Multi-mode reconfigurable intelligent surface assisted cellular network removal system energy efficiency maximization method and system

Abstract

The invention relates to an MO-RIS assisted cellular network removal system energy efficiency maximization method and system. The method comprises the following steps: S1, a base station modulates and sends a signal; s2, taking the signal in the S1 as an input parameter, and calculating a weighted sum rate; s3, calculating transmitting power loss; s4, judging whether the # imgabs0 # obtained in S2 and S3 meets the following relational expression: if the # imgabs1 # is not met, executing S5; otherwise, stopping iteration, skipping to and executing S8; s5, taking the weighted sum rate obtained in the step S4 as an input parameter, executing a beam forming matrix optimization method, and outputting Wopt; s6, taking the Wopt obtained in the step S5 as an input parameter, executing a multimode RIS phase shift matrix optimization method, and outputting theta opt; s7, alternately optimizing the Wopt obtained in the step S5 and the theta opt obtained in the step S6 to serve as input parameters, calculating the current weighted sum rate and power loss, making t = t + 1, and skipping to the step S4; and S8, obtaining an EE optimal value. According to the multi-mode reconfigurable intelligent surface assisted cellular network removal energy efficiency maximization method and system, the user energy efficiency can be remarkably improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of wireless communication and network optimization, and particularly relates to a method and system for maximizing the energy efficiency of a cell-free network system through a multi-mode reconfigurable intelligent surface (MO-RIS). The present invention is particularly applicable to dynamic resource management in 6G ultra-dense networks, large-scale Internet of Things (IoT), and edge computing scenarios. Background Art

[0002] Traditional cell-free networks improve coverage through cooperation among distributed access points (APs), but the cooperative transmission of a large number of APs leads to a sharp increase in transmission power. When the user density exceeds 50 per square kilometer, the energy efficiency (EE) of the system drops by more than 40% due to power loss. Existing solutions (such as Chinese Patent Application No. CN202110789012.3) use static beamforming and do not dynamically adjust power allocation according to the real-time channel state, resulting in resource waste. Existing energy efficiency optimization solutions (such as the Dinkelbach algorithm) are mostly designed for centralized MIMO and are not adapted to the distributed architecture of cell-free networks. The literature (W. Yu et al., IEEE TSP 2020) points out that when the number of users exceeds 10, the complexity of such algorithms explodes due to non-convex problems, and the convergence time is extended by more than 3 times. Existing RIS research mostly focuses on passive reflection or dual-mode reflection / transmission and lacks support for signal amplification functions. In weak signal scenarios (such as rural coverage), traditional RIS cannot effectively improve the signal-to-noise ratio at the receiving end due to the attenuation of the signal in the link from the base station to the RIS through free space propagation and the attenuation of the signal after reflection in the link from the RIS to the user. Summary of the Invention

[0003] In view of the above existing problems, the present invention proposes a method and system for maximizing the energy efficiency of a cell-free network system based on a multi-mode reconfigurable intelligent surface (MO-RIS). The proposed MO-RIS-CellFree system of the present invention combines passive reflection, active signal amplification, and dynamic beamforming technologies. By jointly optimizing the base station transmission strategy and the MO-RIS phase shift configuration, the multiplicative attenuation of the traditional cascaded channel is converted into controllable additive attenuation, and at the same time, multi-user interference is significantly reduced, realizing the coordinated improvement of the system energy efficiency (Energy Efficiency, EE) and the user weighted sum rate (Weighted Sum-Rate, WSR).

[0004] To achieve the above object, the present invention adopts the following technical solutions:

[0005] A method for maximizing the energy efficiency of a multi-mode reconfigurable intelligent surface-assisted cell-free network system is implemented according to the following steps:

[0006] S1: Fix the reflection and refraction phase shift matrices of the multi-mode reconfigurable intelligent surface (MO-RIS); After the base station performs digital modulation on the signal, it sends the signal to the multi-mode reconfigurable intelligent surface and the user terminal;

[0007] S2: The user terminal receives the signal sent by the base station in step S1 and uses it as an input parameter to calculate its weighted sum rate where, ∑ is the summation symbol, λ k ∈R + represents the weight factor of user k, R + is the set of positive real numbers, θ k is the signal-to-noise ratio at user k;

[0008] S3: Calculate the transmission power loss, and the power loss is as follows: where, is the reciprocal of the energy conversion coefficient of base station BS; W U and W BS respectively represent the dissipated power consumed by each user and the base station; W PS is the RIS unit power; w b,k represents the precoding vector of the b-th base station for the k-th user; K is the number of users; B is the number of base stations; RN is the number of RISs;

[0009] S4: Judge the weighted sum rate obtained in step S2 and the power loss obtained in step S3, and calculate whether the energy efficiency value of the current user terminal

[0010] satisfies the following relationship:

[0011] where, ε is a constant; if not satisfied, execute step S5; if satisfied, stop the iteration, jump and execute step S8; opt ; where, N is a natural number representing the number of base station antennas; the superscript opt represents the optimal value of W;

[0012] S6: Use the optimal beamforming matrix W opt obtained in step S5 as an input parameter, execute the RIS phase shift matrix optimization method, and output the RN×RN dimensional optimal phase shift matrix Θ opt ; where, RN is a natural number representing the number of RIS units; the superscript opt represents the optimal value of Θ;

[0013] S7: According to W opt obtained in step S5 and Θopt , calculate the weighted sum rate of the current system user side and power loss Then let n = n + 1, and jump to step S4;

[0014] S8: Take EE = WSR / P obtained in step S4 as the output of the final result, and obtain the optimal value of the system energy efficiency EE of the user.

[0015] Preferably, in the optimization method of the beamforming matrix W in step S5:

[0016] The RIS-assisted cellular-free network system includes B base stations with a positive integer number of antennas M b of, K users with a positive integer number of antennas U k of, R RISs with a positive integer number of antennas N r For simplicity and without loss of generality, assume M b , U k and N r are respectively equal to M, U, and N; Let k be the kth user, then the signal-to-interference-plus-noise ratio θ corresponding to the kth user k is expressed as:

[0017]

[0018] Among them, is a 1×BMK-dimensional column vector, representing the equivalent channel coefficient matrix from the base station to the kth user, and the superscript H represents the conjugate transpose; is an M×1-dimensional column vector, representing the precoding vector at the Bth base station; is a U×M-dimensional vector, representing the frequency-domain channel from base station b to user k; is a U×N-dimensional vector, representing the frequency-domain channel from the RIS to user k; G is an N×M-dimensional matrix, representing the channel coefficient matrix from the base station to the RIS; is an N×N-dimensional diagonal matrix, representing the phase shift matrix at RIS r; δ k is a constant, representing the variance of Gaussian white noise at the user; then the weighted sum rate of K users is:

[0019]

[0020] Among them, λ k is a real constant, representing the rate of the kth user. Then the harvested power of the kth energy receiver is:

[0021]

[0022] Among them, represents the equivalent channel coefficient matrix from the base station to the kth energy receiver; hb,k is an \(N\times1\) dimensional column vector, representing the channel coefficient vector from the base station to the \(k\)th energy receiver; \(h\) r,k is an \(M\times1\) dimensional column vector, representing the channel coefficient vector from the multi-mode RIS to the \(k\)th energy receiver; \(\delta\) Z 2 is a constant, representing the variance of Gaussian white noise at the multi-mode RIS; \(\eta\) k is E the energy harvesting efficiency of \(K\)

[0023] S5.2: Model the problem 1 of the system user - end energy efficiency EE as:

[0024]

[0025] where, is a constant, representing the maximum transmit power of base station \(b\); Apply the Dinkelbach algorithm to convert equation (4) into an equivalent form. The optimal EE satisfies:

[0026]

[0027] Introduce an auxiliary variable Using the fractional programming and quadratic transformation method, the objective function of equation (5) is equivalent to:

[0028]

[0029] Fix the variables \(\{w k \}, \Theta r \}, calculate the partial derivatives of the independent variables, and make equal to zero to obtain the optimal solution as:

[0030]

[0031]

[0032] where, represents the conjugate transpose of the signal - to - noise ratio, \(\delta k is a constant, representing the variance of Gaussian white noise at the user; For the optimal beamforming at the base station, first define and let Fix the variables \(\{\mu, \upsilon, \Theta i \}, and define the intermediate variable matrix:

[0033]

[0034]

[0035] \upsilon k =h k w k(11)

[0036] Then the transmit beamforming matrix problem 2 is formulated as:

[0037]

[0038] where, Since matrices A and are both positive semi - definite, the simplified sub - problem P1 mentioned in equation (12) is a standard quadratic - constrained quadratic programming problem and can be optimally solved using the existing alternating direction method of multipliers method.

[0039] Preferably, in the multi - mode RIS phase - shift matrix Θ optimization method of step S6:

[0040] For optimal RIS precoding, fix the variables W, μ, and υ, and solve for Θ. Based on the given (W * , μ * and υ * ), for the WSR maximization problem P1 equivalent to equation (5) o , the sub - problem at the RIS for RIS precoding design can be equivalently rewritten as:

[0041]

[0042] Reduce the complexity by simplifying the expression of f2 in equation (13). First, define a new auxiliary function with respect to Θ:

[0043]

[0044] Then correct f2 in equation (13) to:

[0045]

[0046] Then we have:

[0047]

[0048] In equation (16), define holds; is the column vector composed of the diagonal elements of RIS; represents an RN×1 - dimensional column vector composed of complex numbers; Substitute equation (16) into equation (15) and simplify to get:

[0049]

[0050] where, we have:

[0051]

[0052]

[0053]

[0054] The active precoding sub-problem reformulated from Equation (13) is further simplified to:

[0055]

[0056] Using popular optimization algorithms, considering a series of solvable sub-problems, the problem is iteratively solved by approximating the objective function and constraint set of Equation (21). Let Then the problem can be reformulated as:

[0057]

[0058] An MO algorithm is proposed to obtain the optimal phase shift of the k-th user. It mainly derives a gradient descent algorithm on the manifold space. Equation (22) can be reformulated as:

[0059]

[0060] where γ is a constant controlling the convergence. Since Equation (23) is equivalent to Equation (22), and the feasible set of Equation (23) is defined as G RN , that is, RN complex circles, and each complex circle can represent The set G can be regarded as a sub-manifold of Χ; the product of RN complex circles corresponds to a sub-manifold of Χ RN . Therefore, the manifold of Equation (23) can be expressed as where z l is the i-th element of the vector z. Next, the steps of the MO algorithm for iteratively solving Equation (23) are described:

[0061] S6.2.1: Search direction: First, define as the objective function at the m-th iteration of Equation (23). Then, set the search direction of Equation (23) to be opposite to the gradient in the Euclidean space:

[0062]

[0063] S6.2.2: Projection search direction on the tangent space: The optimization step on the manifold space finds the Riemannian gradient of f5 at the current tangent space at the point; project the search direction z in the Euclidean space (m) onto to obtain The Riemannian gradient at the point is:

[0064]

[0065] Among them

[0066] S6.2.3: Update and descent of the tangent space: In the tangent space can be updated to:

[0067]

[0068] where ξ is a step size;

[0069] S6.2.4: Retraction operation: Since is not in G RN Therefore is mapped to the manifold G RN through the retraction operation, and each element in is normalized to unity:

[0070]

[0071] The value ranges of the parameters γ and ξ are determined by the following theorem to ensure the convergence of the MO algorithm:

[0072] Let λ D and λ D+γI be the largest eigenvalues of the matrices D and D + γI respectively. If γ and ξ satisfy the following conditions:

[0073]

[0074] then the MO algorithm generates a non-decreasing sequence until convergence.

[0075] The present invention also discloses a method for maximizing the energy efficiency of a MO-RIS assisted cell-free network for performing the above method, including the following modules:

[0076] Signal transmission module: used to fix the reflection and refraction phase shift matrix of the MO-RIS; after the base station performs digital modulation on the signal, the signal is sent to the reconfigurable intelligent surface MO-RIS and the user receiver;

[0077] Weighted sum rate calculation module: used to calculate the weighted sum rate with the signal received by the destination user receiver from the base station as the input parameter where λ k ∈R + represents the weight factor of the k-th user receiver, K represents the number of users, and k is an integer;

[0078] Transmission power loss calculation module: calculates the total power loss where, is the reciprocal of the energy conversion coefficient of the base station BS; W U and W BS respectively represent the dissipated power consumed by each user and the base station; W PS is the power of the RIS unit; K is the number of users; B is the number of base stations; RN is the number of RISs;

[0079] Judgment module: Judge whether the weighted sum rate obtained by the user side and the total power loss P satisfy the following relational expression: where ε is a constant; if not satisfied, it is executed by the beamforming matrix module; if satisfied, the EE optimal value of the user is output;

[0080] Beamforming matrix module: Take the weighted sum rate WSR obtained by the judgment module as the input parameter, execute the optimization method of the beamforming matrix W, and output the optimal beamforming matrix W of dimension MBK×1 opt ; where MBK is a natural number representing the number of antennas of the base station; the superscript opt represents the optimal value of W;

[0081] MO-RIS phase shift matrix optimization module: Take the optimal beamforming matrix W opt obtained by the beamforming matrix and energy matrix optimization module as the input parameter, execute the multi-mode RIS phase shift matrix optimization method, and output the optimal phase shift matrix Θ of dimension RN×RN opt ; where RN is a natural number representing the number of RIS units, and the superscript opt represents the optimal value of Θ;

[0082] Alternating iterative optimization module: Perform alternating iterative optimization on W opt obtained by the beamforming matrix and energy matrix optimization module and Θ opt obtained by the MO-RIS phase shift matrix optimization module, and calculate the energy efficiency of the current user side Then let n=n + 1, and it is executed by the judgment module.

[0083] The present invention also discloses an electronic device, including:

[0084] A processor;

[0085] A memory for storing a program, which when called and executed by the processor, causes the processor to execute the above method or system.

[0086] Some of the prior arts related to the present invention are briefly as follows:

[0087] 1. Digital modulation methods such as binary phase shift keying (BPSK)

[0088] Digital modulation is the core technology of modern communication systems. It realizes efficient and reliable information transmission by converting digital signals (bit streams) into analog waveforms suitable for physical channel transmission. Error control techniques can be used in digital transmission systems to support complex signal conditions and processing techniques such as source coding, encryption techniques, and equalization. In digital modulation, the modulated signal can be represented as a time series of symbols or pulses, where each symbol can have m finite states, and each symbol can be represented by n bits. Currently, the main digital modulation methods include BPSK, QPSK, 8PSK, QAM, etc. For details, see "G. Proakis, M. Salehi. Digital Communications, 5th Edition[J]. 2008".

[0089] 2. Quadratic transformation method

[0090] The fractional programming problem is a class of nonlinear optimization problems with a fractional objective function, and its mathematical model can be expressed as:

[0091]

[0092] where A(x) and B(x) are continuous functions defined on and B(x)>0.

[0093] The quadratic transformation method is the key technology to solve such problems. Its core idea is to introduce an auxiliary variable y to convert the original fractional objective function into an equivalent quadratic form. For the specific algorithm implementation, refer to: "K. Shen and W. Yu, "Fractional Programming for Communication Systems—Part I: Power Control and Beamforming," IEEE Transactions on Signal Processing, vol. 66, no. 10, pp. 2616–2630, May 2018." The quadratic transformation theorem can be stated as:

[0094]

[0095] This method satisfies condition C1: The objective function can be decomposed into g(x, y) = f(A(x))q1(y) + h(B(x))q2(y), which is convenient for alternating optimization; C2: When x * maximizes A(x) / B(x), there exists y * such that (x * , y * ) jointly maximizes g(x, y);

[0096] C3: The optimal auxiliary variable satisfies y * = argmax y g(x, y), where g(x, y * ) = A(x) / B(x); C4: Ensure the convexity of the optimization problem.

[0097] 3. Dinkelbach Algorithm

[0098] Energy efficiency optimization is a core technology in the cellular-free network, aiming to maximize the communication performance (such as throughput) per unit energy consumption. The Dinkelbach algorithm is a classic method for solving fractional programming problems and is applicable to optimization problems with a fractional objective function. The problem is modeled as follows:

[0099]

[0100] where N(x) represents the system throughput (such as the weighted sum rate WSR), and M(x) represents the total power consumption. Parameter initialization: Set the initial energy efficiency value η (0) = 0, and the number of iterations m = 0.

[0101] Iterative optimization:

[0102] Sub-problem solving: In each iteration, solve the linear optimization problem: Update the energy efficiency value: According to the current solution x (m) , update

[0103] Termination condition: When |N(x (m) ) - η (m) M(x m )| < ε, the algorithm converges.

[0104] Application in the present invention - Dynamic energy efficiency optimization: Model the energy efficiency maximization problem of the cellular-free network as a fractional programming problem and decompose it into linear sub-problems through the Dinkelbach algorithm. Avoid directly dealing with the non-convexity of the fractional function and reduce the computational complexity. For details, see: "W. Dinkelbach, “On Nonlinear Fractional Programming,” Management Science, vol. 13, no. 7, pp. 492–498, 1967."

[0105] 4. Subgradient Method

[0106] The subgradient method is an iterative method for solving convex optimization problems of convex functions. It can be used for non-differentiable objective functions. When the objective function is differentiable, for unconstrained problems, the subgradient method has the same search direction as the gradient descent method. In each iteration of the subgradient method, a negative subgradient direction is selected as the search direction, and the search step size is set according to certain rules. The subgradient can be selected as the subgradient of the objective function or the subdifferential of any function that violates the constraints. For non-smooth convex optimization problems, the subgradient method can guarantee global convergence. See specifically "K. Singh, S. Biswas, M.-L. Ku, and M. F. Flanagan, “Transceiver design and power control for full-duplex ultra-reliable low-latency communication systems,” IEEE Trans. Wireless Commun., vol. 21, no. 2, pp. 1392–1406, Feb. 2022."

[0107] The present invention has the following technical effects:

[0108] The present invention realizes the optimal improvement of the energy efficiency of the multi-user MO-RIS cellular-free system. BRIEF DESCRIPTION OF THE DRAWINGS

[0109] Figure 1 It is a model diagram related to the method for maximizing the energy efficiency of the MO-RIS-assisted cellular-free network system according to the preferred embodiment of the present invention;

[0110] Figure 2 It is a flowchart of the method for maximizing the energy efficiency of the multi-mode RIS-assisted multi-user cellular-free network system according to the preferred embodiment of the present invention;

[0111] Figure 3 It is a flowchart of the steps for optimizing the transmit beamforming matrix according to the preferred embodiment of the present invention;

[0112] Figure 4 It is a flowchart of the steps for optimizing the reflection and refraction phase shift matrix of the multi-mode RIS according to the preferred embodiment of the present invention;

[0113] Figure 5 It is a comparison diagram of the algorithm convergence performance under different schemes;

[0114] Figure 6 It is a comparison diagram of the energy efficiency performance as the number of RIS units increases under different schemes;

[0115] Figure 7 It is a system diagram of the MO-RIS-assisted cellular-free network system for maximizing the energy efficiency according to the preferred embodiment of the present invention. Detailed implementation manners

[0116] The present invention will be further described below in conjunction with specific embodiments.

[0117] Figure 1 It is a model diagram involved in the method for maximizing the energy efficiency of a multi-mode MO-RIS assisted de-cellular network system according to a preferred embodiment of the present invention. The system includes a base station with a positive integer b number of antennas, a MO-RIS with a positive integer r number of reflection and refraction units, and k single-antenna users; k is a positive integer.

[0118] Figure 2 It is a flowchart of a method for maximizing the energy efficiency of a multi-mode MO-RIS assisted multi-user de-cellular network system according to an embodiment of the present invention. This embodiment is completed through the following steps:

[0119] Step 1: Fix the reflection and refraction phase shift matrix of the MO-RIS; the base station digitally modulates the transmitted signal, and after passing through free space, transmits it to the MO-RIS and the user receiver;

[0120] Step 2: The destination user receiver receives the transmitted signal in Step S1, and calculates its weighted sum rate as an input parameter In this step, initialize the iteration number n = 1 and the convergence judgment constant ε = 0.01;

[0121] Step 3: Calculate the transmission power loss of each system during the calculation process. The total power loss is as follows:

[0122] Step 4: Judge Whether it holds, where Refers to the weighted sum rate value obtained from the nth iteration calculation. If it holds, jump to Step 8; otherwise, execute Step 5; here Can characterize the convergence performance of this method;

[0123] Step 5: Use the weighted sum rate WSR obtained in Step 4 as an input parameter, and obtain the optimal beamforming matrix W of an MBK×1 dimensional vector according to the beamforming matrix optimization method (n) ;

[0124] Step 6: Use the optimal beamforming matrix W in Step 5 (n) As an input parameter, execute the multi-mode RIS phase shift matrix optimization method to obtain an RN×RN dimensional optimal phase shift matrix

[0125] Step 7: Calculate the weighted sum rate of the current system user end according to the alternating optimization results of W (n) And And let n = n + 1, and then execute Step 4; ​

[0126] Step Eight: Output the maximum energy efficiency EE of the system user opt .

[0127] Figure 3 The flowchart of the steps for optimizing the transmit beamforming matrix in the embodiment of the present invention is mainly completed through the following steps:

[0128] Step One: Initialize the iteration number i = 1, μ (1) = 1, v (1) = 1, R (1) = 0 and the convergence judgment constant ε = 0.01, and input the initial signal value;

[0129] Step Two: Judge whether f1(m) ≤ ε holds, where f1(m) represents the objective value of Equation (6) obtained by the m-th iteration calculation. If it holds, obtain the optimal beam matrix through Equation (12) and jump to Step Seven; otherwise, execute Step Three;

[0130] Step Three: Judge whether the energy efficiency converges. If it converges, execute Step Six; otherwise, execute Step Four;

[0131] Step Four: Solve the energy efficiency E (i) and update the current beamforming matrix W (i) ;

[0132] Step Five: Update the Lagrangian variable μ (i) = μ (1) , and as well as i = i + 1;

[0133] Step Six: Calculate the objective values of Equations (A1) and (A2), denoted as and f1(m) respectively, and m = m + 1, then execute Step Two;

[0134] Step Seven: Finally, output the optimal transmit beamforming vector W opt .

[0135] Figure 4 The flowchart of the steps for optimizing the MO-RIS reflection and refraction phase shift matrix in the embodiment of the present invention is mainly completed through the following steps:

[0136] Step One: Initialize the iteration number m = 0, the initial energy efficiency and the convergence judgment constant ε = 0.01, and input the beamforming matrix W;

[0137] Step 2: Determine if \(f_2(m)\leq\epsilon\), where \(f_2(m)\) represents the objective value of Equation (32) obtained from the \(m\)-th iterative calculation. If it holds, obtain the optimal phase shift matrix through the successive lower bound maximization method and jump to Step 7; otherwise, execute Step 3;

[0138] Step 3: Determine if the weighted sum rate converges. If it converges, execute Step 6; otherwise, execute Step 4;

[0139] Step 4: Solve the weighted sum rate \(R\) (n) and update the current phase shift matrix

[0140] Step 5: Update the intermediate variables and \(m = m + 1\);

[0141] Step 6: Calculate the objective values of Equations (A4) and (A5), denoted as and \(f_2(m)\), respectively, and \(m = m + 1\), then execute Step 2;

[0142] Step 7: Finally, obtain the optimal MO-RIS phase shift matrix \(\Theta\) opt .

[0143] Figure 5 is a comparison chart of the convergence performance of the proposed algorithm under different RIS-assisted cellular-free networks. Among them, the number of users is set to 4, the base station transmit power is set to 0 dBm, the number of RIS elements is 50, and the number of base station antennas is 2. It can be seen from the figure that the EE under different RIS assistances first increases and then levels off as the number of iterations increases. The EE of the multi-mode RIS-assisted scheme proposed in the present invention is always better than the random phase shift, no-RIS scheme, and no-direct link scheme. It converges after about 15 - 20 times, and the convergence value is about 7.5 bps / Hz / J, verifying the effectiveness and convergence of the present invention.

[0144] Figure 6 is a comparison chart of the energy efficiency performance with the increase of the number of RIS units under different schemes. It can be seen that the EE of the proposed RIS-assisted cellular-free network increases with the increase of the number of RIS elements. When the number of RIS reaches 80, the EE reaches the maximum value. When the number of RIS is large, more accurate phase shifts need to be used for passive precoding so that the signals reflected by the RIS can reach the users more accurately. When only the indirect path is provided by the RIS, the optimization effect of the proposed multi-mode RIS-assisted scheme improves with the increase of the number of RIS, and the optimization effect of the MO algorithm of the present invention is about 50% higher than that of the iterative non-convex optimization algorithm (CCM). These results show that introducing multi-mode RIS in the cellular-free network is a technology with broad application prospects that can improve system performance.

[0145] In the process of optimizing the RIS phase shift, the present invention is optimized by an algorithm, and a gradient descent algorithm is used in the manifold space to optimize the phase shift problem.

[0146] As Figure 7 shown, this embodiment discloses a MO-RIS assisted energy efficiency maximization system for a cellular network to execute the above method, which includes the following modules:

[0147] Signal transmission module: used to fix the MO-RIS reflection and refraction phase shift matrix; after the base station performs digital modulation on the signal, the signal is sent to the reconfigurable intelligent surface MO-RIS and the user receiver;

[0148] Weighted sum rate calculation module: used to calculate the weighted sum rate with the signal as the input parameter after the destination user receiver receives the signal sent by the base station where, λ k ∈R + represents the weight factor of the k-th user receiver, and k is an integer;

[0149] Transmission power loss calculation module: calculates the total power loss where, is the reciprocal of the energy conversion coefficient of the base station BS; W U and W BS respectively represent the dissipated power consumed by each user and the base station; W PS is the RIS unit power; K is the number of users; B is the number of base stations; RN is the number of RISs;

[0150] Judgment module: judges whether the user-end weighted sum rate obtained by the weighted sum rate calculation module and the total power loss P satisfy the following relational expression:

[0151] where, ε is a constant; if not satisfied, it is executed by the beamforming matrix module; if satisfied, the EE optimal value of the user is output;

[0152] Beamforming matrix module: takes the weighted sum rate WSR obtained by the judgment module as the input parameter, executes the optimization method of the beamforming matrix W, and outputs the optimal beamforming matrix W of dimension MBK×1 opt ; where, MBK is a natural number representing the number of antennas of the base station; the superscript opt represents the optimal value of W;

[0153] MO-RIS phase shift matrix optimization module: takes the optimal beamforming matrix W opt obtained by the beamforming matrix and energy matrix optimization module as the input parameter, executes the multi-mode RIS phase shift matrix optimization method, and outputs the optimal phase shift matrix Θ of dimension RN×RN opt; where \(R_N\) is a natural number representing the number of RIS units, and the superscript opt represents the optimal value of \(\Theta\);

[0154] Alternating iteration optimization module: Alternately iteratively optimize \(W\) obtained by the beamforming matrix and energy matrix optimization module opt and \(\Theta\) obtained by the MO-RIS phase shift matrix optimization module opt to calculate the energy efficiency of the current user terminal Then let \(n = n + 1\), which is executed by the judgment module.

[0155] For other content of this embodiment, reference can be made to the above method embodiment.

[0156] A preferred embodiment of the present invention also discloses an electronic device, including:

[0157] A processor;

[0158] A memory for storing a program, which when called and executed by the processor, causes the processor to execute the above method or system.

[0159] In summary, the present invention first fixes the MO-RIS reflection and refraction phase shift matrices, and uses the Lagrangian decomposition method to solve the optimal transmit beamforming matrix and energy matrix; secondly, fixes the beamforming matrix and uses the Dinkelbach algorithm to solve the MO-RIS reflection and refraction phase shift matrices; finally, performs alternating optimization to obtain the optimal user weighted sum rate and total system power consumption, and further obtains the optimal energy efficiency.

[0160] Although the embodiments of the present invention have been clearly described. However, for those skilled in the art, without departing from the principle and spirit of the method of the present invention, various changes, modifications, substitutions, and variations can be made to these embodiments. The scope of the present invention is defined by the appended claims and their equivalents, and still belongs to the scope of the method of the present invention and is still regarded as the protection scope of the present invention.

Claims

1. A method for maximizing energy efficiency of a decellularized network system assisted by a multi-mode reconfigurable smart surface, characterized in that: The specific steps are as follows: S1: fix the reflection and refraction phase shift matrix of the multi-mode reconfigurable smart surface; after the base station performs digital modulation on the signal, it sends it to the multi-mode reconfigurable smart surface and the user end; S2: The user end receives the signal sent by the base station in step S1 and uses it as an input parameter to calculate the weighted sum rate Among them, ∑ is the summation symbol, λ k ∈R + represents the weight factor of user k, R + is a set of positive real numbers, θ k is the signal-to-noise ratio at user k; S3: Calculate the transmission power loss. The power loss is as follows: in, is the inverse of the energy conversion coefficient of the base station BS; W U and W BS Represents the dissipated power consumed by each user and base station respectively; W PS is the unit power of the multi-mode reconfigurable smart surface; w b,k represents the precoding vector of the b-th base station for the k-th user; K is the number of users; B is the number of base stations; RN is the number of multi-mode reconfigurable smart surfaces; S4: Determine the weighted sum rate obtained in step S2 and the power loss obtained in step S3 Calculate the energy efficiency value of the current user end Does it satisfy the following relationship: If not, execute step S5; if satisfied, stop iteration, jump and execute step S8; where ε is a constant; S5: Take the weighted sum rate WSR obtained in step S4 as the input parameter, execute the beamforming matrix W optimization method, and output the MBK×1-dimensional optimal beamforming matrix W opt ; Where MBK is a natural number, indicating the number of antennas in the base station; the superscript opt ​​indicates the optimal value of W; S6: The optimal beamforming matrix W obtained in step S5 is opt As input parameters, the RIS phase shift matrix optimization method is executed, and the RN×RN dimension optimal phase shift matrix Θ is output. opt ; Where RN is a natural number, indicating the number of multi-mode reconfigurable smart surface units; the superscript opt ​​indicates the optimal value of Θ; S7: Based on the W obtained in step S5 opt and θ obtained in step S6 opt , calculate the weighted sum rate of the current system user end and power loss Then let n=n+1, and jump to step S4; S8: The obtained EE=WSR / P is output as the result to obtain the optimal value of the user's system energy efficiency EE.

2. The method for maximizing energy efficiency of a decellularized network system assisted by a multi-mode reconfigurable intelligent surface according to claim 1, characterized in that: In the optimization method of the beamforming matrix W in step S5: The system contains B antennas, the number of which is a positive integer M. b The number of base stations and antennas K is a positive integer U k The number of users, R antennas is a positive integer N r RIS, set M b , U k and N r are equal to M, U and N respectively; let k be the kth user, then the signal to interference noise ratio corresponding to the kth user is It is expressed as: in, is a 1×BMK-dimensional column vector, representing the equivalent channel coefficient matrix from the base station to the k-th user, and the superscript H represents the conjugate transpose; is an M×1-dimensional column vector, representing the precoding vector at the Bth base station; is a U×M dimensional vector, representing the frequency domain channel from base station b to user k; is a U×N dimensional vector, representing the frequency domain channel from RIS to user k; G is an N×M dimensional matrix, representing the channel coefficient matrix from the base station to RIS; is an N×N dimensional diagonal matrix, representing the phase shift matrix at RISr; δ k is a constant, representing the variance of Gaussian white noise at the user; then the weighted sum rate of K users is: Among them, λ k is a real constant, representing the rate of the kth user; the harvested power of the kth energy receiving end is: in, represents the equivalent channel coefficient matrix from the base station to the kth energy receiving end; h b,k is an N×1-dimensional column vector, representing the channel coefficient vector from the base station to the kth energy receiving end; h r,k is an M×1-dimensional column vector, representing the channel coefficient vector from the multimode RIS to the k-th energy receiving end; is a constant, representing the variance of Gaussian white noise at the multimode RIS; η k K E Energy collection efficiency of each energy receiving end; The energy efficiency EE problem 1 of the user-side system is modeled as: in, is a constant, which represents the maximum transmission power of base station b. The Dinkelbach algorithm is used to convert equation (4) into an equivalent form, and the optimal EE satisfies: Introducing auxiliary variables Using fractional programming and quadratic transformation methods, the objective function of formula (5) is equivalent to: Fixed variable {{w k },Θ r }, calculate the partial derivative of the independent variable so that Zero, the optimal solution for: in, represents the conjugate transpose of the signal-to-noise ratio, δ k is a constant, representing the variance of Gaussian white noise at the user; for the optimal beamforming at the base station, first define Fixed variables {μ,υ,Θ i }, define the intermediate variable matrix: u k =h k w k (11) Then the transmit beamforming matrix problem 2 is expressed as: in, 3. The method for maximizing energy efficiency of a decellularized network system assisted by a multi-mode reconfigurable intelligent surface according to claim 1, characterized in that: In the RIS phase shift matrix θ optimization method of step S6: Based on the given W * , μ * and * , in the equivalent WSR maximization problem of formula (5) The sub-problem of RIS precoding design at RIS is equivalently written as: Define new helper functions on Θ: Modify f2 in formula (13) to: Then we have: In formula (16), we define Established, is a column vector consisting of the diagonal elements of RIS; represents an RN×1-dimensional column vector composed of complex numbers; Substituting equation (16) into equation (15) to simplify it, we obtain: Among them, there are: The active precoding sub-problem of formula (13) can be further simplified as follows: The problem is solved iteratively by approximating the objective function and constraint set of equation (21), The problem Rephrased as: Using the gradient descent method, equation (22) is reformulated as: Among them, γ is a constant that controls convergence; due to Formula (23) is equivalent to formula (22); the feasible set of formula (23) is defined as That is, RN complex circles, each complex circle represents gather Seen as A substream of ; the product of RN complex circles corresponds to A sub-population of ; therefore, the populace of formula (23) is expressed as Among them, z l is the I-th element of vector z.

4. The method for maximizing energy efficiency of a decellularized network system assisted by a multi-mode reconfigurable intelligent surface according to claim 3, characterized in that: The specific steps of iteratively solving equation (23) are as follows: S6.2.1: Definitions is the objective function of the mth iteration of formula (23), and the search direction of formula (23) is set to be the same as The gradient in Euclidean space is opposite: S6.2.2: Optimization steps on the popular space in the current cut space exist Find the Riemann gradient of f5 at point; change the search direction z in Euclidean space to (m) Project to On, get exist The Riemann gradient at is: in, S6.2.3: Update of the cut space is reduced: In the cutting space Updated to: Among them, ξ is a step size; S6.2.4: Mapping to the popular G through the retraction operation RN In Each element in is normalized to unity: Determine the value range of parameters γ and ξ: Let λ D and λ D+γI are the maximum eigenvalues ​​of matrices D and D+γI respectively. If γ and ξ satisfy the following conditions: A non-increasing sequence is generated until convergence.

5. A multi-mode reconfigurable smart surface-assisted decellularized network system energy efficiency maximization system, used to execute the method according to any one of claims 1 to 4, characterized in that: Includes the following modules: Signal transmission module: used to fix the reflection and refraction phase shift matrix of the multi-mode reconfigurable smart surface; after the base station performs digital modulation on the signal, it sends the signal to the reconfigurable smart surface and the user end; Weighted sum rate calculation module: After the user terminal receives the signal sent by the base station, it uses the signal as an input parameter to calculate the weighted sum rate Among them, λ k ∈R + represents the weight factor of the kth user receiving end, K represents the number of users, and k is an integer; Transmit power loss calculation module: calculate the transmit power loss in, is the inverse of the energy conversion coefficient of the base station BS; W U and W BS Represents the dissipated power consumed by each user and base station respectively; W PS is the RIS unit power; K is the number of users; B is the number of base stations; RN is the number of RIS; Judgment module: judge the weighted sum rate Does the power loss P satisfy the following relationship: Where ε is a constant; if it is not satisfied, it is executed by the beamforming matrix module; if it is satisfied, the optimal value of the user's system energy efficiency EE is output; Beamforming matrix module: takes the weighted sum rate WSR obtained by the judgment module as the input parameter, executes the optimization method of the beamforming matrix W, and outputs the MBK×1-dimensional optimal beamforming matrix W opt ; Where MBK is a natural number, indicating the number of antennas in the base station; the superscript opt ​​indicates the optimal value of W; MO-RIS phase shift matrix optimization module: The optimal beamforming matrix W is obtained opt As input parameters, the multi-mode RIS phase shift matrix optimization method is executed, and the RN×RN dimension optimal phase shift matrix Θ is output. opt ; Where RN is a natural number, indicating the number of RIS units, and the superscript opt ​​indicates the optimal value of Θ; Alternating iterative optimization module: The obtained W opt and θ opt Perform alternating iterative optimization to calculate the energy efficiency of the current user end Then let n=n+1, and the judgment module executes.

Images