Resource allocation method based on downlink cooperative simultaneous wireless information and power transfer-non-orthogonal multiple access (SWIPT-NOMA) system

Abstract

The wide application of the Internet of Things enables spectrum resources to be scarce, and the terminal equipment is urgent for the demand of energy and efficient communication. The invention designs a wireless energy transfer communication (SWIPT)-non-orthogonal multiple access (NOMA) system model based on downlink cooperation. The invention provides an energy efficiency maximization resource allocation algorithm based on cooperation SWIPTT-NOMA by considering a resource allocation problem under the constraints of throughput, energy collection and transmitting power of a user, so as to improve the energy efficiency of a system. Through a Dinkelbach method, variable replacement and an SCA method, a nonlinear non-convex optimization problem is converted into a convex optimization problem, and a globally optimal solution is solved by using a Lagrange duality theory. A simulation result shows that compared with an optimization scheme based on a fixed power allocation factor and a fixed time slot selection factor, the proposed strategy can improve the system energy efficiency on the premise of ensuring the user throughput, and a more efficient resource allocation scheme is provided for a wireless energy supply network.

Description

Technical Field

[0001] The present invention relates to a resource allocation method based on a downlink cooperative wireless power carrying communication-non-orthogonal multiple access (SWIPT-NOMA) system, and proposes a downlink cooperative wireless power carrying communication-non-orthogonal multiple access energy efficiency maximization resource optimization algorithm to solve the problem so as to maximize the energy efficiency of the system. Background Art

[0002] With the widespread application of the Internet of Things (IoT) and large-scale wireless sensor networks, the demand for energy and efficient communication of a large number of terminal devices has become more urgent. Traditional battery-powered methods can no longer meet the needs of long-term operation of these devices. In this context, Simultaneous Wireless Information and Power Transfer (SWIPT) came into being. It transmits data and power simultaneously through wireless radio frequency signals. The device can collect energy while receiving information, which can effectively solve the problem of frequent charging or replacement due to the limited battery capacity of the device. The two major working modes of SWIPT are time-switching energy harvesting (TS-EH) and power-splitting energy harvesting (PS-EH). In the TS-EH mode, the receiver performs information decoding and energy harvesting in different time periods, while in the PS-EH mode, the receiver performs information decoding and energy harvesting at the same time, and balances the two tasks by adjusting the power allocation ratio. Compared with the PS-EH mode, the simplification of hardware implementation is a significant advantage of the TS-EH mode. Secondly, less signal interference is also a major advantage of the TS-EH mode. Since energy collection and information demodulation are performed in different time periods, the energy signal and the information signal will not interfere with each other, which helps to improve the demodulation accuracy and reception efficiency. Finally, the TS-EH mode is more adaptable to different channel conditions. It can flexibly adjust the time allocation ratio of energy collection and data demodulation according to the channel conditions, making it easier to optimize system performance in a changing channel environment. This flexible mechanism enables SWIPT to show extremely high potential in energy-constrained application scenarios, which not only significantly improves the system energy efficiency, but also effectively extends the operating life of the equipment and promotes the construction of a green communication network.

[0003] In order to further improve spectrum utilization, non-orthogonal multiple access (NOMA) technology has gradually become a key technology in the field of future wireless communication networks. NOMA technology is proposed to overcome the spectrum utilization bottleneck of traditional orthogonal multiple access (OMA) technology, allowing multiple users to share the same spectrum resources for data transmission at the same time, and using power domain multiplexing technology to distinguish between users. Specifically, by assigning different power levels to each user, multiple users can transmit data at the same time and under the same frequency resources. This method is particularly suitable for scenarios where the channel conditions between users vary greatly. In a typical downlink NOMA system, users with strong channel gain are assigned lower power, while users with weak channel gain are assigned higher power. Before users with strong channel gain decode their own signals, they also need to perform interference cancellation (SIC) to decode the signals from users with weak channel gain.

[0004] Combining SWIPT and NOMA can achieve a dual improvement in spectrum efficiency and energy efficiency. This combination can achieve efficient data transmission for more users under the same spectrum resources while meeting the energy needs of some users, helping to extend the battery life of devices and improve the utilization of network resources.

[0005] The paper "Jie Tang, Jingci Luo, Mingqian Liu, Daniel KC So, Emad Alsusa, Gaojie Chen, Kai-Kit Wong, and Jonathon A Chambers. Energy efficiency optimization for noma with swipt. IEEE Journal of Selected Topics in Signal Processing, 13(3): 452–466, 2019." designed a NOMA system based on SWIPT, which maximized the overall energy efficiency of the system by jointly optimizing power allocation and time slot selection factors. By using the Dinkelbach method to convert the nonlinear problem into a linear problem for solution, the results show that the performance of the system is better than the traditional OMA scheme. This study provides a strong theoretical basis for the practical application of NOMA and SWIPT in large-scale multi-user communication scenarios in terms of energy efficiency, and further expands the research prospects of energy efficiency optimization. Subsequent studies have proposed more optimization schemes for different scenarios. The paper "Jie Tang, Jingci Luo, Mingqian Liu, Daniel KC So, Emad Alsusa, Gaojie Chen, Kai-Kit Wong, and Jonathon A Chambers. Energy efficiency optimization for noma with swipt. IEEE Journal of Selected Topics in Signal Processing, 13(3): 452–466, 2019." designed a NOMA system based on SWIPT, which maximized the overall energy efficiency of the system by jointly optimizing power allocation and time slot selection factors. By using the Dinkelbach method to convert the nonlinear problem into a linear problem for solution, the results show that the performance of the system is better than the traditional OMA scheme. This study provides a strong theoretical basis for the practical application of NOMA and SWIPT in large-scale multi-user communication scenarios in terms of energy efficiency, and further expands the research prospects of energy efficiency optimization.Subsequent studies have proposed more optimization schemes for different scenarios. The paper "Shuang Li, Pengfei Wu, and Fuhua An. Energy ellicieney of NOMA network under fast fading channel. In 2022 11th lnlernalional Conferenecon Communicalions, Cireuils and Syslems (CCCAS), page 8170175, 2022." proposes a power allocation scheme for maximizing system energy efficiency for NOMA systems under fast time-varying fading channels. Based on the requirements of meeting user service quality, a power allocation optimization problem between users is established to maximize system energy efficiency. The paper "Haixin Cui, Xinnwan Yo, and Fhn You. Energy-ellicient resource allocation for heterogeneous swipt-nomn systems, / N / N eccss, 10:7928179288, 2022." designed a heterogeneous SWIPT-NOMA system, considering the user's minimum rate requirements, energy harvesting, and the transmission power of heterogeneous wireless access technologies, and proposed a hybrid resource allocation framework to maximize energy efficiency performance through joint power and time scheduling management.

[0006] Although the SWIPT-NOMA system has made significant progress in improving system energy efficiency in different scenarios, how to further optimize the energy efficiency in the network remains a challenge. Therefore, user cooperation technology is widely used in the SWIPT-NOMA system as an effective means. On the one hand, user cooperation technology can save spectrum resources by sharing channels between multiple devices. On the other hand, user cooperation simulates a multi-antenna system by cooperatively transmitting and receiving signals to improve signal reliability and transmission efficiency. Collaborative users can act as relay nodes to help other users transmit data, forming a virtual distributed antenna system, and further improving the capacity and coverage of the network. By introducing user cooperation technology into the SWIPT-NOMA system, collaborative users can help other users transmit information while also acting as energy relays to provide additional energy supplements for energy-constrained devices. This not only enhances the reliability and coverage of the signal, but also improves the energy efficiency of the system. Therefore, this combination strategy is worthy of in-depth study in resource-constrained scenarios and has important academic and practical significance. The document "Kamal Agrawal, Shankar Prakriya, and Mark F Flanagan. Ts-based swipt in full-duplex relayed noma with intelligent relay battery management. lEEETansactions on Communications, 2023" studies the collaborative NOMA system based on time slot selection. The base station communicates with the near-end user through a direct link, while the communication of the remote user is assisted by a full-duplex relay with battery-assisted energy harvesting function. The system throughput and average battery energy consumption under the static battery energy (SBE) scheme and the dynamic battery energy (DBE) scheme are considered respectively.Different from the battery-assisted energy harvesting scheme, the paper "Mengru Wu, Qingyang Song, Lei Guo, and Abbas Jamalipour. Joint user pairing and resource allocation in a swipt-enabled cooperative noma system. EEE Transactions on Vehicular Technology, 70(7): 68266840.2021." studies a SWIPT-enabled cooperative NOMA system with power splitting (PS) and time switching (TS) users using SWIPT technology for energy harvesting. By optimizing user pairing and power allocation between far-end and near-end users, and performing PS and TS control at the near-end user, the total rate of the system is maximized while ensuring the user's target data rate and minimum energy harvesting requirements. On this basis, the paper "Yi Yuan, Yanqing Xu, Zheng Yang, Peng Xu, and Zhiguo Ding. Energy efficiency optimization in full-duplex user-aided cooperative swipt noma systems, EEE Tansactions on Communications 67(8):5753-5767, 2019" proposed a cooperative NOMA system, in which the full-duplex users at the center of the system act as relays to help edge users. The energy efficiency of the system is maximized by combining the PS factor and beamforming vector while meeting the minimum throughput of the users. Experimental results show that the proposed strategy significantly improves the energy efficiency in low-power areas compared with the existing strategy. Summary of the invention

[0007] From the above research, it can be seen that the current research on energy-efficient resource allocation for SWIPT-NOMA systems and research on collaborative SWIPT-NOMA systems has made progress, and these designs and optimization strategies have effectively improved system performance. However, in the research on the energy efficiency of collaborative SWIPT-NOMA systems, it has not yet considered improving the system energy efficiency by jointly optimizing the TS factor and the power split ratio. Compared with the above work that only optimizes the PS factor, the joint optimization of the TS factor and the power split ratio can not only more effectively allocate time and power within the same spectrum, maximize the simultaneous access capabilities of users, and improve spectrum utilization, but also reduce interference between different users to ensure that users with low channel quality obtain sufficient information to improve system performance.

[0008] Therefore, this paper introduces a user cooperative communication mechanism, in which the near-end user acts as a relay to assist the base station in transmitting information, and studies the problem of maximizing the energy efficiency resource allocation of the downlink cooperative SWIPT-NOMA system. In order to solve this problem, an energy-efficient resource allocation algorithm based on cooperative SWIPT-NOMA is proposed, which maximizes the energy efficiency of the system by jointly optimizing the power allocation factor, TS factor, and user transmit power. The contributions of this invention are as follows:

[0009] 1.(1) An energy efficiency resource allocation model for the downlink cooperative SWIPT-NOMA system is established, in which the base station uses NOMA technology to simultaneously provide services to two users in the same frequency band. The near-end user collects energy and acts as a relay to help the far-end user receive base station information. The overall energy efficiency of the system is maximized while satisfying the quality of service constraints of all users, energy collection constraints, and maximum transmit power constraints.

[0010] (2) Aiming at the energy efficiency resource allocation problem of the system, a collaborative SWIPT-NOMA energy efficiency resource allocation algorithm is proposed. Specifically, the nonlinear programming problem is first transformed into a linear programming problem through the Dinklebach method, and then the original NP-hard problem is transformed into a deterministic convex optimization problem through variable transformation and the Successive Convex Approximation (SCA) method, and then the Lagrangian dual method is used to obtain the optimal solution of the problem.

[0011] (3) The simulation results show that the proposed optimization scheme not only has good convergence, but also has higher energy efficiency compared with the fixed power allocation factor and fixed time slot selection factor optimization schemes. This study provides a new strategy for efficient resource allocation in wireless power communication networks.

[0012] The technical solution adopted by the present invention is: a downlink cooperative wireless energy-carrying communication-non-orthogonal multiple access energy efficiency maximization resource optimization algorithm, including the following system model.

[0013] (1) In a downlink cooperative SWIPT-NOMA system, there is a base station (BS) and two users. Due to physical obstacles or severe shadow effects, there is no direct link between the BS and the remote user User2. The near user User1 acts as a relay to help the BS transmit information to User2. User1 adopts half-duplex working mode and has limited battery capacity. It collects energy from the BS signal and uses this energy to extend the working time of User1. |h1| 2 and |h2| 2 represents the Rayleigh fading channel gains from BS to User1 and from User1 to User2, respectively, |h i| 2 =|g i | 2 d i -β ,|g1| 2 and |g2| 2 denote the random channel power gains from BS to User1 and from User1 to User2, β denotes the path loss coefficient, and d sn1 and d nf1 Respectively represent the distance from BS to User1 and from User1 to User2.

[0014] (2) The total duration of the transmission frame structure in practical applications is defined as T. The structure is divided into three stages: the energy collection stage of User 1 lasts for τT (0≤τ≤1), the information decoding stage of User 1 lasts for And User2's information decoding phase continues If the BS is powered by s Sending energy harvesting signal, the power received by the receiving end User1 can be expressed as p s |h1| 2 , the energy collected by User1 can be expressed as E h =ηp s |h1| 2 τT, where η represents the energy conversion efficiency of user User1.

[0015] In the second stage, the BS transmits a superimposed signal

[0016]

[0017] Where x1(t) and x2(t) represent the unit power signals sent to User1 and User2, respectively, and α(0≤α≤1) and (1-α) represent the power division ratios of x1(t) and x2(t), respectively. The signal received by User1 can be expressed as

[0018]

[0019] Where h1(t) represents the channel gain between BS and User1, n1(t) represents the additive white Gaussian noise (AWGN) at User1, After receiving the signal from the BS, according to the SIC principle, User1 decodes x2 first and then decodes x1. Therefore, the Signal-to-Interference-plus-Noise Ratio (SINR) of x1 and x2 detected by User1 are respectively

[0020]

[0021] and

[0022]

[0023] In the third phase, User1 uses all the energy collected in the first phase to forward the decoded information x2 to User2 as a relay. Assume that User1 only consumes part of the collected energy when acting as a relay, and User2 can successfully decode the information from User1 continuously. Therefore, in the current time slot, the energy collected by User1 from the BS is not fully utilized. This shows that even if the BS reduces its transmission power p s , User1 and User2 can also successfully decode the information they need, thus saving the energy consumption of the BS on the downlink. In order to make full use of the energy collected at User1, we set User1 to use all its collected energy to relay information. The signal received by User2 can be expressed as

[0024]

[0025] Where h2(t) represents the channel gain between User1 and User2, n2(t) represents the additive white Gaussian noise at User2, Then, the x2 signal-to-noise ratio (SNR) detected by User2 is

[0026]

[0027] Where p represents the transmission power of User 1.

[0028] (3) The throughputs of x1 and x2 from BS to User1 are expressed as

[0029]

[0030] and

[0031]

[0032] The throughput of x2 from User1 to User2 is

[0033]

[0034] Based on the above transmission protocol and Maximum Ratio Combining (MRC) technology, the total system throughput can be calculated as

[0035] Rtotal =R 11 +min{R 12 ,R 22} (10)

[0036] The total energy consumption of the system base station is expressed as

[0037]

[0038] (4) Maximize the overall energy efficiency of all nodes by jointly optimizing the power-time domain resources (α, τ, p). At the same time, we set the constraints of the minimum throughput of User1 and User2, as well as the constraint of ensuring that the energy collected by User1 meets the forwarding requirements. The target problem can be expressed as:

[0039]

[0040] C5:p≤p max

[0041] C6:0≤α≤1,0≤τ

[0042] in and They represent the minimum throughput that User2 and User1 need to achieve, respectively. max represents the maximum transmission power of User1. Constraints (C1) and (C2) ensure that User1 can successfully decode all the information it receives, and constraint (C3) ensures that User2 can also successfully decode the information forwarded from User1. Constraint (C4) ensures that the energy used by User1 for information forwarding does not exceed the collected energy. In addition, User1's transmission power, power allocation factor, and time slot selection factor are constrained by (C5) and (C6).

[0043] Since the variables are mutually coupled, problem P1 is a nonlinear non-convex fractional programming problem, and it is quite difficult to solve it directly. Therefore, an optimization method is proposed, which is to first use the Dinkelbach method to transform the nonlinear problem into a linear problem, then use new variables and perform variable substitution and the SCA method to transform the non-convex optimization problem into a convex optimization problem, and finally use the Lagrange dual method to solve the convex optimization problem to obtain the optimal solution. This process will be repeated until convergence.

[0044] The wireless energy-carrying communication-non-orthogonal multiple access energy efficiency maximization resource optimization algorithm based on downlink cooperation in the above model is obtained according to the following steps:

[0045] (1)Dinkelbach method

[0046] Since both the objective function and the constraints are non-convex, the nonlinear characteristics of the fractional programming problem P1 make it difficult to solve directly. By using the Dinkelbach method, the fractional programming problem can be transformed into the form of numerator-denominator subtraction. First, define an optimization parameter λ * as follows

[0047]

[0048] where α * ,τ * ,p * represents the solution to problem P1.

[0049] Lemma 1: For R total (α,τ,p)>0 and E total (τ,p)>0, there exists α * ,τ * ,p * The original problem P1 is maximized. The maximum system energy efficiency value can be obtained if and only if the following formula is satisfied:

[0050]

[0051] Proof: Here we use the proof by contradiction to prove Lemma 1. According to the above expression, suppose there is another set of solutions (α F ,τ F ,p F ) and λ F ≥λ * According to the definition of the above formula, we can get This is consistent with the initial assumption λ F >λ * conflict.

[0052] Through Lemma 1, the non-convex hierarchical objective function can be converted into a linear form. Then, given a parameter λ, the original problem P1 is converted into a linear problem P2 for analysis.

[0053]

[0054] Based on the above analysis, an iterative algorithm is obtained, which generates a series of λ, which converges to the feasible solution λ * (α * ,τ * ,p * ), the main framework of the proposed solution is summarized in Algorithm 1.

[0055] Solution framework for problem P2: The specific process of the algorithm based on the Dinkelbach method is as follows:

[0056]

[0057] (2) Lagrangianduality method

[0058] Although the Dinkelbach method mentioned above has transformed the nonlinear problem into a linear problem, the optimization problem is still a non-convex optimization problem and it is difficult to directly find the solution to the problem. 12 ,γ 22}, so that the objective function and constraints (C2) and (C3) can be rewritten as

[0059]

[0060] The new variable γ introduced here is coupled with the power allocation factor α, and the Hessian matrix of the bilinear function is neither semi-positive definite nor semi-negative definite, so the function is neither convex nor concave. In order to solve this problem, the coupled part on the left side of formula (19) is transformed

[0061]

[0062] Based on the above transformation, the SCA method is used to replace the coupling part with their convex upper bound function. Specifically, the first-order Taylor expansion expression is used to satisfy the feasible solution (γ (n) ,α (n) ) approximation (γ-α) 2 The lower bound of .

[0063] (γ-α) 2 ≥(γ (n) -α (n) ) 2 +2(γ (n) -α (n) )×[(γ-γ (n) )-(α-α (n) )] (twenty one)

[0064] According to formula (21), the convex upper bound is used to approximate the coupled part on the left side of formula (19). Therefore, formula (19) can be converted into the following feasible form:

[0065]

[0066] Accordingly, problem P2 is re-planned as

[0067]

[0068] Obviously, when all optimization variables satisfy the constraints and are in the convex set, the objective function of problem P2' is a linear concave function. Therefore, P2' is a convex optimization problem and satisfies the Slater condition and can be solved by the Lagrangian Dual Method.

[0069] The Lagrangian dual function of problem P2' is formulated as

[0070]

[0071] where L(φ) is the Lagrangian function of problem P2', φ = {α, τ, p, μ1, μ2, μ3, μ4, μ5, μ6, μ7, μ8} is the set of all initial and dual variables, and μ1, μ2, μ3, μ4, μ5, μ6, μ7, μ8 are non-negative Lagrangian multipliers defined according to the constraints (C1), (C4), (C5), (C6), (16), (17), (21) of problem P2'. L(φ) can be written as

[0072] The dual problem of P3 can be expressed as

[0073]

[0074] If problem P2' is to be solved, first, solve problem P3 by giving the Lagrange multiplier that satisfies the constraints of problem P4 to obtain the optimization variables α, τ, p, γ. Then, solve P4 by updating the Lagrange multiplier to minimize f(μ1, μ2, μ3, μ4, μ5, μ6, μ7, μ8). This iterative method is used to solve problem P2'. By using the KKT (Karush KuhnTucker) condition to solve the optimization variable α under the given Lagrange multiplier opt ,τ opt ,p opt ,γ opt , first we need to find partial derivatives of the variables α, τ, p, γ in the Lagrangian function L(φ) and set their stationary points to zero. The formula is as follows:

[0075]

[0076] where γ1'1 is as follows

[0077]

[0078] The Lagrange multipliers μ1,μ2,μ3,μ4,μ5,μ6,μ7,μ8 are updated by the subgradient method given below

[0079]

[0080] where [z] + =max(0,z), j represents the iteration index, θ represents the iteration step size. The update formula is as follows

[0081]

[0082] Based on the above analysis, the specific algorithm flow of the Lagrangian duality method is as follows

[0083]

[0084] The beneficial effects of the present invention are as follows: In a single base station and multiple user scenario, the present application proposes a downlink collaborative wireless energy-carrying communication-non-orthogonal multiple access system, which maximizes the overall energy efficiency of the system by jointly optimizing power allocation, time slot selection, and user transmission power. The proposed downlink collaborative wireless energy-carrying communication-non-orthogonal multiple access maximum energy efficiency resource optimization method not only reduces the energy consumption of the base station and improves the user's throughput, but also contributes to green communication. Compared with different schemes, it also shows stable adaptability and reliability. This has opened up a new path for the collaborative wireless energy-carrying communication-non-orthogonal multiple access technology system to improve throughput and reduce energy consumption, and is expected to promote technological innovation and practical applications in related fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Figure 1 It is the system model diagram;

[0086] Figure 2 It is a frame structure diagram;

[0087] Figure 3 A comparison chart of the number of iterations of the proposed optimization scheme and other optimization schemes;

[0088] Figure 4 This is a comparison chart of the distance from BS to User1 and the path loss coefficient;

[0089] Figure 5 It is a comparison diagram of path loss coefficient, energy conversion efficiency and minimum throughput constraint;

[0090] Figure 6 It is a comparison chart of the distance from BS to User1 and the distance from User1 to User2 with the minimum throughput constraint;

[0091] Figure 7 It is the relationship diagram between different optimization methods and path loss coefficient;

[0092] Figure 8 It is a relationship diagram between different optimization methods and base station transmission power; DETAILED DESCRIPTION

[0094] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments: The invention proposes a downlink cooperative wireless energy-carrying communication-non-orthogonal multiple access maximum energy efficiency resource optimization algorithm, the system model is as follows Figure 1 As shown in the figure, in a downlink cooperative SWIPT-NOMA system, there is a base station and two users. Due to physical obstacles or severe shadow effects, there is no direct link between the BS and the remote user User2. The near-end user User1 acts as a relay to help the BS transmit information to User2. User1 adopts half-duplex working mode and its battery capacity is limited. It collects energy from the BS signal and uses this energy to extend the working time of User1. |h1| 2 and |h2| 2 Represent the Rayleigh fading channel gains from BS to User1 and from User1 to User2, and |g2| 2 denote the random channel power gains from BS to User1 and from User1 to User2, β denotes the path loss coefficient, and d sn1 and d nf1 Respectively represent the distance from BS to User1 and from User1 to User2.

[0095] The total duration of the transmission frame structure in practical applications is defined as T. The structure is divided into three stages: the energy collection stage of User1 lasts for τT (0≤τ≤1), the information decoding stage of User1 lasts for And User2's information decoding phase continues If the BS is powered by s Sending energy harvesting signal, the power received by the receiving end User1 can be expressed as p s |h1| 2 , the energy collected by User1 can be expressed as E h =ηp s |h1| 2 τT, where η represents the energy conversion efficiency of user User1. In the second stage, the BS transmits a superposition signal

[0096]

[0097] Where x1(t) and x2(t) represent the unit power signals sent to User1 and User2, respectively, and α(0≤α≤1) and (1-α) represent the power division ratios of x1(t) and x2(t), respectively. The signal received by User1 can be expressed as

[0098]

[0099] Where h1(t) represents the channel gain between BS and User1, n1(t) represents the additive white Gaussian noise (AWGN) at User1, After receiving the signal from the BS, according to the SIC principle, User1 decodes x2 first and then decodes x1. Therefore, the Signal-to-Interference-plus-Noise Ratio (SINR) of x1 and x2 detected by User1 are respectively

[0100]

[0101] and

[0102]

[0103] In the third phase, User1 uses all the energy collected in the first phase to forward the decoded information x2 to User2 as a relay. Assume that User1 only consumes part of the collected energy when acting as a relay, and User2 can successfully decode the information from User1 continuously. Therefore, in the current time slot, the energy collected by User1 from the BS is not fully utilized. This shows that even if the BS reduces its transmission power p s , User1 and User2 can also successfully decode the information they need, thus saving the energy consumption of the BS on the downlink. In order to make full use of the energy collected at User1, we set User1 to use all its collected energy to relay information. The signal received by User2 can be expressed as

[0104]

[0105] Where h2(t) represents the channel gain between User1 and User2, n2(t) represents the additive white Gaussian noise at User2, Then, the x2 signal-to-noise ratio (SNR) detected by User2 is

[0106]

[0107] Where p represents the transmission power of User 1.

[0108] (3) The throughputs of x1 and x2 from BS to User1 are expressed as

[0109]

[0110] and

[0111]

[0112] The throughput of x2 from User1 to User2 is

[0113]

[0114] Based on the above transmission protocol and Maximum Ratio Combining (MRC) technology, the total system throughput can be calculated as

[0115] R total =R 11 +min{R 12 ,R 22} (49)

[0116] The total energy consumption of the system base station is expressed as

[0117]

[0118] (4) Maximize the overall energy efficiency of all nodes by jointly optimizing the power-time domain resources (α, τ, p). At the same time, we set the constraints of the minimum throughput of User1 and User2, as well as the constraint of ensuring that the energy collected by User1 meets the forwarding requirements. To simplify the model, we assume that the total time T = 1. The target problem can be expressed as:

[0119]

[0120] in and They represent the minimum throughput that User2 and User1 need to achieve, respectively. max represents the maximum transmission power of User1. Constraints (C1) and (C2) ensure that User1 can successfully decode all the information it receives, and constraint (C3) ensures that User2 can also successfully decode the information forwarded from User1. Constraint (C4) ensures that the energy used by User1 for information forwarding does not exceed the collected energy. In addition, User1's transmission power, power allocation factor, and time slot selection factor are constrained by (C5) and (C6).

[0121] Since the variables are mutually coupled, problem P1 is a nonlinear non-convex fractional programming problem. It is quite difficult to solve this problem directly. Therefore, a downlink cooperative wireless energy-carrying communication-non-orthogonal multiple access energy efficiency maximization resource optimization algorithm is proposed, that is, the Dinkelbach method is first used to transform the nonlinear problem into a linear problem, and then the new variables are cited and the variables are replaced and the SCA method is used to transform the non-convex optimization problem into a convex optimization problem, and finally the Lagrangian dual method is used to solve the convex optimization problem to obtain the optimal solution. This process will be repeated until convergence.

[0122] The energy efficiency maximization resource optimization algorithm based on downlink cooperative wireless energy-carrying communication-non-orthogonal multiple access in the above model is obtained according to the following steps:

[0123] (1)Dinkelbach method

[0124] Since both the objective function and the constraints are non-convex, the nonlinear characteristics of the fractional programming problem P1 make it difficult to solve directly. By using the Dinkelbach method, the fractional programming problem can be transformed into the form of numerator-denominator subtraction. First, define an optimization parameter λ * as follows

[0125]

[0126] where α * ,τ * ,p * represents the solution to problem P1.

[0127] Lemma 1: For R total (α,τ,p)>0 and E total (τ,p)>0, there exists α * ,τ * ,p * The original problem P1 is maximized. The maximum system energy efficiency value can be obtained if and only if the following formula is satisfied:

[0128]

[0129] Proof: Here we use the proof by contradiction to prove Lemma 1. According to the above expression, suppose there is another set of solutions (α F ,τ F ,p F ) and λ F ≥λ * According to the definition of the above formula, we can get This is consistent with the initial assumption λ F >λ * conflict.

[0130] Through Lemma 1, the non-convex hierarchical objective function can be converted into a linear form. Then, given a parameter λ, the original problem P1 is converted into a linear problem P2 for analysis.

[0131]

[0132] Based on the above analysis, an iterative algorithm is obtained, which generates a series of λ, which converges to the feasible solution λ * (α * ,τ * ,p * ), the main framework of the proposed solution is summarized in Algorithm 1.

[0133] Solution framework for problem P2: The specific process of the algorithm based on the Dinkelbach method is as follows:

[0134]

[0135]

[0136] (2) Lagrangianduality method

[0137] Although the Dinkelbach method mentioned above has transformed the nonlinear problem into a linear problem, the optimization problem is still a non-convex optimization problem and it is difficult to directly find the solution to the problem. 12 ,γ 22}, so that the objective function and constraints (C2) and (C3) can be rewritten as

[0138]

[0139] The new variable γ introduced here is coupled with the power allocation factor α, and the Hessian matrix of the bilinear function is neither semi-positive definite nor semi-negative definite, so the function is neither convex nor concave. In order to solve this problem, the coupled part on the left side of formula (58) is transformed

[0140]

[0141] Based on the above transformation, the SCA method is used to replace the coupling part with their convex upper bound function. Specifically, the first-order Taylor expansion expression is used to satisfy the feasible solution (γ (n) ,α (n) ) approximation (γ-α) 2 The lower bound of .

[0142] (γ-α) 2 ≥(γ (n) -α (n) )2 +2(γ (n) -α (n) )×[(γ-γ (n) )-(α-α (n) )] (60)

[0143] According to formula (60), the convex upper bound is used to approximate the coupled part on the left side of formula (58). Therefore, formula (58) can be converted into the following feasible form:

[0144]

[0145] Accordingly, problem P2 is re-planned as

[0146]

[0147] Obviously, when all optimization variables satisfy the constraints and are in the convex set, the objective function of problem P2' is a linear concave function. Therefore, P2' is a convex optimization problem and satisfies the Slater condition and can be solved by the Lagrangian Dual Method.

[0148] The Lagrangian dual function of problem P2' is formulated as

[0149]

[0150] where L(φ) is the Lagrangian function of problem P2', φ = {α, τ, p, μ1, μ2, μ3, μ4, μ5, μ6, μ7, μ8} is the set of all initial and dual variables, and μ1, μ2, μ3, μ4, μ5, μ6, μ7, μ8 are non-negative Lagrangian multipliers defined according to the constraints (C1), (C4), (C5), (C6), (54), (55), (59) of problem P2'. L(φ) can be written as

[0151] The dual problem of P3 can be expressed as

[0152]

[0153] If problem P2' is to be solved, first, solve problem P3 by giving the Lagrange multiplier that satisfies the constraints of problem P4 to obtain the optimization variables α, τ, p, γ. Then, solve P4 by updating the Lagrange multiplier to minimize f(μ1, μ2, μ3, μ4, μ5, μ6, μ7, μ8). This iterative method is used to solve problem P2'. By using the KKT (Karush KuhnTucker) condition to solve the optimization variable α under the given Lagrange multiplier opt ,τopt ,p opt ,γ opt , first we need to find partial derivatives of the variables α, τ, p, γ in the Lagrangian function L(φ) and set their stationary points to zero. The formula is as follows:

[0154]

[0155] where γ1'1 is as follows

[0156]

[0157] The Lagrange multipliers μ1,μ2,μ3,μ4,μ5,μ6,μ7,μ8 are updated by the subgradient method given below

[0158]

[0159] Where [z]+=max(0,z), j represents the iteration index, and θ represents the iteration step size. The update formula is as follows

[0160]

[0161] Based on the above analysis, the specific algorithm flow of the Lagrangian duality method is as follows

[0162]

[0163]

[0164] Experimental analysis:

[0165] The Rayleigh fading channel is used to simulate the actual environment, the path loss coefficient β = 3, and the distance from the BS to the user User1 is set to d sn1 =5m, the distance from user User1 to user User2 is set to d nf1 =10m, the bandwidth B of the overall system is set to 1MHZ, and the noise variables of information decoding at user 1 and user 2 are In order to be closer to the actual situation, the maximum forwarding power P of user User1 is max No more than 0.1W, the minimum throughput limits of the two users are Energy conversion efficiency η=0.7.

[0166] Figure 3The convergence effect of the algorithm under different schemes is demonstrated. First, it can be seen from the figure that the proposed algorithm converges after a small number of iterations, which proves the effectiveness of the proposed algorithm. Secondly, compared with the fixed power allocation factor α scheme and the fixed time slot selection factor τ scheme, the proposed algorithm can significantly improve the system energy efficiency, proving its effectiveness and advantages in improving the system energy efficiency. Finally, in the fixed power allocation factor α scheme, it can be seen that the system energy efficiency can be improved as α increases, but in the fixed time slot selection factor τ scheme, it can be seen that the system energy efficiency decreases as τ increases under the constraints of meeting the energy collection constraints.

[0167] Figure 4 Shows different path loss coefficients β and different distances d from BS to User1 sn1 Iterative performance of system energy efficiency. Due to the Rayleigh fading channel model, the system channel gain |h i | 2 will vary with β and d sn1 It can be seen from the figure that when β=2.75, d sn1 =4m, the system energy efficiency is the largest. When β remains unchanged, the system energy efficiency increases with d sn1 When d sn1 When β is constant, the system energy efficiency decreases as β increases. This indicates that the system channel gain is positively correlated with the system energy efficiency.

[0168] Figure 5 Shows different path loss coefficients β, different energy conversion efficiencies η and different User2 minimum throughput constraints At different BS transmission power p s First, it can be seen from the figure that as p s The energy efficiency of the system decreases gradually as p increases. This is because as p increases s As p increases, the throughput and energy consumption of the system are gradually increasing, but the growth trend of the system throughput is smaller than the growth trend of the system energy consumption. s The energy efficiency of the system decreases as the value increases. Secondly, it can be observed that when β = 2.9, η = 0.9 and When the system energy efficiency is the largest, the system energy efficiency increases as It increases with the decrease of β and η. s Under the condition of , the increase of η will reduce the energy collection time, which means that there will be more time for information decoding and forwarding in a time slot, that is, the increase of η will increase the system throughput and reduce the system energy consumption, increasing the system energy efficiency. Figure 4It has been analyzed in that the system channel gain is positively correlated with the system energy efficiency, so the increase of β will also increase the system energy efficiency. In terms of s Under constant conditions The smaller the system is, the better it can control energy consumption while meeting user needs, thereby increasing system energy efficiency. The smaller it is, the easier it is to meet user throughput requirements, and the algorithm can select a better scheduling solution to improve system energy efficiency.

[0169] Compared to Figure 5 The results of different β, η and In different p s The impact on system energy efficiency under the condition of Figure 6 The different sn1 d nf1 and In different p s The impact of d on system energy efficiency. As can be seen from the figure, when d sn1 =4m,d nf1 =8m and When the system energy efficiency reaches the optimal value, the system energy efficiency increases with d sn1 d nf1 At rated p s When d sn1 d nf1 When the reduction leads to an increase in the system channel gain, Under the same conditions Figure 4 It has been proven that the system channel gain is positively correlated with the system energy efficiency, but Figure 6 Indicated in The system energy efficiency gradually increases under the condition of decreasing. This shows that the system energy efficiency is not only positively correlated with the system channel gain but also It is a negative correlation.

[0170] Figure 7 and Figure 8 The β and p under different schemes were studied respectively. s Impact on system energy efficiency. First, it can be seen from the two figures that the system energy efficiency under all schemes increases with β and p s The system energy efficiency of the proposed optimization scheme is better than that of other schemes. Figure 8 It can be seen that the system energy efficiency under the fixed τ scheme increases with p s The decreasing trend of increasing is greater than that of the proposed optimization scheme and the fixed α scheme, which shows that when p is fixed s Under the condition of , τ affects the system energy efficiency to a greater extent than α.

[0171] Based on the above analysis and discussion of the present invention, it can be known that the method for maximizing energy efficiency based on downlink cooperative wireless energy communication-non-orthogonal multiple access system proposed in the present invention, under the conditions of satisfying user throughput restrictions and energy collection restrictions, uses Dinkelbach, SCA and Lagrange dual methods to jointly optimize power allocation, time slot selection and user transmission power to maximize the overall system energy efficiency. Compared with other methods, the method proposed in the present invention can obtain better system energy efficiency, effectively reduce base station energy consumption and improve user throughput.

Claims

1. A downlink cooperative wireless energy-carrying communication-non-orthogonal multiple access energy efficiency maximization resource optimization algorithm, including the following system models: (1) In a downlink cooperative SWIPT-NOMA system, there is a base station (BS) and two users. Due to physical obstacles or severe shadow effects, there is no direct link between the BS and the remote user User2. The near user User1 acts as a relay to help the BS transmit information to User2. User1 adopts half-duplex working mode and has limited battery capacity. It collects energy from the BS signal and uses this energy to extend the working time of User1. |h1| 2 and |h2| 2 Represent the Rayleigh fading channel gains from BS to User1 and from User1 to User2, |g1| 2 and |g2| 2 denote the random channel power gains from BS to User1 and from User1 to User2, β denotes the path loss coefficient, and d sn1 and d nf1 Respectively represent the distance from BS to User1 and from User1 to User2. (2) The total duration of the transmission frame structure in practical applications is defined as T. The structure is divided into three stages: the energy collection stage of User 1 lasts for τT (0≤τ≤1), the information decoding stage of User 1 lasts for And User2's information decoding phase continues If the BS is powered by s Sending energy harvesting signal, the power received by the receiving end User1 can be expressed as p s |h1| 2 , the energy collected by User1 can be expressed as E h =ηp s |h1| 2 τT, where η represents the energy conversion efficiency of user User1. In the second stage, the BS transmits a superimposed signal Where x1(t) and x2(t) represent the unit power signals sent to User1 and User2, respectively, and α(0≤α≤1) and (1-α) represent the power division ratios of x1(t) and x2(t), respectively. The signal received by User1 can be expressed as Where h1(t) represents the channel gain between BS and User1, n1(t) represents the additive white Gaussian noise (AWGN) at User1, After receiving the signal from the BS, according to the SIC principle, User1 decodes x2 first and then decodes x1. Therefore, the Signal-to-Interference-plus-Noise Ratio (SINR) of x1 and x2 detected by User1 are respectively In the third phase, User1 uses all the energy collected in the first phase to forward the decoded information x2 to User2 as a relay. Assume that User1 only consumes part of the collected energy when acting as a relay, and User2 can successfully decode the information from User1 continuously. Therefore, in the current time slot, the energy collected by User1 from the BS is not fully utilized. This shows that even if the BS reduces its transmission power p s , User1 and User2 can also successfully decode the information they need, thus saving the energy consumption of the BS on the downlink. In order to make full use of the energy collected at User1, we set User1 to use all its collected energy to relay information. The signal received by User2 can be expressed as Where h2(t) represents the channel gain between User1 and User2, n2(t) represents the additive white Gaussian noise at User2, Then, the x2 signal-to-noise ratio (SNR) detected by User2 is Where p represents the transmission power of User 1. (3) The throughputs of x1 and x2 from BS to User1 are expressed as and The throughput of x2 from User1 to User2 is Based on the above transmission protocol and Maximum Ratio Combining (MRC) technology, the total system throughput can be calculated as R total =R 11 +min{R 12 ,R 22 } (10) The total energy consumption of the system base station is expressed as (4) Maximize the overall energy efficiency of all nodes by jointly optimizing the power-time domain resources (α, τ, p). At the same time, we set the constraints of the minimum throughput of User1 and User2, as well as the constraint of ensuring that the energy collected by User1 meets the forwarding requirements. The target problem can be expressed as: C5:p≤p max C6:0≤α≤1,0≤τ Where R1 t and R2 t They represent the minimum throughput that User2 and User1 need to achieve, respectively. max represents the maximum transmission power of User1. Constraints (C1) and (C2) ensure that User1 can successfully decode all the information it receives, and constraint (C3) ensures that User2 can also successfully decode the information forwarded from User1. Constraint (C4) ensures that the energy used by User1 for information forwarding does not exceed the collected energy. In addition, User1's transmission power, power allocation factor, and time slot selection factor are constrained by (C5) and (C6). Since the variables are mutually coupled, problem P1 is a nonlinear non-convex fractional programming problem. It is quite difficult to solve this problem directly. Therefore, an optimization method is proposed, which is to first use the Dinkelbach method to transform the nonlinear problem into a linear problem, then use new variables and perform variable substitution and the SCA method to transform the non-convex optimization problem into a convex optimization problem, and finally use the Lagrange dual method to solve the convex optimization problem to obtain the optimal solution. This process will be repeated until convergence.

2. According to the downlink coordinated wireless energy-carrying communication-non-orthogonal multiple access energy efficiency maximization resource optimization algorithm described in claim 1, the problem is solved: (1)Dinkelbach method Since both the objective function and the constraints are non-convex, the nonlinear characteristics of the fractional programming problem P1 make it difficult to solve directly. By using the Dinkelbach method, the fractional programming problem can be transformed into the form of numerator-denominator subtraction. First, define an optimization parameter λ * as follows where α * ,τ * ,p * represents the solution to problem P1. Lemma 1: For R total (α,τ,p)>0 and E total (τ,p)>0, there exists α * ,τ * ,p * The original problem P1 is maximized. The maximum system energy efficiency value can be obtained if and only if the following formula is satisfied: Proof: Here we use the proof by contradiction to prove Lemma 1. According to the above expression, suppose there is another set of solutions (α F ,τ F ,p F ) and λ F ≥λ * According to the definition of the above formula, we can get This is consistent with the initial assumption λ F >λ * conflict. Through Lemma 1, the non-convex hierarchical objective function can be converted into a linear form. Then, given a parameter λ, the original problem P1 is converted into a linear problem P2 for analysis. Based on the above analysis, an iterative algorithm is obtained, which generates a series of λ, which converges to the feasible solution λ * (α * ,τ * ,p * ), the main framework of the proposed solution is summarized in Algorithm 1. Solution framework for problem P2: The specific process of the algorithm based on the Dinkelbach method is as follows: (2)Lagrangian duality method Although the Dinkelbach method mentioned above has transformed the nonlinear problem into a linear problem, the optimization problem is still a non-convex optimization problem and it is difficult to directly find the solution to the problem. 12 ,γ 22 }, so that the objective function and constraints (C2) and (C3) can be rewritten as The new variable γ introduced here is coupled with the power allocation factor α, and the Hessian matrix of the bilinear function is neither semi-positive definite nor semi-negative definite, so the function is neither convex nor concave. In order to solve this problem, the coupled part on the left side of formula (19) is transformed Based on the above transformation, the SCA method is used to replace the coupling part with their convex upper bound function. Specifically, the first-order Taylor expansion expression is used to satisfy the feasible solution (γ (n) ,α (n) ) approximation (γ-α) 2 The lower bound of . (c-a) 2 ≥(γ (n) -a (n) ) 2 +2(c (n) -a (n) )×[(γ-γ (n) )-(a-a (n) )] (21) According to formula (21), the convex upper bound is used to approximate the coupled part on the left side of formula (19). Therefore, formula (19) can be converted into the following feasible form: Accordingly, problem P2 is re-planned as Obviously, when all optimization variables satisfy the constraints and are in the convex set, the objective function of problem P2' is a linear concave function. Therefore, P2' is a convex optimization problem and satisfies the Slater condition and can be solved by the Lagrangian Dual Method. The Lagrangian dual function of problem P2' is formulated as where L(φ) is the Lagrangian function of problem P2', φ = {α, τ, p, μ1, μ2, μ3, μ4, μ5, μ6, μ7, μ8} is the set of all initial and dual variables, and μ1, μ2, μ3, μ4, μ5, μ6, μ7, μ8 are non-negative Lagrangian multipliers defined according to the constraints (C1), (C4), (C5), (C6), (15), (16), (20) of problem P2'. L(φ) can be written as The dual problem of P3 can be expressed as If problem P2' is to be solved, first, solve problem P3 by giving Lagrange multipliers that satisfy the constraints of problem P4 to obtain the optimization variables α, τ, p, γ. Then, solve P4 by updating the Lagrange multipliers to minimize f(μ1, μ2, μ3, μ4, μ5, μ6, μ7, μ8). This iterative method is used to solve problem P2'. By using the KKT (Karush Kuhn Tucker) condition to solve the optimization variable α under the given Lagrange multiplier opt ,τ opt ,p opt ,γ opt , first we need to find partial derivatives of the variables α, τ, p, γ in the Lagrangian function L(φ) and set their stationary points to zero. The formula is as follows: where γ1'1 is as follows The Lagrange multipliers μ1,μ2,μ3,μ4,μ5,μ6,μ7,μ8 are updated by the subgradient method given below where [z] + =max(0,z), j represents the iteration index, θ represents the iteration step size. The update formula is as follows Based on the above analysis, the specific algorithm flow of the Lagrangian duality method is as follows

Images