A hybrid beamforming method for massive machine-type communications in millimeter wave systems

Abstract

The present invention discloses a hybrid beamforming method for massive machine-type communications in a millimeter wave system. The present invention uses random matrix theory to approximate the system traversal rate into a form that is easy to handle, and converts the hybrid beamforming design problem based on the maximum traversal rate into a series of iterative sub-problems through an alternating optimization method, and uses a semi-positive definite relaxation method to convert the sub-problems into a convex optimization problem for solution. In addition, the present invention uses the central limit theorem to approximate the user signal-to-interference-noise ratio distribution to a Gaussian distribution, and converts the hybrid beamforming design optimization problem based on the maximum user access probability into a series of iterative sub-problems through alternating optimization and Dinklebach transformation methods, and uses a convex hull relaxation method and a conditional gradient descent method to convert the sub-problems into a convex optimization problem and solve it. Compared with traditional hybrid beamforming design methods, the present invention can significantly improve the system traversal rate and user access probability.

Description

Technical Field

[0001] The present invention belongs to the field of communications, and in particular relates to a hybrid beamforming design method for massive machine type communications in a millimeter wave system. Background Art

[0002] Massive machine type communication is one of the key scenarios of the fifth generation (5G) mobile communication, in which a large number of low-cost, low-energy, sparsely active machine type devices request access to the network. With the rapid development of mobile Internet of Things, emerging massive machine type communication application scenarios, such as intelligent transportation and smart cities, have the transmission requirements of large data packets. Therefore, the sixth generation (6G) mobile communication system must support massive connections and broadband services at the same time. Since the spectrum resources in the Sub-6 GHz band are very limited, it is difficult to meet the growing spectrum resource requirements of emerging massive machine type communication scenarios. Considering the rich spectrum resources in the millimeter wave band (30-300GHz), an effective solution is to apply massive machine type communication to the millimeter wave band to achieve massive connections and broadband services.

[0003] In millimeter wave systems, due to the short wavelength of signal carriers, large-scale antenna arrays can be deployed in the receiver, and beamforming technology can be used to increase signal power to combat the severe channel fading of millimeter wave systems. Traditional all-digital beamforming technology requires each antenna to be equipped with an RF link, resulting in high hardware cost and high power consumption. In recent years, hybrid beamforming technology has attracted great attention and has been widely used in millimeter wave systems. Hybrid beamforming technology only requires a small number of RF links to connect to large-scale antenna arrays, achieving performance comparable to all-digital beamforming with less hardware cost and lower power consumption. Summary of the invention

[0004] Technical problem: The purpose of the present invention is to provide a hybrid beamforming design method for massive machine type communications in a millimeter wave system. The present invention can improve the traversal rate of massive machine type communications in a millimeter wave system; improve the system user access probability of massive machine type communications in a millimeter wave system; and dynamically adjust the hybrid beamforming vector as the distribution of machine type devices in the system changes.

[0005] Technical solution: To achieve the above-mentioned invention object, the hybrid beamforming design method for massive machine type communication of a millimeter wave system described in the present invention comprises the following steps:

[0006] Step 1: construct a signal transmission model for massive machine-type communications in a millimeter wave system, and use the statistical information of the user signal arrival angle estimated by the base station to propose optimization problems for hybrid beamforming designs based on maximum ergodic rate and maximum user access probability.

[0007] Step 2: Use random matrix theory to approximate the system ergodic rate into a form that is easy to handle. Use the alternating optimization method to transform the hybrid beamforming design problem based on the maximum ergodic rate into a series of iterative subproblems. Use the semi-positive definite relaxation method to transform the subproblems into convex optimization problems for solution.

[0008] Step 3: Use the central limit theorem to approximate the user signal-to-noise ratio distribution as a Gaussian distribution. Use alternating optimization and Dinklebach transformation to transform the hybrid beamforming design optimization problem based on the maximum user access probability into a series of iterative subproblems. Use the convex hull relaxation method and conditional gradient descent method to transform the subproblems into a convex optimization problem and solve it.

[0009] in:

[0010] In the massive machine-type communication of the millimeter wave system in step 1, the user is equipped with a single antenna, and the base station adopts a fully connected hybrid beamforming structure consisting of K RF links and M r The uplink receive beamforming system is composed of a uniform linear antenna array. Each RF link is equipped with a phase shifter module connected to all antennas. The phase shifter realizes uplink receive beamforming by changing the phase of the received signal of each antenna. The phase shifter coefficient of each antenna corresponding to each RF link is called a hybrid beamforming vector.

[0011] The signal transmission model of the massive machine-type communication of the millimeter wave system in step 1 is as follows:

[0012] Let y k is the received signal vector on the kth RF link, which can be expressed as

[0013]

[0014] The superscript T represents the transpose operation of the matrix, the superscript H represents the conjugate transpose operation of the matrix, and v k represents the hybrid beamforming vector of the kth RF link, X = [x1,…,x N ] T , x n represents the transmitted signal of the nth user, Z represents the system noise, N is the number of system users, H=[ξ1h1,…,ξ N h N ],ξ n is the nth user active indicator variable, ξ n =1 means the nth user is active, ξ n =0 means the nth user is inactive; h n represents the channel response vector between the nth user and the base station antenna array, which can be written as

[0015]

[0016] Where L represents the number of channel paths, represents the channel gain of the lth path of the nth user, which obeys a complex Gaussian distribution with a mean of 0 and a variance of 1. is the arrival angle of the nth user on the lth path, The array response vector represents the signal, and its expression can be written as

[0017]

[0018] in, e is a natural constant, λ is the signal carrier wavelength, d is the antenna spacing of the uniform linear antenna array, and Extending the above formula to K RF links, that is, the received signal Y of K RF links on the base station side can be written as

[0019] Y=VHX+VZ

[0020] Where Y = [y1,…,y K ] T represents the received signal matrix, V = [v1,…,v K ] H Hybrid beamforming matrix.

[0021] The statistical information of the user signal arrival angle in step 1 specifically refers to the probability distribution function of the user signal arrival angle, which is obtained by the base station through an angle estimation algorithm based on the uplink received signal over a long period of time.

[0022] The optimization problem construction method based on the hybrid beamforming design with maximum ergodic rate and maximum user access probability in step 1 comprises the following specific steps:

[0023] Step 1.1, establish the optimization problem of hybrid beamforming design based on maximum ergodic rate. The system ergodic rate R(V) can be expressed as

[0024]

[0025] Among them, det[·] represents the operation of finding the determinant of the matrix, It means taking the expected operation on H, the superscript -1 means taking the inverse operation on the matrix, and log means taking the logarithm with base 2. represents the system noise variance, S represents the covariance matrix of the user's transmitted signal, where S can be approximately regarded as a unit matrix, I K represents the unit matrix of dimension K*K, Therefore, the optimization problem of hybrid beamforming design based on maximum ergodic rate is Can be written as

[0026]

[0027]

[0028] Here, |·| represents the modulus operation on a complex number, arg max represents the corresponding variable value when the objective function takes the maximum value, and st is the abbreviation of subject to, which means "satisfy the following constraints." Symbols This holds true for any value. k (m) represents v k The mth element in .

[0029] Step 1.2, construct an optimization problem based on the hybrid beamforming design with the maximum user access probability. When the signal interference noise ratio of the user signal is greater than a certain threshold, it can be considered to have successfully accessed. The signal interference noise ratio of the nth user can be written as

[0030]

[0031] Among them, ||·||2 represents the operation of taking the two norms, represents the expectation of the square of the mutual correlation coefficient between the nth user and the n'th user, represents the expected operation, therefore, the optimization problem of hybrid beamforming design based on the maximum probability of system user access is Can be written as

[0032]

[0033]

[0034] Among them, r min Indicates the signal-to-interference-noise ratio threshold for successful user access. Represents the probability of an event occurring.

[0035] In step 2, the system ergodic rate is approximated into a form that is easy to handle by using random matrix theory, and the hybrid beamforming design problem based on the maximum ergodic rate is converted into a series of iterative sub-problems by an alternating optimization method. The sub-problems are converted into convex optimization problems by using a semi-positive definite relaxation method for solving, which specifically includes the following steps:

[0036] Step 2.1, use random matrix theory to approximate the system ergodic rate into a form that is easy to handle. According to the deterministic equivalent method and Jensen inequality, the hybrid beamforming design optimization problem based on the maximum ergodic rate can be approximated as

[0037]

[0038]

[0039] in Indicates M r *M r dimensional discrete Fourier transform matrix, the subscript N of the above matrix a Indicates that its value increases with the number of active users N a And changes, the same below, The nth element in and The mth element in Can be written as

[0040]

[0041]

[0042] U T =[u T,1 ,u T,2 ,…,u T,N ]=I N It is an N*N dimensional unit matrix. and The expressions are

[0043]

[0044]

[0045] Diag(·) means converting a vector into a main diagonal matrix. is the channel information statistical matrix, and Respectively represent the length N a The all-1 vector and length is NN a All zero vectors, g can be written as

[0046]

[0047] The superscript * indicates the conjugate operation, and diag(·) indicates the operation of taking the diagonal elements of a matrix to generate a vector;

[0048] Step 2.2, transforming the hybrid beamforming design problem based on the maximum ergodic rate into a series of iterative sub-problems through an alternating optimization method, and transforming the hybrid beamforming design problem based on the maximum ergodic rate into a series of iterative sub-problems. The specific steps are as follows:

[0049] Step 2.2.1, fix the hybrid beamforming matrix V and calculate Given the hybrid beamforming matrix V, it can be found by iterative calculation and The value of is then calculated using the following formula

[0050]

[0051] in, N a ∈{0,1,…,N} represents the number of system users N a The probability of taking a value, α represents the user's active probability. Here, it is assumed that the active probability of each user is the same. It means to take N non-repeating numbers from N numbers. a The number of combinations;

[0052] Step 2.2.2, Fixation Solve the hybrid beamforming matrix V, fix K-1 hybrid beamforming vectors, and let V k =[v1,…,v k-1 ,v k+1 ,…,v K ] H , then the hybrid beamforming vector v k The optimization subproblem of the design can be written as

[0053]

[0054]

[0055] Among them, the superscript Indicates taking a non-negative square root operation on each element in the matrix, X k Can be written as

[0056]

[0057] Step 2.3, using a semi-definite relaxation method to convert the hybrid beamforming vector v in step 2.2.2 of claim 6 k The optimization sub-problem of the design is transformed into a convex optimization problem for solution; The optimization problem It can be relaxed by semidefinite relaxation method to

[0058]

[0059]

[0060] Tr{·} represents taking the trace of the matrix. k (m,m) indicates medium Q k The above optimization problem is a convex optimization problem, which can be solved by common optimization methods such as gradient descent method. express The matrix obtained by solving is The rank is 1, then it can be decomposed into The hybrid beamforming vector v k Can be written as in Represents the matrix element normalization operation, if If the rank is not 1, the hybrid beamforming vector v can be obtained by Gaussian randomization method k .

[0061] In step 3, the user signal to noise ratio distribution is approximated to a Gaussian distribution by using the central limit theorem, and the hybrid beamforming design optimization problem based on the maximum user access probability is transformed into a series of iterative sub-problems by using alternating optimization and Dinklebach transformation method, and the sub-problems are transformed into convex optimization problems and solved by using convex hull relaxation method and conditional gradient descent method, which specifically includes the following steps:

[0062] Step 3.1: Use the central limit theorem to approximate the user signal-to-noise ratio distribution to a Gaussian distribution. At this time, the system user access probability can be written as

[0063]

[0064] Where erf(·) represents the Gaussian error function,

[0065]

[0066]

[0067]

[0068]

[0069]

[0070]

[0071] Step 3.2, the hybrid beamforming design optimization problem based on the maximum user access probability is transformed into a series of iterative sub-problems through alternating optimization and Dinklebach transformation. Assuming that the user access probability is greater than 50%, that is, Fixed K-1 hybrid beamforming vectors, then the hybrid beamforming vector v k The optimization subproblem of the design can be written as

[0072]

[0073]

[0074] in, The above fractional optimization problem can be transformed into

[0075]

[0076]

[0077] in,

[0078]

[0079]

[0080]

[0081]

[0082] The variable y is updated by

[0083]

[0084] [t] represents the tth iteration;

[0085] Step 3.3, using the convex hull relaxation method and the conditional gradient descent method, the series of iterative subproblems in step 3.2 are transformed into convex optimization problems for solution. The convex hull relaxation method can be used to convert the iterative subproblems into convex optimization problems. Relaxation to a convex optimization problem

[0086]

[0087]

[0088] The above problem can be solved by conditional gradient descent method;

[0089] Finally, we need to find the subproblem through the following optimization problem The feasible solution is:

[0090]

[0091]

[0092] The above optimization problem needs to be transformed into a convex optimization problem through a semi-positive definite relaxation method:

[0093]

[0094]

[0095] The above optimization problem is a convex optimization problem, which can be solved by common optimization methods such as gradient descent method. Finally, the hybrid beamforming vector v is obtained by matrix decomposition or Gaussian randomization method. k .

[0096] Beneficial effects: Compared with the prior art, the present invention has the following advantages:

[0097] The present invention can improve the traversal rate of massive machine-type communications in millimeter wave systems;

[0098] The present invention can improve the system user access probability of massive machine-type communications in the millimeter wave system;

[0099] The present invention can dynamically adjust the hybrid beamforming vector as the distribution of machine-type devices in the system changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0100] Figure 1 It is a flow chart of a specific implementation mode of the present invention;

[0101] Figure 2 A structural diagram of a uniform linear antenna array configured at a base station side in a specific implementation manner of the present invention;

[0102] Figure 3 This is a diagram of a fully connected hybrid beamforming structure configured on the base station side in a specific implementation manner of the present invention. DETAILED DESCRIPTION

[0103] The present invention is further explained below in conjunction with the accompanying drawings and specific embodiments. It should be understood that these examples are only used to illustrate the present invention and are not used to limit the scope of the present invention. After reading the present invention, various equivalent forms of modifications to the present invention by those skilled in the art all fall within the scope defined by the claims attached to this application.

[0104] The hybrid beamforming method for massive machine-type communications in a millimeter wave system of the present invention is as follows: Figure 1-3 As shown, the specific steps include:

[0105] Step 1: construct a signal transmission model for massive machine-type communications in a millimeter wave system, and use the statistical information of the user signal arrival angle estimated by the base station to propose optimization problems for hybrid beamforming designs based on maximum ergodic rate and maximum user access probability.

[0106] Step 2: Use random matrix theory to approximate the system ergodic rate into a form that is easy to handle. Use the alternating optimization method to transform the hybrid beamforming design problem based on the maximum ergodic rate into a series of iterative subproblems. Use the semi-positive definite relaxation method to transform the subproblems into convex optimization problems for solution.

[0107] Step 3: Use the central limit theorem to approximate the user signal-to-noise ratio distribution as a Gaussian distribution. Use alternating optimization and Dinklebach transformation to transform the hybrid beamforming design optimization problem based on the maximum user access probability into a series of iterative subproblems. Use the convex hull relaxation method and conditional gradient descent method to transform the subproblems into a convex optimization problem and solve it.

[0108] Preferably: a hybrid beamforming method for massive machine-type communication in a millimeter wave system in step 1, characterized in that: in the massive machine-type communication in the millimeter wave system in step 1, the user is equipped with a single antenna, and the base station adopts a fully connected hybrid beamforming structure, which consists of K radio frequency links and M r uniform linear antenna array, such as Figure 2 Each RF chain is equipped with a phase shifter module connected to all antennas, such as Figure 3 The phase shifter achieves uplink receive beamforming by changing the phase of the signal received by each antenna. The phase shifter coefficients of each antenna corresponding to each RF link are called hybrid beamforming vectors.

[0109] Preferably, the signal transmission model of the massive machine-type communication of the millimeter wave system in step 1 is as follows.

[0110] Let y k is the received signal vector on the kth RF link, which can be expressed as

[0111]

[0112] The superscript T represents the transpose operation of the matrix, the superscript H represents the conjugate transpose operation of the matrix, and v k represents the hybrid beamforming vector of the kth RF link, X = [x1,…,x N ] T , x n represents the transmitted signal of the nth user, Z represents the system noise, N is the number of system users, H=[ξ1h1,…,ξ N h N ],ξ n is the nth user active indicator variable, ξ n =1 means the nth user is active, ξ n =0 means the nth user is inactive; h n represents the channel response vector between the nth user and the base station antenna array, which can be written as

[0113]

[0114] Where L represents the number of channel paths, represents the channel gain of the lth path of the nth user, which obeys a complex Gaussian distribution with a mean of 0 and a variance of 1. is the arrival angle of the nth user on the lth path, The array response vector represents the signal, and its expression can be written as

[0115]

[0116] in, e is a natural constant, λ is the signal carrier wavelength, and d is the antenna spacing of the uniform linear antenna array. Extending the above formula to K RF links, that is, the received signal Y of K RF links on the base station side can be written as

[0117] Y=VHX+VZ

[0118] Where Y = [y1,…,y K ] T represents the received signal matrix, V = [v1,…,v K ] H Hybrid beamforming matrix.

[0119] Preferably, the statistical information of the user signal arrival angle in step 1 specifically refers to the probability distribution function of the user signal arrival angle, which is obtained by the base station through an angle estimation algorithm based on the uplink received signal over a long period of time.

[0120] Preferably, the optimization problem construction method based on the hybrid beamforming design with maximum ergodic rate and maximum user access probability in step 1 comprises the following specific steps:

[0121] Step 1.1, establish the optimization problem of hybrid beamforming design based on maximum ergodic rate. The system ergodic rate R(V) can be expressed as

[0122]

[0123] Among them, det[·] represents the operation of finding the determinant of the matrix, It represents the expectation operation on H, the superscript -1 represents the inverse operation of the matrix, and log represents the logarithm operation with base 2. represents the system noise variance, S represents the covariance matrix of the user's transmitted signal, where S can be approximately regarded as a unit matrix, I K represents the unit matrix of dimension K*K, Therefore, the optimization problem of hybrid beamforming design based on maximum ergodic rate is Can be written as

[0124]

[0125]

[0126] Here, |·| represents the modulus operation on a complex number, arg max represents the corresponding variable value when the objective function takes the maximum value, and st is the abbreviation of subject to, which means "satisfy the following constraints." Symbols This holds true for any value. k (m) represents v k The mth element in .

[0127] Step 1.2, construct an optimization problem based on the hybrid beamforming design with the maximum user access probability. When the signal interference noise ratio of the user signal is greater than a certain threshold, it can be considered to have successfully accessed. The signal interference noise ratio of the nth user can be written as

[0128]

[0129] Among them, ||·||2 represents the operation of taking the two norms, represents the expectation of the square of the mutual correlation coefficient between the nth user and the n'th user, Therefore, the optimization problem of hybrid beamforming design based on the maximum probability of system user access is Can be written as

[0130]

[0131]

[0132] Among them, r min Indicates the signal-to-interference-noise ratio threshold for successful user access. Represents the probability of an event occurring.

[0133] Preferably, in step 2, the system ergodic rate is approximated into a form that is easy to handle by using random matrix theory, the hybrid beamforming design problem based on the maximum ergodic rate is converted into a series of iterative sub-problems by an alternating optimization method, and the sub-problems are converted into convex optimization problems by using a semi-positive definite relaxation method for solving, which specifically includes the following steps:

[0134] Step 2.1, use random matrix theory to approximate the system ergodic rate into a form that is easy to handle. According to the deterministic equivalent method and Jensen inequality, the hybrid beamforming design optimization problem based on the maximum ergodic rate can be approximated as

[0135]

[0136]

[0137] in Indicates M r *M r dimensional discrete Fourier transform matrix. The subscript N of the above matrix is a Indicates that its value increases with the number of active users N a The same applies to changes below. The nth element in and The mth element in Can be written as

[0138]

[0139]

[0140] U T =[u T,1 ,u T,2 ,…,u T,N ]=I N It is an N*N dimensional unit matrix. and The expressions are

[0141]

[0142]

[0143] Diag(·) means converting a vector into a main diagonal matrix. is the channel information statistical matrix, and Respectively represent the length N a The all-1 vector and length is NN a All zero vectors, g can be written as

[0144]

[0145] The superscript * indicates a conjugate operation, and diag(·) indicates an operation of generating a vector by taking the diagonal elements of a matrix.

[0146] Step 2.2, transform the hybrid beamforming design problem based on maximum ergodic rate into a series of iterative sub-problems through the alternating optimization method. The hybrid beamforming design problem based on maximum ergodic rate is transformed into a series of iterative sub-problems. The specific steps are as follows:

[0147] Step 2.2.1, fix the hybrid beamforming matrix V and calculate Given the hybrid beamforming matrix V, it can be found by iterative calculation and The value of is then calculated using the following formula

[0148]

[0149] in, N a ∈{0,1,…,N} represents the number of system users N a The probability of taking a value, α represents the user's active probability. Here, it is assumed that the active probability of each user is the same. It means to take N non-repeating numbers from N numbers. a The number of combinations;

[0150] Step 2.2.2, Fixation Solve the hybrid beamforming matrix V, fix K-1 hybrid beamforming vectors, and let V k =[v1,…,v k-1 ,v k+1 ,…,v K ] H , then the hybrid beamforming vector v k The optimization subproblem of the design can be written as

[0151]

[0152]

[0153] Among them, the superscript Indicates taking a non-negative square root operation on each element in the matrix, X k Can be written as

[0154]

[0155] Step 2.3, using a semi-definite relaxation method to convert the hybrid beamforming vector v in step 2.2.2 of claim 6 k The optimization sub-problem of the design is transformed into a convex optimization problem for solution; The optimization problem It can be relaxed by semidefinite relaxation method to

[0156]

[0157]

[0158] Tr{·} represents taking the trace of the matrix. k (m,m) indicates medium Q k The above optimization problem is a convex optimization problem, which can be solved by common optimization methods such as gradient descent method. express The matrix obtained by solving is The rank is 1, then it can be decomposed into The hybrid beamforming vector v k Can be written as in Represents the matrix element normalization operation, if If the rank is not 1, the hybrid beamforming vector v can be obtained by Gaussian randomization method k .

[0159] Preferably: in step 3, the user signal to interference and noise ratio distribution is approximated to a Gaussian distribution by using the central limit theorem, the hybrid beamforming design optimization problem based on the maximum user access probability is transformed into a series of iterative sub-problems by alternating optimization and Dinklebach transformation method, and the sub-problems are transformed into convex optimization problems and solved by using the convex hull relaxation method and the conditional gradient descent method, which specifically includes the following steps:

[0160] Step 3.1, use the central limit theorem to approximate the user signal to noise ratio distribution to a Gaussian distribution. At this time, the system user access probability can be written as

[0161]

[0162] Where erf(·) represents the Gaussian error function,

[0163]

[0164]

[0165]

[0166]

[0167]

[0168]

[0169] Step 3.2, the hybrid beamforming design optimization problem based on the maximum user access probability is transformed into a series of iterative sub-problems through alternating optimization and Dinklebach transformation. Assuming that the user access probability is greater than 50%, that is, Fixed K-1 hybrid beamforming vectors, then the hybrid beamforming vector v k The optimization subproblem of the design can be written as

[0170]

[0171]

[0172] in, The above fractional optimization problem can be transformed into

[0173]

[0174]

[0175] in,

[0176]

[0177]

[0178]

[0179]

[0180] The variable y is updated by

[0181]

[0182] [t] represents the tth iteration.

[0183] Step 3.3, using the convex hull relaxation method and the conditional gradient descent method, the series of iterative subproblems in step 3.2 are transformed into convex optimization problems for solution. Using the convex hull relaxation method, the iterative subproblems can be Relaxation to a convex optimization problem

[0184]

[0185]

[0186] The above problem can be solved by conditional gradient descent method.

[0187] Finally, we need to find the subproblem through the following optimization problem The feasible solution is:

[0188]

[0189]

[0190] The above optimization problem needs to be transformed into a convex optimization problem through a semi-positive definite relaxation method:

[0191]

[0192]

[0193] The above optimization problem is a convex optimization problem, which can be solved by common optimization methods such as gradient descent method. Finally, the hybrid beamforming vector v is obtained by matrix decomposition or Gaussian randomization method. k .

[0194] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A hybrid beamforming method for massive machine-type communications in a millimeter wave system, characterized by: The following steps are involved: Step 1: construct a signal transmission model for massive machine-type communications in a millimeter wave system, and use the statistical information of the user signal arrival angle estimated by the base station to propose optimization problems for hybrid beamforming designs based on maximum ergodic rate and maximum user access probability. Step 2: Use random matrix theory to approximate the system ergodic rate into a form that is easy to handle. Use the alternating optimization method to transform the hybrid beamforming design problem based on the maximum ergodic rate into a series of iterative subproblems. Use the semi-positive definite relaxation method to transform the subproblems into convex optimization problems for solution. Step 3: The user signal-to-interference-noise ratio distribution is approximated as a Gaussian distribution using the central limit theorem. The hybrid beamforming design optimization problem based on the maximum user access probability is transformed into a series of iterative sub-problems through alternating optimization and Dinklebach transformation. The sub-problems are transformed into a convex optimization problem and solved using the convex hull relaxation method and the conditional gradient descent method. The optimization problem construction method based on the hybrid beamforming design with maximum ergodic rate and maximum user access probability in step 1 comprises the following specific steps: Step 1.1, establish the optimization problem of hybrid beamforming design based on maximum ergodic rate. The system ergodic rate R(V) is expressed as Among them, det[·] represents the operation of finding the determinant of the matrix, It means taking the expected operation on H, the superscript -1 means taking the inverse operation on the matrix, and log means taking the logarithm with base 2. represents the system noise variance, S represents the covariance matrix of the user's transmitted signal, where S is approximately considered as a unit matrix, and I K represents the unit matrix of dimension K*K, Therefore, the optimization problem of hybrid beamforming design based on maximum ergodic rate is Written Among them, |·| represents the modulus operation on complex numbers, arg max represents the corresponding variable value when the objective function takes the maximum value, st is the abbreviation of subject to, which means "satisfy the following constraints", and the symbol This means that for any value, v k (m) represents v k The mth element in ; Step 1.2, construct an optimization problem based on the hybrid beamforming design with the maximum user access probability. When the signal interference noise ratio of the user signal is greater than a certain threshold, it is considered to be successfully accessed. The signal interference noise ratio of the nth user is written as Among them, ||·||2 represents the operation of taking the two norms, represents the expectation of the square of the mutual correlation coefficient between the nth user and the n'th user, represents the expected operation, therefore, the optimization problem of hybrid beamforming design based on the maximum probability of system user access is Written Among them, r min Indicates the signal-to-interference-noise ratio threshold for successful user access. It indicates the probability of an event occurring; In step 2, the system ergodic rate is approximated into a form that is easy to handle by using random matrix theory, which specifically includes: Step 2.1, according to the deterministic equivalence method and Jensen inequality, the hybrid beamforming design optimization problem based on maximum ergodic rate is approximated as in Indicates M r *M r dimensional discrete Fourier transform matrix, the subscript N of the above matrix a Indicates that its value increases with the number of active users N a And changes, the same below, The nth element in and The mth element in Written U T =[u T,1 ,u T,2 ,…,u T,N ]=I N It is an N*N dimensional unit matrix; and The expressions are Diag(·) means converting a vector into a main diagonal matrix. is the channel information statistical matrix, and Respectively represent the length N a The vector of all 1s and the length is NN a All zero vectors, g is written as The superscript * indicates a conjugate operation, and diag(·) indicates an operation of generating a vector by taking the diagonal elements of a matrix.

2. The hybrid beamforming method for massive machine type communications in a millimeter wave system according to claim 1, characterized in that: In the massive machine-type communication of the millimeter wave system in step 1, the user is equipped with a single antenna, and the base station adopts a fully connected hybrid beamforming structure consisting of K RF links and M r The system consists of a uniform linear antenna array, and each RF link is equipped with a phase shifter module connected to all antennas; The phase shifter achieves uplink receive beamforming by changing the phase of the received signal of each antenna. The phase shifter coefficient of each antenna corresponding to each RF link is called a hybrid beamforming vector.

3. The hybrid beamforming method for massive machine-type communications in a millimeter wave system according to claim 1, characterized in that: The signal transmission model of the massive machine-type communication of the millimeter wave system in step 1 is as follows: Let y k is the received signal vector on the kth RF link, which is expressed as The superscript T represents the transpose operation of the matrix, the superscript H represents the conjugate transpose operation of the matrix, and v k represents the hybrid beamforming vector of the kth RF link, X = [x1,…,x N ] T , x n represents the transmitted signal of the nth user, Z represents the system noise, N is the number of system users, H=[ξ1h1,…,ξ N h N ],ξ n is the nth user active indicator variable, ξ n =1 means the nth user is active, ξ n =0 means the nth user is inactive; h n represents the channel response vector between the nth user and the base station antenna array, written as Where L represents the number of channel paths, represents the channel gain of the lth path of the nth user, which obeys a complex Gaussian distribution with a mean of 0 and a variance of 1. is the arrival angle of the nth user on the lth path, The array response vector represents the signal, and its expression is written as in, e is a natural constant, λ is the signal carrier wavelength, d is the antenna spacing of the uniform linear antenna array, and Extending the above formula to K RF links, that is, the received signal Y of K RF links on the base station side can be written as Y=VHX+VZ Where Y = [y1,…,y K ] T represents the received signal matrix, V = [v1,…,v K ] H Hybrid beamforming matrix.

4. The hybrid beamforming method for massive machine-type communications in a millimeter wave system according to claim 1, characterized in that: The statistical information of the user signal arrival angle in step 1 specifically refers to the probability distribution function of the user signal arrival angle, which is obtained by the base station through an angle estimation algorithm based on the uplink received signal over a long period of time.

5. The hybrid beamforming method for massive machine type communications in a millimeter wave system according to claim 1, characterized in that: In step 2, the hybrid beamforming design problem based on the maximum ergodic rate is converted into a series of iterative sub-problems by an alternating optimization method, and the sub-problems are converted into convex optimization problems by a semi-positive definite relaxation method for solving, which specifically includes the following steps: Step 2.2, transforming the hybrid beamforming design problem based on the maximum ergodic rate into a series of iterative sub-problems through an alternating optimization method, and transforming the hybrid beamforming design problem based on the maximum ergodic rate into a series of iterative sub-problems. The specific steps are as follows: Step 2.2.1, fix the hybrid beamforming matrix V and calculate Given the hybrid beamforming matrix V, we can find it through iterative calculation and The value of is then calculated using the following formula in, N a ∈{0,1,…,N} represents the number of system users N a The probability of taking a value, α represents the user's active probability. Here, it is assumed that the active probability of each user is the same. It means to take N non-repeating numbers from N numbers. a The number of combinations of numbers; Step 2.2.2, Fixation Solve the hybrid beamforming matrix V, fix K-1 hybrid beamforming vectors, and let V k =[v1,…,v k-1 ,v k+1 ,…,v K ] H , then the hybrid beamforming vector v k The optimization subproblem of the design is written as Among them, the superscript Indicates taking a non-negative square root operation on each element in the matrix, X k Written Step 2.3, using the semi-definite relaxation method, the hybrid beamforming vector v in step 2.2.2 is k The optimization sub-problem of the design is transformed into a convex optimization problem for solution; The optimization problem The relaxation is done by semidefinite relaxation method. The optimization problem The relaxation is done by semidefinite relaxation method: Tr{·} represents the trace of the matrix; Q k (m,m) indicates medium Q k The element in the mth row and the mth column; the above optimization problem is a convex optimization problem, which can be solved by common optimization methods such as gradient descent method. express The matrix obtained by solving is The rank is 1, then it decomposes into The hybrid beamforming vector v k Written in Represents the matrix element normalization operation, if If the rank is not 1, the hybrid beamforming vector v is obtained by Gaussian randomization method k .

6. The hybrid beamforming method for massive machine type communications in a millimeter wave system according to claim 1, characterized in that: In step 3, the user signal to noise ratio distribution is approximated to a Gaussian distribution by using the central limit theorem, and the hybrid beamforming design optimization problem based on the maximum user access probability is transformed into a series of iterative sub-problems by using alternating optimization and Dinklebach transformation method, and the sub-problems are transformed into convex optimization problems and solved by using convex hull relaxation method and conditional gradient descent method, which specifically includes the following steps: Step 3.1: Use the central limit theorem to approximate the user signal-to-noise ratio distribution to a Gaussian distribution. At this time, the system user access probability is written as Where erf(·) represents the Gaussian error function, Step 3.2, the hybrid beamforming design optimization problem based on the maximum user access probability is transformed into a series of iterative sub-problems through alternating optimization and Dinklebach transformation method. Assuming that the user access probability is greater than 50%, that is, Fixed K-1 hybrid beamforming vectors, then the hybrid beamforming vector v k The optimization subproblem of the design is written as in, The above optimization subproblem is transformed into in, The variable y is updated by [t] represents the tth iteration; Step 3.3, using the convex hull relaxation method and the conditional gradient descent method to transform the series of iterative subproblems in step 3.2 into convex optimization problems for solution, and using the convex hull relaxation method to transform the iterative subproblems Relaxation to a convex optimization problem The above problem is solved by conditional gradient descent method; Finally, we need to find the subproblem through the following optimization problem The feasible solution is: The above optimization problem needs to be transformed into a convex optimization problem through a semi-positive definite relaxation method: The above optimization problem is a convex optimization problem, which is solved by common optimization methods such as gradient descent method. Finally, the hybrid beamforming vector v is obtained by matrix decomposition or Gaussian randomization method. k .

Images