Energy efficiency optimization method for massive MIMO system based on nonlinear energy harvesting

Abstract

The present invention claims a method for optimizing the energy efficiency of a large-scale MIMO system based on nonlinear energy harvesting. The present invention considers using a nonlinear energy harvesting circuit model at the sensor node end (user) to complete energy collection, considers fairness between sensor node ends (users) under the condition of satisfying the user's QoS (quality of service), and constructs an optimization model for maximizing the energy efficiency of the minimum user in a wireless power supply large-scale MIMO (multiple-input multiple-output) system. Since the original problem is non-convex, the present invention first converts the original problem into a convex optimization problem through variable substitution and formula deformation. For the existing non-convex constraints, the first-order Taylor expansion, SCA (continuous convex approximation) algorithm and other methods are used for equivalent transformation, and then the interior point method is combined for solution, and an effective resource allocation strategy is proposed on this basis. Finally, through simulation verification, the proposed method can significantly improve the energy efficiency performance of the minimum user and meet the communication needs of edge users.

Description

Technical Field

[0001] The present invention belongs to the field of resource allocation in a large-scale MIMO system, and specifically, to a method for allocating combined power and time resources in a large-scale MIMO system with wireless power supply based on a nonlinear energy harvesting model. Background Art

[0002] With the rapid development of communication technology, the number of mobile data broadband users has also continued to increase, resulting in a significant increase in the energy consumption of communication systems. The energy consumed by information and communication technology accounts for a gradually increasing proportion of the total energy consumption in various fields around the world. Therefore, in order to obtain higher spectrum efficiency and energy efficiency (EE) in limited wireless spectrum resources, massive MIMO technology has become an important technical means. Due to the large number of antennas, massive MIMO technology can effectively eliminate the interference between users, improve the system frequency efficiency, and achieve the goal of high-speed data transmission in wireless communication networks; on the other hand, in order to solve the problem of energy supply, researchers have conducted research on energy harvesting (EH) technology. EH technology can collect radio frequency (RF) energy in the environment and convert it into usable electrical energy. The way of collecting energy through EH technology can replace traditional wired towers to provide energy for wireless devices. Because of this characteristic, wireless power transmission (WPT) technology that can complete energy harvesting has become an effective way to solve the problem of energy shortage in energy-constrained networks or devices. As an extension of WPT technology, wireless power communication network (WPCN) has also received widespread attention in recent years. WPCN uses a store-and-forward transmission protocol for communication, which can effectively solve the problem of high transmission power consumption in massive MIMO systems.

[0003] Since energy-saving technologies can reduce network energy consumption and extend network life, EE has become an important performance indicator for future wireless networks. Xu J, Zhu P, Li J et al. studied the power allocation problem based on physical layer security in the paper "Secrecy energy efficiency optimization for multi-user distributed massive MIMO systems, IEEE Transactions on Communications, 2020, vol. 68, no. 2, pp. 915-929". They first measured the confidentiality EE using the global average confidentiality energy efficiency measurement method, and then proposed the confidentiality EE maximization problem of joint power allocation, long-distance antenna unit clusters, long-distance antenna unit selection, and artificial noise selection. Finally, they proposed a two-layer iterative algorithm to obtain the approximate optimal solution to the problem.

[0004] Combining massive MIMO technology with WPT technology can realize the function of wireless power supply for the network, and on this basis, an effective resource allocation algorithm is proposed, which helps to meet the energy saving and consumption reduction needs of the network. Wang Z, Lv T studied the energy efficiency optimization problem of multi-cell massive MIMO-NOMA networks based on WPT in the paper "Energy-efficient and distributed resource allocation for WPT enabled multicell massive MIMO-NOMA networks, 2020IEEE 31st Annual International Symposium on Personal, Indoor and Mobile Radio Communications, 2020, pp. 1-6". The paper first proposed a system energy efficiency maximization problem of jointly optimizing power, time, subcarrier allocation and antenna selection, and then used nonlinear programming to transform the original problem into a convex optimization problem. Finally, a distributed resource allocation algorithm based on alternating direction multipliers was proposed to solve the problem. Li B, Dai Y, Dong Z et al. studied a millimeter-wave massive MIMO system based on ultra-dense heterogeneous networks in the paper "Energy-efficient resources allocation with millimeter-wave massive MIMO in ultra dense HetNets by SWIPT and CoMP, IEEE Transactions on Wireless Communications, 2021, vol. 20, no. 7, pp. 4435-4451". Considering the use of SWIPT technology to obtain energy from the environment, a system energy efficiency optimization problem based on multi-point coordinated transmission and reception was first proposed. Then, based on the Dinkelbach algorithm, an iterative optimization algorithm was designed to solve the problem. Finally, simulation verified the effectiveness and fast convergence of the proposed algorithm compared with the baseline scheme.

[0005] Fairness is an important issue in wireless network resource allocation. When allocating power, due to the channel differences between users, only considering the optimization of the global energy efficiency of the system may lead to unfair resource allocation, resulting in users with extremely poor energy efficiency. The maximum-minimum criterion, also known as the Max-Min criterion, was first used to solve the flow control problem in the network. It has outstanding performance in achieving absolute fairness among all users and is also the most commonly used fairness criterion in communication systems. L Zhao, X Wang et al. considered a massive MIMO system in which a base station sends information and energy to multiple energy harvesting receivers in the document "Massive MIMO downlink for wireless information and energy transfer with energy harvesting receivers, IEEE Transactions on Communications, 2019, vol. 67, no. 5, pp. 3309-3322". By jointly designing the power allocation ratio of the base station and the power splitting (or time switching) factor of the receiver, the problems of maximizing the minimum transmission rate and maximizing the system energy efficiency between receivers were studied respectively. Huang K et al. published a paper titled "Max-min energy efficiency optimization algorithm for wireless power transfer enabled massive MIMO systems" at the 2019 IEEE 5th International Conference on Computer and Communications (ICCC) conference. The paper studied the energy-carrying massive MIMO system using the receive-before-send protocol and proposed a power and time allocation algorithm based on an iterative algorithm to maximize the energy efficiency of the minimum user. The simulation results show the effectiveness of the algorithm. In addition, based on the linear EH model, the Chinese invention patent ZL201910058638.3 "Maximization and minimum energy efficiency resource allocation method in wireless power supply massive MIMO network" proposed an energy efficiency resource allocation algorithm for wireless power supply massive MIMO system based on user Max-Min fairness. However, the patent considers linear energy harvesting, and the proposed optimization method cannot directly handle nonlinear energy harvesting systems. At the same time, the invention patent ZL201910058638.3 assumes that the fixed power consumption of the user is a constant value, and the energy consumed by this part of the fixed power consumption comes from the user's own stored electricity, without considering that the fixed power consumption comes from energy harvesting. Therefore, this method cannot guarantee that the system will continue to work sustainably.Therefore, it is urgent to further consider the case of nonlinear energy harvesting, and at the same time consider that the user's transmission power consumption and fixed power consumption are subject to the causal constraints of energy harvesting, and design a method for resource allocation of wireless power supply large-scale MIMO systems that can handle nonlinear energy harvesting.

[0006] In summary, researchers at home and abroad have conducted many studies on wireless power supply large-scale MIMO systems and proposed various resource allocation algorithms to effectively improve the performance of the system. In addition, in the WPCN scenario, the conversion efficiency of the receiving end converting the received RF signal power into direct current (DC) power mainly depends on the EH circuit. At present, most research works are carried out around the linear EH model. However, the linear EH model is an idealized model, which is usually inconsistent with the actual measurement results. The actual measurement results show that the energy conversion of RF-DC is usually nonlinear. Therefore, the present invention considers a wireless power supply large-scale MIMO system based on the nonlinear EH model, and at the same time, considering the need to ensure user fairness, the energy transmission time and power allocation are combined in the wireless power supply large-scale MIMO system to optimize the minimum user energy efficiency. Summary of the invention

[0007] The present invention aims to solve the above problems of the prior art. A method for optimizing the energy efficiency of a wireless large-scale MIMO system based on nonlinear energy harvesting is proposed. The technical solution of the present invention is as follows:

[0008] A method for optimizing energy efficiency of a wireless large-scale MIMO system based on nonlinear energy harvesting comprises the following steps:

[0009] 101. Establish a wireless power supply massive MIMO system that combines massive MIMO technology with a wireless power supply communication network WPCN. In this system, the power beacon PB and the base station BS are equipped with a large number of antennas. In the system, the power beacon PB transmission power constraint, energy transmission time constraint, user minimum QoS constraint, and energy consumption constraint are considered. On this basis, a Max-Min optimization problem model that maximizes the minimum user energy efficiency is constructed. The Max-Min optimization problem model is a non-convex optimization problem.

[0010] 102. Based on a simplified calculation method including variable substitution and first-order Taylor expansion, the non-convex Max-Min energy efficiency optimization problem in step 101 is converted into an equivalent convex optimization problem; the variable substitution is to introduce auxiliary variables to simplify the original objective function and add constraints containing the introduced variables; the first-order Taylor expansion is to use Taylor expansion to perform first-order expansion on non-convex constraints to convert the constraints into convex constraints;

[0011] 103. Then, the interior point method and the continuous convex approximation SCA algorithm are used to solve the convex optimization problem equivalently transformed in the above step 102, and the power allocation strategy is solved, including: the transmission power p allocated by PB to user k k , the transmission power p of user k to transmit information to BS t,k , energy transfer time τ, and thus the optimal energy efficiency based on the minimum number of users under all constraints can be obtained.

[0012] Furthermore, the Max-Min optimization problem of maximizing the minimum user energy efficiency based on the wireless power supply massive MIMO system in step 101 is:

[0013]

[0014] C2:0≤P≤P max ,

[0015] C3:0≤τ≤1,

[0016] C4:p k ≥0,k=1,…,K,

[0017] C5:p t,k ≥0,k=1,…,K,

[0018] C6:E tot,k (p t,k ,τ)≤E hst,k (p k ,τ),k=1,…,K,

[0019]

[0020] In order to simplify the expression, the parameter M is introduced k , and there is p=(p 1 ,p 2 ,…,p K ) represents the power vector allocated by PB to user k, p t =(p t,1 ,p t,2 ,…,p t,K ) represents the transmission power vector of user k sending information to BS, K, M, N represent the number of single-antenna users, BS antennas, and PB antennas, respectively, and min{M,N}>>K; α k , β k Respectively represent the path loss from user k to BS and PB to user k; σ 2 Represents noise power; P max represents the maximum transmit power of PB; P represents the total transmit power of PB; represents the energy consumption of user k, where pr,k represents the fixed power consumption of the receiving circuit of user k, μ is a constant factor reflecting the inefficiency of the power amplifier, and p c,k represents the fixed power consumption of the transmitting circuit of user k; considering the normalization of time unit, τ represents the time for PB to transmit energy to the user node, and 1-τ is the time for the user node to complete information transmission using the collected energy; represents the energy collected by user k through the nonlinear energy collection model, A k Indicates the maximum energy harvesting power at the user end when the energy harvesting circuit is saturated, constant a k and b k are parameters related to the circuit, expressed as circuit sensitivity and leakage current, represents the zero-input zero-output response of the model; Indicates the minimum signal to interference and noise ratio SINR requirement of the user.

[0021] Furthermore, in step 101, constraints C1 and C2 are combined and considered, and the optimization problem is written as:

[0022]

[0023] C2:0≤τ≤1,

[0024] C3:p k ≥0,k=1,…,K,

[0025] C4:p t,k ≥0,k=1,…,K,

[0026] C5:E tot,k (p t,k ,τ)≤E hst,k (p k ,τ),k=1,…,K,

[0027]

[0028] Constraint C1 means that the sum of the power allocated to all users needs to be less than the maximum transmit power P of PB max ; Constraint C2 represents the energy transmission time allocation constraint; C3 and C4 represent the power constraint allocated to user k by PB and the transmission power constraint of user k respectively; C5 represents that the total energy consumed by each user must be less than or equal to the energy collected from PB, where the collected energy is obtained through a nonlinear energy collection circuit; C6 is the minimum SINR constraint, requiring that the SINR of each user must be greater than the minimum SINR constraint It can be seen that the above problem is a non-convex problem.

[0029] Furthermore, the non-convex Max-Min energy efficiency optimization problem in step 101 is converted into an equivalent convex optimization problem based on a simplified calculation method including variable substitution and first-order Taylor expansion, specifically:

[0030] 401. Introduce auxiliary variable t, and let The energy efficiency optimization problem can be rewritten as:

[0031]

[0032]

[0033] C2:x≥0,

[0034] C3:p k ≥0,k=1,…,K,

[0035] C4:p t,k ≥0,k=1,…,K,

[0036]

[0037]

[0038] Among them, C7 is the equivalent treatment of the original objective function, using the auxiliary variable t to be equivalent to the minimum user energy efficiency, then user k, The energy efficiency of must be greater than or equal to the minimum user energy efficiency t, which simplifies the objective function;

[0039] 402. Let t = e s , By variable substitution, we introduce x=e z , Because t = e s is a monotonically increasing function of s, therefore, the optimization problem of step 401 can be rewritten as:

[0040]

[0041]

[0042] C2:p k ≥0,k=1,…,K,

[0043]

[0044]

[0045]

[0046] At this point, the optimization variable considered has been transformed into v = (v 1 ,…,vK ), p=(p 1 ,p 2 ,…,p K ), z, s, where K, M, and N represent the number of single-antenna users, BS antennas, and PB antennas, respectively, and min{M,N}>>K; constraint C1 indicates that the sum of the power allocated to all users needs to be less than the maximum transmit power P of PB. max ; Constraint C2 represents the power constraint allocated by PB to user k; C3 indicates that the total energy consumed by each user must be less than or equal to the energy collected from PB; C4 is the minimum SINR constraint, requiring that the SINR of each user must be greater than the minimum SINR constraint C5 is the constraint that the energy efficiency of any user must be greater than or equal to the minimum user energy efficiency;

[0047] 403. Constraint C1 is a linear function, C1 is a convex constraint; similarly, constraint C2 is also a convex constraint, and constraints C3, C4, and C5 are convex.

[0048] Furthermore, the original problem in step 402 can be equivalent to a convex optimization problem:

[0049]

[0050]

[0051] C2:p k ≥0,k=1,…,K,

[0052] C3:f k (v k ,z)+g k (p k ,z)≤0,k=1,…,K,

[0053]

[0054]

[0055] Among them, in constraint C3, It is represented as the expression introduced in the process of formula transformation, It is expressed as the expression introduced in the process of formula transformation. Since f k (v k ,z) and g k (p k ,z) are all convex functions, so C3 is a convex constraint; in C5, It is expressed as the expression introduced in the process of formula transformation, and L k (z,s,v k ) is a convex constraint, is the intermediate expression in the formula transformation, which can be obtained by k (v k ) is expanded by first-order Taylor to obtain an equivalent expression:

[0056]

[0057] Where n represents the number of iterations, Indicates the expansion point, For about v k The SCA algorithm can be used to approximate the approximate expression to the original function. Therefore, in the equivalent problem, C5 is a convex constraint.

[0058] Furthermore, in step 103, the step of solving the convex optimization problem by using the SCA algorithm and the interior point method includes:

[0059] 501. Initialize all relevant input parameters of the problem: the maximum total transmit power P of PB max , the large-scale fading α from user k to BS k , the large-scale fading β from PB to user k k , the number of PB antennas N, the number of BS antennas M, the number of single-antenna users K, and the fixed power consumption p c,k 、p r,k , the minimum signal-to-interference-noise ratio requirement of the sensor node Gaussian noise power σ 2 , nonlinear energy harvesting circuit parameters A k , a k , b k , Iteration number n = 1, maximum iteration number N max , convergence threshold value ε;

[0060] 502. Solve the problem by the interior point method, that is, construct a new unconstrained objective function - penalty function, define the penalty function in the feasible domain, and solve the penalty function in the feasible domain. The extreme point is always inside the feasible domain. In this way, the solution of the unconstrained optimization problem is always a feasible solution, so the optimal solution of the original optimization problem can be approached successively in the feasible domain.

[0061] 503. The complexity of the algorithm depends on the number of iterations n and the convergence threshold ε. Since the complexity of the interior point method is Where x and y represent the number of optimization variables and the number of constraints respectively; the number of optimization variables in the proposed problem is x 1 =2K+2, the number of constraints is y 1 =4K+1, so the computational complexity of the algorithm is

[0062] The advantages and beneficial effects of the present invention are as follows:

[0063] The present invention establishes an energy efficiency optimization problem based on the Max-Min fairness criterion while considering the nonlinear characteristics of the EH model. In step 102, the original non-convex problem is considered to be converted into an equivalent convex optimization problem by variable substitution, SCA and other methods. Compared with other traditional wireless power supply large-scale MIMO systems, the present invention has the advantages of low polynomial complexity and simple solution. At the same time, the nonlinear characteristics of the energy harvesting model are considered, which is closer to the actual situation in application. In addition, QoS is guaranteed from the perspective of user SINR. While being innovative, the nonlinear EH model and fairness criterion are considered, making the present invention more in line with reality. The present invention is suitable for wireless power supply large-scale MIMO systems based on nonlinear energy harvesting under fairness conditions, and has good feasibility and practicality. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 The present invention provides a preferred embodiment of a large-scale MIMO system model based on nonlinear energy harvesting;

[0065] Figure 2 The minimum user energy efficiency of the present invention under different PB transmission powers compared with the linear EH model;

[0066] Figure 3 The total energy efficiency of users under different PB transmission powers for the present invention and the comparative linear EH model;

[0067] Figure 4 The minimum user energy efficiency of the present invention is compared with that of the comparative linear EH model under different BS antenna numbers (100-150) and the number of users is 9 and 13 respectively;

[0068] Figure 5 The total energy efficiency of users under different BS antenna numbers is compared between the present invention and the comparative linear EH model when the number of users is 9 and 13 respectively;

[0069] Figure 6 This is a preferred flow chart of the present invention. DETAILED DESCRIPTION

[0070] The following will describe the technical solutions in the embodiments of the present invention in detail in conjunction with the accompanying drawings in the embodiments of the present invention. The described embodiments are only part of the embodiments of the present invention.

[0071] The technical solution of the present invention to solve the above technical problems is:

[0072] This embodiment is a method for maximizing the minimum energy efficiency of users in a wireless power supply large-scale MIMO system based on nonlinear energy harvesting under the Max-Min fairness criterion. PB transmits energy to users, and users transmit data to BS. The XY coordinate system is used to fix the position relationship between PB, BS and user groups. The coordinates of PB and BS are located at (25,0)m and (-25,0)m respectively. During initialization, the number of antennas of PB and BS is set to N=60 and M=100 respectively, and the distribution range of K users is within the rectangular area of ​​[-20,20]×[-20,20]. The large-scale fading from PB to the user end is expressed as d PB,k is the distance from PB to user k; let the distance from user k to BS be d BS,k , the large-scale fading from user k to BS is expressed as The path loss factor of the channel is r = 3. The noise power is set to σ 2 =1×10 -10 W. The fixed power consumption of the receiving and transmitting circuits of user k is set to p r,k =1×10 -4 W, p c,k =1×10 -3 W. In particular, in the nonlinear EH model, the maximum EH power A at the user end is set to 24 mW, the circuit parameters a = 150, b = 0.014; and in the linear EH model for comparison, the energy conversion efficiency η is set to linear is 0.5.

[0073] Step 1: Initialize the maximum total transmit power P of PB max , large-scale fading from user k to BS and PB to user k α k , β k , the number of antennas N and M of PB and BS, the number of users K, and the fixed power consumption p c,k 、p r,k , the minimum SINR requirement for sensor nodes Gaussian noise power σ 2 , nonlinear energy harvesting circuit parameters A k , a k , b k , Iteration number n = 1, maximum iteration number N max , convergence threshold value ε;

[0074] Step 2: Establish a wireless power supply large-scale MIMO system that combines large-scale MIMO technology with WPCN (wireless power communication network) and construct a Max-Min optimization problem model that maximizes the minimum user energy efficiency. Introduce auxiliary variables t, and let Rewrite the optimization problem, and the optimization variables are (p, pt ,x,t);

[0075] Step 3: Introduce variables using variable substitution: x = e z , t=e s , Through the first-order Taylor expansion, we get the updated equivalent convex optimization problem, and the optimization variables become (p, v, z, s);

[0076] Step 4: Solve the problem using the interior point method and obtain the optimal solution using the SCA algorithm

[0077] Step 5: Solve for time allocation The transmission power allocated by PB to user k And the transmission power of user k to transmit information to BS Minimum user maximum energy efficiency * =s (i) ;

[0078] Step 6: Get the optimal solution s * , by maximizing the constraint Determine and calculate the final optimization result of the system;

[0079] Furthermore, the Max-Min optimization problem model for maximizing the minimum user energy efficiency based on the wireless power supply large-scale MIMO system in the second step is:

[0080]

[0081]

[0082] C2:x≥0,

[0083] C3:p k ≥0,k=1,…,K,

[0084] C4:p t,k ≥0,k=1,…,K,

[0085]

[0086]

[0087] in, p=(p 1 ,p 2 ,…,p K ) represents the power vector allocated by PB to user k, p t =(p t,1 ,p t,2 ,…,p t,K) represents the transmission power vector of user k sending information to BS, K, M, N represent the number of single-antenna users, BS antennas, and PB antennas, respectively, and min{M,N}>>K; α k , β k Respectively represent the path loss from user k to BS and PB to user k; σ 2 Represents noise power; P max represents the maximum transmit power of PB; P represents the total transmit power of PB; represents the energy consumption of user k, where p r,k represents the fixed power consumption of the receiving circuit of user k, μ is a constant factor reflecting the inefficiency of the power amplifier, and p c,k represents the fixed power consumption of the transmitting circuit of user k; considering the normalization of time unit, τ represents the time for PB to transmit energy to the user node, and 1-τ is the time for the user to complete information transmission using the collected energy; represents the energy collected by user k through the nonlinear EH model, A k Indicates the maximum energy harvesting power at the user end when the energy harvesting circuit is saturated, constant a k and b k are parameters related to the circuit, expressed as circuit sensitivity and leakage current, represents the zero-input zero-output response of the model; Indicates the minimum SINR requirement of the user.

[0088] Constraint C1 means that the sum of the power allocated to all users needs to be less than the maximum transmit power P of PB max ; Constraint C2 represents the energy transmission time allocation constraint; C3 and C4 represent the power constraint allocated to user k by PB and the transmission power constraint of user k respectively; C5 indicates that the total energy consumed by each user must be less than or equal to the energy collected from PB, where the collected energy is obtained through a nonlinear energy collection circuit; C6 requires that the SINR of each user must be greater than the minimum SINR constraint C7 is an equivalent treatment of the original objective function, using the auxiliary variable t to be equivalent to the minimum user energy efficiency.

[0089] Furthermore, in the third step, the equivalent convex optimization problem updated after variable substitution, formula deformation and first-order Taylor expansion is:

[0090]

[0091]

[0092] C2:p k ≥0,k=1,…,K,

[0093] C3:f k(v k ,z)+g k (p k ,z)≤0,k=1,…,K,

[0094]

[0095]

[0096] in

[0097]

[0098]

[0099]

[0100] Where v = (v 1 ,…,v K ), p=(p 1 ,p 2 ,…,p K ), K, M, and N represent the number of single-antenna users, the number of BS antennas, and the number of PB antennas, respectively, and min{M,N}>>K; the constraints are rewritten as follows: Constraint C1 indicates that the sum of the power allocated to all users needs to be less than the maximum transmission power P of PB max ; Constraint C2 represents the power constraint allocated to user k by PB; C3 represents the equivalent constraint of the total energy consumed by each user; C4 is the equivalent minimum SINR constraint, requiring that the SINR of each user must be greater than the minimum SINR constraint C5 is the equivalent constraint that the energy efficiency of any user must be greater than or equal to the minimum user energy efficiency.

[0101] Furthermore, the interior point method in the fourth step solves the equivalent problem and obtains the optimal solution

[0102] Furthermore, the fifth step can obtain: the time allocation parameter PB allocates to the user node The transmission power allocated by PB to user k And the transmission power of user k to transmit information to BS Minimum user maximum energy efficiency * =s (i) .

[0103] Furthermore, the fifth step is to obtain the optimal solution s * , and calculate the final optimization result of the system.

[0104] In this example, Figure 1The present invention provides a preferred embodiment of a wireless power supply large-scale MIMO system model based on nonlinear energy harvesting. In the figure, PB transmits energy to user nodes through wireless power transmission (WPT), and the user uses the harvested energy to send information to the BS (WIT). Figure 2 It shows that when the number of users K is 9 and 13 respectively, the minimum user energy efficiency of the linear and nonlinear EH models varies with the PB transmission power P max of changes. Figure 3 For different PB transmission power P max Under the condition, the total energy efficiency change curves of users of linear and nonlinear EH models are shown when the number of users is 9 and 13; Figure 4 It shows how the minimum user energy efficiency of the linear and nonlinear EH models varies with the number of base station antennas M when the number of users K is 9 and 13 respectively; Figure 5 The total energy efficiency curves of linear and nonlinear EH models are shown in Figure 2 under different base station antenna numbers M, considering the number of users is 9 and 13. Figure 2 It can be seen that the minimum user energy efficiency of the two EH models increases with P max The value of P increases with the increase of P, and finally becomes flat. This is because the EH circuit has a maximum collection power. When it reaches the saturation area, as P max As increases, the minimum user energy efficiency no longer increases. Figure 3 It can be seen that when the number of users is the same, the total energy efficiency of users in the two models increases with P max As the number of users increases, the total energy efficiency of users in the nonlinear EH model increases until it becomes flat. The total energy efficiency of users in the linear model is higher than that in the linear model. When the number of users increases, the total energy efficiency of users is improved in both EH models because of the existence of multi-user diversity gain. Figure 4 It can be seen that the minimum user energy efficiency of the two models increases with the increase of the number of BS antennas M. This is because when the number of antennas increases, the antenna gain at the BS end will increase the user energy efficiency. The minimum user energy efficiency obtained by the nonlinear EH model is significantly higher than that of the linear EH model. When the number of BS antennas is the same, more users will produce higher energy consumption. Therefore, if the number of users increases, the minimum user energy efficiency of the two models will decrease. Figure 5 It can be seen that the total energy efficiency of users increases with the increase in the number of BS antennas. Due to the existence of multi-user diversity gain, the increase in the number of users will improve the total energy efficiency of users in the system.

[0105] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0106] The above embodiments should be understood to be only used to illustrate the present invention and not to limit the protection scope of the present invention. After reading the contents of the present invention, technicians can make various changes or modifications to the present invention, and these equivalent changes and modifications also fall within the scope defined by the claims of the present invention.

Claims

1. A method for optimizing energy efficiency of a large-scale MIMO system based on nonlinear energy harvesting, characterized in that: The following steps are involved:

101. Establish a wireless power supply massive MIMO system that combines massive MIMO technology with a wireless power supply communication network WPCN. In this system, the power beacon PB and the base station BS are equipped with a large number of antennas. In the system, the power beacon PB transmission power constraint, energy transmission time constraint, user minimum QoS constraint, and energy consumption constraint are considered. On this basis, a Max-Min optimization problem model that maximizes the minimum user energy efficiency is constructed. The Max-Min optimization problem model is a non-convex optimization problem.

102. Based on a simplified calculation method including variable substitution and first-order Taylor expansion, the non-convex Max-Min energy efficiency optimization problem in step 101 is converted into an equivalent convex optimization problem; the variable substitution is to introduce auxiliary variables to simplify the original objective function and add constraints containing the introduced variables; the first-order Taylor expansion is to use Taylor expansion to perform first-order expansion on non-convex constraints to convert the constraint conditions into convex constraints; 103. Then, the interior point method and the continuous convex approximation SCA algorithm are used to solve the convex optimization problem equivalently transformed in the above step 102, and the power allocation strategy is solved, including: the transmission power p allocated by PB to user k k , the transmission power p of user k to transmit information to BS t,k , energy transfer time τ, so that the optimal energy efficiency based on the minimum user under all constraints can be obtained; The Max-Min optimization problem of maximizing the minimum user energy efficiency based on the wireless power supply massive MIMO system in step 101 is: C2:0≤P≤P max , C3:0≤τ≤1, C4:p k ≥0,k=1,…,K, C5:p t,k ≥0,k=1,…,K, C6:E tot,k (p t,k ,τ)≤E hst,k (p k ,τ),k=1,…,K, In order to simplify the expression, the parameter M is introduced k , and there is p=(p1,p2,…,p K ) represents the power vector allocated by PB to user k, p t =(p t,1 ,p t,2 ,…,p t,K ) represents the transmission power vector of user k sending information to BS, K, M, N represent the number of single-antenna users, BS antennas, and PB antennas, respectively, and min{M,N}>>K; α k , β k Respectively represent the path loss from user k to BS and PB to user k; σ 2 Represents noise power; P max represents the maximum transmission power of PB; P represents the total transmission power of PB; E tot,k (p t,k ,τ)=p r,k τ+μp t,k (1-τ)+p c,k (1-τ), represents the energy consumption of user k, where p r,k represents the fixed power consumption of the receiving circuit of user k, μ is a constant factor reflecting the inefficiency of the power amplifier, and p c,k represents the fixed power consumption of the transmitting circuit of user k; considering the normalization of time unit, τ represents the time for PB to transmit energy to the user node, and 1-τ is the time for the user node to complete information transmission using the collected energy; A represents the energy expression collected by user k through the nonlinear energy collection model. k Indicates the maximum energy harvesting power at the user end when the energy harvesting circuit is saturated, constant a k and b k are parameters related to the circuit, respectively expressed as circuit sensitivity and leakage current, represents the zero-input zero-output response of the model; represents the minimum signal to interference and noise ratio SINR requirement of user k; In step 101, constraints C1 and C2 are combined and considered, and the optimization problem is written as: C2:0≤τ≤1, C3:p k ≥0,k=1,…,K, C4:p t,k ≥0,k=1,…,K, C5:E tot,k (p t,k ,τ)≤E hst,k (p k ,τ),k=1,…,K, Constraint C1 means that the sum of the power allocated to all users needs to be less than the maximum transmit power P of PB max ; Constraint C2 represents the energy transmission time allocation constraint; C3 and C4 represent the power constraint allocated to user k by PB and the transmission power constraint of user k respectively; C5 represents that the total energy consumed by each user must be less than or equal to the energy collected from PB, where the collected energy is obtained through a nonlinear energy collection circuit; C6 is the minimum SINR constraint, requiring that the SINR of each user must be greater than the minimum SINR constraint It can be seen that the above problem is a non-convex problem; The non-convex Max-Min energy efficiency optimization problem in step 101 is transformed into an equivalent convex optimization problem based on a simplified calculation method including variable substitution and first-order Taylor expansion, specifically:

401. Introduce auxiliary variable t, and let The energy efficiency optimization problem can be rewritten as: C2:x≥0, C3:p k ≥0,k=1,…,K, C4:p t,k ≥0,k=1,…,K, Among them, C7 is the equivalent treatment of the original objective function, using the auxiliary variable t to be equivalent to the minimum user energy efficiency, then user k, The energy efficiency of must be greater than or equal to the minimum user energy efficiency t, which simplifies the objective function; 402. Let t = e s , By variable substitution, we introduce x=e z , Because t = e s is a monotonically increasing function of s, therefore, the optimization problem of step 401 can be rewritten as: C2:p k ≥0,k=1,…,K, The optimization variables considered have been transformed into v = (v1,…,v K )、p=(p1,p2,…,p K ), z, s, where K, M, and N represent the number of single-antenna users, BS antennas, and PB antennas, respectively, and min{M,N}>>K; constraint C1 indicates that the sum of the power allocated to all users must be less than the maximum transmit power P of PB. max ; Constraint C2 represents the power constraint allocated by PB to user k; C3 indicates that the total energy consumed by each user must be less than or equal to the energy collected from PB; C4 is the minimum SINR constraint, requiring that the SINR of each user must be greater than the minimum SINR constraint C5 is the constraint that the energy efficiency of any user must be greater than or equal to the minimum user energy efficiency; 403. The objective function of the optimization problem in step 402 is a linear function, and the constraint C1 is a linear function, C1 is a convex constraint; similarly, constraint C2 is also a convex constraint, and constraints C3, C4, and C5 are convex; By replacing variables and transforming formulas, the original problem in step 402 can be equivalent to a convex optimization problem: C2:p k ≥0,k=1,…,K, C3:f k (v k ,z)+g k (p k ,z)≤0,k=1,…,K, Among them, in constraint C3, It is represented as the expression introduced in the process of formula transformation, It is expressed as the expression introduced in the process of formula transformation. Since f k (v k ,z) and g k (p k ,z) are all convex functions, so C3 is a convex constraint; in C5, It is expressed as the expression introduced in the process of formula transformation, and L k (z,s,v k ) is a convex constraint, is the intermediate expression in the formula transformation, which can be obtained by k (v k ) is expanded by first-order Taylor to obtain an equivalent expression: Where n represents the number of iterations, Indicates the expansion point, For about v k The SCA algorithm can be used to approximate the approximate expression to the original function. Therefore, in the equivalent problem, constraint C5 is a convex constraint.

2. The method for optimizing energy efficiency of a large-scale MIMO system based on nonlinear energy harvesting according to claim 1, characterized in that: The step 103 of solving the convex optimization problem by using the SCA algorithm and the interior point method includes:

501. Initialize all relevant input parameters of the problem: the maximum total transmit power P of PB max , the large-scale fading α from user k to BS k , the large-scale fading β from PB to user k k , the number of PB antennas N, the number of BS antennas M, the number of single-antenna users K, and the fixed power consumption p c,k 、p r,k , the minimum signal-to-interference-noise ratio requirement of the sensor node Gaussian noise power σ 2 , nonlinear energy harvesting circuit parameters A k , a k , b k , Iteration number n = 1, maximum iteration number N max , convergence threshold value ε; 502. Solve the problem by the interior point method, that is, construct a new unconstrained objective function - penalty function, define the penalty function in the feasible domain, and solve the penalty function in the feasible domain. The extreme point is always inside the feasible domain. In this way, the solution of the unconstrained optimization problem is always a feasible solution, so the optimal solution of the original optimization problem can be approached successively in the feasible domain.

503. The complexity of the algorithm depends on the number of iterations n and the convergence threshold ε. Since the complexity of the interior point method is Where x and y represent the number of optimization variables and the number of constraints respectively; the number of optimization variables in the proposed problem is x1=2K+2, and the number of constraints is y1=4K+1, so the computational complexity of the algorithm is

Images