Symbiotic radio system resource allocation method based on full duplex-NOMA under alpha-fairness principle

Abstract

The invention relates to a symbiotic radio system resource allocation method based on full duplex-non-orthogonal multiple access (NOMA) under an alpha-fairness principle, and belongs to the field of Internet of Things. The method comprises the following steps that a full duplex-NOMA symbiotic radio system is established, a full duplex mode is adopted at the receiving end and the transmitting end, and the problem of residual self-interference caused by full duplex is considered; combining nonlinear energy collection and imperfect channel state information conditions, and constructing a resource allocation model with maximized throughput of backscattering equipment (BD) under an alpha-fairness framework; a block coordinate descent method is adopted, and the resource allocation model with the maximum throughput capacity is converted into a convex optimization problem; and solving the optimization problem by using a convex optimization toolbox to obtain time slots among NOMA clusters, a reflection coefficient of the BD and full-duplex transmitting power, namely obtaining a resource allocation scheme. According to the method, the BD reachable throughput is obviously improved in fairness, and the good throughput performance is still achieved even under the strong residual self-interference condition.

Description

Technical Field

[0001] The invention belongs to the field of Internet of Things, and relates to a symbiotic radio system resource allocation method based on full-duplex-non-orthogonal multiple access (NOMA) under the α-fairness principle. Background Art

[0002] With the rapid penetration of the Internet of Things in various fields of social production, the number of sensor nodes in the Internet of Things has exploded. However, the scarcity of spectrum resources and the lack of equipment endurance have become key factors restricting the sustainable development of the Internet of Things. In recent years, symbiotic radio, as an emerging communication paradigm, has been considered as one of the core technologies of the future Internet of Things due to its reciprocal sharing characteristics in the spectrum and energy domains and the ability to achieve highly reliable backscattering communication through joint decoding. It is expected to play a more important role in the future communication field.

[0003] However, existing solutions do not fully consider the relationship between BD transmission performance and fairness, and sacrifice fairness between BDs in exchange for an increase in total throughput, which cannot meet the needs of green and wide coverage advocated by the future Internet of Things. In addition, most of them assume that the channel state is perfectly known, the energy collection of BD is linear, and the residual self-interference problem of full-duplex is ignored. Under ideal settings, some links may be interrupted due to unreasonable resource allocation. Summary of the invention

[0004] In view of this, the object of the present invention is to provide a symbiotic radio system resource allocation method based on full-duplex-NOMA under the α-fairness principle.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] A resource allocation method for a symbiotic radio system based on full-duplex-NOMA under the α-fairness principle, the method comprising the following steps:

[0007] Step 1: Establish a full-duplex-NOMA symbiotic radio system, in which both the transmitting and receiving ends use full-duplex mode, and consider the residual self-interference problem caused by full-duplex;

[0008] Step 2: Combining nonlinear energy harvesting and imperfect channel state information conditions, a resource allocation model that maximizes the throughput of backscatter devices (BD) is constructed under the α-fairness framework;

[0009] Step 3: Use the block coordinate descent method to transform the throughput maximization resource allocation model into a convex optimization problem;

[0010] Step 4: Use the convex optimization toolbox to solve the optimization problem and obtain the time slots between NOMA clusters, the reflection coefficient of BD, and the full-duplex transmission power, that is, the resource allocation plan.

[0011] Further, in step 1, the full-duplex-NOMA symbiotic radio system includes two full-duplex access point devices, denoted by FAP1 and FAP2, respectively, with i representing the number of FAPs, i=1,2, and 2K BDs, which are divided into K NOMA backscattering clusters, each containing 2 BDs. Let the BD reflection coefficient in the K NOMA clusters be z=[z 1,1 ,z 1,2 ,...,z K,1 ,z K,2 ], satisfying z k,1 >z k,2 , k = 1, 2, ..., K, where the first subscript corresponds to the NOMA cluster and the second subscript corresponds to the BD in a NOMA cluster. The residual self-interference of the full-duplex access point devices at both ends of the transmission and reception are denoted by I 1 and I 2 Express, satisfy It means that it obeys the complex Gaussian distribution, 0 is the mean of the distribution, θ 2 P i,k is the variance of the distribution, where θ represents the residual self-interference level, P i,k It represents the power allocation of FAPi in the kth NOMA cluster time slot, where the subscript i corresponds to FAPi and k corresponds to NOMA cluster. For the convenience of analysis, BD(k,m) is used to represent the mth BD in the kth NOMA cluster, where k = 1, 2, ..., K and m = 1, 2.

[0012] Further, in step 2, combining nonlinear energy harvesting and imperfect channel state information conditions, the resource allocation model for maximizing BD throughput under the α-fairness framework is configured as:

[0013]

[0014] in, is the transmit power of FAP, represents a 2K×1 dimensional complex matrix space, is the time slot of the NOMA cluster, is the reflection coefficient of BD; max means the objective function takes the maximum value; is the throughput of BD(k,m), the superscript b indicates the throughput of the corresponding BD, the subscript corresponds to BD(k,m), and u α is the α-fairness utility function; and is the achievable throughput and minimum throughput requirement of the FAPi in the kth NOMA cluster time slot, where The superscript s indicates the throughput of the corresponding FAP, the subscript i corresponds to FAPi, and k corresponds to the NOMA cluster. The superscript min indicates the minimum value of throughput; and is the energy collected and consumed by BD(k,m), where The superscript corresponds to the collected energy, and the subscript corresponds to BD(k,m). The superscript corresponds to the consumed energy, and the subscript corresponds to BD(k,m); P i,k is the power allocation of FAPi in the kth NOMA cluster time slot, P i m is the power consumption required by FAPi, the superscript represents the total power consumption required by FAPi, and the subscript corresponds to FAPi; τ k is the time slot of the kth NOMA cluster, T is the time slot of the kth NOMA cluster and the total time slot; z k,1 and z k,2 are the reflection coefficients of BD in the same NOMA cluster; C 1 Ensure the minimum throughput required by 2 FAPs; C 2 Indicates that the energy collected by each BD is greater than the energy consumed; C 3 Indicates the total transmit power limit of FAP; C 4 Indicates that the sum of TDMA time slots cannot be greater than the system frame length and must be non-negative; C 5 is the reflection coefficient requirement for each BD in the NOMA cluster.

[0015] Furthermore, in step 3, the process of solving the resource allocation model for maximizing throughput includes the following steps:

[0016] a. For NOMA cluster time slot optimization, the following optimization problem is obtained. This problem is a linear programming problem and can be solved directly;

[0017]

[0018] and C 4

[0019] Among them, z (n) and P (n) represent the reflection coefficient and transmit power given in the nth iteration respectively.

[0020] b. For the optimization of the reflection coefficient of BD, the following optimization problem is obtained:

[0021]

[0022] and C5

[0023] Among them, τ (n) represents the NOMA cluster time slot given at the nth iteration.

[0024] c. For the transmit power optimization of full-duplex access points, the following optimization problem is obtained:

[0025]

[0026] and C 3

[0027] The beneficial effects of the present invention are:

[0028] In the face of challenges such as scarcity of spectrum resources and excessive energy consumption in the future Internet of Things and the need for green and wide coverage, this paper integrates full-duplex, symbiotic radio and NOMA technologies. Considering nonlinear energy harvesting, channel estimation errors and full-duplex residual self-interference, a symbiotic radio system resource allocation method based on full-duplex-NOMA under the α-fairness principle is designed. With the goal of maximizing the total throughput of BD under the framework of α-fair utility function, a non-convex optimization function constrained by multiple conditions is constructed. The fast coordinate descent method, quadratic change and concave-convex programming are used to transform it into a solvable convex optimization problem; finally, based on the idea of ​​alternating iteration, the joint optimization of NOMA cluster time slot allocation, BD reflection coefficient and FAP power allocation is realized. The results show that compared with the existing methods, the proposed method has good convergence and effectiveness. Moreover, according to the requirements of actual application scenarios, the α value can be flexibly adjusted to achieve different levels of throughput fairness, ensuring that all BDs in the subsystem can achieve efficient and fair backscattering transmission. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 is a flow chart of the resource allocation method of the present invention;

[0030] Figure 2 It is an overall architecture diagram of the symbiotic radio system based on full-duplex-NOMA under the α-fairness principle of the present invention;

[0031] Figure 3 It is a convergence characteristic curve diagram of the objective function under different fairness levels of the present invention;

[0032] Figure 4 is a curve diagram of the total BD throughput of the present invention and the comparative method at different fairness levels; DETAILED DESCRIPTION

[0033] See also Figure 1-Figure 4 , which is a resource allocation method for a symbiotic radio system based on full-duplex-NOMA under the α-fairness principle.

[0034] The present invention provides a symbiotic radio system resource allocation method based on full-duplex-NOMA under the α-fairness principle, such as Figure 1 As shown, the method comprises the following steps:

[0035] Step 1: Establish a full-duplex-NOMA symbiotic radio system, in which both the transmitting and receiving ends use full-duplex mode, and consider the residual self-interference problem caused by full-duplex;

[0036] Step 2: Combining nonlinear energy harvesting and imperfect channel state information conditions, a resource allocation model that maximizes BD throughput is constructed under the α-fairness framework;

[0037] Step 3: Use the block coordinate descent method to transform the throughput maximization resource allocation model into a convex optimization problem;

[0038] Step 4: Use the convex optimization toolbox to solve the optimization problem and obtain the time slots between NOMA clusters, the reflection coefficient of BD, and the full-duplex transmission power, that is, the resource allocation plan.

[0039] Further, in step 1, if Figure 2 The main system of the system uses a full-duplex access point device for transceiver, with one transmitting and one receiving antenna. The secondary system includes two full-duplex access point devices, represented by FAP1 and FAP2, with 2K BDs. The BDs are divided into K NOMA backscattering clusters, each containing 2 BDs. Let the BD reflection coefficient in the K NOMA clusters be z = [z 1,1 ,z 1,2 ,...,z K,1 ,z K,2 ], satisfying z k,1 >z k,2 , k = 1, 2, ..., K. The residual self-interference of the full-duplex access point device at both ends of the transmission and reception is represented by I 1 and I 2 Express, satisfy Where θ represents the residual self-interference level, i = 1, 2. For ease of analysis, BD(k,m) is used to represent the mth BD in the kth NOMA cluster, where k = 1, 2, ..., K, and m = 1, 2.

[0040] Furthermore, in step 2, the process of constructing a resource allocation model for maximizing BD throughput includes:

[0041] a. Based on the minimum throughput constraint of the primary user, the BD energy collection constraint, the total transmit power limit of the access point for full-duplex, etc., a resource allocation model is established to maximize the throughput of the secondary system under the α-fairness framework;

[0042] b. Combining nonlinear energy harvesting and imperfect channel state information conditions, a more realistic resource allocation model is established.

[0043] Assuming that the channel coefficient is estimated based on the minimum mean square error criterion, the true channel coefficient h and the estimated value satisfy where e h represents the channel estimation error, the subscript h corresponds to the channel h, which is a Gaussian random variable unrelated to h, that is, So there is where d 12 is the distance from FAP1 to FAP2, with superscript β 1 represents the path loss index between FAP1 and FAP2. Similarly, f k,m and g k,m The corresponding estimated value and Satisfying relationship between: The subscripts correspond to BD(k,m), and They represent the estimation errors of the corresponding channel coefficients respectively, and the variances of all channel estimation errors are set to be Among them, the channel coefficient between FAP1 and FAP2 is h, and the channel coefficient between FAP1 and BD(k,m) is f k,m , the channel coefficient between FAP2 and BD(k,m) is g k,m .

[0044] Assume that in the communication time slot T, the symbols sent by FAP1 and FAP2 are s 1 and 2 It is assumed that it obeys the standard cyclic symmetric complex Gaussian distribution, that is, i=1,2. Let the symbol transmitted by BD(k,m) to FAP1 and FAP2 be c k,m , the subscript corresponds to BD(k,m), satisfying c k,m ~CN(0,1), k=1,2,...,K, m=1,2. Therefore, the backscattered signal of BD(k,m) can be expressed as

[0045]

[0046] Among them, P i,k satisfy P i m is the total transmission power of FAPi, i=1,2.

[0047] Considering the limited energy collection capability of BD, its energy collection per unit time can be modeled as:

[0048]

[0049] Among them, Φ(p k,m ) indicates that the variable is p k,m The function of η k,m ∈[0,1] represents the energy collection efficiency of BD(k,m), and the subscript corresponds to BD(k,m). Indicates that when the energy harvesting circuit reaches saturation, the superscript corresponds to the saturation state and the subscript corresponds to BD(k,m). Therefore, the energy harvested by BD(k,m) can be expressed as:

[0050]

[0051] At the same time, when BD is in the energy harvesting state, it also needs to consume circuit energy. The circuit energy consumption of BD(k,m) is modeled as:

[0052]

[0053] in, and They represent the power consumption when BD(k,m) is in the backscattering state and in the energy harvesting state respectively. The superscripts BC and EC correspond to the backscattering state and the energy harvesting state respectively, which are used to distinguish them.

[0054] Under the above setting, in the time slot τ k , the signal received by FAP1 is:

[0055]

[0056] Among them, w 1,k Represents the time slot τ k , the noise at FAP1, subscript 1FAP1, k corresponds to the time slot of the NOMA cluster, it is assumed to obey zero mean, and the variance is Additive Gaussian white noise. Similarly, in the time slot τ k , the signal received by FAP2 is:

[0057]

[0058] Among them, w 2,k Represents the time slot τ k The noise at FAP2 is assumed to have zero mean and variance. Additive Gaussian white noise. Assume k=1,2,...,K.

[0059] FAP1 and FAP2 use the same decoding method. Therefore, the signal at FAP1 is analyzed first. k, FAP 1 first decodes the primary signal s from FAP 2 2 , and its corresponding signal to interference and noise ratio (SINR) is:

[0060]

[0061] in, The superscript corresponds to the FAP in the main system, and the subscript corresponds to FAP1 in the kth NOMA cluster time slot. The subscript corresponds to FAP1, and is related to the interference caused by the sign of BD(k,m) backscattering to FAP1, m = 1, 2. Therefore, for the primary system, FAP1 in time slot τ k The throughput is:

[0062]

[0063] where log 2 (·) represents the logarithmic function with base 2.

[0064] Decode the main signal 2 After that, FAP1 decodes the backscattered signals from the BDs of the k-th NOMA cluster. According to the decoding rule of uplink NOMA, BD(k,1) with higher reflection coefficient and better channel condition is decoded first, and then the backscattered signals of BD(k,2) are decoded. Therefore, FAP1 decodes c k,1 and c k,2 The SINRs are:

[0065]

[0066] in The superscripts correspond to BD in the subsystem, and the subscripts correspond to FAP1 decoding c k,1 ,In 1,m represents the interference caused by channel estimation error, subscript 1 corresponds to FAP1, and m corresponds to the mth BD. For BD(k,m), In 1,m Further expressed as

[0067]

[0068] In the process of BD(k,m) backscattering to FAP1, BD(k,m) in time slot τ k The throughput is:

[0069]

[0070] in, The superscripts correspond to BD in the subsystem, and the subscripts correspond to FAP1 and BD(k,m).

[0071] Similarly, FAP2 decodes the main signal s from FAP1 1 The SINR is:

[0072]

[0073] in, Related to the interference caused by the BD(k,m) backscatter symbol to FAP2, m = 1, 2. Therefore, for the primary system, FAP2 k The throughput is:

[0074]

[0075] Similarly, FAP2 decodes c k,1 and c k,2 The SINRs are:

[0076]

[0077] in m=1,2. Therefore, in the process of BD(k,m) backscattering to FAP2, it is in the time slot τ k The throughput is:

[0078]

[0079] In this system, the total throughput of BD(k,m) in the time slot is the sum of the throughput of the entire process of backscattering to FAP1 and FAP2, which can be expressed as:

[0080]

[0081] The present invention will use the α-fair utility function to analyze the achievable throughput of BD under different fairness levels. The general form of the α-fair utility function is:

[0082]

[0083] Wherein, ln(·) represents the logarithmic function with e as the base.

[0084] The goal of the present invention is to optimize the power allocation of full-duplex access points by jointly optimizing the power allocation of full-duplex access points by P=[P 1,1 ,...,P 1,K ,P 2,1 ,...,P 2,K ], the TDMA time slot allocation of each NOMA cluster τ = [τ 1 ,τ 2 ,...,τ K ] and the reflection coefficient z of BD, the resource optimization problem can be described as:

[0085]

[0086] Among them, the constraint condition C 1 Guarantee the minimum throughput required by FAP1 and FAP2; Constraint C 2 Indicates that the energy collected by each BD is greater than the energy consumed; constraint C 3 Indicates the total transmit power limit of FAP; constraint C 4 Indicates that the sum of TDMA time slots cannot be greater than the system frame length and is non-negative; constraint C 5 is the reflection coefficient requirement for each BD in the NOMA cluster.

[0087] For P1, we first use the block coordinate descent method to transform it into the following five sub-optimization problems:

[0088] a. TDMA time slot optimization of NOMA cluster,P1 is transformed into the following optimization problem, which can be solved directly.

[0089]

[0090] and C 4

[0091] b. Optimization of the reflection coefficient of BD. The optimization problem of P1 transformation is as follows:

[0092]

[0093] and C 5

[0094] For z, the objective function is a non-concave function, and the constraint condition C 1 and C 2 is a non-convex constraint.

[0095] In order to solve P1-2, according to the quadratic transformation method, the objective function is rewritten as:

[0096]

[0097] Among them, f k,m,i (·) is the relationship between z and auxiliary variable y k,m,i function, 1≤k≤K, m∈[1,2], i∈[1,2] correspond to FAP1 and FAP2 respectively, define y k,m =[y k,m,1 ,y k,m,2 ]. When m = 1, f k,1,1 (z,y k,1,1 ) and f k,1,2 (z,y k,1,2 ) can be expressed as:

[0098]

[0099] When m = 2, f k,2,1 (z,y k,2,1 ) and f k,2,2 (z,y k,2,2 ) are respectively expressed as:

[0100]

[0101] Through analysis, k,1,1 The optimal value of for:

[0102]

[0103] in, The superscript of represents the optimal value, and the superscript (n) represents the nth iteration value. The optimal values ​​of other auxiliary variables are in a similar format.

[0104] After the second transformation, we further consider adopting the α-fair utility function. When α≥0, the objective function after the second transformation is concave with respect to the variable z.

[0105] The non-convex constraints of P1-2 are processed. First, C 1 , using the concave-convex programming method, the non-convex function is transformed into an approximate convex function by using Taylor's theorem. Specifically, first and Written in convex form, they are:

[0106]

[0107]

[0108] In order to facilitate analysis, f si Indicates the corresponding bracketed part.

[0109] Next, at the local point z (n) By performing a first-order Taylor series expansion at , we can obtain the lower bounds of the main system throughput for FAP1 and FAP2 respectively. and They are:

[0110]

[0111] in, The superscript L1 indicates the lower bound of the throughput value for FAP1.

[0112] C 2 , by introducing the slack variable r k,m, convert it into a convex constraint, where 1≤k≤K, m∈[1,2]. Through the above transformation, P1-2 becomes P1-2′. Obviously, this problem is a convex optimization problem, so it can be effectively solved using convex optimization (CVX) tools.

[0113] d. Transmit power optimization of full-duplex access points. The optimization problem of P1 conversion is as follows:

[0114]

[0115] and C 3

[0116] Since the objective function and the constraint C 1 and C 2 are non-convex, so similar to the above, the quadratic transformation and Taylor expansion are used to respectively transform the objective function and the constraint condition C 1 After this processing, we get:

[0117]

[0118] in where the subscripts correspond to BD(k,m).

[0119] C 2 , Introducing the slack variable γ k,m ;right Introducing slack variables β j,m ; Through this treatment, problem P1-3 is approximated as P1-3′, which is a convex optimization problem and can be solved using the CVX tool.

[0120] Example

[0121] This example describes the application effect of the present invention in detail in combination with simulation.

[0122] This example verifies the convergence and effectiveness of the method through simulation results. Assume that the number of NOMA clusters is 2, that is, the system contains 4 BDs; the distance between FAP1 and FAP2 is 12m; the distances from FAP1 to the 4 BDs are 7m, 9m, 11m, and 13m respectively; the distances from FAP2 to the 4 BDs are 8m, 10m, 12m, and 13m respectively; other parameters are set as follows, β 1 =3,β 2 =2,β 3 =2, σ e =10 -3 ,θ=0.01,T=2,η k,m =0.6,

[0123] In this embodiment, the proposed symbiotic radio system resource allocation method based on full-duplex-NOMA under the α-fairness principle is transformed from Figure 3 It can be seen that the objective function of this method changes with the number of iterations. In this figure, the ordinate represents the specific value of the objective function, that is, the value after the BD throughput is substituted into the α-fair utility function. As can be seen from the figure, when 0≤α≤1, the value of the objective function is positive; when α>1, the value of the objective function is negative, which corresponds to the form of the utility function. In particular, the objective function tends to be stable after about 5 iterations, that is, convergence is achieved, indicating that after 5 iterations, the BD of this scheme can achieve stable throughput under different fairness level requirements.

[0124] Compared with traditional methods, Figure 4 It can be seen that the total throughput of all schemes decreases with the increase of α, which confirms that maximizing the total throughput usually comes at the expense of reducing fairness between BDs. However, considering the fairness between BDs allows the receiver to fully collect information from BDs, which is more in line with the actual situation. Therefore, maximizing the total throughput of BDs under the α fair utility function can control the balance between throughput performance and fairness, and obtain good performance in terms of overall system performance. At the same time, under the same residual self-interference level, the resource allocation scheme of full-duplex-NOMA (this scheme) has higher throughput performance than the full-duplex-TDMA scheme; compared with the half-duplex-NOMA scheme, there is no self-interference problem in the half-duplex, but this scheme can obtain better throughput performance.

[0125] Finally, it should be noted that the above preferred embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail through the above preferred implementation cases, those skilled in the art should understand that various changes can be made in form and details without departing from the scope defined by the claims of the present invention.

Claims

1. A resource allocation method for a symbiotic radio system based on full-duplex non-orthogonal multiple access (NOMA) under the α-fairness principle, characterized in that: The following steps are involved: Step 1: Establish a full-duplex-NOMA symbiotic radio system, in which both the transmitting and receiving ends use full-duplex mode, and consider the residual self-interference problem caused by full-duplex; Step 2: Combining nonlinear energy harvesting and imperfect channel state information conditions, a resource allocation model that maximizes the throughput of backscatter devices (BD) is constructed under the α-fairness framework; Step 3: Use the block coordinate descent method to transform the throughput maximization resource allocation model into a convex optimization problem; Step 4: Use the convex optimization toolbox to solve the optimization problem and obtain the time slots between NOMA clusters, the reflection coefficient of BD, and the full-duplex transmission power, that is, the resource allocation plan.

2. The method according to claim 1, characterized in that: In step 1, the full-duplex-NOMA symbiotic radio system includes two full-duplex access point devices, denoted by FAP1 and FAP2, respectively, with i representing the number of FAPs, i=1,2, and 2K BDs, which are divided into K NOMA backscattering clusters, each containing 2 BDs. Let the BD reflection coefficient in the K NOMA clusters be z=[z 1,1 ,z 1,2 ,...,z K,1 ,z K,2 ], satisfying z k,1 >z k,2 , k = 1, 2, ..., K, where the first subscript corresponds to the NOMA cluster and the second subscript corresponds to the BD in a NOMA cluster. The residual self-interference of the full-duplex access point device at both ends of the transmission and reception is represented by I1 and I2 respectively, satisfying It means that it obeys the complex Gaussian distribution, 0 is the mean of the distribution, θ 2 P i,k is the variance of the distribution, where θ represents the residual self-interference level, P i,k It represents the power allocation of FAPi in the kth NOMA cluster time slot, where the subscript i corresponds to FAPi and k corresponds to NOMA cluster. For the convenience of analysis, BD(k,m) is used to represent the mth BD in the kth NOMA cluster, where k = 1, 2, ..., K and m = 1, 2.

3. The method according to claim 1, characterized in that: In step 2, combining nonlinear energy harvesting and imperfect channel state information conditions, the resource allocation model for maximizing BD throughput under the α-fairness framework is configured as: in, is the transmit power of FAP, represents a 2K×1 dimensional complex matrix space, is the time slot of the NOMA cluster, is the reflection coefficient of BD; max means the objective function takes the maximum value; is the throughput of BD(k,m), the superscript b indicates the throughput of the corresponding BD, the subscript corresponds to BD(k,m), and u α is the α-fairness utility function; and is the achievable throughput and minimum throughput requirement of the FAPi in the kth NOMA cluster time slot, where The superscript s indicates the throughput of the corresponding FAP, the subscript i corresponds to FAPi, and k corresponds to the NOMA cluster. The superscript min indicates the minimum value of throughput; and is the energy collected and consumed by BD(k,m), where The superscript corresponds to the collected energy, and the subscript corresponds to BD(k,m). The superscript corresponds to the consumed energy, and the subscript corresponds to BD(k,m); P i,k is the power allocation of FAPi in the kth NOMA cluster time slot, P i m is the power consumption required by FAPi, the superscript represents the total power consumption required by FAPi, and the subscript corresponds to FAPi; τ k is the time slot of the kth NOMA cluster, T is the time slot of the kth NOMA cluster and the total time slot; z k,1 and z k,2 They are the reflection coefficients of BDs in the same NOMA cluster; C1 guarantees the minimum throughput required by two FAPs; C2 indicates that the energy collected by each BD is greater than the energy it consumes; C3 indicates the total transmit power limit of FAP; C4 indicates that the sum of each TDMA time slot cannot be greater than the system frame length and the non-negative requirement; C5 is the reflection coefficient requirement for each BD in the NOMA cluster.

4. The method according to claim 1, characterized in that: In step 3, the process of solving the resource allocation model for maximizing throughput includes the following steps: a. For NOMA cluster time slot optimization, the following optimization problem is obtained, which can be solved directly. Among them, z (n) and P (n) represent the reflection coefficient and transmit power given in the nth iteration respectively. b. For the optimization of the reflection coefficient of BD, the following optimization problem is obtained: Among them, τ (n) represents the NOMA cluster time slot given in the nth iteration. For z, the objective function is a non-concave function, and the constraints C1 and C2 are non-convex constraints. According to the quadratic change method, the objective function can be optimized. For C1, the concave-convex programming method is used to transform the non-convex function into an approximate convex function by using Taylor's theorem. For C2, the slack variable r is introduced. k,m , convert it into a convex constraint, where the subscript corresponds to BD(k,m), 1≤k≤K, m∈[1,2]. Thus, the problem can be transformed into a convex optimization problem. c. For the transmission power optimization of the full-duplex access point device, the following optimization problem is obtained: Among them, since the objective function, and the constraints C1 and C2 are all non-convex, similar to the above, the quadratic transformation and Taylor expansion are used to process the objective function and constraint C1 respectively; for constraint C2, Introducing the slack variable γ k,m , the subscript corresponds to BD(k,m); Introducing slack variables β j,m , the subscript corresponds to BD(j,m); transform the problem into a convex optimization problem.

5. The method according to claim 1, characterized in that: In step 4, the optimization problem has been transformed into a convex optimization problem. The convex optimization (CVX) tool can be used directly to solve the optimization problem and obtain the time slots between NOMA clusters, the reflection coefficient of BD, and the full-duplex transmission power, that is, the resource allocation plan.

Images