A method for optimizing downlink resource allocation in NOMA system

Abstract

The present invention discloses a method for optimizing allocation of downlink resources of a NOMA system. The method is as follows: 1) for a NOMA downlink system consisting of one base station and S users, a power allocation model is established for calculating the power allocated to each user according to the channel state information of each user in the NOMA downlink system; 2) an objective function is established with the maximum power of the base station and the maximum energy efficiency to ensure the quality of the user as the allocation target; 3) based on the objective function, a hybrid whale particle swarm optimization algorithm HWPSO is used to search for the allocation of user power and determine the power allocated to each user. The present invention can obtain the optimal power allocation strategy that maximizes the total throughput of the system.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technology, and in particular to improving the spectrum utilization and system capacity of a wireless communication system by effectively allocating wireless resources. Specifically, the present invention proposes a downlink power allocation optimization method for a non-orthogonal multiple access (NOMA) system based on a hybrid whale particle swarm optimization algorithm (HWPSO) to obtain a higher total system throughput. Background Art

[0002] As an emerging wireless access technology, NOMA has higher spectrum efficiency and system capacity than traditional orthogonal multiple access (OMA) technology. In the NOMA system, multiple users can reuse the same time-frequency resource block, and multiple access and user reuse are achieved through non-orthogonality in the power domain, thereby significantly improving spectrum utilization. Therefore, a reasonable power allocation strategy is crucial to fully realize the potential of NOMA technology. For scenarios with few users, traditional algorithms such as equal power allocation and allocation based on channel gain ratio can be used to achieve optimal downlink power allocation of the NOMA system. With the continuous expansion of the user scale, especially in actual large-scale multi-user mobile wireless communication scenarios, the power allocation optimization problem will become more and more complex. In real scenarios, there are a large number of users accessing at the same time, the number of users changes dynamically, and the channel environment is more complex and changeable. For this highly complex large-scale multi-user NOMA system, traditional power allocation algorithms face new challenges and bottlenecks in terms of optimization accuracy, robustness and computational efficiency. Compared with traditional allocation algorithms, bio-inspired intelligent optimization algorithms can effectively reduce the dependence on precise mathematical models and have higher execution efficiency and optimization accuracy. Therefore, they have received extensive attention from researchers in the field of wireless communication resource allocation.

[0003] As a classic biologically inspired intelligent optimization algorithm, the particle swarm optimization (PSO) algorithm has the advantages of few adjustment parameters, strong optimization ability, and easy programming implementation. It can effectively solve the power allocation problem between users. However, the basic PSO algorithm uses the same particle position update strategy in the global exploration stage and the local utilization stage, and only adopts a simple random initialization method in the initialization process, which makes the entire algorithm process highly random and blind. It lacks guidance based on prior knowledge and historical search experience. When solving the complex multivariable optimization problem of downlink power allocation in large-scale multi-user NOMA systems, it faces many challenges such as slow optimization speed and low convergence accuracy. To this end, the patent of this invention proposes an effective NOMA system downlink resource allocation method based on an improved PSO algorithm-HWPSO. Summary of the invention

[0004] In view of the problems existing in the prior art, the object of the present invention is to provide a NOMA system downlink resource allocation method based on the HWPSO optimization algorithm.

[0005] The present invention takes the downlink resource allocation of the NOMA system as the engineering application background. First, aiming at the problem of maximizing the user transmission rate in the downlink of the NOMA system, a corresponding mathematical optimization model is established, involving decision variables such as user power allocation and related constraints. Secondly, on the basis of the PSO algorithm, the local search strategy of the whale optimization (WOA) algorithm is integrated, and the key parameters in the algorithm are adaptively improved to propose a novel hybrid intelligent optimization algorithm-hybrid whale particle swarm optimization (HWPSO) algorithm. The HWPSO algorithm combines the global exploration capability of the PSO algorithm with the local utilization behavior of the WOA algorithm, balances the global search and local search through an adaptive adjustment mechanism, and applies it to the optimal allocation of downlink resources in the NOMA system to obtain the optimal power allocation strategy that maximizes the total system throughput.

[0006] Simulation experiments are used to verify the feasibility and effectiveness of the HWPSO algorithm in solving the downlink power allocation problem of the NOMA system. Experimental results show that the HWPSO algorithm can achieve a better balance between global optimization capability and local fine search. Especially in complex scenarios such as large-scale users and channel state information (CSI) environments, the algorithm shows obvious superior performance.

[0007] The technical solution adopted by the present invention is:

[0008] A method for allocating downlink resources in a NOMA system, comprising the steps of:

[0009] 1) For a NOMA downlink system consisting of 1 base station and S users, according to the channel state information {h1,h2,...,h S} Modeling is used to calculate the power allocation model (p1, p2, .., p S ), where S = M*J, each channel corresponds to a subcarrier, J represents the total number of subcarriers; M represents the number of users on the jth subcarrier, h S is the channel state information of the Sth user, p S is the power allocation model for the Sth user, j = 1~J;

[0010] 2) To meet the maximum power of the base station and the maximum energy efficiency to ensure user quality as the allocation goal, establish the objective function in, P Cis the path loss of downlink signal transmission, P max is the maximum transmission power; p j,m represents the power allocated to the mth user on the jth subcarrier, P max is the maximum transmission power, R j,m represents the downlink transmission rate of the mth user on the jth subcarrier, R min The minimum transmission rate for users;

[0011] 3) Based on the objective function, the hybrid whale particle swarm optimization algorithm HWPSO is used to search for the optimal allocation of user power and determine the power allocated to each user.

[0012] Furthermore, based on the optimization objective function, the method for optimizing the allocation of user power using the hybrid whale particle swarm optimization algorithm HWPSO is as follows:

[0013] Initialize the population size N, the number of users S, and the maximum number of loop iterations max_iter in the hybrid whale particle swarm optimization algorithm HWPSO, where each particle corresponds to the number of users M on the subcarrier; calculate the position vector X of each particle in each population size as the power allocation of each user on the corresponding subcarrier;

[0014] During the main iterative optimization process of the hybrid whale particle swarm optimization algorithm HWPSO, the new velocity V of particle i is calculated. i and the new position X i , the new position X of each particle i i Substitute the objective function to calculate the corresponding fitness value, and record the current minimum fitness value of the population and the corresponding global optimal solution; if the fitness value of the current particle is less than the current minimum fitness value of the population, update the minimum fitness value and use the current position X of particle i i Update the current global optimal solution gBest;

[0015] In the secondary iterative optimization process of the hybrid whale particle swarm optimization algorithm HWPSO, the global optimal solution of the current particle swarm is used as the leader position of the whale group. Whenever a candidate solution better than the current global optimal solution is found, the global optimal value and the leader position are updated; otherwise, the current optimal power allocation strategy remains unchanged;

[0016] When the number of loop iterations of the main iteration reaches the maximum number of loop iterations max_iter, the optimal energy efficiency is output.

[0017] Calculate the position vector X(i) of each particle i in each population size; where mod() represents the modulo operation.

[0018] Furthermore, the new velocity V of particle i is updated according to the following formula: i and the new position X i :

[0019]

[0020] in, represents the velocity component of user s in the i-th particle at the k+1th iteration; w is the inertia weight, rand1 and rand2 are random variables uniformly distributed in [0,1]; c1 and c2 are individual learning factors and social learning factors, respectively, and Pbest is is the optimal solution for user s in the i-th particle; gbest s is the optimal solution found by user s in the entire particle swarm; X k is is the position of user s of the ith particle at the kth iteration; g is a nonlinear function decreasing from 1 to 0; max_iter is the maximum number of main iterations; iter is the current number of main iterations; It is represented as the position of user s in the i-th particle at the k+1th iteration.

[0021] Furthermore, the secondary iteration optimization method is: the maximum number of iterations SM2 of the secondary iteration is set to a decreasing function of the number of main iterations max_iter indicates the maximum number of loop iterations set; iter indicates the current number of loop iterations; a "forced whale" mechanism is introduced, and its mathematical expression is:

[0022]

[0023] Among them, a is used to control The size of it represents the current number of secondary iterations; To control the whale algorithm's choice of search strategy, When forcibly executing the shrinkage and encirclement strategies of the whale algorithm, the algorithm converges faster; is a random number in the range [0,1]. is a coefficient vector used to calculate the distance between the current position and the optimal position;

[0024] when The shrinking and encircling strategy of the whale algorithm is as follows:

[0025]

[0026] in, and The position of user s in the i-th particle at the k-th and k+1-th iterations, It is represented as the optimal solution of user s in the i-th particle in the k-th iteration, It is represented as a random perturbation solution; l is a random variable uniformly distributed in [-1,1]; p is a random variable of the probability of strategy selection;

[0027] when When the exploration strategy of the whale algorithm is

[0028] A server, characterized in that it comprises a memory and a processor, wherein the memory stores a computer program, the computer program is configured to be executed by the processor, and the computer program comprises instructions for executing the above method.

[0029] The present invention mainly includes the following contents:

[0030] 1. Modeling of downlink resource allocation problem in NOMA system:

[0031] The modeling of resource allocation problem mainly includes the establishment of mathematical models for the transmission rate and power allocation model of the NOMA system downlink user. Consider a NOMA downlink system consisting of 1 base station (BS) and S users. The BS is located in the center of the cell, and the users are randomly distributed near the base station. Each subcarrier can occupy multiple users. The number of users occupied on each subchannel j is set to M, and the user set is expressed as ψ={1,2,...M}. All users share the same time / frequency resources, and the base station implements multi-user multiplexing by performing non-orthogonal power allocation in the power domain. So the signal x sent by the BS to each subchannel j is j It can be expressed as:

[0032]

[0033] In formula (1), x j is the signal sent by the base station BS to each subcarrier j; p j,m is the transmission power of user m on subcarrier j; s m is the modulation symbol, E[s m | 2 =1].

[0034] The total bandwidth of the entire system is W, which is divided into J subcarriers, and the bandwidth of each subcarrier is Wsub = W / J. The maximum allocatable power of the base station is P max Taking a specific subcarrier J1 as an example, a set of users is multiplexed, and the user index is represented by {m1,m2,...m t}, where t = M, the power allocated by the base station to the user set M is expressed as P m, assuming that the base station has learned the channel state information CSI of all users. It is known that when the base station sends signals to each user of multiplexed subcarrier J, the signal will be interfered by noise after being transmitted through the wireless channel. Correspondingly, according to formula (1), for the mth user in the user set of j on each subcarrier, the received signal Y j,m The expression is as follows:

[0035]

[0036] In formula (2), h j,m is the channel gain of the mth user on the jth subcarrier. The second summation term is the superposition of interference signals between users on the same carrier, and the third term N j,m is the Gaussian white noise in the transmission path of user m on subcarrier j.

[0037] According to the principle of serial interference cancellation (SIC), users are demodulated in order of channel gain, and users with larger gains are demodulated first. Taking a specific subcarrier J1 as an example, the present invention assumes that the channel gain of the user satisfies h j1,1 ≥h j1,2 ≥h j1,3 ≥....≥h j1,t , the demodulation order is {m j1,1 ,m j1,2 ,m j1 ,3...m j1,t} (where t = M), and eliminate the interference caused by other users in turn. On this basis, the realized transmission rate R of the mth user on subcarrier j j,m It can be expressed as:

[0038]

[0039] In formula (2), h j,m is the channel gain of the mth user on the jth subcarrier; W is the total downlink system bandwidth (this parameter is known); N j,m is Gaussian white noise.

[0040] Based on the above NOMA downlink system model, the present invention aims to maximize the energy efficiency of the entire system and establishes the following optimization objective function:

[0041]

[0042] In formula (4), max() is the maximum value function; η J,S represents the energy efficiency of the entire system; the denominator summation term represents the sum of the transmission rates of the user set; P C Path loss of downlink signal transmission; P max is the maximum transmission power; p j,mrepresents the power allocated to the mth user on the jth subcarrier; R j,m Indicates the downlink transmission rate of the mth user on the jth subcarrier. C1 indicates that the sum of the power allocated to the user set is not greater than the maximum transmit power of the base station; C2 indicates that the power allocated to the user is not less than 0; C3 ensures the transmission data quality (QoS) constraint of the user.

[0043] Since the objective function is a nonlinear programming problem with constraints, the solution usually adopted by the intelligent swarm algorithm is to transform the constraints into the form of penalty functions. Therefore, the final optimization objective function can be obtained from the objective function (4):

[0044]

[0045] In formula (5), max() is the maximum value function; J is the number of base station subcarriers; m is the number of users on carrier J; M is the total number of users on carrier J; the first term is the inverse of the energy efficiency of the NOMA system M(p j,m ) is the power constraint function, and its calculation method is shown in formula (6); H(R j,m ) is the user quality constraint function QOS, and its calculation method is shown in formula (7):

[0046]

[0047] In formula (6), max() is the maximum value function, p j,m represents the power allocated to the mth user on the jth subcarrier, P max is the maximum transmission power; Formula (6) indicates that the sum of the transmission powers of all user sets under a system cannot exceed the set maximum power P max ; R in formula (7) j,m represents the downlink transmission rate of the mth user of the jth subcarrier; R min is the minimum transmission rate of the user. Formula (7) indicates that the transmission rate of the user in the system cannot be lower than the set minimum rate to ensure the data transmission quality of the user.

[0048] 2. Design of NOMA System Resource Allocation Algorithm

[0049] Based on the modeling of the resource allocation problem of the above NOMA system, the HWPSO algorithm is used to optimize the allocation of user power.

[0050] Since the basic particle swarm algorithm has strong global search capabilities, it converges quickly when searching for the global optimal solution, but has a high risk of falling into the local optimal solution, while the whale algorithm focuses on local search and has a more stable convergence. Therefore, the two algorithms are mixed to avoid the problem of falling into the local optimal solution during the optimization process, making full use of the advantages of the two algorithms at different stages and balancing the optimization capabilities of the exploration and utilization stages.

[0051] In the HWPSO algorithm, each particle represents a candidate power allocation scheme for a group of users. If the number of particle populations is N, the position of the i-th particle (i.e., the power of all user sets) is expressed as P i =(p i1, p i2, ...,p iS )(i=1,2,...,N), S represents the user set; firstly, the Circle initialization mapping elite reverse learning strategy is introduced into the algorithm, and the better solution is selected as the initialization population. The initialization formula is as follows:

[0052]

[0053] In formula (8), X(i) represents the input value of the Circle map when generating the position of the i-th particle. For example, when initializing the first particle, i = 0, X(0) is a random value. By calculation, we can get X(1), which corresponds to the user power matrix P1 = (p 11, p 12, ...,p 1S ), which is the Circle mapping input value of particle 1, which is reflected in the code as X(0) = rand(); mod() represents the modulo operation; +0.2 means introducing a positive offset, which makes the individual move closer to a better area; Indicates mapping X(n) to [0,2π], sin(2π×X(i)) indicates the introduction of nonlinearity to avoid capturing the local optimum. The main iterative particle swarm algorithm search strategy of the HWPSO algorithm is as follows:

[0054]

[0055] In formula (9), represents the velocity component of the S user set in the i-th particle at the k+1-th iteration, m is the user on the j-th subcarrier; w is the inertia weight, with a maximum value of 1 and a minimum value of 0.4; rand1 and rand2 are random variables uniformly distributed in [0,1]; c1 and c2 are individual learning factors and social learning factors, respectively, with a maximum value of 2 and a minimum value of 0.4; Pbest is It is represented as the S component of the best solution in the i-th particle - the optimal power solution of the S user set; gbests It is represented as the S components of the optimal solution found by the S user set in the entire particle swarm N; X k iS It is represented as the position of the S component of the ith particle at the kth iteration, and also represents the power matrix P i =(p i1, p i2, ...,p iS ); g is a nonlinear function that decreases from 1 to 0; max_iter is the maximum number of loop iterations, which is set to 100; iter is the number of main iterations of the current HWPSO; It is represented by the position of the S component of the i-th particle in the k+1th iteration (the position here represents the power allocated to the user set). In the secondary iteration stage of the HWPSO algorithm, the current optimal solution obtained by the particle swarm algorithm in the previous stage is used as the leader position of the initialization whale group. Among them, each whale represents a potential user power allocation solution. In order to reduce the complexity of the algorithm, the maximum number of iterations SM2 of the secondary iteration is set to a decreasing function of the number of main iterations, that is:

[0056]

[0057] In formula (10), SM2 represents the maximum number of iterations of the whale algorithm, which is a function that decreases as the number of current iterations of the particle swarm algorithm increases; max_iter represents the maximum number of iterations set by the particle swarm algorithm; iter represents the current number of iterations of the particle swarm algorithm.

[0058] At the same time, this secondary iteration stage also introduces a "forced whale" mechanism, whose mathematical expression is:

[0059]

[0060] In formula (11), parameter a is expressed as the control As the number of iterations increases, the value of a will become smaller and smaller. It will become smaller and smaller, and the value range of a is [0,2]; it represents the current number of iterations of the HWPSO secondary, and gradually increases to SM2; SM2 represents the maximum number of iterations of the whale algorithm; It is expressed as controlling the whale algorithm's choice of search strategy. When forcibly executing the shrinkage and encirclement strategies of the whale algorithm, this can make the algorithm converge faster; is a random number with a value range of [0,1], so The value range of is [-a, a]; Represented as a coefficient vector, it is mainly used to calculate the distance between the current position and the optimal position.

[0061] Therefore, when The shrinking and encircling strategy of the whale algorithm is as follows:

[0062]

[0063] In formula (12), and The position of the S component of the i-th particle at the k-th and k+1-th iterations is also represented by the power matrix P i =(p i1 ,p i2 ,...,p iS ); It is represented as the s-component of the optimal solution of the i-th particle in the k-th iteration—the optimal power solution of the s-user set; It is expressed as the distance or position between the candidate solution and a reference point (selected according to p). When p < 0.5, represents the difference between the whale position (candidate solution) and the current best solution. When p ≥ 0.5, represents the difference between the whale position (candidate solution) and the randomly perturbed solution; Expressed as a random perturbation solution; e jl is a coefficient of the spiral update strategy, where l is a random variable uniformly distributed in [-1,1], j is an imaginary number, which is mainly used for spiral path generation, simulating the behavior of whales making spiral movements around their prey; p is a random variable for the probability of selecting the [0,1] strategy, when p < 0.5, the surround behavior is executed, and when p ≥ 0.5, the spiral pursuit behavior is executed.

[0064] when The exploration strategy of the whale algorithm is as follows:

[0065]

[0066] In the early stage of the HWPSO algorithm's primary iteration, due to the large number of secondary iterations, the value of parameter a will be very small, and is a value between [-1,1]. Therefore, the whale algorithm will be forced to perform shrinking and surrounding behavior, making full use of the local search capability of the whale algorithm.

[0067] The advantages of the present invention are as follows:

[0068] The method of the present invention adopts a hybrid optimization strategy based on the architecture of the PSO algorithm and WOA. By alternately applying different optimization strategies of the two algorithms, the resource allocation of the NOMA system is made more reasonable and efficient, thereby maximizing the overall energy efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1A resource allocation flow chart provided in an embodiment of the present invention.

[0070] Figure 2 This is a resource allocation result when the user S=4 provided by an embodiment of the present invention.

[0071] Figure 3 This is a resource allocation result when the number of users S=10 provided by the embodiment of the present invention.

[0072] Figure 4 This is a resource allocation result when the number of users S=20 provided by the embodiment of the present invention.

[0073] Figure 5 This is a resource allocation result when the number of users S=30 provided by the embodiment of the present invention.

[0074] Figure 6 This is a resource allocation result when the number of users S=40 provided by the embodiment of the present invention.

[0075] Figure 7 This is a resource allocation result when the number of users S=50 provided by the embodiment of the present invention.

[0076] Figure 8 This is a resource allocation result when the number of users S=60 provided by the embodiment of the present invention.

[0077] Fig. 9 This is a resource allocation result when the number of users S=70 provided by the embodiment of the present invention.

[0078] Fig.10 This is a resource allocation result when the number of users S=80 provided by the embodiment of the present invention.

[0079] Fig.11 This is a resource allocation result when the number of users S=90 provided by the embodiment of the present invention.

[0080] Fig.12 This is a resource allocation result when the number of users S=100 provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0081] The present invention is further described in detail below in conjunction with the accompanying drawings. The examples given are only used to explain the present invention but not to limit the scope of the present invention.

[0082] The core idea of ​​the present invention is to transform the resource allocation problem into a minimum optimization problem with constraints through mathematical modeling, and then optimize and solve the problem using the proposed optimization algorithm.

[0083] The detailed steps of the resource allocation method according to the embodiment of the present invention are as follows:

[0084] Step 1: Determine the NOMA system model parameters and obtain the channel state information CSI {h1,h2,...,h S} Model the resource allocation description of downlink users. By initializing particles with Circle, we can obtain the power solution set allocated to each user. Using equation (3), we can calculate the transmission rate corresponding to each user.

[0085] Step 2: Optimization model building. The maximum energy efficiency under the condition of satisfying the maximum power of the base station and ensuring the user quality QoS is taken as the allocation target. The optimization objective function of the HWPSO algorithm is obtained (5), where the constraint violation is calculated according to (6) and (7).

[0086] Step 3: Algorithm parameter setting and initialization. Preset the algorithm population size N, the number of users S, the maximum number of loop iterations max_iter and other parameters (N = 100, max_iter = 100), perform Circle initialization mapping elite reverse learning on all N, and calculate the position vector X of each particle according to formula (8): iS (1), represents a set of power allocation schemes, and the best candidate solution is taken as the initial population position.

[0087] Step 4: Main iterative optimization process. Calculate the new velocity V of the particle according to equation (9): iS and the new position X iS , the new position X of each particle iS Substitute them into formula (5) to calculate the objective function value as the fitness value, that is, the inverse value of energy efficiency. And record the current minimum fitness value of the population and the corresponding global optimal solution. During the iteration process, the algorithm will compare the fitness value of the current particle with the recorded minimum fitness value. If the fitness value of the current particle is smaller, it means that the power allocation strategy corresponding to the current particle can achieve higher energy efficiency. At this time, it is necessary to update the minimum fitness value and use the current particle position X iS Update the current global optimal solution gbest S , otherwise it remains unchanged.

[0088] Step 5: Secondary iterative optimization process. The algorithm takes the global optimal solution of the current particle swarm as the leader position of the whale group, and dynamically calculates the maximum number of iterations of this secondary iteration according to formula (10). In the early stage of the secondary iteration, the algorithm will update the whale position according to the contraction and encirclement strategy in formula (12), and choose whether to execute contraction or encirclement behavior through random probability p, in order to perform fine local search near the leader; as the secondary iteration continues to decrease, the algorithm will appropriately perturb the whale position according to the exploration perturbation mechanism of formula (13), so that it can escape the local area and explore a large range in the entire solution space. Whenever a candidate solution that is better than the current global optimal solution is found, the global optimal value and leader position are immediately updated; otherwise, the current optimal power allocation strategy remains unchanged.

[0089] Step 6: If the number of loop iterations of the algorithm reaches the set maximum number of loop iterations, the optimal energy efficiency is output, otherwise return to step 4 to continue the next round of iterations.

[0090] In order to verify the feasibility and effectiveness of the method of the present invention in solving the downlink resource allocation problem, the present invention is experimentally compared with the basic PSO and WOA algorithms. The relevant parameters are set as follows: the population size of all algorithms is 50, the maximum number of iterations is 100, and simulation experiments are carried out for different numbers of users S = 4, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100.

[0091] The experimental results are as follows:

[0092] When S=4, the evolution curve of the optimization objective function of the algorithm is as follows: Figure 2 As shown; when S = 10, the evolution curve of the optimization objective function of the algorithm is as follows Figure 3 As shown; when S = 20, the evolution curve of the optimization objective function of the algorithm is as follows Figure 4 As shown; when S = 30, the evolution curve of the optimization objective function of the algorithm is as follows Figure 5 As shown; when S = 40, the evolution curve of the optimization objective function of the algorithm is as follows Figure 6 As shown; when S = 50, the evolution curve of the optimization objective function of the algorithm is as follows Figure 7 As shown; when S = 60, the evolution curve of the optimization objective function of the algorithm is as follows Figure 8 As shown; when S = 70, the evolution curve of the optimization objective function of the algorithm is as follows Fig. 9 As shown; when S = 80, the evolution curve of the optimization objective function of the algorithm is as follows Fig.10 As shown; when S = 90, the evolution curve of the optimization objective function of the algorithm is as follows Fig.11 As shown; when S = 100, the evolution curve of the optimization objective function of the algorithm is as follows Fig.12 As shown; the energy efficiency of the algorithm under different users is shown in Table 1:

[0093] Table 1 Energy efficiency of the algorithm under different users

[0094]

[0095]

[0096] according to Figure 2-Figure 12 From the results of the impact of different power allocation strategies on energy efficiency shown in the figure, we can see that the HWPSO algorithm can plan a power allocation scheme that meets the constraints when the number of users ranges from 4 to 100, and the algorithm has the advantages of fast optimization speed and stable and reliable optimization results. When the number of users is in the range of 4-60, the basic PSO algorithm and WOA algorithm can also allocate user power that meets the constraints; however, when the number of users exceeds 60, the optimization ability of the PSO algorithm and WOA algorithm decreases significantly, and it is impossible to allocate power values ​​for users to obtain the maximum energy efficiency. Table 1 shows the statistical results of energy efficiency obtained by allocating power to users, which further verifies the superiority of the HWPSO algorithm in the reliability and stability of optimization results.

[0097] In summary, compared with the traditional PSO algorithm and WOA algorithm, the HWPSO algorithm shows a better global search ability in the downlink resource allocation problem and can achieve higher energy efficiency, which proves that the downlink resource allocation method based on the HWPSO algorithm is an effective and feasible solution. This algorithm can not only quickly obtain a power allocation strategy that meets the constraints when the number of users is small, but also maintain a high optimization performance even when the number of users is large, and allocate power values ​​that can achieve maximum energy efficiency to users, which is beyond the reach of the basic PSO algorithm and WOA algorithm. Figure 2-Figure 12 It can be clearly seen from the results in Table 1 that the HWPSO algorithm has obvious advantages in terms of optimization speed, reliability and stability of optimization results, and can be well applied to the actual scenario of downlink resource allocation.

[0098] Although the specific embodiments of the present invention are disclosed for the purpose of illustration, the purpose is to help understand the content of the present invention and implement it accordingly, those skilled in the art will understand that various substitutions, changes and modifications are possible without departing from the spirit and scope of the present invention and the appended claims. Therefore, the present invention should not be limited to the content disclosed in the best embodiment, and the scope of the present invention is subject to the scope defined in the claims.

Claims

1. A method for allocating downlink resources in a NOMA system, comprising the steps of: 1) For a NOMA downlink system consisting of 1 base station and S users, according to the channel state information {h1,h2,...,h S } Modeling is used to calculate the power allocation model (p1, p2, .., p S ), where S = M*J, each channel corresponds to a subcarrier, J represents the total number of subcarriers; M represents the number of users on the jth subcarrier, h S is the channel state information of the Sth user, p S is the power allocation model for the Sth user, j = 1~J; 2) To meet the maximum power of the base station and the maximum energy efficiency to ensure user quality as the allocation goal, establish the objective function P C is the path loss of downlink signal transmission, P max is the maximum transmission power; p j,m represents the power allocated to the mth user on the jth subcarrier, P max is the maximum transmission power, R j,m represents the downlink transmission rate of the mth user on the jth subcarrier, R min The minimum transmission rate for users; 3) Based on the objective function, the hybrid whale particle swarm optimization algorithm HWPSO is used to search for the optimal allocation of user power and determine the power allocated to each user.

2. The method according to claim 1, characterized in that Based on the optimization objective function, the method of optimizing the allocation of user power using the hybrid whale particle swarm optimization algorithm HWPSO is as follows: Initialize the population size N, the number of users S, and the maximum number of loop iterations max_iter in the hybrid whale particle swarm optimization algorithm HWPSO, where each particle corresponds to the number of users M on the subcarrier; calculate the position vector X of each particle in each population size as the power allocation of each user on the corresponding subcarrier; During the main iterative optimization process of the hybrid whale particle swarm optimization algorithm HWPSO, the new velocity V of particle i is calculated. i and the new position X i , the new position X of each particle i i Substitute the objective function to calculate the corresponding fitness value, and record the current minimum fitness value of the population and the corresponding global optimal solution; if the fitness value of the current particle is less than the current minimum fitness value of the population, update the minimum fitness value and use the current position X of particle i i Update the current global optimal solution gBest; In the secondary iterative optimization process of the hybrid whale particle swarm optimization algorithm HWPSO, the global optimal solution of the current particle swarm is used as the leader position of the whale group. Whenever a candidate solution better than the current global optimal solution is found, the global optimal value and the leader position are updated; otherwise, the current optimal power allocation strategy remains unchanged; When the number of loop iterations of the main iteration reaches the maximum number of loop iterations max_iter, the optimal energy efficiency is output.

3. The method according to claim 2, characterized in that use Calculate the position vector X(i) of each particle i in each population size; where mod() represents the modulo operation.

4. The method according to claim 2 or 3, characterized in that: Update the new velocity V of particle i according to the following formula i and the new position X i : in, represents the velocity component of user s in the i-th particle at the k+1th iteration; w is the inertia weight, rand1 and rand2 are random variables uniformly distributed in [0,1]; c1 and c2 are individual learning factors and social learning factors, respectively, and Pbest is is the optimal solution for user s in the i-th particle; gbest s is the optimal solution found by user s in the entire particle swarm; X k is is the position of user s of the ith particle at the kth iteration; g is a nonlinear function decreasing from 1 to 0; max_iter is the maximum number of main iterations; iter is the current number of main iterations; It is represented as the position of user s in the i-th particle at the k+1th iteration.

5. The method according to claim 2 or 3, characterized in that: The secondary iteration optimization method is: set the maximum number of secondary iterations SM2 to a decreasing function of the number of primary iterations max_iter indicates the maximum number of loop iterations set; iter indicates the current number of loop iterations; a "forced whale" mechanism is introduced, and its mathematical expression is: Among them, a is used to control The size of it represents the current number of secondary iterations; To control the whale algorithm's choice of search strategy, When forcibly executing the shrinkage and encirclement strategies of the whale algorithm, the algorithm converges faster; is a random number in the range [0,1]. is a coefficient vector used to calculate the distance between the current position and the optimal position; when The shrinking and encircling strategy of the whale algorithm is as follows: in, and The position of user s in the i-th particle at the k-th and k+1-th iterations, It is represented as the optimal solution of user s in the i-th particle in the k-th iteration, It is represented as a random perturbation solution; l is a random variable uniformly distributed in [-1,1]; p is a random variable of the probability of strategy selection; when When the exploration strategy of the whale algorithm is 6. A server, characterized in that: The invention comprises a memory and a processor, wherein the memory stores a computer program, the computer program is configured to be executed by the processor, and the computer program comprises instructions for executing the method according to any one of claims 1 to 5.

Images