Passive sensing device based on intelligent reflective surface and sensing method thereof

Abstract

The present invention discloses a passive sensing device based on an intelligent reflective surface and a sensing method thereof. The device includes a base station or a wireless access point, an intelligent reflective surface and a sensor, and each device is connected by a wireless communication signal. The method includes the following steps: the electromagnetic wave emitted by the base station or the wireless 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 relevant parameters of the intelligent reflective surface are adjusted so that the sensor receives the secondary reflected signal, and the sensor processes the signal to realize the perception of the space target; due to the existence of interference signals that directly reach the intelligent reflective surface from the base station or the wireless access point, a method based on the alternating direction multiplier method can be used to suppress the interference signal while reconstructing the effective signal; finally, the perception of the space target is realized based on the MUSIC algorithm. While having good estimation performance, the present invention greatly reduces the computational complexity, which is conducive to engineering implementation.

Description

Technical Field

[0001] The present invention relates to the field of integrated sensing and communication (ISAC) systems, and more specifically, to a passive sensing device based on an intelligent reflective surface and a sensing method thereof. Background Art

[0002] In recent years, with the development of B5G and 6G wireless communications, integrated sensing and communication systems have become a hot topic. In the active mode, the ISAC system needs to add additional signals for the transmission of the sensing function; in the passive mode, the system can only use the communication signal. In addition, passive sensing systems usually have the advantage of convenient implementation and can be equipped without changing the original communication system. In order to perceive the target, the location information can be obtained by the delay or direction of the received signal. For delay estimation, a synchronization reference signal from the base station BS (Base Station) or the wireless access point AP (Access Point) must be used, so the synchronization link from the AP to the sensor must be used in addition; for azimuth estimation, the direction of arrival DOA (Direction of Arrival) can be obtained without a synchronization signal, so it is more practical in system implementation. However, DOA estimation usually has multiple receiving channels, which is costly and complicated. In recent years, RIS has been proposed to improve communication and radar performance, and DOA estimation with only one full-function receiving channel can also be achieved by changing the amplitude or phase of the reflected signal.

[0003] In passive sensing systems, target positioning based on direction finding is achieved by receiving signals through sensors, which is relatively easy to implement and can be well applied in practical systems. For DOA estimation methods, traditional methods are based on Fast Fourier Transform (FFT). However, due to the limited resolution of FFT, some subspace-based super-resolution methods have also been proposed, such as the Multiple Signal Classification (MUSIC) algorithm and the Estimating Signal Parameters Via Rotational Invariance Techniques (ESPRIT). With the development of compressed sensing technology, the sparse reconstruction (CS) theory has also been proposed for DOA estimation. In most CS-based methods, the spatial domain is discretized into a grid, but the target cannot be accurately located on the grid, so the off-grid error occurs. In order to overcome this error, an atomic norm minimization (ANM) method was proposed, which is a continuous domain method based on sparse reconstruction.

[0004] In recent years, the DOA estimation problem of RIS-based communication perception systems has received widespread attention from researchers at home and abroad. The current products have high computational complexity, such as the publication number CN110611527A, the publication date 2019-12-24, the invention name is an array antenna, beamforming method and perception and communication integrated system, this perception does not use intelligent reflective surfaces, and cannot achieve DOA estimation with only one full-function receiving channel by changing the amplitude or phase of the reflected signal; the publication number is CN113673317A, the publication date 2021-11-19, the invention name is a two-dimensional off-grid DOA estimation method based on atomic norm minimization and dimensionality reduction, the invention uses a semi-definite programming-based method to solve the atomic norm problem, the computational complexity is high, and it cannot be highly applicable, so passive sensing devices and sensing methods still have great research value. Summary of the invention

[0005] In view of the shortcomings of traditional optimization algorithms in the background technology, the present invention proposes a passive sensing device based on an intelligent reflective surface and a sensing method thereof, which realizes signal reconstruction and target perception through rapid iteration with low computational complexity, and is conducive to engineering implementation.

[0006] The present invention provides a passive sensing device based on an intelligent reflective surface, which mainly includes a base station or a wireless access point, an intelligent reflective surface and a sensor, and each device is connected by a wireless communication signal. The base station or the wireless access point is used to transmit a synchronization reference signal; the intelligent reflective surface contains a plurality of array elements, and the amplitude and phase of the electromagnetic wave signal reflected by it can be controlled by a controller; the sensor includes an antenna, a radio frequency module, and a signal processing board, etc. The radio frequency module includes a low noise amplifier, a filter, a gain controllable amplifier, an analog-to-digital converter, and an antenna, etc. The signal processing board consists of a single receiving channel and a signal processing module.

[0007] The present invention provides a passive sensing method, which mainly includes the following process: the electromagnetic wave emitted by the base station or wireless 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 reflection surface, and the relevant parameters of the intelligent reflection surface are adjusted so that the sensor receives the secondary reflection signal, and the sensor processes the signal, thereby realizing the perception of the space target. The method mainly includes the following key steps: constructing a signal model; rewriting the target space perception problem into a sparse reconstruction problem, and reconstructing the effective signal while suppressing the interference signal through the ADMM-based method; after obtaining the reconstructed signal, the single-snapshot MUISC method is used to estimate the spatial spectrum, and finally the target space perception is realized by finding the position of the peak of the spatial spectrum.

[0008] The signal model provided by the present invention mainly includes the echo signal from the target and reflected by the intelligent reflection surface, the interference signal directly transmitted by the wireless access point through the intelligent reflection surface, and the noise in the environment where the device is located.

[0009] The signal reconstruction method provided by the present invention is mainly implemented through the following steps: first, the target perception problem is constructed as a sparse reconstruction optimization problem including L2 norm and atomic norm, then a convex optimization problem is constructed based on the Toeplitz matrix, and then the corresponding parameters are introduced to rewrite the optimization problem as an augmented Lagrangian function.

[0010] According to the signal reconstruction method based on the ADMM method provided by the present invention, a step-by-step iterative calculation is performed on the augmented Lagrangian function, and the specific steps of the method are as follows: input the received signal, the maximum number of iterations and the number of measurements, and initialize the relevant estimation parameters; iteratively update the relevant estimation parameters; repeat the iterative update steps, alternately optimize and iteratively solve multiple variables, finally obtain the global optimal solution, and output the reconstructed signal x.

[0011] After obtaining the reconstructed signal x according to the perception method provided by the present invention, a single-snapshot multi-signal classification MUISC method is used to perceive the spatial target. First, the covariance matrix R is obtained from the reconstructed signal x, and then the matrix R is eigenvalue decomposed and the eigenvalues ​​are sorted to determine the signal subspace and the noise subspace. Then, the corresponding wave arrival direction is estimated by finding the position of the spatial spectrum peak.

[0012] The method proposed by the present invention has the following advantages compared with the prior art:

[0013] 1) The present invention realizes target perception in a passive manner.

[0014] 2) The present invention has a lower computational complexity. The present invention is compared with the method based on semi-positive programming and analyzed as follows:

[0015] In the method of the present invention, the computational complexity is determined by the step of eigenvalue decomposition, so its computational complexity is O((M+1) 3 ), which is much lower than the method based on semi-positive programming O((M+1) 4 ), it can be seen that the present invention can effectively estimate the direction of arrival and greatly reduce the computational complexity, making it easier to implement in engineering. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is the MUSIC algorithm flow chart;

[0017] Figure 2 It is a scene diagram for realizing the present invention;

[0018] Figure 3 In the figure, Figure (a) is a general implementation flow chart of the present invention, and Figure (b) is an implementation flow chart of a signal processing board;

[0019] Figure 4 It is a schematic diagram of the estimated performance of different iteration numbers of the present invention;

[0020] Figure 5 Schematic diagram of the estimated performance of the present invention compared with other methods. DETAILED DESCRIPTION

[0021] The present invention is described in detail below with reference to the accompanying drawings and examples:

[0022] The present invention provides a passive sensing device based on an intelligent reflecting surface and a sensing method thereof. The passive sensing device mainly comprises a base station or a wireless access point, an intelligent reflecting surface and a sensor, wherein the base station or the wireless access point is used to transmit a synchronization reference signal; the intelligent reflecting surface contains a plurality of array elements, and the amplitude and phase of the electromagnetic wave signal reflected by the intelligent reflecting surface can be controlled by a controller; the sensor comprises an antenna, a radio frequency module and a signal processing board, etc., and the signal processing board consists of a single receiving channel and a signal processing module.

[0023] The passive sensing system based on the intelligent reflective surface provided by the present invention mainly comprises the following steps:

[0024] Step 1: A base station or a wireless access point transmits an electromagnetic wave signal to a space containing multiple targets;

[0025] Step 2: Build a signal model. The electromagnetic waves emitted by the base station or wireless access point are reflected by the target after propagating through free space. The reflected signal is reflected twice when passing through the intelligent reflective surface. The relevant parameters of the intelligent reflective surface are adjusted so that the sensor receives the secondary reflected signal. In the case of N time slots and K targets in the perception system, the received signal in the sensor is:

[0026] r=GA(θ)s+Ga(ψ)q+w

[0027] In the formula, is a plural set, is the reflected signal, is an interference signal. The variance is Additive Gaussian white noise, is the measurement matrix, N is the number of measurements, and M represents the number of elements in the RIS. The sensor receives the interference signal from the direct path between the wireless access point and the smart reflector. The direction of arrival of the kth target (k = 0, 1, ..., K-1) and the smart reflector is estimated as θ k , the angle between the wireless access point and the smart reflective surface is ψ. is the steering matrix, where the steering vector is defined as is an imaginary number, and d is the distance between adjacent array elements normalized by the wavelength.

[0028] Step 3: Rewrite the target space perception problem into a sparse reconstruction problem, and reconstruct the effective signal while suppressing the interference signal through the ADMM-based method. The signal reconstruction method mainly includes the following steps:

[0029] 1) Construct a sparse reconstruction optimization problem involving L2 norm and atomic norm;

[0030] 2) Construct convex optimization problem based on Toeplitz matrix;

[0031] 3) Introduce corresponding parameters and rewrite the optimization problem into an augmented Lagrangian function.

[0032] Step 4: Perform step-by-step iterative calculation on the augmented Lagrangian function in step 3 based on the ADMM algorithm. The iterative calculation mainly includes the following steps:

[0033] a) Initialize relevant parameters: Input received signal r, maximum number of iterations and set number of measurements M.

[0034] b) Iteratively update relevant parameters.

[0035] c) Repeat step b), optimize multiple variables alternately and solve them iteratively, and finally obtain the global optimal solution and output the effective reconstructed signal x.

[0036] Step 5: After obtaining the effective reconstructed signal x, the single-snapshot MUISC method can be used to perceive the target. Figure 1 The main steps of the MUSIC algorithm are shown in Figure 1: First, the covariance matrix R is obtained from the reconstructed signal x, and then the matrix R is decomposed by eigenvalues, the eigenvalues ​​are sorted, the number of signal sources is determined, and the signal subspace and noise subspace are determined. Then the corresponding wave arrival direction is estimated by finding the position of the spatial spectrum peak.

[0037] All steps and conclusions of the present invention are verified and given based on the scientific computing software Matlab. In the simulation, the number of RIS array elements M = 64, the number of measurements N = 32, the number of targets K = 3, the normalized distance d = 0.5; the spatial angle range is [-90°, 90°].

[0038] To illustrate the effectiveness of this method, Figure 2 is the implementation scene diagram of the present invention, Figure 3 In the figure, (a) is a general implementation flow chart of the present invention, (b) is an implementation flow chart of a signal processing board, Figure 4 A schematic diagram of the estimated performance of the method with different numbers of iterations is given. Figure 5 A schematic diagram of the estimated performance of this method compared with other methods is given.

[0039] according to Figure 4 , the estimation performance is measured by the root mean square error RMSE; according to Figure 5, and compared four methods, namely, the FFT method, the L1 norm method, the OMP method, and the SDP method. It can be seen from the comparison with other methods that the present invention can obtain better performance at a higher signal-to-noise ratio and avoid off-grid errors. Therefore, by improving the signal-to-noise ratio, the estimation performance of the proposed method can be further improved. However, due to the limited number of iterations, there is a performance gap between the present invention and the SDP method, but because the computational complexity is greatly reduced, the computational time of the present method is less than 1 / 10 of that of the SDP method, so it is a method suitable for engineering applications.

Claims

1. A passive sensing device based on an intelligent reflective surface, characterized in that: The passive sensing device includes a base station or a wireless access point, an intelligent reflective surface and a sensor, and the devices are connected by wireless communication signals; the base station or the wireless access point is used to transmit a synchronization reference signal; the intelligent reflective surface contains a plurality of array elements, and the amplitude and phase of the electromagnetic wave signal reflected by the intelligent reflective surface are controlled by a controller; the sensor processes the reflected signal of the intelligent reflective surface, thereby realizing the perception of the space target; The sensing method based on the passive sensing device comprises the following steps: 1) Construct a signal model; 2) The target space perception problem is rewritten as a sparse reconstruction problem, and the ADMM algorithm is used to suppress interference signals while reconstructing valid signals; 3) After obtaining the reconstructed signal, estimate the spatial spectrum to achieve target perception.

2. A passive sensing device based on a smart reflective surface according to claim 1, characterized in that: The sensor includes an antenna, a radio frequency module and a signal processing board; the radio frequency module includes a low noise amplifier, a filter, a gain controllable amplifier, an analog-to-digital converter and an antenna; the signal processing board includes a single receiving channel and a signal processing module.

3. The passive sensing device based on the intelligent reflective surface according to claim 1, characterized in that: The signal model in step 1) includes: an echo signal from the target and reflected by the smart reflective surface, an interference signal directly transmitted by the wireless access point through the smart reflective surface, and noise in the environment where the device is located.

4. The passive sensing device based on the intelligent reflective surface according to claim 1, characterized in that: In step 2), a sparse reconstruction optimization problem including L2 norm and atomic norm is constructed for the target perception problem, and then a convex optimization problem is constructed based on the Toeplitz matrix. Then, the corresponding parameters are introduced to rewrite the optimization problem into an augmented Lagrangian function; then, a step-by-step iterative calculation is performed on the augmented Lagrangian function based on the ADMM algorithm.

5. The passive sensing device based on the intelligent reflective surface according to claim 4, characterized in that: The specific steps for performing step-by-step iterative calculation of the augmented Lagrangian function based on the ADMM algorithm are as follows: a) Inputting the received signal, the maximum number of iterations and the number of measurements, and initializing the relevant estimation parameters; b) iteratively updating relevant estimated parameters; c) Repeat step b), optimize multiple variables alternately and solve them iteratively, and finally obtain the global optimal solution and output the reconstructed signal x.

6. The passive sensing device based on the intelligent reflective surface according to claim 1, characterized in that: Step 3) After obtaining the reconstructed signal x, the single snapshot multiple signal classification (MUISC) method is used to estimate the spatial spectrum.

7. The passive sensing device based on the intelligent reflective surface according to claim 6, characterized in that: The steps of estimating the spatial spectrum include: firstly, obtaining the covariance matrix R from the reconstructed signal x, then performing eigenvalue decomposition on the matrix R, sorting the eigenvalues ​​to determine the signal subspace and the noise subspace, and finally estimating the corresponding direction of arrival by finding the position of the spatial spectrum peak.

Images