System secrecy rate maximization design in massive MIMO communication

By transforming the optimization problem in large-scale MIMO communication into unconstrained problems on complex manifolds, and using the Riemann gradient descent method combined with momentum information, the problem of slack error in the existing technology is solved, significantly improving the confidentiality of the system Rate.

CN119070867APending Publication Date: 2024-12-03UNIV OF ELECTRONICS SCI & TECH OF CHINA +1
View PDF 0 Cites 3 Cited by

Patent Information

Application Number
CN202411044148.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-31
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

Optimization problem of constant mode beamforming in existing large-scale MIMO communications. Due to the fractional form of CMC and objective function, it leads to non-convexity and NP difficulties. The existing methods solve the problem by relaxing constraints or objective functions, introducing slack errors.

Method used

The optimization problem is transformed into unconstrained problems on the complex manifold, and the descent direction is derived through the Riemann gradient descent method combined with momentum information to avoid slack mode and improve convergence speed.

Benefits of technology

It significantly improves the confidentiality rate of large-scale MIMO communication systems, avoids slack errors, and improves the efficiency of optimization solutions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119070867A_ABST
    Figure CN119070867A_ABST
Patent Text Reader

Abstract

The invention discloses a system secrecy rate maximization design in large-scale MIMO communication, and belongs to the field of MIMO communication. The method comprises the following steps: firstly, constructing an optimization objective function taking maximization of a system secrecy rate as an optimization objective, and then constructing a manifold # imgabs0 # based on a weighted vector of a transmitting end to meet constant modulus constraint, so that the constructed optimization objective function is projected to a manifold space and is converted into an optimization objective function with an unconstrained score; according to the method, a Dinkelbach method is used for converting an optimization objective function of an unconstrained fraction into an integer form, finally, an optimization result of a weighted vector of a base station is obtained based on iterative solution, a manifold method is used for solving constant modulus constraints, and a manifold optimization mode combined with momentum is provided. Compared with the prior art, the method has the advantage that the secrecy rate of the communication system is optimized with lower time complexity.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of MIMO communication, and in particular to a design method for maximizing system confidentiality rate in large-scale MIMO communication. Background Art

[0002] Massive MIMO beamforming plays an important role in improving the physical layer security of communication systems. Constant modulus beamforming technology can greatly save costs and is of great significance in engineering implementation. Therefore, the application of massive MIMO constant modulus beamforming in enhancing physical layer security has aroused widespread interest.

[0003] Due to the CMC (constant modulus constraint) and the fractional form of the objective function, the optimization problem is non-convex and NP-hard. Existing methods mainly solve the problem by relaxing the constraints or the objective function. They can be divided into two categories: relaxing the constraints and relaxing the objective function.

[0004] Among them, in the constraint relaxation method, a typical reference is the semidefinite relaxation (SDR) method proposed in the literature "G.Zhou, C.Pan, H.Ren, K.Wang and Z.Peng,"Secure Wireless Communication in RIS-Aided MISO System With HardwareImpairments," in IEEE Wireless Communications Letters, vol.10, no.6, pp.1309-1313, June 2021". This method first optimizes the covariance matrix of the beamforming weight vector. Then, the phase angle of the main eigenvector is approximated as the beamforming weight vector, which will lead to relaxation errors; the literature "G.Zhou, C.Pan, H.Ren, K.Wang and Z.Peng,"Secure Wireless Communication in RIS-Aided MISO System With Hardware Impairments," in IEEE Wireless Communications Letters, vol.10, no.6, pp.1309-1313, June 2021." noted that the covariance matrix is ​​subject to a rank-one constraint, and proposed an optimization method that approximates the rank-one constraint as a smooth function. However, due to function approximation, this method inevitably introduces errors.

[0005] In the relaxed objective function, the alternating direction multiplier method (ADMM) proposed in the literature "H.Niu, Z.Chu, F.Zhou, Z.Zhu, M.Zhang and K.-K.Wong,"Weighted Sum Secrecy Rate Maximization Using Intelligent ReflectingSurface," in IEEE Transactions on Communications, vol.69, no.9, pp.6170-6184, Sept.2021." relaxes the objective function by using some Lagrangian penalty parameters. The convergence of ADMM depends on the selection of Lagrangian penalty parameters, but choosing appropriate parameters is challenging. In order to avoid parameter selection, the paper "Y.Sun, P.Babu and DPPalomar,"Majorization-Minimization Algorithms in Signal Processing, Communications, and Machine Learning," in IEEE Transactions on Signal Processing, vol.65, no.3, pp.794-816, 1Feb.1, 2017." proposed a majorization-minimization (MM) method, which introduces a surrogate function to replace the objective function. However, it is difficult to select a suitable surrogate function.

[0006] That is, in the above-mentioned relaxation of constraints and objective functions to deal with the optimization problem of CMC, the relaxation method adopted introduces relaxation errors. Therefore, it is necessary to consider a method that does not require relaxation to solve the optimization problem in the constant mode beamforming process. Summary of the invention

[0007] The purpose of the present invention is to provide a design method for maximizing the confidentiality rate of a large-scale MIMO communication system based on Riemann conjugate gradient descent. In view of the problem of relaxation error introduced by solving the optimization problem by relaxing constraints or objective functions in the existing constant modulus beamforming processing, the present invention realizes constant modulus beamforming based on complex circular manifolds. First, the optimization problem is converted into an unconstrained problem on the complex circular manifold, and then the descent direction is derived by incorporating momentum into gradient information to realize constant modulus beamforming, thereby accelerating the convergence process of the optimization solution to improve the confidentiality rate of the large-scale MIMO communication system.

[0008] The technical solution adopted by the present invention is:

[0009] The system confidentiality rate maximization design in massive MIMO communication is used for a communication system including a base station, several legitimate communicators and several eavesdroppers, and includes the following steps:

[0010] Construct an optimization objective function with maximizing the system confidentiality rate as the optimization goal:

[0011]

[0012] Among them, the weight vector of the base station W = [w 1 ,...,w i ,...,w N ],w i represents the weight coefficient of the i-th transmitting antenna of the base station, N is the number of transmitting antennas of the base station, P represents the total transmitting power, σ 2 represents the noise power; denote the channels from the base station to the mth legitimate communicator and the lth eavesdropper respectively, represents the complex domain, M represents the number of legitimate communicators, L represents the number of eavesdroppers, N e represents the number of antennas of each eavesdropper, (·) H represents the conjugate transpose operation;

[0013] Constructing manifolds To satisfy the constant modulus constraint:

[0014] Project the constructed optimization objective function into the manifold space and transform it into an optimization objective function of unconstrained scores: Among them, f 1 (W) and f 2 (W) are two auxiliary quantities about the weighted vector W, and their expressions are:

[0015]

[0016] Among them, β 1 , β 2 are the preset weights of the legitimate communicator and the eavesdropper respectively, and I is the unit matrix;

[0017] The Dinkelbach method is used to convert the unconstrained score optimization objective function into an integer form, and then the optimization result of the weight vector W of the base station is obtained based on the iterative solution.

[0018] Furthermore, the Dinkelbach method is used to convert the unconstrained score optimization objective function into an integer form:

[0019]

[0020] Among them, φ(W) represents the optimization objective function in integer form, the initial value of the auxiliary parameter η is the preset value, η k+1 represents the updated value of the auxiliary parameter η at the k+1th iteration;

[0021] Based on the weight vector W updated after the kth iteration k The corresponding auxiliary amount f 1 (W k ), f 2 (W k ) to obtain the updated value of the auxiliary parameter η at the k+1th iteration

[0022] Based on η k+1 Get the weighted vector after the k+1th iteration update: W k+1 =f 1 (W)-η k+1 f 2 (W).

[0023] Furthermore, when iteratively solving the weighted vector W, the search direction (descent direction) is updated in combination with the momentum information and the gradient information, specifically including:

[0024] Update the search direction based on the descent of the Riemann gradient:

[0025] γ k = -grad(φ(W k ))+ζ k-1 *Trans(γ k-1 )

[0026] Among them, γ k represents the search direction at the kth iteration, grad(φ(W k )) indicates that φ(W) at point W k The Riemann gradient at, Trans(γ k-1 ) represents the search direction γ at the k-1th iteration k-1 In the corresponding tangent space

[0027] Among them, the Riemann gradient grad(φ(W k )) is expressed as:

[0028]

[0029] Where Re(·) represents the real part extraction function, and the Euclidean gradient of φ(W) Among them, ▽f 1 (W) and The auxiliary amount f 1 (W) and f 2 The gradient of (W);

[0030] Conjugate gradient parameter ζ k =grad(φ(W)) H *diff / Re(diff H *Trans(γ k-1 )), where the expression of the auxiliary parameter diff is: diff = grad (φ (W k ))-Trans(grad(φ(W k-1 ))));

[0031] Based on the current updated search step α new =c n *α, according to the formula W k ' +1 =W k +α new *γ k Get the weighted vector W after the k+1th iteration update k ' +1 , and then project it onto the manifold to obtain the final weighted vector W after the current iteration update k+1 ; Where α is the initial optimization step size of the current iteration, the value range of parameter c is (0,1), and n represents the number of searches per iteration;

[0032] When the preset iterative convergence condition is met, the most recently obtained weighted vector W k+1 As the optimization result of the weight vector W of the base station.

[0033] Furthermore, the adaptive step size method is used to update the initial optimization step size α for the next iteration. If the search times n = 1, then α = 2α new ; If the number of searches n = 2, then α = α new ; If the number of searches n is greater than 2, then α = 2α new .

[0034] Furthermore, the iterative convergence condition is |φ(W k+1 )-φ(W k )|Does not exceed the preset convergence threshold.

[0035] The technical solution provided by the present invention brings at least the following beneficial effects:

[0036] The present invention projects the optimization problem of modulus beamforming of a large-scale MIMO communication system onto a complex circular manifold to satisfy the constant modulus constraint, and uses the Dinkelbach method to solve the fractional problem after projection, and then designs the transmitter weight vector of the MIMO communication system based on the Riemann gradient descent method, thereby significantly improving the confidentiality rate of the large-scale MIMO communication system. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0038] Figure 1 is the relationship between the system confidentiality rate and the number of antennas in the embodiment;

[0039] Figure 2 is the relationship between the system confidentiality rate and the legal communicator in the embodiment;

[0040] Figure 3 is the relationship between the system confidentiality rate and the eavesdropper in the embodiment;

[0041] Figure 4 It is the relationship between the system confidentiality rate and the transmission power in the embodiment. DETAILED DESCRIPTION

[0042] To make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be described in detail and completely in conjunction with the drawings in the implementation of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings can be arranged and designed using different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the present application for protection, but merely represents selected embodiments of the present invention.

[0043] When implementing constant modulus beamforming for massive MIMO to enhance physical layer security, the resulting problem is non-convex and NP-hard due to the constant modulus constraint. Existing methods mainly solve the problem by relaxing constraints or objective functions, which inevitably introduces relaxation errors. The inventors of the embodiments of the present invention noticed during the implementation that the complex circular manifold naturally satisfies CMC. Furthermore, the optimization process on the complex circular manifold can be regarded as the movement of particles on a high-dimensional plane. Based on this feature, the embodiments of the present invention propose a manifold optimization method (MOCM) combined with momentum, that is, first converting the original problem into an unconstrained problem on the complex circular manifold, that is, projecting the optimization problem into the manifold space to avoid relaxation; then, the resulting problem is solved by updating the descent direction by combining gradient information and momentum information, so as to help maintain the descending trend of particles during the optimization process and accelerate the convergence speed, that is, by incorporating momentum into the gradient information to derive the descent direction, thereby accelerating the convergence process of the optimization solution of the transmitter weight vector of the MIMO communication system under the constant modulus constraint.

[0044] For the convenience of description, the following definitions are given first:

[0045] The transpose, conjugate, and conjugate transpose operations are respectively represented by (·) T ,(·) * ,(·) H express; Represent the complex number domain and the real number domain respectively; diag([x 1 ,x 2 ,…,x N ]) represents a diagonal matrix, where the diagonal elements are x 1 ,x 2 ,…,x N , N represents the number of diagonal elements; represents the expectation of a random variable; tr{·},vec(·),(·) -1 Represent the trace, column vectorization and inverse of the matrix respectively; Represent the real and imaginary parts of a complex-valued scalar / vector / matrix, respectively; ||·|| F ,|·| respectively represent the matrix binorm and the vector element-wise modulus; ⊙ represents the Hadamard product; the matrices I,1(0) respectively represent the identity matrix and the matrix of all 1s(0).

[0046] The design for maximizing the system confidentiality rate in large-scale MIMO communication provided by an embodiment of the present invention includes: first, converting the original optimization problem into an unconstrained fractional problem on a manifold; then, using the Dinkelbach method to convert the difficult fractional problem into an integer problem; in addition, in order to optimize the convergence performance, the descent direction is updated by combining momentum information and gradient information.

[0047] Consider a communication system with a base station, M legitimate communicators, and L eavesdroppers. The base station uses N transmitting antennas to send confidential information to the single-antenna legitimate communicator. In order to steal confidential information, each eavesdropper is equipped with N e Antennas.

[0048] The signal sent by the base station can be expressed as: x=Ws(t), where x represents the signal sent by the base station, s(t) represents the transmission waveform, t represents time, and W represents a weighting vector.

[0049] Therefore, the signal received by each legitimate communicator and each eavesdropper is:

[0050]

[0051] in, They represent the channels from the base station to the mth legitimate communicator and the lth eavesdropper respectively. represents additive Gaussian white noise, σ 2Represents the noise power.

[0052] The achievable rates of legitimate communicators and eavesdroppers can be expressed as:

[0053] R b,m =log(1+γ b,m );R e,l =log(1+γ e,l )

[0054] Among them, γ b,m ,γ e,l They represent the SNR (signal-to-noise ratio) of the legitimate communicator and the eavesdropper respectively.

[0055]

[0056] The confidentiality rate of this communication system is:

[0057]

[0058] Therefore, the optimization problem of maximizing the system confidentiality rate by massive MIMO technology and constant modulus beamforming can be expressed as:

[0059]

[0060] Among them, w i represents the weight coefficient of the i-th transmitting antenna, that is, the weight vector W = [w 1 ,...,w i ,...,w N ], P represents the total transmission power.

[0061] Due to the CMC and maximum functions in the definition of confidentiality rate, the optimization problem is non-convex and NP-hard. In order to solve this problem, the above optimization problem is transformed into:

[0062]

[0063] Among them, f 1 (W) and f 2 (W) are two auxiliary quantities about the weighted vector w, and their expressions are:

[0064]

[0065] Among them, β 1 , β 2 are the preset weights of legitimate communicators and eavesdroppers respectively, where β 1 , β 2 are all greater than 0, preferably, β 1 The order of magnitude is set to 10 -4 , β 2The order of magnitude is set to 10 -3 , for example, 1 =10 -4 , β 1 =8×10 -3 .

[0066] When solving the problem, first construct the manifold To satisfy the constant modulus constraint:

[0067] So The tangent space for: Among them, 0 N Represents an N-dimensional all-zero vector;

[0068] After projecting the original problem into the manifold space, it is transformed into an unconstrained fraction problem:

[0069] Applying Dinkelbach's method, the fraction problem can be reformulated as:

[0070]

[0071] The iterative update process of the parameter η and the weight vector W is as follows:

[0072]

[0073]

[0074] The subscripts of the parameter η and the weight vector W are used to identify the number of iterations.

[0075] Regarding the optimization process of W, in the embodiment of the present invention, a descent method based on Riemann gradient is proposed to achieve it, and the descent direction of the method is given by the following formula:

[0076] γ k = -grad(φ(W k ))+ζ k-1 *Trans(γ k-1 )

[0077] Among them, γ k represents the descent direction (i.e., search direction) at the kth iteration, grad(φ(W k )) indicates that φ(W) at point W k The Riemann gradient at, Trans(γ k-1 ) represents the descending direction γ at the k-1th iteration k-1 In the corresponding tangent space That is, momentum information during the optimization process;

[0078] Among them, the calculation formula of Riemann gradient is:

[0079]

[0080] Where Re(·) represents the real part extraction function, represents the Euclidean gradient of φ(W), which is expressed as:

[0081]

[0082] Among them, the auxiliary amount f 1 (W) and f 2 Gradient of (W) and is defined as:

[0083]

[0084]

[0085] In the descending direction, by introducing momentum information (Trans(γ k-1 ), the particle can accumulate velocity in the search space, thus maintaining a downward trend and converging quickly, ζ k Represents the Hestenes-Stiefel parameter (i.e., conjugate gradient parameter), which is used to balance the influence of momentum information. Its expression is:

[0086] ζ k =grad(φ(W)) H *diff / Re(diff H *Trans(γ k-1 ))

[0087] where diff is given by:

[0088] diff=grad(φ(W k ))-Trans(grad(φ(W k-1 ))))

[0089] Then, an adaptive method is used to obtain the search step size (i.e., the adaptive step size method), and a new iteration point is determined according to the descending direction and the search step size. In the descending direction, the new iteration point is:

[0090] W k+1 =W k +α new *γ k

[0091] Among them, α new Represents the Armijo search step, represented by α new =c n*α is given, where α is the initial optimization step size of the current iteration, c is the Armijo parameter, c needs to be greater than 0 and less than 1. Preferably, it is set to 0.5 in this embodiment, and n is the number of searches. Check whether the following inequality is satisfied during each search iteration. If not, add 1 to n.

[0092] φ(W k )<φ(W k-1 )+α new (grad(φ(W k )) H *γ k )

[0093] In order to speed up the search, the embodiment of the present invention updates the initial search parameters according to the number of searches n. If n = 1, the initial search step is too short. Set α = 2α new To obtain a faster descent speed; if n = 2, then α = α new ; If n>2, the linear search ends at α new Very small, set α = 2α new , to ensure the next descending speed.

[0094] Preferably, in the embodiment of the present invention, the pseudo code of the adaptive step size method is shown in Table 1:

[0095] Table 1

[0096]

[0097]

[0098] Finally, the output iteration points are projected onto the manifold:

[0099] W k+1 ←unt(W k+1 )

[0100] The function unt(·) represents:

[0101] unt([x 1 ,x 2 ,...,x N ] T )=[x 1 / |x 1 |,x 2 / |x 2 |,...,x N / |x N |] T

[0102] Among them, x 1 ,x 2 ,...,xN Corresponding to the iterative update output W k+1 N weight coefficients.

[0103] Preferably, in the embodiment of the present invention, the pseudo code of the optimization process for maximizing the system confidentiality rate is shown in Table 2:

[0104] Table 2

[0105]

[0106]

[0107] Example

[0108] In order to verify the optimization performance of the system confidentiality rate maximization design in the massive MIMO communication proposed in the embodiment of the present invention, the transmitting end weight vector is optimized using the existing scheme and the scheme proposed in the embodiment of the present invention.

[0109] The simulation experiment conditions are set as follows: consider a large-scale MIMO communication system, there are multiple legitimate communicators and eavesdroppers in the system, the number of transmitting antennas of the MIMO transmitter is N, and the channel of the communication system is the Rayleigh channel.

[0110] In order to evaluate the performance of the proposed method, this example compares the following methods:

[0111] (1) The method of using the minimax method for signal processing, referred to as the MM method: Reference "Y. Sun, P. Babu and DP Palomar, "Majorization-Minimization Algorithms in Signal Processing, Communications, and Machine Learning," in IEEE Transactions on Signal Processing, vol. 65, no. 3, pp. 794-816, 1Feb. 1, 2017."

[0112] (2) The method of using semi-definite relaxation method to design system confidentiality rate, referred to as SDR method: reference "S. Fan, J. Xu, J. Xue and Y. Zhang,"One-Bit Statistical ECSI Learning for Single-Group Multicast Secure Beamforming," in IEEE Communications Letters, vol. 26, no. 11, pp. 2552-2556, Nov. 2022.".

[0113] (3) The method of using the alternating direction multiplier method to maximize the system confidentiality rate, referred to as the ADMM method: Reference "Q.Li, C.Li and J.Lin,"Constant Modulus Secure Beamforming for Multicast MassiveMIMO Wiretap Channels," in IEEE Transactions on Information Forensics and Security, vol. 15, pp. 264-275, 2020.".

[0114] Figure 1 The relationship between the confidentiality rate and the total power transmitted by the base station is shown. It can be seen that the increase in antenna transmission power leads to a corresponding increase in the confidentiality rate of the entire system. In contrast, the performance of the proposed method (design for maximizing the confidentiality rate of the system in massive MIMO communication, proposed method) is better than MM, SDR and ADMM.

[0115] Figure 2 In the figure, the relationship between the number of base station transmitting antennas and the total confidentiality rate is shown. The total transmit power is set to 6dB. As the number of antennas increases, the average power on each antenna will decrease. However, as the number of antennas increases, the system confidentiality rate will increase. At the same time, the performance of the method proposed in the present invention is better than MM, SDR and ADMM.

[0116] Figure 3 and Figure 4 The performance gains for different numbers of legitimate communicators and eavesdroppers are shown respectively. The confidentiality rate decreases with the number of Bob and Eve. The system confidentiality rate maximization design in large-scale MIMO communication proposed in the embodiment of the present invention maintains the best performance in a variety of scenarios. The method proposed in the embodiment of the present invention uses the Dinkelbach method to solve the fractional problem, the manifold method to solve the constant modulus constraint, and the manifold optimization method combined with momentum to solve the weight vector optimization problem at the transmitter. Experiments show that compared with the existing method, the scheme adopted in the embodiment of the present invention achieves superior performance with lower time complexity.

[0117] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

[0118] The above are only some embodiments of the present invention. For those skilled in the art, several modifications and improvements can be made without departing from the creative concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A design for maximizing system confidentiality rate in massive MIMO communication, for a communication system including a base station, several legitimate communicators and several eavesdroppers, characterized in that: The following steps are involved: Construct an optimization objective function with maximizing the system confidentiality rate as the optimization goal: s.t.|w i |=1,i=1,...N, m=1,...,M l=1,...,L Among them, the weight vector of the base station W = [w1,...,w i ,...,w N ],w i represents the weight coefficient of the i-th transmitting antenna of the base station, N is the number of transmitting antennas of the base station, P represents the total transmitting power, σ 2 represents the noise power; denote the channels from the base station to the mth legitimate communicator and the lth eavesdropper respectively, represents the complex domain, M represents the number of legitimate communicators, L represents the number of eavesdroppers, N e represents the number of antennas of each eavesdropper, (·) H represents the conjugate transpose operation; Constructing manifolds To satisfy the constant modulus constraint: Project the constructed optimization objective function into the manifold space and transform it into an optimization objective function of unconstrained scores: Among them, f1(W) and f2(W) are two auxiliary quantities about the weighted vector W, and their expressions are: Among them, β1 and β2 are the preset weights of the legitimate communicator and the eavesdropper respectively, and I is the unit matrix; The Dinkelbach method is used to convert the unconstrained score optimization objective function into an integer form, and then the optimization result of the weight vector W of the base station is obtained based on the iterative solution.

2. The system confidentiality rate maximization design in massive MIMO communication as claimed in claim 1, characterized in that: The Dinkelbach method is used to convert the unconstrained score optimization objective function into an integer form: Among them, φ(W) represents the optimization objective function in integer form, the initial value of the auxiliary parameter η is the preset value, η k+1 represents the updated value of the auxiliary parameter η at the k+1th iteration; Based on the weight vector W updated after the kth iteration k The corresponding auxiliary amount f1(W k )、f2(W k ) to obtain the updated value of the auxiliary parameter η at the k+1th iteration Based on η k+1 Get the weighted vector after the k+1th iteration update: W k+1 =f1(W)-η k+1 f2(W).

3. The system confidentiality rate maximization design in massive MIMO communication as claimed in claim 2, characterized in that: When iteratively solving the weighted vector W, the search direction is updated by combining momentum information and gradient information, including: Update the search direction based on the descent of the Riemann gradient: c k =-grad(φ(W k ))+g k-1 *Trans(g k-1 ) Among them, γ k represents the search direction at the kth iteration, grad(φ(W k )) indicates that φ(W) at point W k The Riemann gradient at, Trans(γ k-1 ) represents the search direction γ at the k-1th iteration k-1 In the corresponding tangent space Among them, the Riemann gradient grad(φ(W k )) is expressed as: Where Re(·) represents the real part extraction function, and the Euclidean gradient of φ(W) in, and are the gradients of the auxiliary quantities f1(W) and f2(W) respectively; Conjugate gradient parameter ζ k =grad(φ(W)) H *diff / Re(diff H *Trans(γ k-1 )), where the expression of the auxiliary parameter diff is: diff = grad (φ (W k ))-Trans(grad(φ(W k-1 )))); Based on the current updated search step α new =c n *α, according to the formula W k ' +1 =W k +α new *γ k Get the weighted vector W after the k+1th iteration update k ' +1 , and then project it onto the manifold to obtain the final weighted vector W after the current iteration update k+1 ; Where α is the initial optimization step size of the current iteration, the value range of parameter c is (0,1), and n represents the number of searches per iteration; When the preset iterative convergence condition is met, the most recently obtained weighted vector W k+1 As the optimization result of the weight vector W of the base station.

4. The system confidentiality rate maximization design in massive MIMO communication as claimed in claim 3, characterized in that: Adaptive step size method is used to update the initial optimization step size α for the next iteration. If the search times n = 1, then α = 2α new ; If the number of searches n = 2, then α = α new ; If the number of searches n is greater than 2, then α = 2α new .

5. The system confidentiality rate maximization design in massive MIMO communication as claimed in claim 1, characterized in that: The iterative convergence condition is |φ(W k+1 )-φ(W k )|Does not exceed the preset convergence threshold.

6. The system confidentiality rate maximization design in massive MIMO communication as claimed in claim 1, characterized in that: The order of magnitude of β1 is set to 10 -4 , the order of magnitude of β2 is set to 10 -3 .

Citation Information

Cited By

  • Physical layer safety communication method and system based on fluid antenna

    CN121173339A

  • Coding matrix solving method based on iterative conjugate gradient for digital communication

    CN121864265A

  • An iterative conjugate gradient based encoding matrix solution method for digital communication

    CN121864265B