Combined precoding and beamforming method for RIS-assisted cellular-removal large-gauge MIMO under electromagnetic interference

Abstract

The invention discloses a joint distributed precoding and beamforming method for RIS-assisted cellular removal large-gauge MIMO (Multiple Input Multiple Output) under electromagnetic interference (EMI). Specifically, the method comprises the following steps: constructing an RIS-assisted cellular-removal large-scale MIMO downlink model under electromagnetic interference; a distributed processing framework based on joint optimization is provided, and a combined vector, a transmitting end active precoding matrix and RIS passive beam forming are received through collaborative design; an efficient distributed alternating optimization algorithm is developed, and a minimum system weighted mean square error (MSE) is taken as a target function. Compared with a traditional centralized algorithm, the scheme provided by the invention has the advantage that the calculation complexity is reduced while the similar performance is kept. A simulation experiment shows that compared with other reference schemes, the scheme can effectively improve the system performance under electromagnetic interference, and the reliability and practicability of the method in a complex electromagnetic environment are verified.

Description

Technical Field

[0001] This invention belongs to the field of communication technology, specifically relating to a joint precoding and beamforming method for RIS-assisted decellularization of large-scale MIMO under electromagnetic interference. Background Technology

[0002] With the rapid development of wireless communication technology, decellularized massive MIMO and reconfigurable smart surface (RIS) technology have become key technologies for realizing future fifth-generation (B5G) and even sixth-generation (6G) networks. Decellularized massive MIMO effectively solves the edge effects and interference problems in traditional cellular networks by distributively deploying a large number of access points and utilizing coherent signal processing technology to provide seamless coverage and high spectral efficiency for multiple user devices. Meanwhile, RIS, as an emerging electromagnetic wave modulation technology, consists of a large number of low-cost, programmable passive reflective elements, which can dynamically reconfigure the wireless propagation environment, significantly improving signal coverage and suppressing multi-user interference. Therefore, this invention considers combining the advantages of decellularized massive MIMO and RIS to further improve communication performance.

[0003] It is worth noting that most existing research on RIS-assisted decellularized massive MIMO systems is often based on idealized electromagnetic environment assumptions, neglecting the impact of electromagnetic interference (EMI) prevalent in real-world scenarios. This necessitates considering EMI in addition to noise from inter-user interference and other noise sources. Therefore, in the design and analysis of RIS-assisted decellularized massive MIMO systems, fully considering the impact of EMI and researching corresponding interference suppression and channel enhancement strategies are crucial for improving the system's robustness and performance in real-world environments. Summary of the Invention

[0004] This invention addresses the technical problems existing in the prior art by providing a joint precoding and beamforming method for RIS-assisted decellularized large-scale MIMO under electromagnetic interference. This method incorporates consideration of the impact of electromagnetic interference and aims to provide a novel joint distributed precoding and beamforming method for the downlink of RIS-assisted decellularized large-scale MIMO under electromagnetic interference. This addresses the performance impact caused by electromagnetic interference between RISs, improves the sum and rate at the user end, and makes the communication system more practical and robust.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A joint precoding and beamforming method for RIS-assisted decellularization of large-scale MIMO under electromagnetic interference includes the following steps:

[0007] Step (1) Establish a RIS-assisted decellularized large-scale MIMO communication system model under electromagnetic interference;

[0008] Step (2) Based on the RIS-assisted decellularized large-scale MIMO communication system model under electromagnetic interference, the signal-to-interference-plus-noise ratio of user k is calculated, and the mean square error of user k is calculated.

[0009] Step (3) takes minimizing the mean square error of the user as the objective function and jointly optimizes the passive beamforming parameters of the precoding vector, the merging vector and the RIS.

[0010] Step (4) Design a distributed joint precoding and beamforming framework, solve the joint optimization problem, and determine the passive beamforming parameters of the precoding vector, merging vector and RIS to optimize the system energy efficiency under electromagnetic interference conditions.

[0011] Preferably, the RIS-assisted decellularized large-scale MIMO communication system under electromagnetic interference in step (1) includes: a physical model of electromagnetic interference at the RIS: Where A r This represents the area of ​​each RIS element. R represents the EMI power at RIS. r Representing the spatial correlation matrix of RIS, the downlink transmission channel model from the l-th AP to the k-th user: in, It is the direct channel from APl to user k; It is the channel from RISr to user k; It is the channel from AP1 to RISr; Φ r =diag{φ r,1 ,…,φ r,M} is the phase shift matrix of the r-th RIS. φ r,M Let |φ| represent the phase shift of the m-th element of the r-th RIS. For an ideal RIS, |φ| r,m |≤1. We assume the local channel state information (CSI) at the l-th access point, i.e. The location is completely known, (·) H Indicates the conjugate transpose symbol. Let H represent the set of complex numbers. Further, we simplify the equivalent channel, i.e., the H of a single-antenna user k. (l,k) ,get Where we define vec(·) represents the vector operator.

[0012] Preferably, in step (2), we assume that the symbol sent to user k is s. k Transmission power In the downlink, sk Precoding is performed at the l-th access point (AP) initially. Therefore, the signal expression after precoding at the l-th AP is: Meanwhile, to meet the transmit power constraints of each access point, the precoding vector w (l,k) Must meet: Among them, P l,max Let be the maximum transmit power of the l-th AP, and all APs have the same maximum transmit power. This represents the expected operator. The signal expression y received by the k-th user. k The specific formula is as follows: The forms comprising the desired signal, inter-user interference, and noise are clearly defined: Where we define n k The additive white Gaussian noise (AWGN) at user k is... I n This represents an n-dimensional identity matrix. For the desired signal, Interference between users and n k These represent the EMI noise interference reflected at the RIS and the noise interference at the user's location, respectively. Furthermore, user k receives signal y... k Then, use combined vectors. After signal processing, we obtain: Substitute y k get:

[0013] Preferably, the mean square error (MSE) at user k in step (3) can be expressed as: in, Let the real part operation be represented. The relationship between the user mean square error and the user rate is known as follows: Therefore, we can solve the weighted sum rate (WSR) maximization problem by obtaining a locally optimal solution that minimizes the user's weighted mean square error. The minimum mean square error optimization problem P1 for user k is expressed as:

[0014]

[0015] Wherein, the precoding matrix at AP is Φ = diag{Φ1,…,Φ R} indicates that all elements in the RIS achieve beamforming of the RIS through phase shifting.

[0016] Preferably, the distributed joint precoding and beamforming framework proposed in step (4) includes the following steps:

[0017] Step (4.1) uses an alternating optimization algorithm to decompose the energy efficiency optimization problem after conversion into three sub-problems and solve the optimization solutions for each iteration;

[0018] Step (4.2) solve for u for a given (w, Φ). k Omitting the irrelevant terms of the objective function, the minimization problem (P1) of the MSE can be reformulated as follows: To obtain the MMSE receive vector u k The optimal closed-form solution, differentiated and with the gradient set to 0, yields: Where, a k b k c k Defined as:

[0019] Step (4.3) for a given (u) k ,Φ), solve w l,k In solving the precoding vector w l,k The distributed projective gradient descent (DPGD) method was used in the process. Regarding the precoding w... l,k The MSE subproblem is denoted as:

[0020]

[0021] To adapt to the distributed framework architecture, each Apl only optimizes the local precoding vector w. l,k The objective function can be decomposed into: Among them, w l =[w l,1 ,…,w l,K ] is the local precoding matrix of AP1, for w l,k The gradient is obtained as follows: First, an initial pre-encoding is randomly obtained for each AP1 and user k. satisfy Gradient updates are performed. After the i-th iteration, APl calculates the local gradient and updates the precoder: Where η is the step size, which can be set using Armijo line search or a fixed value. To ensure that the power constraint condition is met, the updated precoder is normalized; Projected onto its power constraint set ||w l,i || 2 ≤P l,max ,have When the precoding change in adjacent iterations satisfies It may stop when the maximum number of iterations is reached.

[0022] Step (4.4) gives (u) k w l,k To solve for Φ, in order to simplify the MSE subproblem, we introduce the equivalent channel of the combined downlink: Where d d d r They are defined as follows: The phase adjustment information of the RIS phase shift matrix is ​​compressed into a row vector to facilitate subsequent calculations. Define...

[0023] Furthermore, the ADMM algorithm is used to solve the constant mode constraint problem of RIS. We redefine the subproblem of beamforming for passive RIS as...

[0024]

[0025] in, This is the CSI required by the AP. For ease of calculation, we define:

[0026]

[0027] To further solve the aforementioned RIS beamforming subproblem P3 with constant mode constraints, this invention introduces the augmented Lagrangian method to construct a new optimization objective and employs the penalty function dual decomposition (PDD) algorithm to alternately update the variables to obtain an approximate optimal solution. This is achieved by introducing new auxiliary variables. This method transforms the original objective function into the following form.

[0028]

[0029] Here, μ is the dual variable, and ξ>0 is a pre-defined penalty factor. This multivariate problem can be solved by alternately updating each variable while keeping other variables constant. The specific steps are as follows:

[0030] Update θ: given μ and When the problem is a convex quadratic form, it can be solved efficiently using an optimizer: renew Given θ and μ, then Each element can be obtained through phase projection: Update μ: given θ and The dual variable is updated as follows: Benefiting from the rearrangement of the objective function, the beamforming design of passive RIS can be computed centrally on the CPU.

[0031] Step (4.5) Repeat steps (4.2)-(4.5). Terminate the loop when the objective function converges to obtain the optimal solution (w, Φ, u). k The parameters are used to calculate user k information and rate.

[0032] Compared with the prior art, the advantages of the present invention are as follows:

[0033] (1) This invention differs from the existing RIS-assisted decellularized large-scale MIMO joint optimization method by incorporating the impact of electromagnetic interference on the channel, thereby enabling the modeled RIS-assisted decellularized large-scale MIMO model to achieve system performance that is closer to the actual situation, thus making it more practical.

[0034] (2) Based on the established communication model, this invention calculates the signal-to-interference-plus-noise ratio received by user k under electromagnetic interference, thereby calculating the mean square error of user k. With minimizing the mean square error as the objective, a corresponding joint optimization problem is established and the corresponding problem is solved, thereby realizing the quantitative evaluation and optimization of system performance indicators under electromagnetic interference.

[0035] (3) Most existing technologies use traditional centralized joint optimization frameworks to solve problems, which have high computational complexity. This invention proposes a joint distributed precoding and beamforming optimization framework. By iteratively optimizing three key parameters—precoding vector, merging vector, and RIS passive beamforming—it obtains local optimal solutions to subproblems while ensuring algorithm convergence. This effectively maximizes the weighted sum rate while reducing algorithm complexity.

[0036] (4) Simulation experiments verify that the optimization method proposed in this invention can converge quickly under different numbers of users and different RIS scales, and can significantly improve the downlink speed and rate of the system under electromagnetic interference. At the same time, compared with the existing benchmark scheme, this method can effectively suppress electromagnetic interference while maintaining low computational complexity, and is more suitable for deployment in actual wireless communication scenarios. Attached Figure Description

[0037] Figure 1 This is a schematic diagram of a RIS-assisted decellularized massive MIMO downlink system model under electromagnetic interference according to the method of the present invention;

[0038] Figure 2 This is a flowchart of the method for joint optimization of electrical distributed systems proposed in this invention;

[0039] Figure 3 The graph shows the relationship between the downlink speed and the number of iterations.

[0040] Figure 4 This is a graph showing the relationship between average downlink speed and the number of users.

[0041] Figure 5 This is a comparison chart of downlink speed convergence in large-scale and small-scale scenarios. Detailed Implementation

[0042] The present invention will now be described in further detail.

[0043] Example: This invention specifically provides a RIS-assisted decellularized massive MIMO joint precoding and beamforming method under electromagnetic interference, effectively suppressing the effects of electromagnetic interference. A detailed description is provided below with reference to the accompanying drawings.

[0044] Please see Figure 1 This invention relates to a RIS-assisted decellularized massive MIMO downlink system under electromagnetic interference conditions. The invention considers a RIS-assisted decellularized MIMO communication system with K users, R RIS sets, and L AP sets. The user set, RIS set, and AP set are denoted as follows: Where k∈K, r∈R, Let N represent the index sets of users, RIS, and APs, respectively. The number of antennas for the l-th AP and the number of antennas for the k-th user are N and N, respectively. t N r All APs are connected to the CPU via a fronthaul link and provide services to all users using the same time-frequency resources. Each RIS has M elements used for phase-shifting the incident signal. Furthermore, this invention posits that in communication systems, in addition to general additive noise, electromagnetic interference noise also exists at the RIS.

[0045] Furthermore, we established a physical model of electromagnetic interference (EMI). Spatial dependency is eliminated only when the elements of the RIS are arranged in a straight line and the spacing between adjacent elements is an integer multiple of λ / 2. However, this configuration is unrealizable in practical two-dimensional RIS. Therefore, the effect of spatial dependency between RIS elements on EMI needs to be considered. Both the RIS size and the spatial dependency between elements affect the intensity of EMI, and the stronger the spatial dependency, the greater the EMI. In RIS-assisted decellularized massive MIMO systems, EMI at the RIS has a significant and non-negligible impact on communication performance. The EMI at the r-th RIS is defined as:

[0046]

[0047] Among them, A r This represents the area of ​​each RIS element. R represents the EMI power at RIS. r Represents the spatial correlation matrix of RIS. This represents a cyclically symmetric complex Gaussian distribution.

[0048] Furthermore, we derive the downlink transmission portion from the l-th AP to the k-th user as follows:

[0049]

[0050] in, It is the direct channel from APl to user k; It is the channel from RISr to user k; It is the channel from AP1 to RISr; Φ r =diag{φ r,1 ,…,φ r,M} is the phase shift matrix of the r-th RIS. φ r,M Let |φ| represent the phase shift of the m-th element of the r-th RIS. For an ideal RIS, |φ| r,m |≤1. We assume the local channel state information (CSI) at the l-th AP, i.e. The location is completely known, (·) H Indicates the conjugate transpose symbol. The set of complex numbers is represented by diag{·}, and the matrix is ​​represented by diagonal matrix.

[0051] Furthermore, simplifying the equivalent channel, i.e., the H of a single-antenna user k. (l,k) ,get

[0052]

[0053] Where we define vec(·) represents the vector operator, (·) T This represents the conjugate transpose symbol.

[0054] Furthermore, we assume that the symbol sent to user k is s. k Transmission power In the downlink, s k Precoding is performed at the l-th access point (AP) initially. Therefore, the signal expression after precoding at the l-th AP is:

[0055]

[0056] Meanwhile, to meet the transmit power constraints of each access point, the precoding vector w (l,k) Must meet:

[0057]

[0058] Among them, P l,maxLet be the maximum transmit power of the l-th AP, and all APs have the same maximum transmit power. This indicates the expected operator.

[0059] Furthermore, the signal expression y received by the k-th user k The specific formula is as follows:

[0060]

[0061] The forms comprising the desired signal, inter-user interference, and noise are clearly defined:

[0062]

[0063] Where we define n k The additive white Gaussian noise (AWGN) at user k is... I n This represents an n-dimensional identity matrix. For the desired signal, Interference between users and n k These represent EMI noise interference reflected from the RIS and noise interference at the user's location, respectively.

[0064] Furthermore, user k receives signal y k Then, receive the merged vector. After signal processing, we obtain:

[0065]

[0066] Substitute y k get:

[0067]

[0068] Furthermore, the signal-to-interference-plus-noise ratio (SINR) at user k under electromagnetic interference is:

[0069]

[0070] in It is the desired signal power. It is multi-user interference power. It is the noise interference power of EMI. It is additive noise power, ||·|| 2 It is the Euclidean norm.

[0071] Since the above SINR expression is non-convex, the precoding vector w and the merging vector u are jointly optimized. kThe RIS phase shift matrix Φ is typically difficult to solve. Fortunately, inspired by transceiver design algorithms and the known relationship between user mean square error (MSE) and user rate, i.e. We can adopt a compromise approach by using an alternating optimization algorithm to find the local optimum of the weighted sum rate (WSR) problem by minimizing the user-weighted mean square error (MSE). The specific optimization problem is described below.

[0072] Furthermore, the MSE at user k can be expressed as:

[0073]

[0074] in, This indicates the operation of the real part.

[0075] Furthermore, this invention proposes a novel joint precoding and beamforming algorithm framework, whereby the problem of minimizing the weighted sum of MSE at user k can be formulated as follows:

[0076]

[0077] Wherein, the precoding matrix at AP is Φ = diag{Φ1,…,Φ R} indicates that all elements in the RIS achieve beamforming of the RIS through phase shifting.

[0078] The specific steps are as follows:

[0079] Regarding the receiving merge vector: With (w, Φ) fixed, solve for u. k

[0080] Given (w, Φ) and omitting irrelevant terms, the minimum problem P1 of the MSE can be reformulated as:

[0081]

[0082] To obtain the MMSE receive vector u k The optimal closed-form solution, differentiated and with the gradient set to 0, yields:

[0083]

[0084] Among them, a k b k c k They are defined as follows:

[0085]

[0086] Regarding active precoding: fixed (u k ,Φ), solve wl,k

[0087] Solving the precoding vector w l,k The distributed projective gradient descent (DPGD) method was employed during the process. This method conforms to the decellularized massive MIMO communication model, requiring no global information exchange; each AP only needs its local CSI and the feedback from the user terminal. The user terminal's feedback regarding the precoding w... l,k The MSE subproblem is denoted as:

[0088]

[0089] To adapt to the distributed framework architecture, each Apl only optimizes the local precoding vector w. l,k The objective function can be decomposed into:

[0090]

[0091] Among them, w l =[w l,1 ,…,w l,K ] is the local precoding matrix of AP1.

[0092] Next to w l,k The gradient is obtained as follows:

[0093]

[0094] First, each AP1 and user k obtains the initial pre-encoding. And satisfy power constraints After the i-th iteration, APl calculates the local gradient and updates the precode as follows:

[0095]

[0096] Where η is the step size, which can be determined using Armijo line search or set to a fixed value.

[0097] To meet power constraints, the updated precoding is normalized; Projected onto the power constraint set ||w l,i || 2 ≤P l,max Specifically:

[0098]

[0099] When the precoding change in adjacent iterations satisfies Alternatively, stop iterating when the maximum number of iterations is reached.

[0100] Regarding beamforming: Fixed (u k w l,kSolve for Φ

[0101] To simplify the MSE subproblem, an equivalent channel combining the downlink is introduced:

[0102]

[0103] Where d d d r They are defined as follows: The phase adjustment information of the RIS phase shift matrix is ​​compressed into a row vector for easier subsequent calculations. Define...

[0104] Next, we use the ADMM algorithm to solve the constant mode constraint problem of RIS. We redefine the subproblem of beamforming for passive RIS as:

[0105]

[0106]

[0107] in, This represents the CSI required from the AP. For simplified calculation, it is defined as follows:

[0108]

[0109] To further solve the aforementioned RIS beamforming problem with constant mode constraints, this paper introduces the augmented Lagrangian method to construct a new optimization objective function and employs the penalty function dual decomposition (PDD) algorithm to obtain an approximate optimal solution by alternately updating the variables. This is achieved by introducing new auxiliary variables. The original objective function P3 is transformed into the following form.

[0110]

[0111] Where μ is the dual variable and ξ>0 is the preset penalty factor.

[0112] This multivariate problem P4 can be solved by alternately updating each variable while keeping other variables constant. The specific steps are as follows:

[0113] Update θ: given μ and When the problem is a convex quadratic form, it can be solved efficiently using an optimizer:

[0114]

[0115] renew Given θ and μ, then Each element can be obtained through phase projection:

[0116]

[0117] Update μ: given θ and The dual variable is updated as follows:

[0118]

[0119] Substitute the obtained optimal solution into the rate formula to calculate the user's downlink rate.

[0120] The simulation experiment in this embodiment was conducted using MATLAB R2023a software. In this simulation, the user is positioned at (0, 0, 0)m by default, and five APs equipped with four antennas are deployed, evenly distributed along the x-axis in a plane with y = -50m and z = 3m. The x-coordinate increases from -80m to 80m in 40m increments, and two RISs equipped with 100 reflective elements are located at coordinates (30, 10, 6)m and (-30, 10, 6)m.

[0121] See Figure 3 The electromagnetic interference (EMI) model incorporated in this invention significantly impacts the downlink rate. Furthermore, under EMI conditions, the proposed distributed joint optimization method effectively improves the system rate, demonstrating good interference robustness and optimization performance. Moreover, the method achieves rapid and stable convergence within approximately five iterations, verifying its low computational complexity while maintaining performance, making it suitable for real-world, fast-response communication scenarios.

[0122] See Figure 4 Comparing downlink and speed with different numbers of users, the proposed distributed joint optimization method can significantly improve downlink and speed when the number of users is small. However, as the number of users increases, the interference level increases positively correlated with the number of users. Limited by electromagnetic interference, the gain in downlink and speed gradually weakens and eventually tends to saturate.

[0123] See Figure 5 The proposed joint optimization method under electromagnetic interference can be configured in both small-scale and large-scale systems, with 2 users, 2 APs, and 16 reflective elements in a small-scale system, and 10 users, 10 APs, and 4 RIS with 100 reflective elements in a large-scale system. This demonstrates the high feasibility of the proposed optimization method.

[0124] It should be noted that the above embodiments are not intended to limit the scope of protection of the present invention. Equivalent transformations or substitutions made based on the above technical solutions all fall within the scope of protection of the claims of the present invention.

Claims

1. A joint precoding and beamforming method for RIS-assisted decellularization of large-scale MIMO under electromagnetic interference, characterized in that, Includes the following steps: Step (1) Establish a RIS-assisted decellularized large-scale MIMO communication system model under electromagnetic interference; Step (2) Based on the RIS-assisted decellularized large-scale MIMO communication system model under electromagnetic interference, the signal-to-interference-plus-noise ratio of user k is calculated, and the mean square error of user k is calculated. Step (3) takes minimizing the mean squared error of the user as the objective function and jointly optimizes the distributed precoding vector w and the merging vector u. k The passive beamforming parameter Φ of RIS; Step (4) Design the joint distributed precoding and beamforming framework, solve the joint optimization problem, and determine the distributed precoding vector w and the merging vector u. k The passive beamforming parameter Φ of RIS is used to optimize system performance under electromagnetic interference conditions.

2. The method for joint precoding and beamforming of RIS-assisted decellularization of large-scale MIMO under electromagnetic interference as described in claim 1, characterized in that, The RIS-assisted decellularized large-scale MIMO communication system under electromagnetic interference in step (1) includes: a physical model of electromagnetic interference at the RIS: Where A r This represents the area of ​​each RIS element. R represents the EMI power at RIS. r Represents the spatial correlation matrix of RIS; downlink transmission channel model from the l-th AP to the k-th user: in, It is the direct channel from APl to user k; It is the channel from RISr to user k; It is the channel from AP1 to RISr; Φ r =diag{φ r,1 ,…,φ r,M } is the phase shift matrix of the r-th RIS; φ r,M Let |φ| represent the phase shift of the m-th element of the r-th RIS. For an ideal RIS, |φ| r,m |≤1; Assume the local channel state information (CSI) at the l-th AP, i.e. The location is completely known, (·) H Indicates the conjugate transpose symbol. Let represent the set of complex numbers, and let diag{·} denote a diagonal matrix; further, simplify the equivalent channel, i.e., the H of a single-antenna user k. (l,k) ,get Where, definition vec(·) represents the vector operator, (·) T This represents the conjugate transpose symbol.

3. The method for joint precoding and beamforming of RIS-assisted decellularization of large-scale MIMO under electromagnetic interference as described in claim 1, characterized in that, In step (2), it is assumed that the symbol sent to user k is s. k Transmission power In the downlink, s k Precoding is performed at the l-th access point (AP) initially. Therefore, the signal expression after precoding at the l-th AP is: Meanwhile, to meet the transmit power constraints of each AP, the precoding vector w (l,k) Must meet: Among them, P l,max Let be the maximum transmit power of the l-th AP, and all APs have the same maximum transmit power. This represents the expected operator, the signal expression y received by the k-th user. k It can be represented as: The forms comprising the desired signal, inter-user interference, and noise are clearly defined: Among them, the definition n k For additive noise at user k, I n Represents an n-dimensional identity matrix. For the desired signal, Interference between users and n k These represent the EMI noise interference reflected from the RIS and the noise interference at the user's location, respectively. The user k receives the signal y... k Then, use combined vectors. Signal processing yields: Substitute y k get: The mean square error (MSE) at user k is expressed as: in, This indicates the operation of the real part.

4. The method for joint precoding and beamforming of RIS-assisted decellularization of large-scale MIMO under electromagnetic interference as described in claim 1, characterized in that, The minimum mean square error problem P1 for user k in step (3) is expressed as follows: Wherein, the precoding matrix at AP is Φ = diag{Φ1,…,Φ R } indicates that all elements in the RIS achieve beamforming of the RIS through phase shifting.

5. The method for joint precoding and beamforming of RIS-assisted decellularization of large-scale MIMO under electromagnetic interference as described in claim 1, characterized in that, The proposed layout-based joint precoding and beamforming framework in step (4) includes the following steps: Step (4.1) uses an alternating optimization algorithm to decompose the energy efficiency optimization problem after conversion into three sub-problems and solve the optimization solutions for each iteration; Step (4.2) solve for u for a given (w, Φ). k Omitting the irrelevant terms of the objective function, the minimization problem (P1) of the MSE can be reformulated as follows: To obtain the MMSE receive vector u k The optimal closed-form solution, differentiated and with the gradient set to 0, yields: Where, a k b k c k Defined as: Step (4.3) for a given (u) k ,Φ), solve w l,k In solving the precoding vector w l,k In the process, the Distributed Projective Gradient Descent (DPGD) method was used, regarding the precoding w l,k The MSE subproblem is denoted as: To adapt to the distributed framework architecture, each Apl only optimizes the local precoding vector w. l,k The objective function can be decomposed into: Among them, w l =[w l,1 ,…,w l,K ] is the local precoding matrix of AP1, for w l,k The gradient is obtained as follows: First, an initial pre-encoding is randomly obtained for each AP1 and user k. satisfy Gradient updates are performed. After the i-th iteration, APl calculates the local gradient and updates the precoder: Where η is the step size, which is set using Armijo line search or a fixed value. To ensure that the power constraint condition is met, the updated precoder is normalized; Projected onto its power constraint set ||w l,i || 2 ≤P l,max ,have When the precoding change in adjacent iterations satisfies Or stop when the maximum number of iterations is reached; Step (4.4) for a given (u) k w l,k To solve for Φ, in order to simplify the MSE subproblem, an equivalent channel combining the downlink is introduced: Where d d d r They are defined as follows: The phase adjustment information of the RIS phase shift matrix is ​​compressed into a row vector to facilitate subsequent calculations. Define... By using the ADMM algorithm to solve the constant mode constraint problem of RIS, the subproblem of beamforming of passive RIS is redefined as... in, It is the CSI required by AP, which defines: To further solve the aforementioned RIS beamforming subproblem P3 with constant mode constraints, an augmented Lagrangian method is introduced to construct a new optimization objective. The penalty function dual decomposition (PDD) algorithm is then used to alternately update the variables to obtain an approximate optimal solution. This is achieved by introducing new auxiliary variables. This method transforms the original objective function into the following form. Where μ is the dual variable and ξ>0 is the preset penalty factor, this multivariate problem is solved by alternately updating each variable while keeping other variables fixed. The specific steps are as follows: Update θ: given μ and When the optimization problem is a convex quadratic problem, an optimizer can be used to solve it efficiently. renew Given θ and μ, then Each element can be obtained through phase projection: Update μ: given θ and The dual variable is updated as follows: Benefiting from the rearrangement of the objective function, the beamforming design of passive RIS can be computed centrally on the CPU. Step (4.5) Repeat steps (4.2)-(4.5). Terminate the loop when the objective function converges to obtain the optimal solution (w, Φ, u). k The parameters are used to calculate user k information and rate.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements a joint precoding and beamforming method for RIS-assisted decellularization of large-scale MIMO under electromagnetic interference as described in any one of claims 1 to 5.

7. A computer-readable storage medium storing computer instructions thereon, characterized in that, When executed by the processor, the computer instructions implement a joint precoding and beamforming method for RIS-assisted decellularization of large-scale MIMO under electromagnetic interference as described in any one of claims 1-5.

Images