Multiple intelligent reflectors aided MIMO channel estimation method based on spatial correlation

Through the multi-intelligent reflective surface assisted MIMO channel estimation method based on spatial correlation, the channel stacking problem in the multi-intelligent reflective surface system is solved, more accurate channel estimation and signal coverage are achieved, and system performance is improved.

CN118869406BActive Publication Date: 2024-12-03INTELLIGENT LOCK RES INST OF WENZHOU UNIV +2
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411328395.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-24
Publication Date
2024-12-03
Estimated Expiration
2044-09-24

AI Technical Summary

Technical Problem

In multi-intelligent reflective surface systems, channel estimation faces channel stacking problems, which makes it difficult to achieve channel estimation accuracy, especially when multiple intelligent reflective surfaces are deployed, the challenge caused by the signal reaching the base station at the same time.

Method used

Using the multi-intelligent reflective surface-assisted MIMO channel estimation method based on spatial correlation, a multi-user communication system assisted by multi-intelligent reflective surfaces is constructed to establish a spatially correlated channel model, design a channel estimation protocol containing two stages, and use the maximum posterior probability. The method constructs a cascading channel optimization objective function, and decomposes it into subproblems between common link variables and user-specific variables through alternating optimization methods, and jointly optimizes the overall phase shift matrix of multi-intelligent reflective surfaces to separate the channel matrix of each intelligent reflective surface.

Benefits of technology

Improve the accuracy and performance of channel estimation. Through the maximum posterior probability method and alternating optimization method, the channel state information of multiple intelligent reflection surfaces can be more accurately estimated, and the corresponding channels of each intelligent reflection surface are separated, improving signal coverage and system performance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118869406B_ABST
    Figure CN118869406B_ABST
Patent Text Reader

Abstract

A multi-intelligent reflector-assisted MIMO channel estimation method based on spatial correlation belongs to the technical field of channel estimation in wireless communication. It constructs a multi-user communication system and establishes a spatial correlation channel model; designs a two-stage channel estimation protocol, and constructs a pilot structure and a receiving signal model in each training time slot; constructs a new set of cascade channel auxiliary variables and establishes a cascade channel optimization objective function; optimizes common link variables and user-specific variables respectively, and jointly optimizes the overall phase shift matrix of the multi-intelligent reflector to estimate the multi-intelligent reflector channel; uses the principle of spatial correlation to compare the strength of the correlation between channel columns, separates the channel matrix, and obtains the reflection channel corresponding to each intelligent reflector. Based on the cascade channel estimation and channel separation of multiple intelligent reflectors, the corresponding channel of each intelligent reflector is estimated, which effectively reduces the computational complexity and pilot overhead and improves the channel estimation accuracy.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of channel estimation in wireless communications, and in particular to a multi-intelligent reflection surface assisted MIMO channel estimation method based on spatial correlation. Background Art

[0002] As an emerging wireless communication technology, the Intelligent Reflecting Surface (IRS) technology has received widespread attention in recent years. IRS consists of a large number of controllable reflection units, each of which can adjust the phase and amplitude of the incident signal to achieve intelligent reflection and beamforming of the signal. By deploying intelligent reflection surfaces in the signal transmission path, the propagation path of the wireless signal can be dynamically adjusted to improve signal coverage, system capacity and spectrum efficiency. Compared with traditional relay stations, intelligent reflection surfaces have the advantages of low cost, low energy consumption and easy deployment, and are considered to be an important part of future 5G and 6G communication systems.

[0003] In modern wireless communication systems, multiple-input multiple-output (MIMO) technology is widely used. By configuring multiple antennas at the transmitting and receiving ends, multiple data streams can be transmitted and received simultaneously, significantly improving the system's spectrum efficiency and data transmission rate. However, the performance of MIMO systems is heavily dependent on accurate channel state information (CSI). However, compared with traditional active devices, smart reflectors do not have sensor elements or RF chains, so they do not have the baseband processing capabilities to perform active transmission and reception, resulting in channel estimation being a challenging task for systems assisted by smart reflectors. At present, most studies on channel estimation assisted by smart reflectors focus on single smart reflector systems. For multi-intelligent reflector systems, multiple smart reflectors are deployed in the communication environment. By coordinating the reflection characteristics of these reflectors, the signal propagation path can be more flexibly adjusted, thereby enhancing signal coverage and improving system performance. However, the deployment of multiple smart reflectors will cause the signals reflected by different reflectors to arrive at the base station almost at the same time. When the base station estimates all reflected channels, channel stacking occurs, which brings challenges to accurate channel estimation for multi-intelligent reflector systems. Summary of the invention

[0004] In view of the shortcomings of the prior art, the object of the present invention is to provide a multi-intelligent reflection surface assisted MIMO channel estimation method based on spatial correlation.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A multi-intelligent reflection surface assisted MIMO channel estimation method based on spatial correlation comprises the following steps:

[0007] S1 constructs a multi-user communication system assisted by multiple intelligent reflectors, and establishes a spatial correlation channel model based on the spatial correlation matrix of the reflective elements of each intelligent reflector;

[0008] S2 designs a two-stage channel estimation protocol and constructs the user sending pilot structure and the base station receiving signal model in each training time slot;

[0009] S3 constructs a new set of auxiliary variables of the cascade channel according to the observation value signals received in the two stages, and establishes the cascade channel optimization objective function using the maximum a posteriori probability method;

[0010] S4 decomposes the cascade channel optimization objective function into two sub-problems: optimizing common link variables and user-specific variables. It obtains the local optimal solution of the sub-problems through an alternating optimization method, and jointly optimizes the overall phase shift matrix of the multi-intelligent reflector to obtain the multi-intelligent reflector stacking channel.

[0011] S5 uses the principle of spatial correlation between reflective elements of the same intelligent reflective surface to compare the strength of the correlation between the columns of channels of multiple intelligent reflective surfaces, separates the channel matrix of multiple intelligent reflective surfaces, and obtains the corresponding channels of each intelligent reflective surface.

[0012] Furthermore, in S1,

[0013] S11 is based on the known azimuth angle of incidence on each smart reflective surface. α With elevation β , and get its array response signal , and the correlation coefficient matrix between the reflective elements of each intelligent reflective surface is obtained from it P ,

[0014] ,

[0015] in, u a and u b Indicates the first a and b The position vector of each reflective element relative to the central reference reflective element, l Refers to the l Block intelligent reflective surface;

[0016] S12 estimates all intelligent reflective surfaces as a whole, that is, the Kth user-intelligent reflective surface and the channel of the mth element of the smart reflector - the base station Represented as a cascade channel ,

[0017] ,

[0018] Where diag[·] means constructing a diagonal matrix.

[0019] Furthermore, in S2,

[0020] S21 obtains several independent measurements with different phase shifts θ during training, and the number of independent measurements is greater than the channel unknown value the number of

[0021] S22 makes full use of the spatial correlation between the reflective elements of the smart reflective surface to From the superimposed channel matrix Separate from;

[0022] S23 keeps the reflective elements of all smart reflective surfaces in the open state, and K users send different pilot signals in two stages. The total training time slot is ,

[0023] In the first phase of training time slot ,in Indicates rounding up, K users send K pilots at the same time to form a pilot sequence, denoted as S ;

[0024] The second phase includes training slots , K users send 1 pilot each time, and the pilot sequence is the first stage pilot S The first column of ;

[0025] The S24 base station receiving signal model is trained in the first stage. , where G contains all smart reflector-base station channels, H r represents all channels of the user-intelligent reflector surface, is the additive white Gaussian noise at the base station at time t, Expressed as noise power, (·) T is the transposed matrix and diag[·] is constructed The diagonal matrix of ;

[0026] The base station receiving signal model in the second stage of training is ;

[0027] The signals received in the two stages are combined to obtain a base station receiving signal model that includes all time slots.

[0028] .

[0029] Furthermore, in S3,

[0030] S31 constructs a channel containing a common link R g and user-specific channel R u A set of auxiliary variables,

[0031]

[0032] in, and Represent the user-specific channel R u The kth column and the common link channel R g The mth column of

[0033] S32 uses the auxiliary variable set obtained in S31 to record the cascade channel estimation problem as ,

[0034] ,

[0035] The corresponding optimization objective function

[0036]

[0037] where ||·|| F , ||·||2,(·) T , (·) H and(·) -1 Respectively expressed as Frobenius norm, two-norm, transpose, conjugate transpose, and inverse matrix, where , and the first stage noise power and the second stage noise power .

[0038] Furthermore, in S4, It is divided into two independent convex sub-problems, which are solved by an effective alternating optimization method as follows:

[0039] Before the S41 iteration begins, the received signal including the two-stage training time slot is used to adjust the common link variable R g Initialize, and then fix R g variables, for user-specific variables R u Solve the optimization objective function;

[0040] S42 Fixed R u Variables, R g Optimize the solution of the objective function by continuously iterating and optimizing the objective function until the objective functions of the two variables converge;

[0041] S43 After a finite number of iterations, the channel estimation of the entire cascade channel obtains a local optimal solution.

[0042] Furthermore, in S5,

[0043] S51 obtains a channel matrix including channel state information of all smart reflection surfaces, and separates a channel matrix corresponding to each smart reflection surface from the obtained channel matrix by using a method of comparing the strength of correlation;

[0044] After channel estimation, S52 solves the correlation matrix between two adjacent columns of the estimated channel matrix, establishes an index factor with the minimum channel correlation, finds the column with the weakest correlation as a dividing line, and divides the channel matrices of different intelligent reflection surfaces.

[0045] The beneficial effects of the present invention are as follows: by utilizing the correlation between smart reflective surface elements to establish a spatial correlation channel model, it is considered that multiple reflective surfaces are regarded as a whole block and the maximum a posteriori probability method is used for channel estimation and optimization, and the overall phase shift optimization of multiple smart reflective surfaces is comprehensively considered to further improve the channel estimation accuracy; after obtaining the precise channel containing multiple smart reflective surfaces, the different correlations between the reflective surfaces are used to find the channel matrix dividing line to separate the channels corresponding to each smart reflective surface. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is an overall flow chart of the multi-intelligent reflector-assisted multi-user communication system based on spatially correlated channels of the present invention.

[0047] Figure 2 This is a model diagram of the multi-user communication system assisted by multiple intelligent reflective surfaces of the present invention.

[0048] Figure 3 It is a performance comparison diagram of normalized mean square error of different algorithms at different transmission powers in a specific application example of the present invention. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] It should be noted that all directional indications in the embodiments of the present invention (such as up, down, left, right, front, back, etc.) are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0051] like Figure 1As shown, the present invention provides a multi-intelligent reflection surface assisted MIMO channel estimation method based on spatial correlation, which includes the following steps:

[0052] S1 constructs a multi-user communication system assisted by multiple intelligent reflectors, and establishes a spatial correlation channel model based on the spatial correlation matrix of the reflective elements of each intelligent reflector;

[0053] S2 designs a two-stage channel estimation protocol and constructs the user sending pilot structure and the base station receiving signal model in each training time slot;

[0054] S3 constructs a new set of auxiliary variables of the cascade channel according to the observation value signals received in the two stages, and establishes the cascade channel optimization objective function using the maximum a posteriori probability method;

[0055] S4 decomposes the cascade channel optimization objective function into two sub-problems: optimizing common link variables and user-specific variables. It obtains the local optimal solution of the sub-problems through the alternating optimization method, and jointly optimizes the overall phase shift matrix of the multi-intelligent reflector to obtain the multi-intelligent reflector stacking channel.

[0056] S5 uses the principle of spatial correlation between reflective elements of the same intelligent reflective surface to compare the strength of the correlation between the columns of channels of multiple intelligent reflective surfaces, separates the channel matrix of multiple intelligent reflective surfaces, and obtains the corresponding channels of each intelligent reflective surface.

[0057] Figure 2 An embodiment is provided, which uses two intelligent reflective surfaces. Based on this embodiment, S1 constructs a multi-user communication system assisted by multiple intelligent reflective surfaces. Assume that in a user cell, a base station equipped with M antennas and K single-antenna users are deployed, and there are large buildings or obstacles between the two that block electromagnetic signal transmission, resulting in a weak direct line-of-sight link signal from the base station to the user; parallel deployment l The number of random-sized smart reflective surface reflective elements is N l It is used to assist the information transmission between the base station and the single-antenna user, where both the base station and the smart reflective surface adopt uniform planar arrays.

[0058] By knowing the azimuth angle α and elevation angle β incident on each smart reflector, the array response signal can be obtained. , from which the correlation coefficient matrix between the elements of each intelligent reflective surface can be calculated P .

[0059]

[0060] in, u a and u b Indicates the firsta and b The position vector of the reflective element relative to the central reference reflective element is obtained by multiplying a correlation coefficient matrix on the basis of the original channel. P , establish a spatial correlation channel model. It should be noted that the base station also adopts a multi-antenna model, and there is also a correlation coefficient matrix. The base station-intelligent reflection surface channel also needs to be multiplied by the base station's correlation coefficient matrix. In addition (·) T ,|·|,j, [·]and[·] ab They represent transpose, modulus, imaginary unit, expectation and the matrix's ( a , b ) elements.

[0061] Because there are multiple smart reflectors in the system, the signals received by the base station will be stacked. In order to reduce the amount of calculation and pilot overhead, all smart reflectors are first estimated as a whole. That is, the phase shifts θ of all smart reflectors are put in a matrix for unified design. In this way, the initial channel estimation is equivalent to estimating a smart reflector channel. Then the Kth user-smart reflector can be and the channel of the mth element of the smart reflector - the base station Represented as a cascade channel

[0062]

[0063] Where diag[·] denotes the construction of a diagonal matrix. It is much easier to obtain the prior knowledge of the cascaded channel than to obtain the prior distribution from the user to the smart reflector and from the smart reflector to the base station separately.

[0064] In S2, in order to recover the multi-intelligent reflector cascade channel from the received signal , enough independent measurements with different phase shifts θ should be obtained during training to ensure that the number of independent measurements is greater than the channel unknowns Then, the spatial correlation between the elements of the smart reflective surface is fully utilized to divide the channel of each smart reflective surface into From the superimposed channel matrix Keep all the reflective elements of the smart reflector open, and K users send different pilot signals in two stages. The total training time slot is .

[0065] In the first phase of training time slot ,in Indicates rounding up, K users send K pilots at the same time, and the pilot sequence is recorded as S ;

[0066] The second phase includes training slots , K users send 1 pilot each time, and the pilot sequence is the first stage pilot S The first column of The base station receiving signal model in the first stage of training is , where G contains all smart reflector-base station channels, H r represents all channels of the user-intelligent reflector surface, is the additive white Gaussian noise at the base station at time t, Expressed as noise power, (·) T is the transposed matrix and diag[·] is constructed The diagonal matrix of .

[0067] In the second stage, the signal received by the base station is recorded as Since the pilot signal sent in the second stage is the first column of the pilot signal in the first stage, the signals received in the two stages can be combined to obtain a received signal model containing all time slots:

[0068]

[0069] In S3, as described in step 2, the received signal model is obtained by obtaining the covariance function of the cascaded channel , [·]and[·] H Denote expectation and conjugate transpose respectively. The maximum a posteriori probability method is used to construct the objective optimization function of the cascade channel. However, since the maximum a posteriori probability method requires separate user-intelligent reflector and intelligent reflector-base station covariances, which are generally difficult to obtain from the cascade channel covariance, we first construct a common link channel R g and user-specific channel R u The auxiliary variable set of

[0070]

[0071] in, and Represent the user-specific channel R u The kth column and the common link channel R g With this new set of auxiliary variables, the cascade channel estimation problem can be denoted as P1 and expressed as:

[0072]

[0073] The corresponding optimization objective function is:

[0074]

[0075] In the formula ||·|| F , ||·||2, (·) T , (·) H and(·) -1 They represent the Frobenius norm, the second norm, the transpose, the conjugate transpose and the inverse matrix respectively. , and the first stage noise power and the second stage noise power .

[0076] In S4, the optimization variables in the two likelihood functions of the cascaded channel objective optimization function obtained in the third step above are highly coupled, which makes it extremely difficult to solve P1. In order to solve this problem, P1 is regarded as a typical double convex problem and divided into two independent convex sub-problems, which can be solved by an effective alternating optimization method. The specific process is:

[0077] Before the iteration starts, the common link variable R is first adjusted using the received signal including the two-stage training time slot. g Initialize, and then fix R g variables, for user-specific variables R u Solve the optimization objective function; then fix R u Variables, R g The objective function is optimized by continuously iterating the optimization objective function until the objective functions of the two variables converge; and the function converges very quickly. After a limited number of iterations, the channel estimation of the entire cascade channel obtains a local optimal solution.

[0078] In order to further improve the channel estimation accuracy and performance, the overall phase shift matrix of all smart reflectors was optimized. The discrete Fourier transform matrix was used and a target optimization function with an additional steering vector was introduced. The optimal phase shift design was obtained by solving the target function through a continuous convex approximation algorithm.

[0079] To further improve the channel estimation performance, we use a phase shift configuration based on the discrete Fourier transform (DFT) matrix to simultaneously design the phase shifts of the two IRSs, and then rewrite the phase shift matrix ψ as:

[0080]

[0081] Where F is the discrete Fourier transform matrix expressed as follows (its tth column is expressed as ):

[0082]

[0083] and is the additional steering vector, and the phase shift matrix of the tth time slot is obtained for

[0084]

[0085] Since the total received signal power required for channel estimation is a function, so here we can use the maximum total power of the required received signal to establish the phase variable The objective optimization function is a function with a modulus constraint of 1.

[0086]

[0087] P is the covariance matrix and discrete Fourier transform matrix of the channel variables known at time t The above objective optimization function can be found to be a non-convex optimization problem. Here, a continuous convex approximation algorithm is used to obtain The optimal solution of the phase shift matrix Optimization.

[0088] In S5, according to the above method, a channel matrix including the channel state information (CSI) of all smart reflection surfaces can be obtained. The channel matrix corresponding to each smart reflection surface can be separated from the obtained channel matrix by using a method of comparing the strength of the correlation. When modeling the channel in S1, a spatially correlated channel model is established by multiplying the smart reflection surface correlation coefficient matrix. After channel estimation, this correlation still exists. By solving the correlation matrix between each adjacent column of the estimated channel matrix, an index factor with the minimum channel correlation is established, and the column with the weakest correlation is found as the dividing line to divide the channel matrices of different smart reflection surfaces.

[0089] For example, assuming that the system contains two IRS blocks, first, use the Pearson correlation coefficient to calculate the correlation strength between each column, and obtain the channel matrix R containing two IRS blocks obtained previously k Then, according to the correlation strength of all adjacent columns, we find the two columns with the weakest correlation and get the boundary line that distinguishes the CSI of the two cascaded channels. If the correlation between the nth column and the n+1th column is the weakest, then the entire R k It can be divided into two sub-matrices, corresponding to two different IRS, that is, R1=R k [:, 1:n]; R2=R k [:, n+1:N1+N2]. Among them, R k [:, 1:n]: indicates that the first column to the nth column of the matrix R2=R k[:, n+1:N1+N2]: indicates columns n+1 to N1+N2.

[0090] An indoor scenario in a three-dimensional Cartesian coordinate system is considered, which includes BS, IRS1 and IRS2. The centers of these entities are located at (10, 0, 3), (0, 1.5, 3) and (0, 20, 3), respectively. They jointly serve K = 8 users, randomly distributed in a 5m × 5m cubic area. In addition, we generate the channel coefficients using model parameters based on the 3GPP ray tracing model. The distance-dependent channel path loss for BS-IRS and IRS user links is modeled as 40 + 17.3lg d, where d represents the link distance. The transmit power of all users is assumed to be equal, and the noise power separation density is set to -170 dBm / Hz. The parameters are set to M = 8, N1= 32 and N2= 32.

[0091] The simulation results are shown in Figure 3. The performance of the bilinear least squares algorithm is the worst because it does not utilize the prior knowledge of the cascaded channel. In contrast, the performance of the traditional linear minimum mean square error algorithm remains stable as the user transmit power increases because it is mainly limited by the number of training time slots, which is less than the number of dual IRS elements in the simulation. It is worth noting that the bilinear least squares method optimized based on the maximum a posteriori probability (MAP) method shows similar performance to the linear minimum mean square error estimation algorithm at low signal-to-noise ratio, but as the user transmit power increases, its performance deteriorates and eventually becomes consistent with the unoptimized bilinear least squares algorithm. It is obvious that compared with other channel estimation algorithms, the scheme proposed in the present invention combined with the optimized channel estimation algorithm can achieve higher gain.

[0092] The embodiments should not be regarded as limiting the present invention, but any improvements made based on the spirit of the present invention should be within the protection scope of the present invention.

Claims

1. A multi-intelligent reflector-assisted MIMO channel estimation method based on spatial correlation, characterized in that: It includes the following steps: S1 constructs a multi-user communication system assisted by multiple intelligent reflectors, and establishes a spatial correlation channel model based on the spatial correlation matrix of the reflective elements of each intelligent reflector; S2 designs a two-stage channel estimation protocol and constructs the user sending pilot structure and the base station receiving signal model in each training time slot; S3 constructs a new set of auxiliary variables of the cascade channel according to the observation value signals received in the two stages, and establishes the cascade channel optimization objective function using the maximum a posteriori probability method; S4 decomposes the cascade channel optimization objective function into two sub-problems: optimizing common link variables and user-specific variables. It obtains the local optimal solution of the sub-problems through the alternating optimization method, and jointly optimizes the overall phase shift matrix of the multi-intelligent reflector to obtain the multi-intelligent reflector stacking channel. S5 uses the principle of spatial correlation between reflective elements of the same intelligent reflective surface to compare the correlation strength between the columns of channels on multiple intelligent reflective surfaces, separate the channel matrix of multiple intelligent reflective surfaces, and obtain the corresponding channels of each intelligent reflective surface. In S1, S11 is based on the known azimuth angle of incidence on each smart reflective surface. α With elevation β , and get its array response signal , and the correlation coefficient matrix between the reflective elements of each intelligent reflective surface is obtained from it P , , in, u a and u b Indicates the first a and b The position vector of each reflective element relative to the central reference reflective element, l Refers to the l Block intelligent reflective surface; S12 estimates all intelligent reflective surfaces as a whole, that is, the Kth user-intelligent reflective surface and the channel of the mth element of the smart reflector - the base station Represented as a cascade channel , , Among them, diag[·] means constructing a diagonal matrix, In S2, S21 obtains several independent measurements with different phase shifts θ during training, and the number of independent measurements is greater than the channel unknown value the number of S22 makes full use of the spatial correlation between the reflective elements of the smart reflective surface to From the superimposed channel matrix Separate from; S23 keeps the reflective elements of all smart reflective surfaces in the open state, and K users send different pilot signals in two stages. The total training time slot is , In the first phase training time slot ,in Indicates rounding up, K users send K pilots at the same time to form a pilot sequence, denoted as S ; The second phase includes training slots , K users send 1 pilot each time, and the pilot sequence is the first stage pilot S The first column of ; The S24 base station receiving signal model is trained in the first stage. , where G contains all smart reflector-base station channels, H r represents all channels of the user-intelligent reflector surface, is the additive white Gaussian noise at the base station at time t, Expressed as noise power, (·) T is the transposed matrix and diag[·] is constructed The diagonal matrix of ; The base station receiving signal model in the second stage of training is ; The signals received in the two stages are combined to obtain a base station receiving signal model that includes all time slots. 。 2. The method for multi-intelligent reflector-assisted MIMO channel estimation based on spatial correlation according to claim 1, characterized in that: In S3, S31 constructs a channel containing a common link R g and user-specific channel R u A set of auxiliary variables, in, and Represent the user-specific channel R u The kth column and the common link channel R g The mth column of S32 uses the auxiliary variable set obtained in S31 to record the cascade channel estimation problem as , , The corresponding optimization objective function where ||·|| F , ||·||2,(·) T , (·) H and(·) -1 Respectively expressed as Frobenius norm, two-norm, transpose, conjugate transpose, and inverse matrix, where, , and the first stage noise power and the second stage noise power .

3. The method for multi-intelligent reflector-assisted MIMO channel estimation based on spatial correlation according to claim 2, characterized in that: In S4, It is divided into two independent convex sub-problems, which are solved by an effective alternating optimization method as follows: Before the S41 iteration begins, the received signal including the two-stage training time slot is used to adjust the common link variable R g Initialize, and then fix R g variables, for user-specific variables R u Solve the optimization objective function; S42 Fixed R u Variables, R g Optimize the solution of the objective function by continuously iterating and optimizing the objective function until the objective functions of the two variables converge; S43 After a finite number of iterations, the channel estimation of the entire cascade channel obtains a local optimal solution.

4. The multi-intelligent reflector-assisted MIMO channel estimation method based on spatial correlation according to claim 2, characterized in that: In S5, S51 obtains a channel matrix including channel state information of all smart reflection surfaces, and separates a channel matrix corresponding to each smart reflection surface from the obtained channel matrix by using a method of comparing the strength of correlation; After channel estimation, S52 solves the correlation matrix between two adjacent columns of the estimated channel matrix, establishes an index factor with the minimum channel correlation, finds the column with the weakest correlation as a dividing line, and divides the channel matrices of different intelligent reflection surfaces.

Citation Information

Patent Citations

  • Wireless light MIMO communication system LED number modulation method based on intelligent reflection surface

    CN115776339A

  • Channel estimation optimization scheme based on double-IRS collaborative MISO system

    CN118200082A

  • Wireless communication system assisted by intelligent reflecting surface

    WO2023020080A1

  • Joint sparse channel estimation method based on intelligent reflecting surface-assisted radar communication

    WO2023221375A1