A hybrid precoding method for massive MIMO
By constructing a uniform linear antenna array response vector codebook matrix At and optimizing channel state information, a hybrid precoder is designed, which solves the problem of high computational complexity of the hybrid precoder and achieves high spectral efficiency performance while reducing computational complexity.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HARBIN INST OF TECH
- Filing Date
- 2023-12-08
- Publication Date
- 2026-07-07
AI Technical Summary
Existing technologies, while ensuring high performance of hybrid precoders, have high computational complexity, making hardware implementation difficult.
A hybrid precoding method for massive MIMO is adopted. By constructing a codebook matrix At of a uniform linear antenna array response vector, and combining singular value decomposition of channel state information and Euclidean distance optimization, a digital precoder FBB and an RF chain analog precoder FRF are designed to reduce computational complexity.
It achieves spectral efficiency performance similar to existing technologies, while the computational complexity does not increase with the number of transmitted data streams, reducing additional computational overhead and making the method more promising in cases with more transmitted data streams.
Smart Images
Figure CN117895979B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of wireless communication technology, specifically relating to a hybrid precoding method for massive MIMO (Multiple-Input Multiple-Output) technology. Background Technology
[0002] To meet the ever-increasing demand for high data rates in wireless communication systems, the millimeter-wave (mmWave) band is widely considered the most practical solution. While mmWave offers higher transmission bandwidth, it suffers from high path loss. Massive MIMO technology can significantly mitigate this loss by improving beam directivity and achieving a high signal-to-noise ratio (SNR) of the received signal. However, implementing massive MIMO presents practical hardware challenges, as each antenna element requires a dedicated radio frequency (RF) chain, greatly increasing hardware complexity. A potential solution is to use a hybrid architecture that controls a large number of antennas with only a few RF chains. In a hybrid beamforming architecture, all-digital precoders and combiners are distributed between low-dimensional baseband and high-dimensional RF components and solved under unity-modulus constraints of analog precoders and combiners. Calculating precoders and combiners typically requires knowledge of channel state information, making this process feasible only at the receiver. The calculated precoder values are then fed back to the transmitter. In hybrid beamforming, the dimension of the RF (or analog) precoder depends on the number of transmit antennas. In massive MIMO systems, the number of transmit antennas can be very large, resulting in significant transmission overhead for the precoder feedback. To overcome these challenges, finite feedback systems are used, quantizing the analog beamforming vector using a dictionary or codebook. The use of a codebook is also important because the number of phase shifter modes available for simulating beamforming is limited.
[0003] Orthogonal codebooks like the DFT codebook and uniformly quantized codebooks have been widely used in the design of hybrid precoders. The OMP (Orthogonal Matched Tracking) algorithm can be used to calculate the coefficient matrix of the digital precoder and find the RF chain parameters of the analog precoder from the codebook. These methods have been extensively analyzed and applied to beamforming in several systems. Increasing the codebook size is an effective way to improve the performance of hybrid precoders. Uniform codebooks or other non-orthogonal codebooks can be used to design hybrid precoders using compressed sensing algorithms such as OMP, but the additional computational complexity is unavoidable.
[0004] In summary, existing technologies require high computational complexity to ensure high performance of hybrid precoders. Therefore, it is essential to propose a new hybrid precoding method. Summary of the Invention
[0005] The purpose of this invention is to solve the problem of high computational complexity in existing technologies while ensuring high performance of hybrid precoders, and to propose a hybrid precoding method for massive MIMO.
[0006] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows:
[0007] A hybrid precoding method for massive MIMO, the method specifically includes the following steps:
[0008] Step 1: Construct the codebook matrix A of the uniform linear antenna array response vector based on the transmitting antenna array. t ;
[0009] Step 2: Using the uniform linear antenna array response vector codebook matrix A t The codebook provides an equivalent formulation of the hybrid precoder design problem;
[0010] Step 3: Calculate the digital precoder F based on the equivalent formulation of the hybrid precoder design problem in Step 2. BB and RF chain analog pre-encoder F RF .
[0011] Furthermore, the specific process of step 1 is as follows:
[0012] Step 11: When the directions of all transmitting antennas satisfy the constraints, record the azimuth code of the antenna array as follows: It is by n = 0, ..., bN t A vector consisting of -1s, where b is a positive integer greater than or equal to 1, and N... t This refers to the number of transmitting antennas;
[0013] The constraint is that the direction of each transmitting antenna is within the range of [-π / 2, π / 2];
[0014] Step 12: Prepare the azimuth code book bN t Substituting each azimuth angle into the array response vector of the uniform linear antenna array Where φ represents the azimuth angle, and bN t The array response vectors corresponding to each azimuth angle are combined to form a uniform linear antenna array response vector codebook matrix A. t .
[0015] Furthermore, the uniform linear antenna array response vector codebook matrix A t for:
[0016]
[0017] in, e is the base of the natural logarithm, j is the imaginary unit, and k is an intermediate variable.
[0018] Furthermore, the uniform linear antenna array response vector codebook matrix A t The l-th column is:
[0019]
[0020] Among them, a t (l) is the codebook matrix A of the uniform linear antenna array response vector. t The lth column.
[0021] Furthermore, the specific process of step 2 is as follows:
[0022] Step 21: The receiver obtains channel state information H through channel estimation, and then performs singular value decomposition on the channel state information H:
[0023] H=UΣV H
[0024] Where U is a dimension of N r The unitary matrix of size ×rank(H), where rank(·) denotes the rank of the matrix, N r V is the number of receiving antennas, Σ is a diagonal matrix obtained by arranging the singular values of the channel state information H in descending order, and the dimension of Σ is rank(H)×rank(H). V is a matrix of dimension N. t The unitary matrix of ×rank(H);
[0025] Step 22: Represent Σ and V as block matrices:
[0026]
[0027] Where Σ1 and Σ2 are the block matrices of Σ, and V1 and V2 are the block matrices of V;
[0028] Take F opt =V1 is the target optimal hybrid precoder;
[0029] Step 23: Treat the design problem of the target optimal hybrid precoder as minimizing F opt With F RF F BB The Euclidean distance problem:
[0030]
[0031]
[0032]
[0033] Among them, F BB It is a digital precoder, F RF It is an RF chain analog pre-encoder. F It is the F-norm. Is to make F opt With F RF F BB The Euclidean distance is minimized in both digital pre-encoders and RF analog pre-encoders. F represents RF Each element in the equation has a modulus of 1, which represents the constant mode constraint of the phase shifter. It is a power constraint for the hybrid pre-encoder;
[0034] Minimize F opt With F RF F BB The sparse equivalence representation of the Euclidean distance problem is:
[0035]
[0036]
[0037]
[0038] in, It is for F BB The dimension obtained by padding with zeros is bN t ×N s The matrix, N s It is the length of the data stream, and satisfies... N RF It represents the number of RFs, and ||·||0 represents the 0 norm;
[0039] Step 24: Analyze the codebook matrix A of the uniform linear antenna array response vector. t By expanding, we obtain the auxiliary matrix.
[0040]
[0041] Step 25, in Multiplying the left side by the diagonal matrix Λ yields bN t ×bN t DFT (Discrete Fourier Transform) matrix
[0042]
[0043] in,
[0044] Step 26: Rewrite the problem described in Step 23 as an equivalent problem as follows:
[0045]
[0046]
[0047]
[0048] Here, the superscript H represents the conjugate transpose of the matrix, and the intermediate variable is the auxiliary matrix.
[0049] Furthermore, the specific process of step 3 is as follows:
[0050] Step 31, through the analysis of... Obtain by performing IDFT operation
[0051]
[0052] in, yes The inverse matrix;
[0053] Step 32, according to For each column vector, calculate the energy and distribution, then find all peak points in the energy and distribution, and finally sort the peak points in descending order, recording the top N. RF The peak point of the position is at The row index k in the data;
[0054]
[0055] Where diag(·) represents a diagonal matrix;
[0056] Step 33: From codebook matrix A t Find all column vectors with column index k to form the RF chain analog pre-encoder F. RF :
[0057]
[0058] Step 34, using row index k and Obtain digital precoder F BB :
[0059]
[0060] in, Representative from Select the matrix consisting of all row vectors with row index k, and F BB Satisfying the energy constraints of the transmitting antenna
[0061] Furthermore, the specific process of step 3 is as follows:
[0062] Step 31: Introduce an auxiliary matrix make
[0063] Step 32, for Obtain by performing IDFT operation
[0064]
[0065] Step 33: Find them separately For each column vector in the algorithm, find the peak point of the energy distribution, then sort the peak points of each column vector in descending order and record the top N. RF The column vector index k corresponding to the peak point of the bit:
[0066]
[0067] Step 34: From the uniform linear antenna array response codebook matrix A t In the process, all column vectors with column index k are found to form the analog pre-encoder F. RF From the auxiliary matrix Find all column vectors with column index k to form an auxiliary matrix.
[0068]
[0069] Step 35, index k using column vector and Obtain digital precoder F BB :
[0070]
[0071] in, Representative from Select the matrix composed of all column vectors with column index k;
[0072] Step 36: Update the auxiliary matrix:
[0073]
[0074] in, It is the updated auxiliary matrix;
[0075] The difference matrix D is calculated as follows:
[0076]
[0077] Step 37: Determine whether the iteration stopping condition has been met;
[0078] If the iteration stopping condition is not met, return to step 32 using the updated auxiliary matrix;
[0079] If the iteration stopping condition is met, then the F obtained from the last iteration will be... RF As an analog pre-encoder for the RF chain, the F obtained in the last iteration BB As a digital precoder.
[0080] Furthermore, the iteration stopping condition is reaching the set maximum number of iterations or It is less than the set threshold.
[0081] The beneficial effects of this invention are:
[0082] Compared with existing hybrid precoder design methods, the method of this invention has very close spectral efficiency performance, but the computational complexity does not increase with the increase of transmission data streams. Therefore, the method of this invention does not require additional computational complexity overhead. Compared with the prior art, this invention effectively reduces computational complexity, which makes the non-orthogonal precoder proposed in this invention more promising in the case of more transmission data streams. Attached Figure Description
[0083] Figure 1 For millimeter-wave massive MIMO hybrid pre-encoder models;
[0084] Figure 2 To demonstrate the spectral efficiency achieved by different codebooks for millimeter-wave systems using different antenna arrays in a single data stream scenario;
[0085] Figure 3 The spectral efficiency achieved by a millimeter-wave system using a 256×64 antenna array with different codebooks under different data stream transmission conditions;
[0086] Figure 4 For transmitting data stream N in a millimeter-wave system using a 256×64 antenna array s When = 2, the spectral efficiency achieved by different codebooks under different RF chain conditions;
[0087] Figure 5 For transmitting data stream N in a millimeter-wave system using a 256×64 antenna array s When = 4, the spectral efficiency achieved by different codebooks under different RF chain conditions;
[0088] Figure 6 For transmitting data stream N s When = 4, the CPU overhead required by different algorithms under different RF chain conditions. Detailed Implementation
[0089] The present application will now be described in further detail with reference to specific embodiments and accompanying drawings. Obviously, the described embodiments are merely a part of the embodiments of the present invention, and not all of them. Other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are all within the scope of protection of the present invention.
[0090] Specific Implementation Method 1: Combination Figure 1 This embodiment describes a hybrid precoding method for massive MIMO, which specifically includes the following steps:
[0091] Step 1: Construct the codebook matrix A of the uniform linear antenna array response vector based on the transmitting antenna array. t ;
[0092] Step 2: Using the uniform linear antenna array response vector codebook matrix A t The codebook provides an equivalent formulation of the hybrid precoder design problem;
[0093] Step 3: Calculate the digital precoder F based on the equivalent formulation of the hybrid precoder design problem in Step 2. BB and RF chain analog pre-encoder F RF .
[0094] Specific Implementation Method Two: This implementation method differs from Specific Implementation Method One in that the specific process of step 1 is as follows:
[0095] Step 11: When the directions of all transmitting antennas satisfy the constraints, record the azimuth code of the antenna array as follows: It is by n = 0, ..., bN t A vector consisting of -1s, where b is a positive integer greater than or equal to 1, and N... t This refers to the number of transmitting antennas;
[0096] The constraint is that the direction of each transmitting antenna is within the range of [-π / 2, π / 2];
[0097] Step 12: Prepare the azimuth code book bN t azimuth angles (i.e.) n = 0, ..., bN t -1) Substitute the array response vectors of the uniform linear antenna array (ULA) into the corresponding vectors. (Assuming the antenna element spacing is half a wavelength), where φ represents the azimuth angle, let bN t The array response vectors corresponding to each azimuth angle are combined to form a uniform linear antenna array response vector codebook matrix A. t .
[0098] The other steps and parameters are the same as in Specific Implementation Method 1.
[0099] Specific Implementation Method Three: This implementation method differs from Specific Implementation Method Two in that the uniform linear antenna array response vector codebook matrix A... t for:
[0100]
[0101] in, e is the base of the natural logarithm, j is the imaginary unit, and k is an intermediate variable.
[0102] The other steps and parameters are the same as in Specific Implementation Method Two.
[0103] Specifically, when b=1, the codebook matrix A t for:
[0104]
[0105] Specific Implementation Method Four: This implementation method differs from Specific Implementation Method Three in that the uniform linear antenna array response vector codebook matrix A... t The l-th column is:
[0106]
[0107] Among them, a t (l) is the codebook matrix A of the uniform linear antenna array response vector. t The lth column.
[0108] The other steps and parameters are the same as in Specific Implementation Method 3.
[0109] Specific Implementation Method Five: This implementation method differs from Specific Implementation Method Four in that the specific process of step 2 is as follows:
[0110] Step 21: The receiver obtains channel state information H through channel estimation, and then performs singular value decomposition on the channel state information H:
[0111] H=UΣV H
[0112] Where U is a dimension of N r The unitary matrix of size ×rank(H), where rank(·) denotes the rank of the matrix, N r V is the number of receiving antennas, Σ is a diagonal matrix obtained by arranging the singular values of the channel state information H in descending order, and the dimension of Σ is rank(H)×rank(H). V is a matrix of dimension N. t The unitary matrix of ×rank(H);
[0113] Step 22: Represent Σ and V as block matrices:
[0114]
[0115] Where Σ1 and Σ2 are the block matrices of Σ, and V1 and V2 are the block matrices of V;
[0116] Take F opt =V1 is the target optimal hybrid precoder;
[0117] Step 23: Treat the design problem of the target optimal hybrid precoder as minimizing F opt With F RF F BB The Euclidean distance problem:
[0118]
[0119]
[0120]
[0121] Among them, F BB It is a digital precoder, F RF It is an RF chain analog pre-encoder. F It is the F-norm. Is to make F opt With F RF F BB The Euclidean distance is minimized in both digital pre-encoders and RF analog pre-encoders. F represents RF Each element in the equation has a modulus of 1, which represents the constant mode constraint of the phase shifter. It is a power constraint for the hybrid pre-encoder;
[0122] Minimize F opt With F RF F BB The sparse equivalence representation of the Euclidean distance problem is:
[0123]
[0124]
[0125]
[0126] in, It is for F BB The dimension obtained by padding with zeros is bN t ×N s The matrix, N s It is the length of the data stream, and satisfies... N RF It represents the number of RFs, and ||·||0 represents the 0 norm;
[0127] Step 24: Analyze the codebook matrix A of the uniform linear antenna array response vector. t By expanding, we obtain the auxiliary matrix.
[0128]
[0129] Step 25, in Multiplying the left side by the diagonal matrix Λ yields bN t ×bN t DFT (Discrete Fourier Transform) matrix
[0130]
[0131] in,
[0132] Step 26: Rewrite the problem described in Step 23 as an equivalent problem as follows:
[0133]
[0134]
[0135]
[0136] Here, the superscript H represents the conjugate transpose of the matrix, and the intermediate variable is the auxiliary matrix. The other steps and parameters are the same as in Specific Implementation Method Four.
[0137] Specific Implementation Method Six: This implementation method differs from Specific Implementation Method Five in that the specific process of step 3 is as follows: Step 31, through the... Perform an IDFT operation (which can be implemented using the FFT algorithm) to obtain
[0138]
[0139] in, yes The inverse matrix;
[0140] Step 32, according to For each column vector, calculate the energy and distribution, then find all peak points in the energy and distribution, and finally sort the peak points in descending order, recording the top N. RF The peak point of the position is at The row index k in the data;
[0141]
[0142] Where diag(·) represents a diagonal matrix;
[0143] Step 33: From codebook matrix A t Find all column vectors with column index k to form the RF chain analog pre-encoder F. RF :
[0144]
[0145] Step 34, using row index k and Obtain digital precoder F BB :
[0146]
[0147] in, Representative from Select the matrix consisting of all row vectors with row index k, and F BB Satisfying the energy constraints of the transmitting antenna
[0148] The other steps and parameters are the same as in Specific Implementation Method 5.
[0149] Specifically, when b = 1, step 32 can be replaced by the following solution:
[0150] turn up The top N column vector energies and distributions RF There are N maximum points, recorded. RF The maximum point is at The row index k in the data;
[0151]
[0152] Specific Implementation Method Seven: This implementation method differs from Specific Implementation Method Five in that the specific process of step 3 is as follows:
[0153] Step 31: Introduce an auxiliary matrix make
[0154] Step 32, for Perform an IDFT operation (which can be implemented using the FFT algorithm) to obtain
[0155]
[0156] Step 33: Find them separately For each column vector in the algorithm, find the peak point of the energy distribution, then sort the peak points of each column vector in descending order and record the top N.RF The column vector index k corresponding to the peak point of the bit:
[0157]
[0158] Step 34: From the uniform linear antenna array response codebook matrix A t In the process, all column vectors with column index k are found to form the analog pre-encoder F. RF From the auxiliary matrix Find all column vectors with column index k to form an auxiliary matrix.
[0159]
[0160] Step 35, index k using column vector and Obtain digital precoder F BB :
[0161]
[0162] in, Representative from Select the matrix composed of all column vectors with column index k;
[0163] Step 36: Update the auxiliary matrix:
[0164]
[0165] in, It is the updated auxiliary matrix;
[0166] The difference matrix D is calculated as follows:
[0167]
[0168] Step 37: Determine whether the iteration stopping condition has been met;
[0169] If the iteration stopping condition is not met, return to step 32 using the updated auxiliary matrix;
[0170] If the iteration stopping condition is met, then the F obtained from the last iteration will be... RF As an analog pre-encoder for the RF chain, the F obtained in the last iteration BB As a digital precoder.
[0171] The other steps and parameters are the same as in Specific Implementation Method 5.
[0172] Specific Implementation Method Eight: This implementation method differs from Specific Implementation Method Seven in that the iteration stopping condition is reaching the set maximum number of iterations or... It is less than the set threshold.
[0173] The other steps and parameters are the same as in Specific Implementation Method Seven.
[0174] Figure 2 and Figure 3 This demonstrates that the method proposed in this invention has greater advantages than traditional methods in large-scale antenna and multi-data stream transmission systems, and that the spectral efficiency advantage of the method proposed in this invention will become increasingly greater as the number of antennas and data streams increase.
[0175] Figure 4 and Figure 5 This demonstrates the necessity of increasing the number of RF chains in multi-data stream scenarios. The system performance will be effectively improved if the number of RF chains is greater than the number of data streams. Furthermore, the method proposed in this invention can achieve the same spectral efficiency performance as the traditional OMP precoding method using uniform quantization codebook. Figure 6 This demonstrates that the method proposed in this invention has a significant advantage in computational complexity under multi-data-stream (multi-RF-chain) scenarios.
[0176] The above examples of the present invention are merely illustrative of the computational model and process of the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is impossible to exhaustively list all possible implementations here. Any obvious variations or modifications derived from the technical solutions of the present invention are still within the scope of protection of the present invention.
Claims
1. A hybrid precoding method for massive MIMO, characterized in that, The method specifically includes the following steps: Step 1: Construct the codebook matrix of the uniform linear antenna array response vector based on the transmitting antenna array. ; Step 2: Using the uniform linear antenna array response vector codebook matrix The codebook provides an equivalent formulation of the hybrid precoder design problem; The specific process of step 2 is as follows: Step 21: The receiver obtains channel state information through channel estimation. Then, the channel state information Perform singular value decomposition: in, It is a dimension of unitary matrix, Describes the rank of a matrix. It refers to the number of receiving antennas. Channel state information The diagonal matrix obtained by arranging the singular values in descending order. The dimension is , It is a dimension of unitary matrix; Step 22, and Represented as a block matrix: in, and yes The block matrix, and yes The block matrix; Pick As the target of optimal hybrid precoder; Step 23: Treat the design problem of the target optimal hybrid precoder as minimization and The Euclidean distance problem: in, It is a digital precoder. It is an RF chain analog pre-encoder. yes Norm, Is to make and The Euclidean distance is minimized in both digital pre-encoders and RF analog pre-encoders. express Each element in the array has a modulus of 1. It is a power constraint for the hybrid pre-encoder; Minimize and The sparse equivalence representation of the Euclidean distance problem is: in, Yes The dimension obtained by padding with zeros is The matrix, It is the length of the data stream, and satisfies... , It is the number of RFs. It is a zero norm; Step 24: Analyze the codebook matrix of the uniform linear antenna array response vector. By expanding, we obtain the auxiliary matrix. : Step 25, in Multiply the left side by the diagonal matrix ,get DFT matrix : in, ; Step 26: Rewrite the problem described in Step 23 as an equivalent problem as follows: Here, the superscript H represents the conjugate transpose of the matrix, and the intermediate variable is the auxiliary matrix. ; Step 3: Calculate the digital precoder based on the equivalent formulation of the hybrid precoder design problem in Step 2. and RF chain analog pre-encoder .
2. The hybrid precoding method for massive MIMO according to claim 1, characterized in that, The specific process of step 1 is as follows: Step 11: When the directions of all transmitting antennas satisfy the constraints, record the azimuth code of the antenna array as follows: , It is by , The vector formed It is a positive integer greater than or equal to 1. This refers to the number of transmitting antennas; The constraint is that the directions of each transmitting antenna are all within... Within the range; Step 12: Prepare the azimuth code book In Substituting each azimuth angle into the array response vector of the uniform linear antenna array ,in To indicate the azimuth angle, The array response vectors corresponding to each azimuth angle are combined to form a uniform linear antenna array response vector codebook matrix. .
3. The hybrid precoding method for massive MIMO according to claim 2, characterized in that, The uniform linear antenna array response vector codebook matrix for: in, , It is the base of the natural logarithm. It is the imaginary unit. It is an intermediate variable.
4. A hybrid precoding method for massive MIMO according to claim 3, characterized in that, The uniform linear antenna array response vector codebook matrix The l-th column is: in, It is the codebook matrix of the response vector of a uniform linear antenna array. The lth column.
5. A hybrid precoding method for massive MIMO according to claim 4, characterized in that, The specific process of step 3 is as follows: Step 31, through the analysis of... Obtain by performing IDFT operation : in, yes The inverse matrix; Step 32, according to Calculate the energy and distribution for each column vector, then find all peak points in the energy and distribution, and finally sort the peak points in descending order, recording the first one. The peak point of the position is at row index in ; in, Represents a diagonal matrix; Step 33: From the codebook matrix Find column vector index as All column vectors constitute the RF chain analog pre-encoder : Step 34: Using row index as well as Obtain digital precoder : in, Representative from Select row index as The matrix consisting of all row vectors, and Satisfying the energy constraints of the transmitting antenna .
6. A hybrid precoding method for massive MIMO according to claim 4, characterized in that, The specific process of step 3 is as follows: Step 31: Introduce an auxiliary matrix ,make ; Step 32, for Obtain by performing IDFT operation : Step 33: Find them separately The peak points of the energy distribution of each column vector in the array are then sorted in descending order, and the peak points of each column vector are recorded. The column vector index corresponding to the peak point of the bit. : Step 34: From the response codebook matrix of the uniform linear antenna array In the middle, find the column vector index as All column vectors constitute the analog pre-encoder From the auxiliary matrix Find column vector index as All column vectors constitute the auxiliary matrix : Step 35: Indexing by column vector as well as Obtain digital precoder : in, Representative from Select column vector index as A matrix consisting of all column vectors; Step 36: Update the auxiliary matrix: in, , It is the updated auxiliary matrix; Calculate the difference matrix for: Step 37: Determine whether the iteration stopping condition has been met; If the iteration stopping condition is not met, return to step 32 using the updated auxiliary matrix; If the iteration stopping condition is met, then the result obtained in the last iteration will be... As an analog pre-encoder for the RF chain, the last iteration obtained As a digital precoder.
7. A hybrid precoding method for massive MIMO according to claim 6, characterized in that, The iteration stopping condition is when the set maximum number of iterations is reached or It is less than the set threshold.