that satisfy the above constraints are iteratively solved by semi-positive definite programming.

[0038] The interference removal method based on measurement matrix optimization suppresses interference signals in a specific direction, which includes the following steps: modeling the measurement matrix optimization problem as an optimization problem of semi-positive definite programming; solving the aforementioned semi-positive definite programming problem to obtain a matrix, and performing eigendecomposition on the matrix; using the eigenvector obtained by the eigendecomposition to represent a row of the measurement matrix.

[0039] Step 3: The target signal reconstruction method based on Hankel matrix calculates the Hankel matrix of the target signal, takes the Hankel matrix of the target signal as the reconstructed signal, uses the MUSIC method to decompose the reconstructed signal to obtain the signal subspace and the noise subspace, and uses the noise subspace to estimate the spatial spectrum, thereby realizing the perception of the target space.

[0040] The signal reconstruction method based on the Hankel matrix only requires a snapshot of the received signal.

[0041] The signal reconstruction method based on the Hankel matrix takes out multiple groups of ML elements from the denoised M*1-dimensional vector to reconstruct the matrix. The dimension of the reorganized matrix is ​​(ML)*L, where M is the number of array elements of the smart metasurface and L is the number of rows of the reconstructed signal matrix.

[0042] After using the MUSIC method to perform singular value decomposition on the reconstructed matrix, several singular values ​​are obtained, among which some singular values ​​are several orders of magnitude larger than the remaining singular values. The singular vectors corresponding to these singular values ​​constitute the signal subspace. The number of these singular values ​​depends on the number of targets. The singular vectors corresponding to other singular values ​​that are several orders of magnitude smaller than these results constitute the noise subspace.

[0043] The spatial spectrum is composed of the ratio of the L2 norm square of the steering vector of the smart reflective surface to the L2 norm square of the aforementioned noise subspace, wherein the steering vector is used to describe the spatial phase difference of the smart reflective surface. The spatial phase difference indicates that there is a certain difference in phase between signals received by array elements with different positions in space. The smart reflective surface is composed of multiple array elements. The structure of the steering vector is determined by the relative positions of the array elements that constitute the smart reflective surface. The specific value of the steering vector is determined by the angle between the target and the smart reflective surface.

[0044] A specific embodiment of a sensing method of a passive sensing device based on an intelligent reflective surface of the present invention is as follows:

[0045] The following steps are involved:

[0046] In step 1, a received signal model is created, which obeys the following representation:

[0047] r=Gξ+Ga(θ AR )η+w

[0048] Where r represents the received signal, G represents the measurement matrix of the smart reflective surface, ξ represents the target signal, η represents the interference signal from the wireless communication access point directly transmitted through the smart reflective surface, and θ AR represents the angle between the wireless communication access point and the smart reflective surface, a(θ AR) represents the guidance vector of the smart reflection surface corresponding to the aforementioned angle, and w represents the ambient noise.

[0049] In step 2, the interference signal and environmental noise in the received signal are suppressed by the interference removal method based on the atomic norm and measurement matrix optimization, thereby completing the extraction of the target signal; wherein, the goal of the interference removal method based on the atomic norm is to minimize the sum of the square of the L2 norm and the atomic norm, which satisfies the following form

[0050] min ξ，η ||r-Gξ-Ga(θ AR )η||2 2 +ρ||ξ|| A

[0051] min means to find the minimum value that satisfies certain constraints, ||*|| 2 2 represents the square of the L2 norm, ρ represents a constant, ||*|| A Represents the atomic norm; the optimization of the measurement matrix is ​​performed iteratively to optimize each row, where the optimization target of the measurement matrix is ​​determined by the following formula:

[0052] min g ||g H a(θ AR )|| 2 2

[0053] st|g m,n |≤1

[0054] g represents a row of the measurement matrix, (*) H represents the Hermitian matrix, st represents the need to meet the subsequent conditions, g m，n Represents the element in the mth row and nth column, where m and n are constants less than the number of rows and columns of the measurement matrix, and |*| represents an absolute value.

[0055] In step 3, the received signal is reconstructed. The reconstruction method is based on the Hankel matrix. The main reconstruction operation is to take out multiple groups of ML elements from the denoised M*1-dimensional vector to reconstruct the matrix. The dimension of the reconstructed matrix is ​​(ML)*L, where M is the number of array elements of the smart metasurface and L is the number of rows of the reconstructed signal matrix. Then, the MUSIC algorithm is used to perform singular value decomposition on the reconstructed signal. After decomposition, several singular values ​​are obtained, among which some singular values ​​are several orders of magnitude larger than the remaining singular values. The singular vectors corresponding to these singular values ​​constitute the signal subspace. The number of these singular values ​​depends on the number of targets. The singular vectors corresponding to other singular values ​​that are several orders of magnitude smaller than the results of this part constitute the noise subspace. The spatial spectrum is composed of the ratio of the L2 norm square of the guiding vector of the smart reflective surface to the L2 norm square of the aforementioned noise subspace. The spatial spectrum obeys the following form:

[0056] g(θ)=||a(θ)|| 2 2 / ||a(θ) H U|| 2 2

[0057] g(θ) represents the spatial spectrum result at an angle of θ, and U represents the noise subspace.

[0058] Figure 2 This is the estimation result of the spatial spectrum of the present invention when the signal-to-noise ratio is 20 dB. In addition, the figure also includes the results of some other spatial spectrum estimation methods for comparison, mainly including fast Fourier transform (FFT), fast Fourier transform of the signal after interference filtering, orthogonal matching pursuit method (OMP), L1 norm method, and actual target orientation. The closer to the actual target orientation, the more accurate the estimation of the target. The result curve proves that the present invention has better DOA estimation performance.

[0059] Figure 3 and 4 They respectively represent the performance of DOA estimation without and with measurement matrix optimization. The results prove that the measurement matrix optimization method proposed in the present invention can significantly improve the performance of DOA estimation.

[0060] Figure 5 is the root mean square error of the present invention under different wireless communication access points and sensor positions, where the noise variance is 0.01, Figure 5 In the first figure, the wireless access point is located at (0m, 0m), the smart reflector is located at (20m, 20m), and the sensor is located at (20m, 0m). The picture shows the lower bound of the root mean square error of DOA estimation. The area close to the wireless access point and the smart reflector has better performance than other areas. Figure 5In the second picture, the wireless communication access point is located at (0m, 0m), the smart reflective surface is located at (20m, 20m), and the sensor is located at (20m, 17m). Figure 5 In the third picture, the wireless communication access point is located at (20m, 0m), the smart reflective surface is located at (20m, 20m), and the sensor is located at (20m, 17m). Figure 5 In the fourth picture, the wireless communication access point is located at (20m, -20m), the smart reflective surface is located at (20m, 20m), and the sensor is located at (20m, 17m).

[0061] It is to be understood that the present invention is described by some embodiments, and it is known to those skilled in the art that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present invention. In addition, under the teachings of the present invention, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the scope of protection of the present invention.

Claims

1. A sensing method based on a passive sensing device of an intelligent reflective surface, characterized in that: It includes an intelligent reflective surface and a sensor, wherein the intelligent reflective surface controls the amplitude and phase of the electromagnetic wave signal reflected by it through a controller, and the sensor includes an antenna and a data processing device, and the data processing device is composed of a radio frequency module and a signal processing unit, wherein the radio frequency module has only a single radio frequency channel; The method realizes the perception of the spatial position of the target, and comprises the following steps: Step 1: Create a receiving signal model; Step 2: The received signal model mainly includes the target signal, the interference signal and the environmental noise. The interference signal and the environmental noise in the received signal are suppressed by the interference removal method based on the atomic norm and the measurement matrix optimization, so as to complete the extraction of the target signal. Step 3: The target signal reconstruction method based on the Hankel matrix calculates the Hankel matrix of the target signal, takes the Hankel matrix of the target signal as the reconstructed signal, decomposes the reconstructed signal using the MUSIC method to obtain the signal subspace and the noise subspace, and uses the noise subspace to estimate the spatial spectrum, thereby realizing the perception of the target space; The interference removal method in step 2 includes an interference removal method based on the atomic norm and an interference removal method based on measurement matrix optimization, wherein the interference removal method based on the atomic norm calculates the atomic norm of the target signal to obtain a denoised vector after removing the interference signal, and the interference removal method based on measurement matrix optimization realizes the removal of interference signals in a specific direction by adjusting the measurement matrix; the interference removal method based on the atomic norm specifically includes the following steps: constructing a sparse reconstruction optimization model including the L2 norm and the atomic norm; for the sparse reconstruction optimization model, constructing a convex optimization problem of semi-positive definite programming based on the Toeplitz matrix of the model; solving the above convex optimization problem by an iterative method to obtain a denoised vector; the sparse reconstruction optimization model including the L2 norm and the atomic norm is essentially to minimize the sum of the square of the L2 norm and the atomic norm, wherein the square of the L2 norm is the square of the L2 norm of the difference between the received signal minus the target signal and then minus the interference signal, and the atomic norm is the atomic norm of the target signal; A convex optimization problem of semi-positive definite programming is constructed based on the Toeplitz matrix. This process is a method for solving the atomic norm of the target signal. In this process, the atomic norm of the target signal is expressed as the sum of the trace of the Toeplitz matrix composed of a vector u to be solved and a variable v to be solved, wherein the constraints that the vector u to be solved and the variable v to be solved need to satisfy are that the Toeplitz matrix composed of the vector u to be solved, the matrix jointly composed of the variable v to be solved and the target signal needs to be a semi-positive definite matrix, and the vector u and the variable v that satisfy the above constraints are iteratively solved by semi-positive definite programming; In the step 3, after performing singular value decomposition on the reconstructed matrix using the MUSIC method, several singular values ​​are obtained, wherein the values ​​of some singular values ​​are several orders of magnitude larger than the remaining singular values, and the singular vectors corresponding to the singular values ​​constitute the signal subspace, the number of the singular values ​​of the part depends on the number of targets, and the singular vectors corresponding to the other singular values ​​that are several orders of magnitude smaller than the partial results in value constitute the noise subspace; In step 3, the spatial spectrum is composed of the ratio of the L2 norm square of the steering vector of the smart reflection surface to the L2 norm square of the aforementioned noise subspace, wherein the steering vector is used to describe the spatial phase difference of the smart reflection surface, and the spatial phase difference indicates that there is a certain difference in phase between signals received by array elements at different positions in space. The smart reflection surface is composed of multiple array elements, and the structure of the steering vector is determined by the relative positions of the array elements constituting the smart reflection surface. The specific value of the steering vector is determined by the angle between the target and the smart reflection surface.

2. The sensing method of the passive sensing device based on the intelligent reflective surface according to claim 1 is characterized in that: The received signal model in step 1 includes a target signal from the target and reflected by the smart reflective surface, an interference signal from a wireless communication access point directly propagated through the smart reflective surface, and environmental noise.

3. The sensing method of the passive sensing device based on the intelligent reflective surface according to claim 1 is characterized in that: The interference removal method based on measurement matrix optimization suppresses interference signals in a specific direction, and includes the following steps: modeling the measurement matrix optimization problem as an optimization problem of semi-positive definite programming; solving the aforementioned semi-positive definite programming problem to obtain a matrix, and performing eigendecomposition on the matrix; using the eigenvector obtained by the eigendecomposition to represent a row of the measurement matrix.

4. The sensing method of the passive sensing device based on the intelligent reflective surface according to claim 1 is characterized in that: The signal reconstruction method based on the Hankel matrix in step 3 only requires a snapshot of the received signal.

5. The sensing method of the passive sensing device based on the intelligent reflective surface according to claim 1 is characterized in that: In the step 3, the signal reconstruction method based on the Hankel matrix takes out multiple groups of ML elements from the denoised M*1-dimensional vector to reconstruct the matrix. The dimension of the reorganized matrix is ​​(ML)*L, where M is the number of array elements of the smart metasurface, and L is the number of rows of the reconstructed signal matrix.

Images