Beam forming method of active RIS-assisted green communication network

Abstract

A beam forming method of an active RIS auxiliary green communication network comprises the following steps: constructing a network IREE index of a maximized active RIS auxiliary system based on outdated CSI according to a downlink transmission system, wherein the network IREE index comprises two beam forming matrixes and is used as an optimization problem; and splitting the optimization problem into two sub-problems which respectively comprise a beam forming matrix, and alternately solving the two sub-problems through an alternate optimization algorithm to finally obtain the optimized beam forming matrixes at the base station and the active RIS. According to the method, the beamforming at the base station and the RIS is jointly optimized under the background of considering outdated CSI, so that the centralized energy efficiency of the whole system is maximized, and the user demand and the capacity can be well matched.

Description

Technical Field

[0001] The present invention relates to a technology in the field of wireless communication, in particular to a beamforming method of an active reconfigurable intelligent surface (RIS) assisted green communication network. Background Art

[0002] The rapid development of communication technology has driven the growth of energy-saving network architecture and performance indicators. As a key component of green network architecture, active RIS can reflect and amplify signals, overcome double multiplicative fading, and improve system performance. However, the performance indicators of existing RIS-assisted communication systems (such as interruption probability, signal-to-noise ratio, traditional energy efficiency (EE), etc.) fail to fully meet the future needs of green communications. In actual networks, user traffic demands vary greatly, and traditional indicators are difficult to effectively match traffic demands with network capacity, which may lead to energy waste. In addition, it is challenging to obtain accurate channel state information (CSI), especially in rapidly changing channel environments. Existing studies mostly use dual-time scale channel estimation schemes, which are difficult to meet the needs and development of future green communications. Summary of the invention

[0003] Aiming at the problem that the prior art cannot handle the impact of outdated CSI, cannot match the actual network capacity and user needs, and cannot solve the problem that the actual reflection coefficient of RIS cannot be continuously adjusted, the present invention proposes an active RIS-assisted beamforming method for green communication network, which jointly optimizes the beamforming at the base station and RIS under the background of outdated CSI, thereby maximizing the centralized energy efficiency of the entire system and achieving a good match between user needs and capacity.

[0004] The present invention is achieved through the following technical solutions:

[0005] The invention relates to a beamforming method for an active RIS-assisted green communication network. After constructing a network IREE index of an active RIS-assisted system based on outdated CSI and containing two beamforming matrices as an optimization problem according to a downlink transmission system, the optimization problem is split into two sub-problems, each containing a beamforming matrix, and the two sub-problems are alternately solved through an alternating optimization algorithm, so as to finally obtain the optimized beamforming matrices at a base station and an active RIS.

[0006] The downlink transmission system is an active RIS-assisted downlink transmission system, comprising: a base station equipped with multiple transmitting antennas and a RIS having multiple reflecting elements, and the base station communicates with multiple single-antenna users. Technical Effects

[0007] Based on the average achievable rate under outdated CSI conditions, the present invention constructs a network IREE index optimization framework for an active RIS-assisted communication system and decomposes complex non-convex problems into iteratively solvable sub-problems by alternating optimization successive approximation algorithms, using quadratic transformation, relaxation technology and successive approximation, thereby achieving the optimization of RIS reflection coefficients in discrete domains. Compared with the prior art, the present invention can effectively reduce the impact of outdated CSI, effectively match user traffic demand with network capacity, avoid the problem of uneven resource allocation under traditional EE indicators, significantly improve the resource utilization efficiency of the system, reduce unnecessary energy waste, and can quickly converge to performance close to continuous domain optimization under the condition of discrete adjustment of RIS reflection coefficients, reduce the complexity of hardware implementation, and ensure the optimality of system performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 This is a schematic diagram of the active RIS-assisted green communication scenario;

[0009] Figure 2 is a flow chart of an embodiment;

[0010] Figure 3 Optimize the flow chart for alternation;

[0011] Figure 4 is a flow chart of step b;

[0012] Figure 5 This is a schematic diagram for comparing the performance of different algorithms;

[0013] Figure 6 It is a schematic diagram of performance comparison under different quantization levels;

[0014] Figure 7 It is a schematic diagram of performance comparison under different channel coefficients;

[0015] Figure 8 This is a diagram comparing the QOS of the EE and IREE indicators. DETAILED DESCRIPTION

[0016] like Figure 2 As shown, this embodiment relates to a beamforming method for an active RIS-assisted green communication network, including:

[0017] Step 1: construct an optimization problem according to the downlink transmission system;

[0018] like Figure 1 As shown, the downlink transmission system includes: N t A base station with K transmitting antennas, K single-antenna users communicating simultaneously, and N RIS The active RIS of the kth user Signal to Interference and Noise Ratio Corresponding to its transmission rate R k (Θ, W) = BW × log 2 (1+γ k ),in: is the channel parameter of the link from the base station to the RIS, is the channel parameter between RIS and k users, and denote the additive white Gaussian noise (AWGN) of the active RIS and the kth user, respectively, and x k represents the normalized transmitted symbol of the kth user, and denote the beamforming matrices of the base station and RIS respectively; BW is the bandwidth of the corresponding system.

[0019] Due to the timeliness of the channel and the error in channel estimation, the estimated channel response and the true channel response h ru，k Satisfy between: Where: ρ∈(0,1) represents the correlation coefficient, Represents the estimation error. Based on the above settings, the average achievable rate of user k can be obtained in:

[0020] The total power consumption of the downlink transmission system includes the base station power consumption P BS and RIS power consumption P RIS , specifically: in: and P u They represent the static power consumption of the base station, active RIS and active RIS unit, μ BS and μ RIS Indicates the power amplification factor of the base station and RIS.

[0021] To ensure the quality of service for each user, a minimum average transmission rate is guaranteed Right now

[0022] Since the total transmission power of the base station and RIS is limited, that is, and

[0023] Due to the hardware limitations of the RIS components, the reflection coefficient {θ n}From size 2 L The discrete set Ω θ Select in.

[0024] The optimization problem is to maximize the network IREE index of the active RIS-assisted system based on outdated CSI, that is, the throughput demand of the k-th user, specifically: in: Θ and W are beamforming matrices respectively.

[0025] Step 2: Solve the optimization problem in step 1 based on the alternating optimization algorithm, specifically including:

[0026] 2.1 Use the quadratic transformation method to transform the original problem into Among them: y is an auxiliary variable;

[0027] 2.2 Further transform the objective function into: in: Slack variable u=(u 1 ,...,u k , ..., u K );

[0028] 2.3 Constraints Translates to: Finally, by applying the quadratic transformation method again, the optimization problem is transformed into: in: Where: slack variable v = (v 1 ,...,v k , ..., v K ), slack variable z=(z 1 , ..., z k , ..., z K );

[0029] 2.3 As Figure 3 As shown, the optimization problem in step 1 is converted into two sub-problems, and the optimal solution is obtained through iteration by alternately optimizing the two sub-methods, that is, optimizing W and Θ.

[0030] The sub-question 1 refers to:

[0031] The sub-question 2 refers to:

[0032] like Figure 3As shown, the alternating optimization includes:

[0033] Step a: Optimize W while keeping Θ fixed. Specifically, fix all variables except y and solve subproblem 1 to obtain the optimal value. Then, fix all variables except z and solve Problem 1 to obtain the optimal value z * , After fixing y, z, and θ, subproblem 1 becomes a concave problem and can be solved using toolboxes such as CVXPY.

[0034] Step b: Figure 4 As shown, optimizing Θ while keeping W fixed includes:

[0035] b1 relaxes subproblem 2, j relaxes Θ to the continuous domain, and then optimizes Θ while keeping W fixed, and maps Θ from the continuous domain back to the discrete domain through the successive approximation algorithm, and constrains The discrete values ​​in are relaxed to continuous values, i.e. |θ n |≤α max , define T ki ′=Q ki diag(H br w i )and Sub-problem 2 is formulated as sub-problem 3, which is:

[0036] b2 fixes all variables except y and solves subproblem 3 to obtain the optimal value Then, fix all variables except z and solve the problem to obtain the optimal slack variable value z * , At this point, after fixing y, z and Θ, the above becomes a concave problem, which can be solved using toolboxes such as CVXPY.

[0037] b3 maps Θ from the continuous domain back to the discrete domain by successive approximation, thereby achieving the goal of solving subproblem 2 in step b. Specifically, solve the subproblem in step b2 to obtain the beamforming matrix Θ * After that, it is mapped to the discrete domain to obtain the discretized value Specifically: calculate The continuous value before mapping The distance between them is normalized and obtained Choose the beamforming matrix Θ * The index of the half with the smallest distance is added to the set S for the next iteration. In subsequent iterations, select Θ* The index of the element in S remains unchanged, and then the above process is repeated until the predetermined number of iterations Q is reached. Finally, the updated Θ is obtained * .

[0038] Step c: Determine η IREE Whether it converges, when Exit the loop, otherwise repeat steps a, b, and c until convergence.

[0039] The effectiveness of the present invention is verified by simulation numerical experiments: a simulation system is composed of 12 transmitting antennas, an active RIS containing 20 elements and 8 single-antenna users. The base station is located at the origin, and the active RIS is deployed along the x-axis, 10 meters away from the base station. The 8 users are evenly distributed along a quarter circle with a radius of 80 meters and centered on the base station. In addition, due to the low penetration ability of millimeter waves and the influence of obstacles, the direct link between the base station and the user is blocked.

[0040] In the simulation system, the channel parameters are configured according to the Saleh-Valenzuela (SV) channel model, a typical millimeter wave communication model, specifically: The power consumption of each RIS unit is P for active RIS. u =P DC +P r , for passive RIS, it is P u =P r , where P r The power consumption for 1-bit, 2-bit, and 3-bit phase quantization levels is 1.5mW, 3mW, and 4.5mW, respectively. In addition, the DC power consumption of the amplifier of the active RIS is 20.5mW. The quantization levels of amplitude and phase are set to a=2 and b=2, respectively. The total power budget of the system is set to in Minimum user requirements C min,k Set to 0.1D k , the maximum limit is 1Mbit / s. The traffic demand of each user is D k from a lognormal distribution.

[0041] Two different traffic scenarios are considered in the following experiments: Scenario 1: high average traffic, variance is 10Mbit / s, and mean is 60Mbit / s. Scenario 2: low average traffic, variance is 10Mbit / s, and mean is 100Mbit / s.

[0042] The overall experimental results are obtained by averaging multiple Monte Carlo samplings. The specific parameters are shown in Table 1.

[0043] Table 1 Experimental parameters Base station static power consumption 9dBW RIS static power consumption 1.5W RIS and base station power amplifier energy efficiency coefficient 1.25 Channel correlation coefficient 0.95 Amplifier maximum gain factor 23dB RIS antenna gain 19.8dBi Transmitting antenna gain 12.98dBi Receiving antenna gain 5.51dBi

[0044] The scheme for comparing with the above-mentioned present invention includes: the continuous condition is to assume that the beamforming matrix of RIS can be adjusted continuously; the EE benchmark is to calculate the corresponding η under the same experimental conditions IREE ; The number of elements of passive RIS is set to N max =100 to ensure the same power consumption as active RIS.

[0045] like Figure 5 As shown in the figure, the experimental results show that compared with the passive RIS benchmark, the present invention has significant advantages in both scenario 1 and scenario 2. In addition, the present invention has also significantly improved over the EE benchmark, with improvements of 63.1% and 40.5% in scenario 1 and scenario 2, respectively. This is because the traditional EE indicator fails to effectively match user demand and traffic, resulting in low IREE performance. The present invention significantly improves IREE performance by effectively matching user demand with the provided capacity.

[0046] Finally, compared with the continuous condition, the overall performance of the present invention is closer because in each iteration, the present invention gradually approaches the optimal solution in the discrete domain and can obtain results close to the ideal situation at a low quantization level.

[0047] like Figure 6 As shown in Figure 1, the effect of quantization levels a and b on IREE is shown. The experimental results show that when a=2 and b=2, the performance gap is only 6.5% and 10.7% compared to the ideal situation. It is worth noting that the gap in scenario 2 is slightly larger, mainly because the higher traffic demand requires more energy to compensate for the impact of the discretization of the reflection coefficient. The present invention approximates the Θ value before discretization multiple times to ensure that even if the reflection coefficient of RIS cannot be adjusted continuously, the optimized IREE value is still close to the ideal value. The experimental results verify that when using the present invention, only a moderate quantization level is required to obtain performance close to the ideal situation, thereby greatly reducing the complexity of hardware adjustment.

[0048] like Figure 7 The figure shows the impact of outdated CSI on IREE. By comparing the optimization without considering outdated CSI (using Rk) and considering outdated CSI (using ) shows that as ρ decreases, the impact of outdated CSI gradually increases, resulting in a decrease in system performance. This is mainly because outdated CSI adds additional noise and reduces the power of useful signals, thereby reducing SINR and ultimately causing system performance to deteriorate.

[0049] In addition, considering outdated CSI can significantly improve performance compared to ignoring outdated CSI. For example, when ρ = 0.9, scenario 1 and scenario 2 are improved by 26.7% and 29.4% respectively. This improvement is due to the fact that considering outdated CSI can effectively alleviate the interference of other users and the noise caused by outdated CSI, thereby improving the received SNR and improving system performance.

[0050] This experiment further evaluates the improvement of IREE compared to EE by using the network utility index as a measure of the quality of service (QoS) of the communication system. This index is used to evaluate the degree of match between user demand and actual system capacity. When the actual capacity completely matches the user demand, QoS = 1; conversely, when the match is completely inconsistent, QoS = 0. Figure 8 As shown, the use of IREE significantly improves QoS performance compared to EE. For example, in scenario 1, QoS is improved by as much as 187.5%. This improvement is mainly because IREE can effectively match user demand and system capacity, avoiding over-provisioning of capacity to users with lower demand, while users with higher demand can obtain more capacity. In contrast, traditional EE indicators fail to effectively balance user demand and capacity, resulting in poor performance. In addition, in scenario 1, since user demand is relatively low and relatively easy to meet, QoS performance is better than scenario 2. This also shows that under the same power allocation, IREE can better match user demand, thereby reducing unnecessary energy waste. However, the EE indicator performs poorly under different traffic conditions, with almost no difference between scenario 1 and scenario 2, indicating that EE fails to effectively match user demand with traffic distribution.

[0051] In summary, compared with the prior art, the present invention introduces the average achievable rate model based on outdated CSI and the network IREE index, combined with the alternating optimization successive approximation algorithm, which can effectively maximize the IREE index while better matching the network capacity of the RIS-assisted millimeter wave communication system with the user traffic demand, taking into account both the matching of user demand and network capacity and the maximization of energy efficiency.

[0052] The above-mentioned specific implementation can be partially adjusted in different ways by those skilled in the art without departing from the principle and purpose of the present invention. The protection scope of the present invention shall be based on the claims and shall not be limited by the above-mentioned specific implementation. Each implementation scheme within its scope shall be subject to the constraints of the present invention.

Claims

1. An active RIS-assisted green communication network beamforming method, characterized in that: After constructing the network IREE index of the active RIS assisted system based on outdated CSI, which contains two beamforming matrices according to the downlink transmission system, as an optimization problem, the optimization problem is divided into two sub-problems, each containing a beamforming matrix, and the two sub-problems are solved alternately through an alternating optimization algorithm, and finally the optimized beamforming matrices at the base station and the active RIS are obtained; The downlink transmission system is an active RIS-assisted downlink transmission system, comprising: a base station equipped with multiple transmitting antennas and a RIS having multiple reflecting elements, and the base station communicates with multiple single-antenna users.

2. The active RIS-assisted green communication network beamforming method according to claim 1, characterized in that: The downlink transmission system specifically includes: t A base station with K transmitting antennas, K single-antenna users communicating simultaneously, and N RIS The active RIS of the kth user Signal to Interference and Noise Ratio Corresponding to its transmission rate R k (Θ,W)=BW×log2(1+γ k ),in: is the channel parameter of the link from the base station to the RIS, is the channel parameter between RIS and k users, and denote the additive white Gaussian noise (AWGN) of the active RIS and the kth user, respectively, and x k represents the normalized transmitted symbol of the kth user, and represents the beamforming matrix of the base station and RIS respectively; BW is the bandwidth of the corresponding system; the estimated channel response and the true channel response h ru,k Satisfy between: Where: ρ∈(0,1) represents the correlation coefficient, Represents the estimation error. Based on the above settings, the average achievable rate of user k is obtained. in: The total power consumption of the downlink transmission system includes the base station power consumption P BS and RIS power consumption P RIS , specifically: in: and P u They represent the static power consumption of the base station, active RIS and active RIS unit, μ BIS and μ RIS Indicates the power amplification factor of the base station and RIS.

3. The active RIS-assisted green communication network beamforming method according to claim 2, characterized in that: The optimization problem is to maximize the network IREE index of the active RIS-assisted system based on outdated CSI, that is, the throughput demand of the k-th user, specifically: in: Θ and W are beamforming matrices respectively.

4. The active RIS-assisted green communication network beamforming method according to claim 2, characterized in that: The splitting refers to: 2.1 Use the quadratic transformation method to transform the original problem into Among them: y is an auxiliary variable; 2.2 Further transform the objective function into: in: Slack variables u=(u1,…,u k ,…,u K ); 2.3 Constraints Translates to: Finally, by applying the quadratic transformation method again, the optimization problem is transformed into: in: Where: slack variable v = (v1,…,v k ,…,v K ), slack variables z=(z1,…,z k ,…,z K ); 2.3 By converting the optimization problem in step 1 into two sub-problems, the two sub-methods are optimized alternately, that is, W and Θ are optimized, and the optimal solution is obtained through iteration; The sub-question 1 refers to: The sub-question 2 refers to:

5. The beamforming method of active RIS-assisted green communication network according to claim 2 or 4, characterized in that: The alternating optimization includes: Step a: Optimize W while keeping Θ fixed. Specifically, fix all variables except y and solve subproblem 1 to obtain the optimal value. Then, fix all variables except z and solve Problem 1 to obtain the optimal value z * , ; After fixing y, z and Θ, subproblem 1 becomes a concave problem and is solved using toolboxes such as CVXPY; Step b, optimizing θ while keeping W fixed, specifically includes: b1 relaxes subproblem 2, j relaxes Θ to the continuous domain, and then optimizes Θ while keeping W fixed, and maps Θ from the continuous domain back to the discrete domain through the successive approximation algorithm, constraining θ n ∈Ω θ , The discrete values ​​in are relaxed to continuous values, i.e. |θ n |≤α max , define T ki '=Q ki diag(H br w i )and Sub-problem 2 is formulated as sub-problem 3, which is: b2 fixes all variables except y and solves subproblem 3 to obtain the optimal value Then, fix all variables except z and solve the problem to obtain the optimal slack variable value z * , At this time, after fixing y, z and θ, use the CVXPY toolbox to solve; b3 maps Θ from the continuous domain back to the discrete domain by successive approximation, thereby achieving the goal of solving subproblem 2 in step b. Specifically, solve the subproblem in step b2 to obtain the beamforming matrix Θ * After that, it is mapped to the discrete domain to obtain the discretized value Specifically: calculate The continuous value before mapping The distance between them is normalized and obtained Choose the beamforming matrix Θ * The index of the half with the smallest distance is added to the set S for the next iteration. In subsequent iterations, select Θ * The index of the element in S remains unchanged, and then the above process is repeated until the predetermined number of iterations Q is reached, and finally the updated Θ is obtained * ; Step c: Determine η IREE Whether it converges, when Exit the loop, otherwise repeat steps a, b, and c until convergence.

Images