Resource Allocation Method for Cognitive Radio Non-Orthogonal Multiple Access Backscatter Networks

Abstract

For the downlink communication of a non-orthogonal multiple access (NOMA) backscatter cognitive radio (NB-CR) network with multiple backscatter devices (BDs), the present invention proposes a resource allocation (RA) scheme that jointly optimizes the power allocation coefficient and the reflection coefficient (RC) to maximize the energy efficiency (EE) under energy and quality of service (QoS) constraints. The steps of the present invention mainly include 1) communication network modeling; 2) RA problem modeling; and 3) problem solving. After modeling the communication network, a non-convex optimization problem is proposed. To solve this non-convex problem, it is decomposed into two sub-problems: the power allocation optimization sub-problem and the RC optimization sub-problem. For the power allocation optimization sub-problem, an optimization power allocation based on Lagrangian (LOPA) scheme is proposed, which uses the Lagrangian and subgradient iteration method to solve. For the RC optimization sub-problem, an improved particle swarm RC (TPSORC) scheme is designed to determine the optimal RC.

Description

Technical Field

[0001] The present invention belongs to the field of cognitive radio (CR) non - orthogonal multiple access (NOMA) backscatter communication. The proposed resource allocation (RA) algorithm can allocate power for users to achieve the maximum energy efficiency, taking into account the throughput and energy consumption of user decoding. Background Art

[0002] Applications and wireless devices continuously access wireless networks, leading to an increasing demand for spectrum. NOMA backscatter networks have advantages such as low power consumption and high spectrum utilization. NOMA technology improves spectrum efficiency by transmitting more user data in the same frequency band and dynamically allocating spectrum according to the needs of different users. A report by the Federal Communications Commission (FCC) of the United States points out that the allocated spectrum resources are not used efficiently. The emergence of CR has improved the utilization of unused spectrum. The main idea of CR is to allow secondary users (SUs) to coexist with primary users (PUs) and allow SUs to use the unused spectrum space without authorization. To further improve spectrum efficiency, CR is introduced into the NOMA backscatter network to form a cognitive radio network based on NOMA backscatter (NB - CR).

[0003] In the NB - CR network, to improve the data rate of users and reduce the outage probability, we need an effective resource allocation algorithm to allocate resources for different users on the same frequency resources. Currently, a large number of studies focus on the RA problem in the NB - CR network. The optimization objective of this problem is usually the total system rate. Common optimization variables include power allocation variables, time allocation variables, etc. Power allocation means allocating higher power to users with poor channel conditions. Time allocation means allocating time for activities of backscatter devices (BDs) such as backscattering and energy harvesting (EH).

[0004] Energy efficiency (EE), as a measurement criterion different from the outage probability and total rate, measures the balance among energy, rate, and energy consumption. It can further evaluate the performance of communication networks. For RA algorithms with EE as the optimization objective, mathematical methods, heuristic methods, or reinforcement learning methods are usually used to solve the problem and find the optimal optimization variables. Summary of the Invention

[0005] The present invention mainly studies and designs a resource allocation algorithm under the NB - CR network. Considering the interference of the primary network on the secondary network and under the constraints of energy and signal - to - interference - plus - noise ratio, an optimization problem of power allocation for NOMA users is proposed with the goal of maximizing energy efficiency, and the optimal power allocation coefficient and backscatter coefficient are obtained for this problem. It includes the following steps:

[0006] Step 1, Communication Network Modeling: In the network model, the secondary transmitter (ST), BD, and users are all equipped with a single antenna, and all channel coefficients follow independent Rayleigh distributions. This network is mainly divided into a primary network and a secondary network. The primary network mainly consists of primary users (PU) and is mainly interfered with by the ST. The secondary network mainly includes the ST, two secondary users (SU), and N BDs. The ST sends a superimposed signal to the two SUs based on NOMA technology. The SUs decode the signal based on successive interference cancellation (SIC) technology. The BDs receive a part of the signal for energy harvesting (EH) to support their circuit operations, and the other part for backscattering. The BDs use their signals to modify the received signal and then reflect it to the two users and the remaining BDs.

[0007] Step 2, Resource Allocation (RA) Problem Modeling: The optimization objective of this problem is to maximize the energy efficiency, that is, the ratio of the total throughput of the users to the energy consumption of the system. The throughput of the system is mainly calculated by R = log2(1 + SINR), where SINR represents the signal-to-interference-plus-noise ratio for user decoding. The energy consumption of the system mainly includes the transmission power of the ST, and the circuit energy consumption of the BDs and SUs. Among them, according to the maximum interference that the PU can accept, the maximum transmission power of the ST is obtained. The constraints of the RA problem mainly include: the reflection coefficient r and the power allocation coefficient need to take values in the range of (0, 1), the sum of the power allocation coefficients assigned to the two users is 1, that is, a1 + a2 = 1, the harvested energy needs to be greater than the circuit energy consumption, and each user needs to meet the corresponding quality of service (QoS) requirements (the SINR during decoding shall not be lower than the specified minimum value). Under the above optimization conditions, maximize EE to obtain the optimal power allocation coefficient and backscattering coefficient.

[0008] Step 3, Problem Solving: Due to the coupling between the power allocation coefficient and the backscattering coefficient, the problem to be solved is a non-convex problem. To solve this problem, it is divided into two sub-problems for separate solutions: the power allocation coefficient sub-problem and the backscattering coefficient sub-problem.

[0009] Step 3.1, Power Allocation Coefficient Optimization Sub-problem: For this sub-problem, a Lagrangian-based power allocation (LOPA) algorithm is proposed. Through the relationship between the power allocation coefficients, a2 = 1 - a1 is obtained, and then the power allocation optimization sub-problem is obtained. This sub-problem is a convex problem, so the Lagrangian method is used in combination with the KKT conditions to solve the power allocation coefficients, and the sub-gradient iteration method is used to iteratively update the Lagrangian multiplier. When the difference in the function values of the iteration is less than the convergence accuracy, or when the maximum number of iterations is reached, the iteration is stopped.

[0010] Step 3.2, Backscattering Coefficient Optimization Sub - problem: For this sub - problem, an improved particle swarm optimization backscattering coefficient (TPSORC) optimization algorithm is proposed. By solving the previous sub - problem, the power allocation coefficient is expressed in terms of the backscattering coefficient, and the second sub - problem is transformed into a function about the backscattering coefficient. Combining the particle swarm optimization (PSO) algorithm with the LOPA algorithm to solve this problem, the optimal backscattering coefficient is obtained.

[0011] Maximize EE: Through the proposed algorithms above, based on maximizing EE, the optimal power allocation coefficient and backscattering coefficient are obtained, realizing the resource allocation problem for two NOMA users in the network, and the final resource allocation scheme is obtained.

[0012] The present invention has the following significant advantages compared with the existing technologies:

[0013] 1. A NB - CR network is proposed, where the ST interferes with the PU within the acceptable interference range of the PU. This network contains multiple BDs and two NOMA users, where the ST communicates with the SU based on NOMA technology, and the BDs perform EH in a time - slot - sharing manner and backscatter based on the time - division multiple access (TDMA) technology.

[0014] 2. The present invention proposes a RA optimization problem to jointly optimize the power allocation coefficient and the backscattering coefficient. The purpose of this problem is to maximize EE under the constraints of multi - time - slot energy causality and user QoS.

[0015] 3. To solve this problem, we design a LOPA algorithm, which uses the Lagrangian and sub - gradient iterative optimization algorithms to solve the power allocation optimization sub - problem. For the backscattering coefficient optimization sub - problem, we design the TPSORC algorithm to obtain the optimal backscattering coefficient.

[0016] 4. For comparison, benchmark models are introduced: the orthogonal multiple access (OMA) backscattering CR model and the pure NOMA CR model. Under different models, comparative experiments are carried out using the TPSORC algorithm and the average power scheme proposed in the present invention. The experimental results show that the scheme proposed in the present invention can improve the system performance more compared with the benchmark schemes. Brief Description of the Drawings

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present specification or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments in the embodiments of the present specification, and other drawings can also be obtained based on these drawings.

[0018] Figure 1It is the overall flowchart of the resource allocation algorithm for the NOMA backscatter cognitive radio network provided by the embodiments of this specification.

[0019] Figure 2 It is the network model diagram of the resource allocation algorithm for the NOMA backscatter cognitive radio network provided by the embodiments of this specification.

[0020] Figure 3 It is the energy harvesting time slot diagram of the resource allocation algorithm for the NOMA backscatter cognitive radio network provided by the embodiments of this specification.

[0021] Figure 4 It is the particle swarm optimization process diagram of the resource allocation algorithm for the NOMA backscatter cognitive radio network provided by the embodiments of this specification.

[0022] Figure 5 It is the flowchart of the LOPA algorithm of the resource allocation algorithm for the NOMA backscatter cognitive radio network provided by the embodiments of this specification.

[0023] Figure 6 It is the flowchart of the TPSORC algorithm of the resource allocation algorithm for the NOMA backscatter cognitive radio network provided by the embodiments of this specification. Detailed implementation manners

[0024] The applicant found that the disadvantages of the RA algorithm in the NB-CR network are as follows: 1) Most of the work focuses on optimizing the total rate and outage probability, and little attention is paid to the energy efficiency (EE); 2) A small amount of EE optimization work is either for uplink communication or only considers one backscatter device (BD). In this paper, for the downlink communication of the NB-CR network with multiple BDs, an RA problem of jointly optimizing the power allocation coefficient and the reflection coefficient (RC) under energy and quality of service (QoS) constraints is proposed to maximize the EE.

[0025] Therefore, according to the disadvantages of the existing RA algorithm, the applicant proposed an RA algorithm for the multi-BD, dual-NOMA user network of NOMA backscatter cognitive radio, and explored how the base station allocates power to NOMA users in the downlink to maximize the EE.

[0026] In order to more clearly understand the above objects, features and advantages of the present invention, the following combines the accompanying drawings to illustrate the technical solutions provided by the embodiments of this specification.

[0027] Figure 1 It is the overall flowchart of a resource allocation scheme applicable to cognitive radio NOMA backscatter, including: communication network modeling, resource allocation problem modeling, problem solving, power allocation coefficient optimization sub-problem, backscatter coefficient optimization sub-problem, maximizing EE. The communication network modeling is mainly reflected inFigure 2 , problem solving is mainly reflected in Figure 4 and Figure 5 . The problem modeling of resource allocation is mainly reflected as follows:

[0028]

[0029] s.t. C1: 0 < r n ≤ 1,

[0030] C2: 0 < a1(n) < 1, 0 < a2(n) < 1,

[0031] C3: a1(n) + a2(n) = 1,

[0032] C4: EH(n) ≥ Ecc(n),

[0033] C5: γ 1→1 ≥ γ 1min , γ 2→2 ≥ γ 2min

[0034] As described in the above formula, the maximization objective is EE, that is, the ratio of the throughput of the system to the energy consumption. The main constraints include five. Among them, γ1min and γ2min represent the minimum SINR thresholds of SU1 and SU2 respectively. C1 is the constraint of the backscattering coefficient (RC). C2 and C3 are the constraints of the power allocation coefficient. C4 is the constraint between the collected energy and the energy consumption. The energy collected from the previous time slot can be used to support the power consumption of the next time slot. C5 is the constraint to ensure the QoS of each user.

[0035] Figure 2 Model the cognitive radio NOMA backscatter network. In this network model, it is divided into a primary network and a secondary network. The primary network consists of PUs and is mainly interfered by ST. The secondary network includes ST, two SUs, and N BDs. The single-antenna ST sends a superimposed signal to the two SUs based on NOMA technology and sends a signal to the BDs. The SUs decode the signal based on successive interference cancellation (SIC) technology. A part of the signal received by the BDs is used for EH to support its circuit operation, and the other part is used for backscattering. The BDs perform backscattering based on TDMA. During the backscattering time slot, they not only backscatter to the NOMA users but also backscatter signals to the remaining BDs that have not been backscattered.

[0036] Figure 3 For the time slot model of energy harvesting. In each time slot, ST sends a signal to the BDs. The BDs perform backscattering based on TDMA and backscatter to the NOMA users and the BDs that have not been backscattered.

[0037] The first BD has two states: active state and dormant state. The Nth BD also has two states: active state and waiting state. The remaining BDs have three states: active, waiting, and sleeping. When the BD backscatters, the BD is in the active state. When the BD has not completed backscattering, the BD is in the waiting state. When the BD completes its backscattering time slot, the ST will no longer send signals to the BD, and the backscattered signal from the BD is too weak to reach the EH threshold and the signal reception threshold. Therefore, at this time, the BD neither backscatters nor harvests energy and enters the sleeping state.

[0038] Figure 4 The process of the particle swarm optimization algorithm. The steps of the particle swarm optimization algorithm are mainly divided into three steps: i) Initialization: Initialize the positions and velocities of all particles. ii) Iterative update: In each iteration, update the positions and velocities of each particle, as well as the historical best position p and the global best position g of each particle. iii) Stopping criterion: When the iteration reaches the maximum number of iterations T or the objective function converges, stop the iteration.

[0039] The position of each particle is a vector composed of the reflection coefficient r. Therefore, the number of reflection coefficients r is the particle dimension D. We set the maximum and minimum values of the position between [0,1]. Set the historical optimal value as pbest, the optimal value as gbest, the optimal value array is represented by gb, and M represents the number of swarm particles.

[0040] In the iteration, continuously update the historical and global optimal positions and optimal values. Compare the fitness value of the particle with the historical best value. If it is better, update the historical best value and the historical best position. The global best value and the best position are updated according to the historical best values and best positions of all particles. Subsequently, update the position and velocity of the particle and perform boundary processing. When the particle exceeds the set velocity boundary Vmin or Vmax, limit the velocity between Vmin and Vmax and randomly generate a new velocity value. Similarly, if the particle exceeds the position boundary Xmin or Xmax, limit the position between Xmin and Xmax and randomly generate a position value.

[0041] Figure 5 Describes the LOPA algorithm for solving the power allocation coefficient. This algorithm updates Ps and a1 through Formula 1 and Formula 2. The interference of the ST to the PU is mainly represented by PI, Psb represents the maximum transmission power of the ST, h_SR represents the channel gain from the ST to the PU, and the channel gain is calculated through calculation, where represents the path loss exponent, and d represents the distance between communication nodes. Among them

[0042] where a1 *= max(0, min(1, a1)). According to a1 + a2 = 1, we can get a2 * = 1 - a1 * .

[0043] It is judged whether the difference of the iteration values is less than the convergence value through abs(a1.*A - noise.*rmin)>ε and abs(B - a1.*B.*(1 + rmin))>ε. Formulas 3 and 4 are used to update the Lagrange multipliers:

[0044] α t+1 = α t - s1 t (a1 * A - δ 2 γ 2min ) (3)

[0045] β t+1 = β t - s2 t *(B - a1 * B(1 + γ 2min ) - δ 2 γ 2min ) (4)

[0046] Where t represents the number of iterations, and s1 and s2 respectively represent the positive step sizes of the Lagrange multipliers α and β. When the number of iterations reaches the maximum number of iterations or the difference of the iteration values is less than the convergence value, the iteration stops; otherwise, the Lagrange multipliers are continuously updated for iteration. The execution steps of the specific algorithm can be expressed as follows:

[0047] S100: Input initial parameters: PI, γmin, etc.;

[0048] S101: Initialize the Lagrange multipliers and the iteration step size;

[0049] S102: Iteratively update Ps and a1 according to the above formulas;

[0050] S103: Iteratively update the step size according to Formulas 3 and 4. During the iteration, Ps and a1 can be updated according to the updated step size;

[0051] S104: Judge whether to continue the iteration according to whether the values of two consecutive iterations are less than the convergence value or reach the convergence number of times. The iteration repetition steps are S102 - S103, where the iteration judgment condition is: abs(a1.*A - noise.*rmin)>ε & abs(B - a1.*B.*(1 + rmin))>ε;

[0052] S105: When the iteration ends, obtain the final optimal solution, that is, the optimal power allocation coefficient.

[0053] Figure 6 It describes the final resource allocation algorithm for the problem model. The optimization of the backscattering coefficient adopts the particle swarm optimization method. The specific optimization steps can be seen Figure 3 , and it is combined with the LOPA algorithm to obtain a new algorithm, the TPSORC algorithm. The specific steps of this algorithm are as follows:

[0054] S200: Initialize parameters,

[0055] S201: Update the velocity and position of the particle swarm

[0056] S202: Update the global and historical best positions

[0057] S203: Process the particles that exceed the boundary:

[0058] S204: Update Ps and a1 according to iteration formulas 1 and 2.

[0059] S205: Under the constraints that the energy consumption should be less than the energy collected by BD and the signal-to-interference-plus-noise ratio decoded by the user should not be lower than the minimum signal-to-interference-plus-noise ratio requirement, judge the value of EE. If it meets the conditions, calculate normally; if it does not meet the conditions, set EE to 0.

[0060] S106: Determine whether to continue updating the step size according to whether the difference between two iterations is less than the convergence value. The specific iteration steps are: S201~S205. When the difference between two iterations is less than the convergence value, a global optimal solution is obtained. When the number of iterations exceeds the maximum number of iterations, the loop ends.

[0061] The above embodiments are mainly applied to the communication network of multiple BDs in cognitive radio based on NOMA backscattering. In fact, the embodiments of this description can be applied to more communication network scenarios. Next, this description will describe the application of the above algorithm to the NOMA backscattering network.

[0062] In the NOMA backscattering network, first, for the modeling of the network model, it is no longer divided into the primary network and the secondary network, and the transmit power of the base station is no longer restricted by the interference of the primary users.

[0063] For problem modeling, it is similar to the problem modeling in the cognitive radio network based on NOMA backscattering. The maximization objective is still EE, that is, the ratio of the system throughput to the energy consumption. The main constraint conditions include five. Among them, γ1min and γ2min respectively represent the minimum SINR thresholds of SU1 and SU2. C1 is the constraint of the backscattering coefficient (RC). C2 and C3 are the constraints of the power allocation coefficient. C4 is the constraint between the collected energy and the energy consumption. The energy collected in the previous time slot can be used to support the power consumption in the next time slot. C5 is the constraint to ensure the QoS of each user. The specific modeling problem is as follows:

[0064]

[0065] s.t.C1:0 < r n ≤1,

[0066] C2:0 < a1(n) < 1, 0 < a2(n) < 1,

[0067] C3:a1(n) + a2(n) = 1,

[0068] C4:EH(n) ≥ Ecc(n),

[0069] C5:γ 1→1 ≥ γ 1min , γ 2→2 ≥ γ 2min

[0070] In the energy harvesting time slot problem, BD (excluding the first and the last) still has three states for energy harvesting and backscattering. Energy harvesting plays an important role in the energy efficiency optimization of the system. Backscattering to BD can be used to enhance the energy collected by BD, and backscattering to the user can be used to enhance the signal received by the user and improve the signal-to-interference-plus-noise ratio (SINR) for user decoding.

[0071] When the above communication network model, problem model, and energy harvesting model are determined, the LOPA algorithm and TPSORC algorithm can be continued to be used to solve the optimization variables to obtain the optimal power allocation coefficient and backscattering coefficient, and the steps are as described above.

[0072] The algorithm proposed in this description is also applicable to orthogonal multiple access (OMA) backscatter cognitive radio with multiple BD networks. Therefore, the third embodiment is mainly applied to orthogonal multiple access (OMA) backscatter cognitive radio.

[0073] In the OMA backscatter cognitive radio with multiple BD networks, the original dual NOMA users become OMA users. Therefore, the signals transmitted by the base station and the signals received by BD and the user will change, and accordingly, the problem model will also change. The original optimization variables change from the power allocation coefficient and backscattering coefficient to a single backscattering coefficient optimization variable. During the problem modeling process, the power allocation coefficient can be set to 1. On this basis, the TPSORC algorithm is used to solve the backscattering coefficient.

Claims

1. A cognitive radio non-orthogonal multiple access backscatter network resource allocation method, characterized in that: include: S1. Communication network modeling: The primary network consists of the primary user PU; the secondary network consists of ST, two secondary users SU and N BDs; ST sends superimposed signals to the two SUs based on NOMA technology; SU decodes the signals based on continuous interference cancellation technology; S2. Resource allocation RA problem modeling: According to the ratio of the total throughput of the user to the energy consumption of the system, solve the maximum energy efficiency; calculate the system throughput through R = log2 (1 + SINR), where SINR is the signal to interference noise ratio of the user decoding; the energy consumption of the system includes the transmission power of the ST, the circuit energy consumption of the BD and the SU; according to the maximum acceptable interference of the PU, the maximum transmission power of the ST is obtained; maximize the energy efficiency and obtain the optimal power allocation coefficient and backscattering coefficient. The problem modeling is: S3. Solve the power allocation coefficient optimization sub-problem: through the relationship between the power allocation coefficients, we get a2=1-a1, use the Lagrangian method combined with the KKT condition to solve the power allocation coefficient, and use the sub-gradient iteration method to iteratively update the Lagrangian multiplier; when the difference of the iterative function value is less than the convergence accuracy or reaches the maximum number of iterations, stop the iteration; Solve the backscattering coefficient optimization subproblem: By solving the power allocation coefficient optimization subproblem, the power allocation coefficient is expressed as the backscattering coefficient: in, PI is the interference of ST to PU, P sb is the maximum transmission power of ST, h SR is the channel gain from ST to PU; The backscattering coefficient optimization subproblem is transformed into a function of the backscattering coefficient. The particle swarm optimization algorithm is combined with the LOPA algorithm to solve the problem and obtain the optimal backscattering coefficient.

Images