An RIS model and QoS constraint-based MU-MISO system energy efficiency optimization method

Abstract

The application discloses a kind of based on RIS model and QoS constraint MU-MISO system energy efficiency optimization method, comprising: constructing RIS assisted channel model;According to RIS assisted channel model, actual discrete RIS power consumption model is constructed;System total power consumption model is constructed;The joint optimization problem with system weighted energy efficiency maximization as target and meeting quality of service constraint and maximum transmit power constraint is constructed;Optimize beamforming matrix using improved gradient descent algorithm;Optimize the genetic algorithm of grey scale coding discrete phase shift matrix;Judge whether the joint optimization problem converges, if convergence, then output the optimal solution obtained at present, obtain maximum energy efficiency, optimal phase shift matrix and RIS phase configuration;Otherwise, re-iterative optimization beamforming matrix and discrete phase shift matrix, until the joint optimization problem converges or reaches maximum iteration number.The application significantly improves system energy efficiency while ensuring user QoS by jointly optimizing beam matrix and RIS phase configuration.

Description

Technical Field

[0001] This invention belongs to the field of communication technology, and in particular relates to an energy efficiency optimization method for MU-MISO systems based on the RIS model and QoS constraints. Background Technology

[0002] With the development of 5G and future 6G communication technologies, multi-user MISO (Multiple-Input Single-Output) systems have shown great potential in improving spectral efficiency and system capacity. However, traditional communication systems are limited by channel fading and multi-user interference, making it difficult to achieve a balance between energy efficiency and Quality of Service (QoS). Reconfigurable Smart Surfaces (RIS), as an emerging technology, can effectively enhance signal coverage and reduce power consumption by intelligently regulating the wireless propagation environment. However, in RIS-assisted MISO systems, jointly optimizing the beamforming matrix of the base station and the phase shift matrix of the RIS results in a strongly coupled non-convex optimization problem. Furthermore, actual RIS reflector units typically have discrete phase constraints, posing significant challenges to the optimization problem in terms of computational complexity and global convergence. Existing research often employs methods such as alternating optimization or continuous relaxation, but these methods still suffer from problems such as mismatch between energy efficiency models and actual hardware power consumption, and insufficient efficiency in discrete phase optimization, making it difficult to meet the urgent energy efficiency requirements of future green communication networks. Summary of the Invention

[0003] To address the aforementioned shortcomings in existing technologies, this invention provides an energy efficiency optimization method for MU-MISO systems based on a RIS model and QoS constraints, which solves the problems of mismatch between the energy efficiency model and actual hardware power consumption, and insufficient discrete phase optimization efficiency.

[0004] To achieve the aforementioned objectives, the technical solution adopted by this invention is: an energy efficiency optimization method for a MU-MISO system based on a RIS model and QoS constraints, comprising:

[0005] Construct RIS-assisted channel models, including modeling composite channels of direct and reflected links and defining effective channels;

[0006] Construct a practical discrete RIS power consumption model based on the RIS-assisted channel model;

[0007] A total system power consumption model is constructed based on a practical discrete RIS power consumption model.

[0008] A joint optimization problem is constructed with the goal of maximizing the weighted energy efficiency of the system and satisfying the service quality constraint and the maximum transmit power constraint. The joint optimization problem is solved by jointly optimizing the base station beamforming matrix and the RIS discrete phase shift matrix.

[0009] The beamforming matrix is ​​optimized using an improved gradient descent algorithm; the discrete phase shift matrix is ​​optimized using a gray-scale encoded genetic algorithm.

[0010] Determine whether the joint optimization problem has converged. If it has converged, output the current optimal solution, which maximizes energy efficiency, the optimal phase shift matrix, and the RIS phase configuration. Otherwise, iterate again to optimize the beamforming matrix and discrete phase shift matrix until the joint optimization problem converges or the maximum number of iterations is reached.

[0011] The beneficial effects of this invention are as follows:

[0012] 1. By jointly optimizing transmit beamforming and RIS phase configuration, the system energy efficiency is significantly improved while ensuring user QoS constraints, achieving dual optimization of performance and service quality.

[0013] 2. An improved gradient descent method is adopted to efficiently solve non-convex optimization problems. The discrete RIS phase is optimized by combining grayscale encoding and genetic algorithm, which balances performance and feasibility.

[0014] 3. The introduction of a practical RIS power consumption modeling method can more accurately reflect the energy consumption characteristics of the system in actual operation, providing a scientific basis for the design and energy efficiency optimization of green communication systems, and solving the problem of the disconnect between traditional continuous models and engineering applications. Attached Figure Description

[0015] Figure 1 A flowchart of an energy efficiency optimization method for a MU-MISO system based on a RIS model and QoS constraints is provided for an embodiment.

[0016] Figure 2 This shows the trend of energy efficiency changes with different iteration rounds.

[0017] Figure 3 This represents the average energy efficiency of the system under different population sizes and resolutions. Detailed Implementation

[0018] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0019] like Figure 1 As shown, in one embodiment of the present invention, an energy efficiency optimization method for a MU-MISO system based on a RIS model and QoS constraints includes the following steps:

[0020] S1. Construct a RIS-assisted channel model, including modeling a composite channel of direct and reflected links and defining an effective channel.

[0021] The expression for an effective channel is:

[0022]

[0023] in, This represents the valid channel for the k-th user. This represents the direct channel from the base station to the k-th user, i.e., the direct link. This represents the channel from the RIS surface to the k-th user. It is a discrete phase shift matrix. This indicates a reflected link. Direct links use the Rayleigh fading model, while reflected links use the Rice fading model.

[0024] S2. Construct a practical discrete RIS power consumption model based on the RIS-assisted channel model;

[0025] The expression for a practical discrete RIS power consumption model is as follows:

[0026]

[0027] in, This represents the power consumption of a practical discrete RIS. This represents the static power consumption of the RIS hardware, including the control board and driver circuitry. This represents the total power consumption of all RIS units.

[0028] The expression for the total power consumption of all RIS units is:

[0029]

[0030] in, This indicates whether the i-th bit of the n-th RIS unit is in the "on" state, with a value of 1 representing on and a value of 0 representing off. This represents the power consumption of the RIS unit when it is in the "on" state. Indicates the total number of RIS units. This represents the total number of bits.

[0031] S3. Construct a total system power consumption model based on a practical discrete RIS power consumption model;

[0032] The system's total power consumption model includes the base station transmit power, user equipment power consumption, and the static and dynamic power consumption of the RIS (Residual Power Supply). It covers the discrete power consumption characteristics of the base station, user equipment, and RIS, and its expression is as follows:

[0033]

[0034] in, This indicates the total power consumption of the system. It represents all power consumption related to wireless communication transmission, including the power consumed by base stations and mobile devices during signal reception, decoding and transmission, and the energy consumption of base stations and user equipment for signal transmission transmission. This indicates the power loss caused by noise.

[0035] All power consumption related to wireless communication transmission includes the power consumed by base stations and mobile devices during signal reception, decoding, and transmission, as well as the energy consumption of base stations and user equipment for signal transmission transmission, expressed as:

[0036]

[0037] in, This indicates the total power consumption of the base station system. Indicates the base station's transmit power. This indicates the total power consumption of the user equipment.

[0038] S4. Construct a joint optimization problem with the goal of maximizing the weighted energy efficiency of the system while satisfying the service quality constraint and the maximum transmit power constraint.

[0039] The joint optimization problem is solved by jointly optimizing the base station beamforming matrix and the RIS discrete phase shift matrix.

[0040] The expression for the joint optimization problem is:

[0041]

[0042] in, Let be the objective function. Beamforming matrix, It is a discrete phase shift matrix; Represents the user weight matrix, Indicates channel capacity; Indicates constraints. This indicates the average power of the transmitted signal; This represents the signal vector transmitted by the base station; Represents the emission covariance matrix; Indicates the maximum transmission power; Indicates the signal-to-interference-plus-noise ratio; This represents the minimum acceptable speed threshold for the k-th user.

[0043] S5. Optimize the beamforming matrix using an improved gradient descent algorithm; optimize the discrete phase shift matrix using a gray-scale encoded genetic algorithm.

[0044] The specific method for optimizing the beamforming matrix is ​​as follows:

[0045] Using a first-order Taylor expansion to linearize the non-convex terms at the current iteration point (making a first-order approximation of the interference terms or similar bilinear terms), the original constraints are approximated as linear inequalities, thereby constructing a convex subproblem that is solvable at the current iteration point.

[0046] Based on the WMSE weighting factor and the current channel state, the gradient expression of the energy efficiency target with respect to the beamforming matrix is ​​derived, and these expressions are accumulated to form the beam update direction matrix. This direction physically represents the adjustment direction with the fastest energy efficiency improvement in the current iteration.

[0047] Within the current iteration, an Armijo linear backtracking search is performed along the direction beam update direction to determine the optimal step size for this update, ensuring that energy efficiency is monotonically improved after the update.

[0048] The beamforming matrix is ​​updated based on the optimal step size and then normalized in power to ensure that the beamforming matrix satisfies the maximum power constraint.

[0049] Determine if the beamforming matrix has converged. If it has converged, end the beamforming matrix update process and output the current iteration result; otherwise, repeat the iteration update.

[0050] The specific method for optimizing the discrete phase shift matrix is ​​as follows:

[0051] The discrete RIS phase shift values ​​are mapped to gray-scale encoded chromosomes, a population is randomly generated, and a fitness function is constructed.

[0052] Genetic operators are applied to the population, and a perturbation with a Hamming distance ≤ 2 is applied to the best individual in the current iteration. Local search is then used to improve convergence efficiency.

[0053] Determine if the population individuals satisfy the convergence condition; if so, output the decoded result of the current best individual. Otherwise, iterative updates are performed based on the current population.

[0054] S6. Determine whether the joint optimization problem has converged. If it has converged, output the current optimal solution to obtain the maximum energy efficiency, the optimal phase shift matrix, and the RIS phase configuration. Otherwise, iterate again to optimize the beamforming matrix and the discrete phase shift matrix until the joint optimization problem converges or the maximum number of iterations is reached.

[0055] To further verify the effectiveness and performance of the method proposed in this invention, the following simulation experiments were conducted.

[0056] 1. Experimental setup

[0057] System configuration: Number of base station antennas Number of users RIS unit number .

[0058] Power constraint: Maximum transmit power of the base station noise power .

[0059] Channel model: Rayleigh fading is used for direct links, and Ricean fading is used for reflected links (Rician factor). , ).

[0060] Optimization Algorithm: Population Size of Genetic Algorithm Crossover probability Probability of mutation .

[0061] 2. Experimental Procedure

[0062] The joint optimization algorithm (RLA-AW-GD-PRAC) proposed in this invention under the actual RIS power consumption model is compared with the following benchmark methods:

[0063] Benchmark 1: Using the joint optimization algorithm (RLA-AW-GD-FIXED) under the ideal RIS power consumption model;

[0064] Benchmark 2: Randomized RIS phase configuration (RANDOM_THETA_PRAC);

[0065] Benchmark 3: No RIS deployment (NO_RIS_BASE).

[0066] 3. Experimental Results

[0067] Convergence performance: Figure 2 The energy efficiency (EE) trend with different iteration numbers is shown to evaluate the convergence performance of the algorithm. It can be observed that under all system configurations, the EE increases monotonically with the number of iterations and tends to stabilize within a finite number of iterations, verifying the good convergence of the proposed algorithm.

[0068] Energy efficiency performance: The energy efficiency (AVGEE) of the algorithm is evaluated as a function of the number of iterations under different combinations of RIS unit number N and base station antenna number M. Convergence: The algorithm converges within 10 iterations, and the energy efficiency increases monotonically with the number of iterations.

[0069] Algorithm and power consumption modeling performance: Compared with traditional RIS models, stochastic optimization algorithms, and baseline algorithms without RIS, the RLA-AW-GD-PRAC algorithm using actual RIS power consumption modeling consistently achieves the highest energy efficiency across the entire N range, and also achieves the highest maximum transmit power limit for AVGEE. The increase in power consumption exhibits a monotonically rising trend, eventually reaching saturation. This verifies the strong adaptability and optimization effect of the proposed algorithm under actual hardware power consumption constraints.

[0070] As can be observed from Table 1, during the stage of lower transmission power (such as...), Under all algorithms, system energy efficiency significantly improves with increasing power, demonstrating the rate gain-dominant effect of power enhancement. However, with... The energy efficiency growth continues to increase, especially after exceeding 30 dBm, gradually slowing down and eventually approaching saturation. It is worth noting that RLA–AW–GD–PRAC does not force power saturation: iteration and projection ensure that the effective transmit power falls within the budget, thereby avoiding ineffective energy consumption and maintaining a more stable WEE under high power conditions; it also avoids the unrealistic degradation that occurs in idealized power consumption models, further highlighting the necessity of state-dependent RIS power consumption modeling.

[0071] Table 1

[0072]

[0073] Resolution impact: Figure 3 The average energy efficiency (AVGEE) of the proposed algorithm under different population sizes and 1-bit, 2-bit, and 3-bit RIS phase resolutions is presented with a weighting factor η=0.4. The overall trend shows that system energy efficiency significantly improves with increasing RIS phase resolution, especially in the 3-bit configuration, which exhibits optimal performance across all population sizes. Specifically, at the maximum population size (2400), the 3-bit scheme achieves approximately 4.56% and 18.69% EE improvements compared to 2-bit and 1-bit, respectively, validating the potential advantages of high-resolution RIS in energy efficiency optimization. This is attributed to the finer phase control capabilities brought about by the increased resolution, which significantly enhances reflection path gain and beamforming effects.

[0074] In summary, this invention significantly improves system energy efficiency while ensuring user QoS by jointly optimizing the beam matrix and RIS phase configuration. Experiments verify the algorithm's superiority in convergence, power consumption modeling accuracy, and resolution impact, providing a feasible solution for future green communication networks.

Claims

1. A method for optimizing the energy efficiency of a MU-MISO system based on a RIS model and QoS constraints, characterized in that, include: Construct RIS-assisted channel models, including modeling composite channels of direct and reflected links and defining effective channels; Construct a practical discrete RIS power consumption model based on the RIS-assisted channel model; A total system power consumption model is constructed based on a practical discrete RIS power consumption model. A joint optimization problem is constructed with the goal of maximizing the weighted energy efficiency of the system and satisfying the service quality constraint and the maximum transmit power constraint. The joint optimization problem is solved by jointly optimizing the base station beamforming matrix and the RIS discrete phase shift matrix. The beamforming matrix is ​​optimized using an improved gradient descent algorithm; the discrete phase shift matrix is ​​optimized using a gray-scale encoded genetic algorithm. Determine whether the joint optimization problem has converged. If it has converged, output the current optimal solution, which maximizes energy efficiency, the optimal phase shift matrix, and the RIS phase configuration. Otherwise, iterate again to optimize the beamforming matrix and discrete phase shift matrix until the joint optimization problem converges or the maximum number of iterations is reached. The expression for the joint optimization problem is: in, Let be the objective function. Beamforming matrix, It is a discrete phase shift matrix; Represents the user weight matrix, Indicates channel capacity; Indicates the total power consumption of the system; Indicates constraints. This indicates the average power of the transmitted signal; This represents the signal vector transmitted by the base station; Represents the emission covariance matrix; Indicates the maximum transmission power; Indicates the signal-to-interference-plus-noise ratio; This represents the minimum acceptable speed threshold for the k-th user; The specific method for optimizing the beamforming matrix is ​​as follows: Linearize the non-convex terms at the current iteration point using a first-order Taylor expansion, approximate the constraints as linear inequalities, and thus construct a convex subproblem that is solvable at the current iteration point. Based on the WMSE weighting factor and the current channel state, the gradient expression of the energy efficiency target with respect to the beamforming matrix is ​​derived, and these expressions are accumulated to form the beam update direction matrix. This direction physically represents the adjustment direction with the fastest energy efficiency improvement in the current iteration. Within the current iteration, an Armijo linear backtracking search is performed along the direction beam update direction to determine the optimal step size for this update, ensuring that energy efficiency is monotonically improved after the update. The beamforming matrix is ​​updated based on the optimal step size and then normalized in power to ensure that the beamforming matrix satisfies the maximum power constraint. Determine if the beamforming matrix has converged. If it has converged, end the beamforming matrix update process and output the current iteration result; otherwise, repeat the iteration update. The specific method for optimizing the discrete phase shift matrix is ​​as follows: The discrete RIS phase shift values ​​are mapped to gray-scale encoded chromosomes, a population is randomly generated, and a fitness function is constructed. Genetic operators are applied to the population, and a perturbation with a Hamming distance ≤ 2 is applied to the best individual in the current iteration. Local search is then used to improve convergence efficiency. Determine whether an individual in the population satisfies the convergence condition. If it does, output the discrete phase shift matrix decoded from the current best individual; otherwise, repeat the iterative update based on the current population.

2. The method according to claim 1, characterized in that, The expression for an effective channel is: in, This represents the valid channel for the k-th user. This represents the direct channel from the base station to the k-th user. This represents the channel from the RIS surface to the k-th user. For discrete phase shift matrix, This represents the reflection link channel from the base station to the RIS.

3. The method according to claim 1, characterized in that, The expression for a practical discrete RIS power consumption model is as follows: in, This represents the power consumption of a practical discrete RIS. This indicates the static power consumption of the RIS hardware. This represents the total power consumption of all RIS units.

4. The method according to claim 3, characterized in that, The expression for the total power consumption of all RIS units is: in, This indicates whether the i-th bit of the n-th RIS unit is in the "on" state, with a value of 1 representing on and a value of 0 representing off. This represents the power consumption of the RIS unit when it is in the "on" state. This indicates the total number of RIS units. This represents the total number of bits.

5. The method according to claim 4, characterized in that, The total system power consumption model includes base station transmit power, user equipment power consumption, and the static and dynamic power consumption of the RIS (Residual Power Supply). Its expression is as follows: in, This indicates the total power consumption of the system. This represents the power consumption of a practical discrete RIS. It represents all power consumption related to wireless communication transmission, including the power consumed by base stations and mobile devices during signal reception, decoding and transmission, and the energy consumption of base stations and user equipment for signal transmission transmission. This represents the total power consumption of all RIS units; This indicates the power loss caused by noise.

6. The method according to claim 5, characterized in that, The expression for all power consumption related to wireless communication transmission is: in, This indicates the total power consumption of the base station system. Indicates the base station's transmit power. This indicates the total power consumption of the user equipment.

Images