Intelligent reflector-assisted adaptive blind beam forming method

Abstract

The invention belongs to the technical field of wireless communication / perception, and particularly relates to a self-adaptive blind beam forming method assisted by an intelligent reflecting surface. According to the method, modeling is carried out on a desired signal and an interference signal received by a receiver, a signal to interference plus noise ratio is calculated, a signal to interference plus noise ratio problem after maximization of approximation of the receiver is constructed by optimizing a reflection coefficient of an intelligent reflection surface, and then the problem is solved to obtain the reflection coefficient of the intelligent reflection surface. According to the intelligent reflector-assisted wireless sensing architecture and the blind adaptive beam forming method provided by the invention, the problem of quickly improving the signal to interference plus noise ratio when strong interference signals exist in a wireless sensing system and interference channels are completely unknown can be solved. Experiments show that interference elimination is successfully realized with a small amount of limited average received signal power information in an unknown strong interference signal environment, so that the signal to interference plus noise ratio can be effectively improved, and the quality of expected received signals is ensured.

Description

Technical Field

[0001] The present invention belongs to the field of wireless communication / perception technology, and in particular relates to an adaptive blind beamforming method assisted by an intelligent reflecting surface. Background Art

[0002] Smart reflective surfaces, as an emerging technology, show great potential in next-generation wireless communication systems. Their core advantage lies in their ability to cost-effectively reshape the wireless propagation environment, intelligently controlling signal propagation by adjusting the amplitude and phase of reflective elements. Smart reflective surfaces typically consist of a large number of low-cost reflective elements that operate independently, significantly improving the overall performance of wireless communication systems.

[0003] In recent years, the application of smart reflective surfaces in wireless sensing and integrated sensing and communication systems has attracted widespread attention. By carefully designing the reflection coefficient, smart reflective surfaces can significantly enhance the performance of target detection and target estimation in multipath environments, especially non-line-of-sight paths. In addition, passive beamforming techniques based on smart reflective surfaces have also shown effectiveness in clutter suppression. However, the application of these techniques usually relies on prior knowledge of clutter information, which needs to be obtained through training and calibration. Whether the interference sources are artificially set or unintentionally introduced, due to their non-cooperative nature, it is extremely difficult to obtain channel state information of these interference sources, which poses a major challenge to passive beamforming in smart reflective surface-assisted wireless sensing systems. Summary of the Invention

[0004] The present invention aims to provide an adaptive blind beamforming method assisted by smart reflectors. This method, when the interference channel is completely unknown and only limited information about the desired signal direction is known, can rapidly determine the smart reflector's reflection coefficient using only a small amount of received signal power measurements, thereby improving the system's signal-to-interference-and-noise ratio (SINR) and effectively countering both malicious and non-malicious interference. Specifically, the present invention structuredly designs the smart reflector's reflection coefficient and uses it as a measurement vector. Using the corresponding received signal average power information, the method rapidly infers a set of smart reflector reflection coefficients that effectively improve the SINR, thereby ensuring efficient adaptive passive beamforming even in situations with short channel coherence times.

[0005] The technical solution of the present invention is:

[0006] In the presence of unknown strong interference signals, when a directional sensing receiver assisted by a smart reflector receives the desired signal, an adaptive blind beamforming method is used to optimize the reflection coefficient of the smart reflector by maximizing the signal-to-interference-noise ratio. The number of reflection units of the smart reflector in the system is , the number of unknown strong interference signals is , and satisfies The proposed scheme is divided into two stages: measurement and beamforming, including:

[0007] Model the desired signal and interference signal received by the receiver respectively and calculate the signal-to-interference-and-noise ratio, including:

[0008] The desired transmitter signal is expressed as , whose mean is 0 and variance is ; The signal of the expected signal reaching the smart reflector is expressed as ,in represents the complex path gain from the desired transmitter to the smart reflector, is the array response vector of the smart reflective surface, represents the direction of arrival of the desired transmitter signal relative to the smart reflector; the desired signal reflected by the smart reflector and received by the sensor receiver is expressed as ,in represents the line-of-sight channel between the smart reflector and the receiver, and represent the complex path gain and the departure angle relative to the smart reflector, is the reflection coefficient matrix, is the reflection coefficient vector, represents the cascade complex coefficient, is the cascade array response vector relative to the smart reflector related to the desired signal, represents the Hadamard product, is with and Related cascade angles;

[0009] No. The interference signal of an interference source is expressed as , whose mean is 0 and variance is ; Receiver receives The superposition of interference signals is ,in Indicates the The cascade channel from the interference source to the receiver through the smart reflector, Indicates the The channel between the interference source and the smart reflective surface; then the superposition of the desired signal reflected by the smart reflective surface and the interference signal received by the receiver is expressed as

[0010]

[0011] in Represents received noise, with a mean of 0 and a variance of ;

[0012] Assume that all signal sources transmit signals Uncorrelated with each other, the concatenated array response vector and cascade channels Assuming linear independence, the signal-to-interference-plus-noise ratio (SINR) can be calculated as:

[0013]

[0014] in , , the approximately equal sign is because the interference signal is assumed to be much stronger than the receiving noise and is therefore ignored;

[0015] Assume the direction of the desired signal is known, the complex gain and the interference covariance matrix is unknown, by optimizing the reflection coefficient of the smart reflective surface , maximize the approximate signal-to-interference-and-noise ratio of the receiver, the problem is stated as:

[0016]

[0017] There are two ways to solve the problem:

[0018] The first way to solve the problem is: Without considering the unit mode constraint In the case of Usually much smaller than the number of reflective units on a smart reflective surface , It is usually a low-rank matrix, and the optimal solution to the problem must satisfy:

[0019]

[0020] in is a non-negative preset constant, usually representing the gain in the desired direction;

[0021] The average received power received by the receiver Finding the optimal solution, including:

[0022] During the measurement phase, first Generate a set of reflection vectors , and configure them on the smart reflective surface in turn. When the smart reflective surface is configured as the reflection coefficient , calculate the average received power based on the received signal samples:

[0023]

[0024] in is the additive noise used to characterize the estimation error, the matrix Defined as , is a randomly generated vector satisfying ;vector is a non-zero coefficient vector; ;

[0025] Will Written in compact form: ,in and for The vectorized form of The measurement data are stacked to obtain: ,in , , ;

[0026] Random Generation and ensure , then the matrix With probability 1, the full column rank is restored using the least squares method : , after the reorganization and recovery get estimates;

[0027] right Perform truncated eigenvalue decomposition (EVD): ,in Contains the eigenvector, for a diagonal matrix with non-zero eigenvalues;

[0028] Only when When singular, therefore, the matrix bundle The eigenvalues of Estimates: ,in Representing a matrix bundle The eigenvalue extraction operator of

[0029] get After that, calculate : ;

[0030] Based on the obtained matrix , and get the optimal solution ,in for The basis vectors of the null space, corresponding to the reflection coefficient vectors, are expressed as , scaling factor Need to ensure ;

[0031] The second solution is to consider the unit module constraint , the resulting measurement vector and the optimal solution Subject to the following constraints:

[0032] The first constraint By appropriately generating to satisfy;

[0033] The second constraint Translate into feasibility questions, including:

[0034] set up is The null space The matrix consists of basis vectors, then , is the coefficient vector; we get ,definition , then the second constraint The problem statement is:

[0035]

[0036] in represents the infinite norm. This problem is a convex optimization problem and can be solved efficiently using convex optimization algorithms.

[0037] Through the above two methods, we can get the amplitude constraint , the reflection coefficient of the smart reflective surface is designed to be This can achieve adaptive elimination of unknown strong interference signals, and the receiver can receive the desired signal in an almost interference-free environment.

[0038] The beneficial effects of the present invention are as follows: the adaptive blind beamforming method of the intelligent reflective surface assisted perception of the present invention has strong practicality, can achieve interference elimination in an unknown strong interference signal environment with a small amount of limited average received signal power information, ensure the quality of the expected received signal, can effectively improve the signal-to-interference-noise ratio, and ensure reliable signal transmission. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 Schematic diagram of the wireless sensing system assisted by the intelligent reflective surface considered in the present invention.

[0040] Figure 2 Schematic diagram of the two-stage method proposed in the present invention.

[0041] Figure 3 The SINR (dB) values of the present invention and the comparative method vary with the number of samples. Schematic diagram of the change curve.

[0042] Figure 4 SINR (dB) values of the invention and the comparison method vary with the number of interference sources Schematic diagram of the change curve.

[0043] Figure 5 The spatial dimensions are zero and When the number of measured values of the present invention is Schematic diagram of the change curve of SINR. DETAILED DESCRIPTION

[0044] The present invention is described in detail below with reference to the accompanying drawings and simulation examples to demonstrate the practicability of the present invention.

[0045] The present invention considers the problem of adaptive blind beamforming in which a smart reflector-assisted directional sensing receiver uses a small amount of received signal power measurement values to quickly determine the reflection coefficient of the smart reflector. Figure 1 The figure shows a wireless sensing system assisted by a smart reflector, which includes a single-antenna legal transmitter, K interference sources, a directional sensing receiver, and a smart reflector placed near the receiver. The figure marks the channel from the interference source to the smart reflector and the channel from the smart reflector to the sensing receiver. The number of reflection units of the smart reflector in the system is , the number of unknown strong interference signals is , and satisfies The proposed scheme is divided into two stages: measurement and beamforming. First, the desired signal and interference signal received by the receiver are modeled respectively, and the signal-to-interference-noise ratio is calculated.

[0046] The desired transmitter signal is expressed as , whose mean is 0 and variance is ; The signal of the expected signal reaching the smart reflector is expressed as ,in represents the complex path gain from the desired transmitter to the smart reflector, is the array response vector of the smart reflective surface, represents the direction of arrival of the desired transmitter signal relative to the smart reflector; the desired signal reflected by the smart reflector and received by the sensor receiver is expressed as ,in represents the line-of-sight channel between the smart reflector and the receiver, and represent the complex path gain and the departure angle relative to the smart reflector, is the reflection coefficient matrix, is the reflection coefficient vector, represents the cascade complex coefficient, is the cascade array response vector relative to the smart reflector related to the desired signal, represents the Hadamard product, is with and Related cascade angles;

[0047] No. The interference signal of an interference source is expressed as , whose mean is 0 and variance is ; Receiver receives The superposition of interference signals is ,in Indicates the The cascade channel from the interference source to the receiver through the smart reflector, Indicates the The channel between the interference source and the smart reflective surface; then the superposition of the desired signal reflected by the smart reflective surface and the interference signal received by the receiver is expressed as

[0048]

[0049] in Represents received noise, with a mean of 0 and a variance of ;

[0050] Assume that all signal sources transmit signals Uncorrelated with each other, the concatenated array response vector and cascade channels Assuming linear independence, the signal-to-interference-plus-noise ratio (SINR) can be calculated as:

[0051]

[0052] in , , the approximately equal sign is because the interference signal is assumed to be much stronger than the receiving noise and is therefore ignored;

[0053] Assume the direction of the desired signal is known, the complex gain and the interference covariance matrix is unknown, by optimizing the reflection coefficient of the smart reflective surface , maximize the approximate signal-to-interference-and-noise ratio of the receiver, the problem is stated as:

[0054]

[0055] Without considering the unit module constraint In the case of Usually much smaller than the number of reflective units on a smart reflective surface , It is usually a low-rank matrix, and the optimal solution to the problem must satisfy:

[0056]

[0057] in is a non-negative preset constant, usually representing the gain in the desired direction;

[0058] The average received power received by the receiver Finding the optimal solution, including:

[0059] During the measurement phase, first Generate a set of reflection vectors , and configure them on the smart reflective surface in turn. When the smart reflective surface is configured as the reflection coefficient , calculate the average received power based on the received signal samples:

[0060]

[0061] in is the additive noise used to characterize the estimation error, the matrix Defined as , is a randomly generated vector satisfying ;vector is a non-zero coefficient vector; ;

[0062] Will Written in compact form: ,in and for The vectorized form of The measurement data are stacked to obtain: ,in , , ;

[0063] Random Generation and ensure , then the matrix With probability 1, the full column rank is restored using the least squares method : , after the reorganization and recovery get estimates;

[0064] right Perform truncated eigenvalue decomposition (EVD): ,in Contains the eigenvector, for a diagonal matrix with non-zero eigenvalues;

[0065] Only when When singular, therefore, the matrix bundle The eigenvalues of Estimates: ,in Representing a matrix bundle The eigenvalue extraction operator of

[0066] get After that, calculate : ;

[0067] Based on the obtained matrix , and get the optimal solution ,in for The basis vectors of the null space, corresponding to the reflection coefficient vectors, are expressed as , scaling factor Need to ensure ;

[0068] Considering the unit module constraint , the resulting measurement vector and the optimal solution Subject to the following constraints:

[0069]

[0070] The first constraint By appropriately generating to satisfy;

[0071] The second constraint Translate into feasibility questions, including:

[0072] set up is The null space The matrix consists of basis vectors, then , is the coefficient vector; we get ,definition , then the second constraint The problem statement is:

[0073]

[0074] in represents the infinite norm. This problem is a convex optimization problem and can be solved efficiently using convex optimization algorithms.

[0075] The reflection coefficient of the smart reflective surface is designed to be This can achieve adaptive elimination of unknown strong interference signals, and the receiver can receive the desired signal in an almost interference-free environment.

[0076] In the simulation, the expected signal variance is considered , the number of reflection units of the smart reflection surface is , , the noise power is ; Unless otherwise specified, the number of interference signals is In order to demonstrate the effectiveness of the present invention, the present invention is compared with three existing methods. The first existing comparison method is denoted as Non-ABF, that is, the reflection coefficient vector is set to ; The second existing comparative method is denoted as CS, which uses the gradient descent method to estimate , then determine according to the remaining steps of the method of the present invention , need to know in advance The Frobenius norm of is used to ensure convergence; the third existing comparative method is denoted as LS, which directly applies the least squares method to recover , finally confirmed , requiring at least times measurement.

[0077] Figure 1 This figure describes a wireless sensing system assisted by a smart reflector, as considered in this invention. The system includes a single-antenna legitimate transmitter, K interference sources, a directional sensing receiver, and a smart reflector placed near the receiver. The figure shows the channels from the interference source to the smart reflector and from the smart reflector to the sensing receiver.

[0078] Figure 2 The schematic diagram of the two-stage method proposed in the present invention is described. The first stage is the measurement stage, in which the matrix , generating the measurement vector , and finally collect the average power value ; The second stage is the optimization stage, which restores the matrix based on the collected measurements , optimized to get Finally, we get the reflection coefficient vector .

[0079] Figure 3 Describes the SINR (dB) value of the present invention and the comparative method with the number of samples The change curve of the interference source is set to dB. The null space dimension of the present invention , measurement number , covariance sketching (denoted as CS) and least squares (denoted as LS) methods respectively use and This demonstrates that the present invention not only uses a small number of measurements, but also increases with the number of samples. The growth of ,significantly improves the SINR (dB) value.

[0080] Figure 4 Describes the SINR (dB) value of the invention and the comparative method as the number of interference sources The power of the interference source is 20dB, and the number of samples is fixed at The zero space dimension of the present invention , measurement number , CS and LS methods used and It can be seen that the present invention not only uses fewer measurement values, but also its SINR value is always better than the comparison method as the number of interference sources increases.

[0081] Figure 5 The null space dimensions are described as and When the number of measured values of the present invention is The interference power is set to 20dB, and the number of measurement values for the CS and LS methods is fixed to and We see that when the measurement number When the present invention adopts the zero space dimension The SINR is higher than However, for the measured data , zero spatial dimension SINR ratio This is because the smaller null space dimension Allows for more efficient recovery of matrices , thus reducing the estimation error. However, when the number of measurements is large enough, the larger the null space dimension This provides more degrees of freedom to improve numerical stability and ultimately enhance overall performance.

Claims

1. The adaptive blind beamforming method assisted by the smart reflector is used to optimize the reflection coefficient of the smart reflector by maximizing the signal-to-interference-noise ratio when a directional sensing receiver assisted by the smart reflector receives the desired signal in the presence of an unknown strong interference signal. The number of reflective units of the smart reflector is defined as , the number of unknown strong interference signals is , and satisfies ; It is characterized in that, include: Model the desired signal and interference signal received by the receiver respectively and calculate the signal-to-interference-and-noise ratio, including: The desired transmitter signal is expressed as , whose mean is 0 and variance is ; The signal of the expected signal reaching the smart reflector is expressed as ,in represents the complex path gain from the desired transmitter to the smart reflector, is the array response vector of the smart reflective surface, represents the direction of arrival of the desired transmitter signal relative to the smart reflector; the desired signal reflected by the smart reflector and received by the sensor receiver is expressed as ,in represents the line-of-sight channel between the smart reflector and the receiver, and represent the complex path gain and the departure angle relative to the smart reflector, is the reflection coefficient matrix, is the reflection coefficient vector, represents the cascade complex coefficient, is the cascade array response vector relative to the smart reflector related to the desired signal, represents the Hadamard product, is with and Related cascade angles; The first The interference signal of an interference source is expressed as , whose mean is 0 and variance is ; Receiver receives The superposition of interference signals is ,in Indicates the The cascade channel from the interference source to the receiver through the smart reflector, Indicates the The channel between the interference source and the smart reflective surface; The superposition of the desired signal reflected by the smart reflector and the interference signal received by the receiver is expressed as: ， in Represents received noise, with a mean of 0 and a variance of ; Set the signal transmitted by all signal sources Uncorrelated with each other, the concatenated array response vector and cascade channels Linearly independent, the signal-to-interference-and-noise ratio (SINR) is calculated as: ， in , , the approximately equal sign is because the interference signal is assumed to be much stronger than the receiving noise and is therefore ignored; Desired signal direction is known, the complex gain and the interference covariance matrix is unknown, by optimizing the reflection coefficient of the smart reflective surface , maximize the approximate signal-to-interference-and-noise ratio of the receiver, the problem is stated as: ； There are two ways to solve the problem: The first solution to the problem is to ignore the unit module constraint. In the case of The number of reflective units is much smaller than that of smart reflective surfaces , Is a low-rank matrix, the optimal solution to the problem must satisfy: ， in is a non-negative preset constant representing the gain in the desired direction; The average received power received by the receiver Finding the optimal solution, including: During the measurement phase, first Generate a set of reflection vectors , and configure them on the smart reflective surface in turn. When the smart reflective surface is configured as the reflection coefficient When , the average received power is calculated based on the received signal samples: ， in is the number of sample points required to calculate an average power, is the additive noise used to characterize the estimation error, the matrix Defined as , is a randomly generated vector satisfying ;vector is a non-zero coefficient vector; ; Will Written in compact form: ,in and for The vectorized form of The measurement data are stacked to obtain: ,in , , ; Random Generation and ensure , then the matrix With probability 1, the full column rank is restored using the least squares method : , after the reorganization and recovery get estimates; right Perform truncated eigenvalue decomposition (EVD): ,in Contains the eigenvector, for a diagonal matrix with non-zero eigenvalues; Only when When singular, therefore, the matrix bundle The eigenvalues of Estimates: ,in Representing a matrix bundle The eigenvalue extraction operator of get After that, calculate : ; Based on the obtained matrix , and get the optimal solution ,in for The basis vectors of the null space, corresponding to the reflection coefficient vectors, are expressed as , scaling factor Need to ensure ; The second solution is to consider the unit module constraint , the resulting measurement vector and the optimal solution Subject to the following constraints: ， The first constraint By appropriately generating to satisfy; The second constraint Translate into feasibility questions, including: set up is The null space The matrix consists of basis vectors, then , is the coefficient vector; we get ,definition , then the second constraint The problem statement is: ， in represents the infinite norm. This problem is a convex optimization problem and can be solved efficiently using a convex optimization algorithm. By solving the problem Then, the reflection coefficient of the smart reflective surface is designed as This can achieve adaptive elimination of unknown strong interference signals, and the receiver can receive the desired signal in an almost interference-free environment.

Images