A downlink channel estimation method based on a smart reflector-assisted MIMO communication system

Abstract

This invention discloses a downlink channel estimation method based on a smart reflector-assisted MIMO communication system. It constructs a downlink system model of a smart reflector-assisted large-scale MIMO communication system; constructs a row sparse matrix and substitutes it into the downlink system model, thereby transforming the smart reflector-assisted large-scale MIMO downlink concatenated channel estimation into a standard row sparse recovery problem; it uses an off-network sparse Bayesian algorithm to iteratively update the noise accuracy, the accuracy vector of E, and the off-network gap parameter; it sets an iteration update termination condition; it sets a threshold and uses this threshold to select the effective angle set of the channel; it obtains accurate channel state information based on the effective angle set; and it uses the normalized mean square error (NMSE) to determine the accuracy of the calculated concatenated channel. This invention solves the problem of reduced output signal-to-noise ratio causing performance loss in channel estimation caused by the decoupling process of converting concatenated channel estimation into a sparse signal recovery problem through pseudo-inverse operations in existing technologies.

Description

Technical Field

[0001] This invention relates to the field of MIMO communication technology, and in particular to a downlink channel estimation method based on a smart reflector-assisted MIMO communication system. Background Technology

[0002] As a key technology for 5G wireless networks, massive MIMO (Multiple-input Multiple-output) has attracted widespread attention due to its ability to improve spectral efficiency and meet the capacity requirements of next-generation wireless communication. However, the high cost, high energy consumption, and complex signal processing caused by the deployment of a large number of antennas remain challenges for current research.

[0003] Intelligent reflecting surfaces (IRS) can be easily applied to massive MIMO systems due to their low power consumption and low hardware complexity. Specifically, an IRS consists of a large number of low-cost passive reflective elements, dynamically reconfiguring the wireless propagation environment to achieve a cost-effective wireless communication system with high spectral and energy efficiency. Accurate channel state information is essential to obtain the passive beamforming gain provided by the IRS. However, the lack of signal processing capabilities in IRS makes directly obtaining the channel state information of IRS-assisted massive MIMO communication systems a significant challenge.

[0004] Existing methods include a l1-norm-based channel estimation method proposed in the paper (X. Rao and V.K. Lau, Distributed compressive CSI Testing and feedback for FDD multi-user massive MIMO systems, IEEE Transactions on Signal Processing, vol. 62, no. 12, pp. 3261–3271, June, 2014); and a channel estimation method for large-scale MIMO communication systems based on off-network sparse Bayesian learning proposed in Chinese patent application CN201810042694.3. However, when dealing with IRS-assisted large-scale MIMO communication systems, these methods require decoupling through pseudo-inverse operations to transform the cascaded channel estimation into a sparse signal recovery problem. However, for any non-orthogonal matrix, pseudo-inverse operations inevitably lead to a decrease in the output signal-to-noise ratio, resulting in a loss in channel estimation performance. Summary of the Invention

[0005] Purpose of the invention: The purpose of this invention is to provide a downlink channel estimation method for intelligent reflector-assisted MIMO communication systems that improves channel estimation accuracy.

[0006] Technical Solution: To achieve the above objectives, the present invention provides a downlink channel estimation method for a smart reflector-assisted MIMO communication system, comprising the following steps:

[0007] Step S1: Deploy a downlink of a G-group intelligent reflector-assisted massive MIMO communication system and construct the downlink system model Y of the intelligent reflector-assisted massive MIMO communication system;

[0008] Step S2: Construct the row sparse matrix E and input it into the downlink system model Y, thereby transforming the intelligent reflector-assisted large-scale MIMO downlink concatenated channel estimation into a standard row sparse recovery problem;

[0009] Step S3: Iteratively update the noise accuracy α using the off-network sparse Bayes algorithm. (m) The precision vector γ of E (m) Off-grid gap β (m) parameter;

[0010] Step S4: Set the iterative update termination condition. If the update termination condition is not met, proceed to step S3.

[0011] Step S5: Set a threshold and use this threshold to select the effective angle set of the channel;

[0012] Step S6: Obtain accurate channel state information based on the effective angle set;

[0013] Step S7: Use the normalized mean square error (NMSE) to determine the accuracy of the desired cascaded channel.

[0014] In step S1, deploying a G-group intelligent reflector-assisted massive MIMO communication system downlink refers to equipping the base station with a uniform linear array of N antennas, and the mobile user terminal with a single antenna, in M... f Within the snapshot, the base station transmits a signal matrix X with pilot number T and a reflection coefficient matrix S, constructing the signal received by the mobile user terminal into a downlink system model Y of a large-scale MIMO communication system assisted by a smart reflector:

[0015] Y = XH BIU S+N=Φ(β)WS+N BIU ;

[0016] Where Φ(β)=XA(β) is the measurement matrix, H BIU=A(β)W is the downlink concatenated channel of a smart reflector-assisted massive MIMO communication system, A(β) is the array flow matrix of the channel between the base station and the smart reflector, and W is a row sparse matrix representing an L×G dimensional concatenated channel on the measurement matrix Φ(β), where L represents the number of grids, and S is a G×M f The reflection coefficient matrix of the N-dimensional smart reflective surface BIU Given a T×M dimensional noise matrix, N BIU The vector of Gaussian white noise with a mean of 0 and a precision of α in each column.

[0017] The cascaded channel of the intelligent reflector-assisted massive MIMO communication system, wherein:

[0018] W = [w1h] IU,1 w2h IU,2 ...w G h IU,G ],in, It is the channel between the intelligent reflector and the user. It is a sparse vector, and its non-zero elements correspond to the actual departure angle of the base station.

[0019] A(β)=[a(θ1+β1),a(θ2+β2),...,a(θ L +β L )],in,

[0020] Let λ represent the downlink array response vector, λ represent the operating wavelength of the electromagnetic wave, and d represent the distance between adjacent sensors. Uniformly divide the angle domain at the base station end L grid points, i.e. Indicates the gap between the wires.

[0021] Specifically, step S2 involves constructing the row sparse matrix E and substituting it into the downlink system model Y. This involves defining E = WS and substituting it into the downlink system model Y to construct the row sparse recovery problem.

[0022] Y=Φ(β)E+N BIU ,

[0023] Where Φ(β) is the measurement matrix, N BIU Let W be the noise matrix, and let W be the row sparse matrix representing a cascaded channel sparsely on the measurement matrix Φ(β).

[0024] Since S is a known reflection coefficient matrix of the smart reflector, which is either a random or orthogonal matrix, E and W maintain the same row sparsity property, thus transforming the smart reflector-assisted large-scale MIMO downlink concatenated channel estimation into a standard row sparsity recovery problem.

[0025] In step S3, the off-network sparse Bayesian algorithm is used to iteratively update the noise accuracy, the accuracy vector of E, and the off-network gap β parameter. During the iteration process, since E follows a complex Gaussian distribution, the mean and variance of E are directly calculated to avoid the performance loss caused by pseudo-inverse operations. Specifically, it includes the following sub-steps:

[0026] Step S301: Let the count variable for the number of iterations be m = 1, 2, ..., M, (·) (m) This represents the result of the m-th update.

[0027] Set initial values, let m = 1, and the precision vector of E. Each element in the value is 1, and the noise precision α (1) =1, and set β at the same time (1) For all zero elements, Indicates the gap between the wires.

[0028] Step S302: Fix γ (m) β (m) Update α (m) :

[0029]

[0030] Where tr(·) represents the trace of the matrix, ||·|| F Let F denote the F-norm of the matrix, (·) H Let μ represent the conjugate transpose, a = b = 0.00001, and the mean of E be μ. (m) (α,γ,β)=α (m) Σ (m) (α,γ,β)Φ H (β)Y, variance Σ (m) (α,γ,β)=(α (m) Φ H (β)Φ(β)+diag(γ (m) )) -1 diag(·) denotes a diagonal operation matrix;

[0031] Step S303: Fix α (m+1) β (m) Update γ (m) :

[0032]

[0033] Among them, Ξ (m) (α,γ,β)=μ (m) (α,γ,β)(μ (m) (α,γ,β)) H +MΣ (m)(α,γ,β), [·] ll This represents the l-th diagonal element of the matrix;

[0034] Step S304: Fix α (m+1) γ (m+1) Update β (m) :

[0035] β (m) =(P (m) ) -1 v (m) ;

[0036] Among them, P (m) =α (m) Re{(Φ'(β)) H Φ(β)⊙(μ (m) (α,γ,β)(μ (m) (α,γ,β)) H +MΣ (m) (α,γ,β))},

[0037]

[0038] Φ'(β)=XB(β), B(β)=[b(θ1+β1),b(θ2+β2),...,b(θ L +β L )],b(θ l +β l )=(a(θ l +β l ))' is a(θ l +β l Regarding θ l The derivative of .

[0039] The iteration update termination condition mentioned in step S4 refers to setting an iteration update termination condition. It is determined whether convergence has been achieved by judging whether the iteration count variable m has reached the upper limit M or whether the current update result is equal to the previous update result. If neither condition is met, the iteration count variable m = m + 1, and step S3 is re-executed.

[0040] In step S5, setting a threshold and using that threshold to select the effective angle set of the channel means that the threshold is set to... Where (μ) i ) 2 Represents (μ(α,δ,β)) 2 The row with the largest mean is used to select the effective angle set Ω={i(μ i ) 2 ≥η,i=1,2,...,L}.

[0041] In step S6, obtaining accurate channel state information based on the effective angle set refers to estimating the channel state information using the effective angle set Ω as follows:

[0042]

[0043] in(·) Ω This represents a submatrix consisting of the column vectors corresponding to the set Ω in the matrix. This indicates a pseudo-inverse operation.

[0044] In the final downlink concatenated channel estimation expression, the mean μ(α,γ,β) of the obtained row sparse matrix E is used to estimate the downlink concatenated channel, where... This does not affect the process of restoring row sparse matrices.

[0045] In step S7, determining the accuracy of the desired cascaded channel using the normalized mean square error (NMSE) refers to determining the accuracy of the desired cascaded channel H based on the NMSE value. BIU The Monte Carlo method, which involves calculating the mean squared error multiple times through numerous trials and then taking its average, makes the NMSE results more accurate.

[0046]

[0047] The number of Monte Carlo experiments is m. c =1,2,...,M c , Indicates the m-th c Channel matrix of sub-Monte Carlo experiment The estimated value.

[0048] Beneficial effects: The present invention has the following advantages:

[0049] 1. This invention utilizes the channel sparsity property generated by the local scattering effect at the base station, and avoids the performance loss caused by using pseudo-inverse operations by combining the reflection coefficient with the row sparse matrix.

[0050] 2. This invention further explores the channel sparsity of IRS-assisted large-scale MIMO communication systems. Compared with existing methods, it can greatly improve the channel estimation accuracy regardless of whether the reflection coefficient matrix S is a random matrix or an orthogonal matrix. Attached Figure Description

[0051] Figure 1 This is a flowchart of the method of the present invention;

[0052] Figure 2A comparative diagram of the normalized root mean square error (NMSE) of the channel estimation method of the present invention and the pseudo-inverse-based l1-norm and pseudo-inverse-based off-network sparse Bayesian learning methods under the condition of 200 Monte Carlo experiments, with a signal-to-noise ratio of 0dB, S being a random matrix, and the pilot frequency T varying from 30 to 60.

[0053] Figure 3 This diagram compares the normalized root mean square error (NMSE) of the channel estimation method of the present invention with that of the l1-norm method based on pseudo-inverse and the off-network sparse Bayesian learning method based on pseudo-inverse when the signal-to-noise ratio is 0dB under the condition of 200 Monte Carlo experiments, S is an orthogonal matrix, and the pilot frequency T varies from 30 to 60. Detailed Implementation

[0054] The technical solution of the present invention will be described in detail below with reference to the embodiments and accompanying drawings.

[0055] like Figure 1 As shown, a downlink channel estimation method based on a smart reflector-assisted MIMO communication system includes the following steps:

[0056] Step S1: Deploy a downlink of a G-group intelligent reflector-assisted massive MIMO communication system, and construct the downlink system model Y of the intelligent reflector-assisted massive MIMO communication system, specifically as follows:

[0057] The base station is equipped with a uniform linear array of N antennas, while the mobile user terminal is equipped with a single antenna. f Within the snapshot, the base station transmits a signal matrix X with pilot number T and a reflection coefficient matrix S, constructing the signal received by the mobile user terminal into a downlink system model Y of a large-scale MIMO communication system assisted by a smart reflector:

[0058] Y = XH BIU S+N=Φ(β)WS+N BIU ;

[0059] Where Φ(β)=XA(β) is the measurement matrix, H BIU =A(β)W is the downlink concatenated channel of a smart reflector-assisted massive MIMO communication system, A(β) is the array flow matrix of the channel between the base station and the smart reflector, and W is a row sparse matrix representing an L×G dimensional concatenated channel on the measurement matrix Φ(β), where L represents the number of grids, and S is a G×M f The reflection coefficient matrix of the N-dimensional smart reflective surface BIU Given a T×M dimensional noise matrix, N BIU The vector of Gaussian white noise with a mean of 0 and a precision of α in each column.

[0060] In the cascaded channel of a smart reflector-assisted massive MIMO communication system:

[0061] W = [w1h] IU,1 w2h IU,2 ...w G h IU,G ],in, It is the channel between the intelligent reflector and the user. It is a sparse vector, and its non-zero elements correspond to the actual departure angle of the base station.

[0062] A(β)=[a(θ1+β1),a(θ2+β2),...,a(θ L +β L )],in, Let λ represent the downlink array response vector, λ represent the operating wavelength of the electromagnetic wave, and d represent the distance between adjacent sensors. Uniformly divide the angle domain at the base station end L grid points, i.e. Indicates the gap between the wires.

[0063] Step S2: Construct the row sparse matrix E and substitute it into the downlink system model Y, thereby transforming the intelligent reflector-assisted large-scale MIMO downlink concatenated channel estimation into a standard row sparse recovery problem, specifically:

[0064] Define E = WS, and substitute it into the downlink system model Y to construct the row sparse recovery problem:

[0065] Y=Φ(β)E+N BIU ,

[0066] Where Φ(β) is the measurement matrix, N BIU Let W be the noise matrix, and let W be the row sparse matrix representing a cascaded channel sparsely on the measurement matrix Φ(β).

[0067] Since S is a known reflection coefficient matrix of the smart reflector, which is either a random or orthogonal matrix, E and W maintain the same row sparsity property, thus transforming the smart reflector-assisted large-scale MIMO downlink concatenated channel estimation into a standard row sparsity recovery problem.

[0068] Step S3: Iteratively update the noise accuracy α using the off-network sparse Bayes algorithm. (m) The precision vector γ of E (m ), Off-grid gap β (m) The parameters specifically include the following sub-steps:

[0069] Step S301: Let the count variable for the number of iterations be m = 1, 2, ..., M, (·) (m) This represents the result of the m-th update.

[0070] Set initial values, let m = 1, and the precision vector of E. Each element in the value is 1, and the noise precision α (1) =1, and set β at the same time (1) For all zero elements, Indicates the gap between the wires.

[0071] Step S302: Fix γ (m) β (m) Update α (m) :

[0072]

[0073] Where tr(·) represents the trace of the matrix, ||·|| F Let F denote the F-norm of the matrix, (·) H Let μ represent the conjugate transpose, a = b = 0.00001, and the mean of E be μ. (m) (α,γ,β)=α (m) Σ (m) (α,γ,β)Φ H (β)Y, variance Σ (m) (α,γ,β)=(α (m) Φ H (β)Φ(β)+diag(γ (m) )) -1 diag(·) denotes a diagonal operation matrix;

[0074] Step S303: Fix α (m+1) β (m) Update γ (m) :

[0075]

[0076] Among them, Ξ (m) (α,γ,β)=μ (m) (α,γ,β)(μ (m) (α,γ,β)) H +MΣ (m) (α,γ,β), [·] ll This represents the l-th diagonal element of the matrix;

[0077] Step S304: Fix α (m+1) γ (m+1) Update β (m) :

[0078] β (m) =(P (m) ) -1 v (m) ;

[0079] Among them, P (m) =α (m) Re{(Φ'(β)) H Φ(β)⊙(μ (m) (α,γ,β)(μ (m) (α,γ,β)) H +MΣ (m) (α,γ,β))},

[0080]

[0081] Φ'(β)=XB(β), B(β)=[b(θ1+β1),b(θ2+β2),...,b(θ L +β L )],b(θ l +β l )=(a(θ l +β l ))' is a(θ l +β l Regarding θ l The derivative of .

[0082] Step S4: Set the iteration end condition. If the update end condition is not met, proceed to step S3, specifically:

[0083] Set the termination condition for the iteration update. Determine whether the iteration count variable m has reached the upper limit M or whether the current update result is equal to the previous update result. If neither condition is met, then the iteration count variable m = m + 1, and return to re-execute step S3.

[0084] Step S5: Set a threshold and use this threshold to select the effective angle set of the channel, specifically:

[0085] Threshold setting Where (μ) i ) 2 Represents (μ(α,δ,β)) 2 The row with the largest mean is used to select the effective angle set Ω={i|(μ i ) 2 ≥η,i=1,2,...,L}.

[0086] Step S6: Obtain accurate channel state information based on the effective angle set, specifically as follows:

[0087] Using the effective angle set Ω, the estimate is expressed as:

[0088]

[0089] in(·) Ω This represents a submatrix consisting of the column vectors corresponding to the set Ω in the matrix. This indicates a pseudo-inverse operation.

[0090] In the final downlink concatenated channel estimation expression, the mean μ(α,γ,β) of the obtained row sparse matrix E is used to estimate the downlink concatenated channel, where... This does not affect the process of restoring row sparse matrices.

[0091] Step S7: Determine the accuracy of the cascaded channel by using the normalized mean square error (NMSE). Specifically:

[0092] Based on the required cascaded channel H BIU The Monte Carlo method, which involves calculating the mean squared error multiple times through numerous trials and then taking its average, makes the NMSE results more accurate.

[0093]

[0094] The number of Monte Carlo experiments is m. c =1,2,...,M c , Indicates the m-th c Channel matrix of sub-Monte Carlo experiment The estimated value.

[0095] This embodiment further illustrates the effectiveness of the method described in this invention through simulation experiments:

[0096] Assume the base station uses a uniform linear array with N=100 antennas, the downlink operating frequency is 2170MHz, the wireless channel is randomly generated by the 3GPP spatial channel model (SCM), each element of the base station transmits pilot signal matrix X follows an independent Gaussian distribution with zero mean and unit variance, and the background noise is assumed to be Gaussian white noise.

[0097] Experimental conditions:

[0098] Using this invention, the channel was estimated 200 times with a signal-to-noise ratio of 0dB and a pilot frequency T varying from 30 to 60, with 200 grid lines. The simulation results are as follows. Figure 2 and Figure 3 As shown.

[0099] Experimental Analysis:

[0100] from Figure 2It can be seen that the present invention can accurately estimate the downlink channel information of IRS-assisted large-scale MIMO communication system when S is a random matrix, and its NMSE performance is significantly better than the off-network sparse Bayes method based on pseudo-inverse and the l1-norm method based on pseudo-inverse.

[0101] from Figure 3 As can be seen, the present invention can accurately estimate the downlink channel information of an IRS-assisted large-scale MIMO communication system when S is an orthogonal matrix.

Claims

1. A downlink channel estimation method based on a smart reflector-assisted MIMO communication system, characterized in that, Includes the following steps: Step S1: Deploy one The downlink of a large-scale MIMO communication system assisted by a smart reflector is constructed, and a downlink system model Y of the large-scale MIMO communication system assisted by a smart reflector is built. Step S2: Construct a row sparse matrix This is then incorporated into the downlink system model Y, thereby transforming the intelligent reflector-assisted large-scale MIMO downlink concatenated channel estimation into a standard line sparse recovery problem, specifically: Define Substitute this into the downlink system model Y to construct the row sparse recovery problem: ， in For the measurement matrix, For the noise matrix, For a cascaded channel in the measurement matrix A row sparse matrix in the sparse representation above; because It is the known reflection coefficient matrix of the intelligent reflector. This matrix is ​​either a random matrix or an orthogonal matrix, therefore and Maintaining the same row sparsity property, the intelligent reflector-assisted large-scale MIMO downlink concatenated channel estimation is transformed into a standard row sparsity recovery problem; Step S3: Set the initial value, let , row sparse matrix precision vector Each element in the value is 1, indicating noise precision. At the same time, set For all zero elements, Representing the off-network gap, the noise accuracy is iteratively updated using the off-network sparse Bayes algorithm. The precision vector of E Off-grid gap parameter; Step S4: Set the iterative update termination condition. If the update termination condition is not met, proceed to step S3. Step S5: Set a threshold and use this threshold to select the effective angle set of the channel; Step S6: Obtain accurate channel state information based on the effective angle set; Step S7: Use the normalized mean square error (NMSE) to determine the accuracy of the desired cascaded channel.

2. The downlink channel estimation method for a smart reflector-assisted MIMO communication system according to claim 1, characterized in that, Deploying a [unit / item] as described in step S1 The downlink of a group of intelligent reflectors-assisted massive MIMO communication systems refers to the configuration of a group of intelligent reflectors at the base station end. A uniform linear array of antennas is used, with a single antenna at the mobile user terminal. Within the snapshot, the number of pilot signals transmitted by the base station is: signal matrix Reflection coefficient matrix The signal received by the mobile user terminal is used to construct a downlink system model Y for a large-scale MIMO communication system assisted by a smart reflector: ； in, For the measurement matrix, For the downlink concatenation channel of a smart reflector-assisted massive MIMO communication system, This is the array flow matrix of the channel between the base station and the smart reflector. For one Dimensional cascaded channels in the measurement matrix The row sparse matrix of the sparse representation above, where Indicates the number of grid cells. For one The reflection coefficient matrix of the intelligent reflective surface. For one A noise matrix of dimension 1. The mean of each column is 0, and the precision is 0. Gaussian white noise vector.

3. The downlink channel estimation method for a smart reflector-assisted MIMO communication system according to claim 2, characterized in that, The cascaded channel of the intelligent reflector-assisted massive MIMO communication system, wherein: ,in, It is the channel between the intelligent reflector and the user. It is a sparse vector, and its non-zero elements correspond to the actual departure angle of the base station. ,in, Represents the downlink array response vector. Indicates the operating wavelength of electromagnetic waves. Indicates the distance between adjacent sensors. Uniformly divide the angle domain at the base station end of Grid points, i.e. , Indicates the gap between the wires.

4. The downlink channel estimation method for a smart reflector-assisted MIMO communication system according to claim 3, characterized in that, Step S3 describes using an off-grid sparse Bayesian algorithm to iteratively update the noise precision, the precision vector of E, and the off-grid gap β parameter. During the iteration process, since the row sparse matrix E follows a complex Gaussian distribution, the mean and variance of E are directly calculated to avoid performance loss caused by pseudo-inverse operations. Specifically, this includes the following sub-steps: Step S301: Let the count variable for the number of iterations be... , Indicates the first The result of this update is that M represents the maximum value of the count variable; Step S302: Fix , ,renew : ， in, Represents the trace of a matrix. Denotes the F-norm of a matrix. This indicates the conjugate transpose. , mean ,variance , Represents a diagonal operation matrix. This indicates the number of pilot signals transmitted by the base station; Step S303: Fix , ,renew : ； in, , Represents the first of the matrix One diagonal element, Indicates the number of grid cells; Step S304: Fix , ,renew : ； in, , ， , , This indicates that the number of pilot signals transmitted by the base station is... The signal matrix, yes about The derivative of This represents the parameters to be estimated, including noise accuracy, off-grid gap, and row sparse matrix E-precision vector.

5. The downlink channel estimation method for a smart reflector-assisted MIMO communication system according to claim 1, characterized in that, The iteration update termination condition mentioned in step S4 refers to setting the iteration update termination condition by judging the iteration count variable. Has the upper limit been reached? or The convergence is determined by whether the current update result is equal to the previous update result. If neither condition is met, the iteration count variable... Then return to re-execute step S3.

6. The downlink channel estimation method for a smart reflector-assisted MIMO communication system according to claim 1, characterized in that, Step S5, which involves setting a threshold and using that threshold to select the effective angle set of the channel, refers to setting the threshold to... ,in express No. The row with the largest mean is selected, and this threshold is used to select the effective angle set of the channel. , This indicates the number of pilot signals transmitted by the base station. Indicates the number of grid cells. Represents a row sparse matrix The mean.

7. The downlink channel estimation method for a smart reflector-assisted MIMO communication system according to claim 6, characterized in that, Step S6, obtaining accurate channel state information based on the effective angle set, refers to using the effective angle set... The estimate is expressed as: ； in Represents the set in the matrix The submatrix formed by the corresponding column vectors, This indicates a pseudo-inverse operation. This is the array flow matrix of the channel between the base station and the smart reflector. Represented by matrix Set of effective angles The submatrix formed by the corresponding column vectors, For one A dimensional intelligent reflective surface reflection coefficient matrix.

8. The downlink channel estimation method for a smart reflector-assisted MIMO communication system according to claim 7, characterized in that, In the final downlink concatenated channel estimation expression, the obtained row sparse matrix is ​​used. mean Estimate the downlink concatenated channel, where This does not affect the process of restoring row sparse matrices.

9. The downlink channel estimation method for a smart reflector-assisted MIMO communication system according to claim 1, characterized in that, Step S7, which describes using the normalized mean square error (NMSE) to determine the accuracy of the desired cascaded channel, refers to determining the accuracy of the desired cascaded channel based on the NMSE value. The Monte Carlo method, which involves calculating the mean squared error multiple times through numerous trials and then taking its average, makes the NMSE results more accurate. ， Among them, the number of Monte Carlo experiments was 100. , Indicates the first Channel matrix of sub-Monte Carlo experiment The estimated value.

Images