A joint optimization method for user grouping and power allocation in uplink NOMA systems

Abstract

The present invention discloses a joint optimization method for user grouping and power allocation in an uplink NOMA system, and belongs to the field of wireless communication multiple access. The present invention jointly optimizes user grouping and power allocation. First, a system model is constructed, and the problem is formulated on this basis. Secondly, according to the difference in channel gain between users, users are grouped by a grouping algorithm of channel equalization matching. Finally, an adaptive particle swarm optimization algorithm based on chaotic mapping is introduced to obtain a power allocation matrix, and the power of system users is allocated to complete the performance optimization of the uplink NOMA system. The present invention further improves the throughput performance and throughput of the system, and compared with the particle swarm algorithm under the same conditions, the present invention has a faster convergence speed and optimization capability.

Description

Technical Field

[0001] The present invention belongs to the field of wireless communication multiple access, and in particular relates to a joint optimization method for uplink NOMA system user grouping and power allocation. Background Art

[0002] Due to the increasing scarcity of spectrum resources, orthogonal multiple access technology has many limitations. Non-Orthogonal Multiple Access (NOMA) technology not only supports large-scale user access but also significantly improves spectrum efficiency. Currently, there are few research methods for uplink optimization of NOMA systems. These methods either only consider user power allocation and ignore the performance improvement brought by user grouping, or use simple user grouping schemes and ignore performance optimization in scenarios where user gain differences are not significant, or consider both but have limited ability to find the optimal solution. Therefore, there is a lack of a joint optimization method that comprehensively considers user grouping and power allocation and has low complexity. Summary of the Invention

[0003] To address these issues, the present invention proposes a joint optimization method for user grouping and power allocation in an uplink NOMA system. This method jointly optimizes user grouping and power allocation, first grouping users based on differences in channel gain between them. Then, an adaptive particle swarm optimization algorithm based on chaotic mapping is introduced to allocate power among users. The adaptive algorithm dynamically adjusts based on the distance between particles and the optimal point, enhancing the particle swarm algorithm's optimization capabilities. The randomness and ergodic nature of the chaotic mapping strengthen the particle swarm algorithm's global search capabilities, avoiding the dilemma of being stuck in a local optimum.

[0004] The present invention comprises the following steps:

[0005] S1, build system model

[0006] Consider a single-cell uplink NOMA system. Assume that the system has N subchannels and N×K users, represented by the sets N1={1,2,…,n,…,N} and M1={1,2,…,m,…,M}, respectively, where K is the maximum number of users contained in each subchannel. When different users send messages to the base station through NOMA technology, there is interference between users in the same subchannel. In order to avoid severe interference caused by channel multiplexing, which leads to a decrease in user communication quality, it is necessary to limit the maximum number of users K contained in each subchannel. According to the successive interference cancellation (SIC) technical solution in the uplink transmission scenario, the base station first decodes the user signal with higher channel gain, regards the signals of other users as interference, and then removes the decoded signal from the interference items of other users. Assume that the order of subchannel n gain is defined as H n ={h n,1 ≥h n,2 ≥…≥h n,K}, then the signal-to-noise ratio of user k on subchannel n is:

[0007]

[0008] Among them, p n is the transmit power allocated to subchannel n, α n,k is the user power allocation coefficient, h n,k represents the channel gain of user k on base station subchannel n, σ 2 Represents the power of additive white Gaussian noise (AWGN).

[0009] According to Shannon's formula, the uplink data rate of user k in the nth subchannel is expressed as:

[0010] R n,k =Blog2(1+SINR n,k ) (2)

[0011] Where B represents the fixed bandwidth of each subchannel.

[0012] Derived from formula (2), the sum of all users and data rates of the nth subchannel is:

[0013]

[0014] S2, Problem Formulation

[0015] The objective function of the uplink NOMA system joint user grouping and power allocation optimization to maximize the system sum rate problem can be described as:

[0016]

[0017] Among them, constraint C1 is the total allocated power limit on subchannel n, constraint C2 is the constraint of the minimum value of user power coefficient, and constraint C3 is to ensure that the total allocated power of all users is less than the maximum value of the total power, where P tot Indicates the maximum total transmit power of all users.

[0018] S3, user grouping scheme

[0019] Considering the benefits of edge users and the total system throughput, a PF proportional fairness expression is constructed in the uplink NOMA system:

[0020]

[0021] Among them, Γ m represents the cumulative average rate of user m, t c is the average time interval.

[0022] Assume Γ m The initial value is 0, and the gain factor of user m in subchannel n is defined as G m =R n,m / Γm. Set the benchmark value of user channel gain difference according to the actual NOMA system situation judge Gain factor difference between users The size relationship of Then the difference in channel gain between the two users is small, and the mutual matching benefit is low.

[0023] Taking the number of sub-channel users K = 2 as an example, the grouping algorithm is as follows:

[0024] (1) First, the gain factor G of all users m Arrange in descending order, select the two users m1 and m2 with the largest gain factor difference and calculate the gain factor difference. This means that there is a large difference in channel gain between user m1 and user m2. Users m1 and m2 are grouped into a subchannel n and removed from the user set M1. Update Γ m , execute (3); if This means that the channel gain difference between users m1 and m2 is small. The larger one is selected and placed in subchannel n, and it is removed from the user set M1. Update Γ m , execute (2).

[0025] (2) Select another user pair for subchannel n. Randomly select a user from the user set M1. Calculate the sum and data rates of all users in subchannel n according to formula (3). Select the user with the largest sum and data rate and place it in subchannel n. Remove the user from the user set M1 and update Γ. m .

[0026] (3) Let n = n + 1. If n > N, the algorithm ends and outputs the user grouping solution; otherwise, jump to execution (1).

[0027] S4, System user power allocation scheme for joint user grouping

[0028] The basic idea of ​​the particle swarm algorithm is to use each particle to learn its own trajectory during the iteration process, and also learn the optimal value of the trajectory of other particles to achieve the optimal value-oriented algorithm. The particles self-recognize and self-optimize during the movement process, and this process can be formulated as:

[0029] V i =wV i +c1r1(pbest-P i )+c2r2(gbest-P i ) (6)

[0030] P i =P i +V i (7)

[0031] Formula (6) can be called the step length of particle movement, P i is the position information of the i-th particle, also known as the power allocation matrix. w is called the inertia factor, which represents the impact of the previous movement step on this one. c1 and c2 are acceleration factors, which respectively represent the degree to which the particle is affected by its own historical trajectory and the historical trajectory of the population during its movement. Parameters r1 and r2 are random numbers between (0, 1). Equation (4) is defined as the fitness function. The fitness value of each particle is used to determine the quality of the particle. The position of a particle with the largest fitness value in the current iteration is recorded as the individual optimal value pbest, and the position of the population with the largest fitness value in the current iteration is recorded as the global optimal value gbest.

[0032] Formula (7) represents the position update during the particle iteration process. Based on the limitation of the total transmission power of the user, it is necessary to determine whether the position of the particle exceeds the total transmission power of the user after each position update. If it exceeds the total transmission power, it needs to be scaled down by the same proportion. The formula is as follows:

[0033]

[0034] The present invention constructs an adaptive algorithm so that the inertia factor not only decreases with the increase in the number of iterations, but also increases with the increase in the distance from the global optimal point. That is, the inertia factor changes dynamically with the increase in the number of iterations. The inertia factor formula is as follows:

[0035]

[0036] Among them, w max and w min Represent the maximum and minimum values ​​of the inertia factor, f avg and f max They represent the average and maximum values ​​of the fitness function of the current population respectively.

[0037] This paper proposes a chaotic particle swarm optimization (CPSO) algorithm based on the logistic equation. This algorithm adds a chaotic perturbation operator to the adaptive weighted particle swarm optimization algorithm, enabling the algorithm to search locally within a small neighborhood close to the optimal solution, more comprehensively covering solutions near the optimal solution and avoiding the loss of the optimal solution. The particle position update with the addition of the chaotic perturbation operator can be expressed as follows:

[0038] P i =P i +V i +μP i (1-P i ) (10)

[0039] Among them, μ is the control variable, which is usually set to 4, at which time the randomness of the chaotic perturbation operator is the best.

[0040] The power allocation algorithm based on chaotic particle swarm optimization is as follows:

[0041] (1) Initialize the particle swarm size L, the maximum number of algorithm iterations T, the acceleration factor and other related parameters.

[0042] (2) Initialize the position of the particle swarm, i.e. the power allocation matrix of the system. At the same time, initialize the initial motion step of each particle.

[0043] (3) Calculate the fitness value of each particle according to the fitness function, and update the individual optimal value pbest and the population optimal value gbest according to the fitness value.

[0044] (4) Determine whether the current number of iterations t is less than the maximum number of iterations T. If so, continue execution; if not, execute (8).

[0045] (5) Set the inertia factor w according to formula (9).

[0046] (6) Update the motion step and spatial position of each particle according to equations (6) and (10).

[0047] (7) Update the number of iterations t = t + 1 and jump to execution (3).

[0048] (8) The algorithm ends and the power allocation matrix is ​​output.

[0049] In S5, based on the user grouping obtained in S3 and the optimal power allocation matrix obtained in S4, the objective function (4) is substituted into the objective function to obtain the optimal value of the system and rate, thus completing the performance optimization of the uplink NOMA system.

[0050] The beneficial effects of the present invention are:

[0051] 1. Compared with the algorithm that only considers power allocation, the present invention jointly optimizes user grouping and power allocation, further improving the system throughput performance.

[0052] 2. Compared with simple near-far pairing, this method takes into account the pairing of users with similar channel gain differences, further improving the system throughput under the premise of the same algorithm complexity.

[0053] 3. The adaptive particle swarm algorithm based on chaotic mapping optimizes power distribution, uses an adaptive algorithm to dynamically adjust the inertia factor, and based on the ergodicity and randomness of chaotic mapping, enables the algorithm to more comprehensively cover solutions near the optimal solution. Compared with the particle swarm algorithm under the same conditions, the present invention has faster convergence speed and optimization ability. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a specific flow chart of the present invention. DETAILED DESCRIPTION

[0055] In order to facilitate those skilled in the art to understand and implement the present invention, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the embodiments described herein are only used to illustrate and explain the present invention, but are not used to limit the present invention.

[0056] Since the NOMA system transmits signals of multiple users on the same time-frequency resources, this will cause interference between users on the same sub-channel. Reasonable sub-channel user grouping and user power allocation will not only reduce interference between users and improve system throughput, but also reduce the complexity of base station receiver decoding. Therefore, it is very necessary to study the selection and power allocation of users transmitting on the same sub-channel.

[0057] For details of the specific process, see Figure 1 The present invention provides a joint optimization method for uplink NOMA system user grouping and power allocation, the steps of which are as follows:

[0058] S1: Divide the N×K users in the cell into N clusters, each containing K users. Use the sets N1 = {1, 2, …, n, …, N} and M1 = {1, 2, …, m, …, M} to represent the number of subchannel clusters and the total number of users, respectively. K users on the same subchannel share the same time-frequency resource, and different subchannels are orthogonal to each other. Assume that K users

[0059] The channel gain of 222 households transmitting on subchannel n satisfies the condition h n,1 ≥h n,2 ≥…≥h n,K , then the signal-to-noise ratio of the kth user on the subchannel n of the cell is:

[0060]

[0061] Among them, p n is the transmit power allocated to subchannel n, α n,k is the user power allocation coefficient, h n,k represents the channel gain of user k on base station subchannel n, σ 2 Represents the power of additive white Gaussian noise (AWGN).

[0062] According to Shannon's formula, the uplink data rate of user k in the nth subchannel is expressed as:

[0063] R n,k =Blog2(1+SINR n,k ) (2)

[0064] Where B represents the fixed bandwidth of each subchannel.

[0065] Derived from formula (2), the sum of all users and data rates of the nth subchannel is:

[0066]

[0067] S2: The present invention optimizes the total system throughput by user grouping and power allocation. From formula (3), we can see that in order to improve the system sum rate, two aspects must be considered: user grouping and power allocation. Therefore, the objective function of the uplink NOMA system combined with user grouping and power allocation to optimize the maximum system sum rate problem can be described as:

[0068]

[0069] Among them, constraint C1 is the total allocated power limit on subchannel n, constraint C2 is the minimum value constraint of user power coefficient, and constraint C3 is to ensure that the total allocated power of all users is less than the maximum total power. tot Indicates the maximum total transmit power of all users.

[0070] S3: Determine the user allocation relationship for each subchannel. As shown in step 1, the channel gain difference between subchannel users determines the upper limit of NOMA system performance. The greater the channel gain difference, the simpler the design of the SIC receiver at the receiving end. Based on this characteristic and considering the benefits of edge users, this paper proposes a channel equalization matching grouping algorithm.

[0071] Construct a PF fairness criterion expression in the NOMA system:

[0072]

[0073] Among them, Γ m is the cumulative average rate of user m, t c is the average time interval.

[0074] Assume Γ m The initial value is 0, and the gain factor of user m in subchannel n is defined as G m =R n,m / Γ m . Set the benchmark value of user channel gain difference according to the actual NOMA system situation judge Gain factor difference between users The size relationship of Then the difference in channel gain between the two users is small, and the mutual matching benefit is low.

[0075] Taking the number of sub-channel users K=2 as an example, this step can be divided into the following sub-steps:

[0076] (1) First, the gain factor G of all users m Arrange in descending order, select the two users m1 and m2 with the largest gain factor difference and calculate the gain factor difference. This means that there is a large difference in channel gain between user m1 and user m2. Users m1 and m2 are grouped into a subchannel n and removed from the user set M1. Update Γ m , execute (3); if This means that the channel gain difference between users m1 and m2 is small. The larger one is selected and placed in subchannel n, and it is removed from the user set M1. Update Γ m , execute (2).

[0077] (2) Select another user pair for subchannel n. Randomly select a user from the user set M1. Calculate the sum and data rates of all users in subchannel n according to formula (3). Select the user with the largest sum and data rate and place it in subchannel n. Remove the user from the user set M1 and update Γ. m .

[0078] (3) Let n = n + 1. If n > N, the algorithm ends and outputs the user grouping solution; otherwise, jump to execution (1).

[0079] S4: Determine the system user power allocation plan

[0080] In a NOMA system, after the transmitter matches users to subchannel groups, it is necessary to allocate power to the users multiplexed on the subchannels. Appropriate power allocation creates power differences among users within a NOMA group, thereby reducing interference with user signal demodulation at the receiver, improving SIC performance and system throughput. This paper considers a global search optimization based on a chaotic map adaptive particle swarm algorithm, using the user power allocation matrix as the initial position of the population.

[0081] The basic idea of ​​the particle swarm algorithm is to achieve an optimal value-oriented algorithm by using each particle to learn its own trajectory during the iteration process and also learn the optimal value of the trajectory of other particles. The power allocation algorithm of the adaptive particle swarm based on chaotic mapping is detailed as follows:

[0082] First, initialize the particle swarm of size L, the maximum number of iterations T, the acceleration factor and other related parameters. The particle swarm P = {P1, P2, ..., P l ,…,P L}, where P l is the position information of the lth particle, that is, the power allocation matrix. The power allocated by the base station to each user is a particle. According to the objective function (4) as the fitness function, the quality of the particle is judged by calculating the size of the fitness value of each particle. The maximum fitness of a particle in the iteration process so far is defined as the individual optimal value pbest, and the corresponding fitness value is recorded as R l , which represents the process of particle self-cognition. The maximum value of all individual optimal values ​​of all particles in the iteration process so far is defined as the group optimal value gbest, and the corresponding fitness value is recorded as R g , that is, R g =max{R1,R2,…,R L}.

[0083] Secondly, at the beginning of the next iteration, each particle searches for its next position information based on pbest and gbest. This optimization process can be expressed as:

[0084] V i =wV i +c1r1(pbest-P i )+c2r2(gbest-P i ) (6)

[0085] P i =P i +V i (7)

[0086] Formula (6) can be called the step length of particle movement, P iis the position information of the i-th particle, that is, the power allocation matrix. w is called the inertia factor, which represents the impact of the previous movement step on this one. c1 and c2 are acceleration factors, which respectively represent the degree to which the particle is affected by its own historical trajectory and the historical trajectory of the population during the movement. Parameters r1 and r2 are random numbers between two (0, 1). Formula (4) is defined as the fitness function. The fitness value of each particle is used to determine the quality of the particle. The position of a particle with the largest fitness value in the iteration so far is recorded as the individual optimal value pbest, and the position of the population with the largest fitness value in the iteration so far is recorded as the global optimal value gbest.

[0087] Formula (7) represents the position update during the particle iteration process. Based on the limitation of the total transmission power of the user, it is necessary to determine whether the position of the particle exceeds the total transmission power of the user after each position update. If it exceeds the total transmission power, it needs to be scaled down by the same proportion. The formula is as follows:

[0088]

[0089] Formula (6) shows that if the algorithm is to converge quickly, the inertia factor w needs to be increased. However, this may cause the algorithm to "mature" and fall into the local optimum too early. To avoid this shortcoming, the present invention constructs an adaptive algorithm so that the inertia factor not only decreases with the increase in the number of iterations, but also increases with the increase in the distance from the global optimum. In other words, the inertia factor changes dynamically with the increase in the number of iterations. The inertia factor formula is as follows:

[0090]

[0091] Among them, w max and w min Represent the maximum and minimum values ​​of the inertia factor, f avg and f max They represent the average and maximum values ​​of the fitness function of the current population respectively.

[0092] The adaptive weighted particle swarm optimization algorithm has improved performance compared to the particle swarm optimization algorithm, but it may still fall into a local optimum. Based on this situation, the present invention proposes a chaotic particle swarm optimization scheme (CPSO) based on the logistic equation. On the basis of the adaptive weighted particle swarm optimization algorithm, a chaotic perturbation operator is added, so that the algorithm can search locally in a small neighborhood close to the optimal solution, more comprehensively covering solutions near the optimal solution, and avoiding the loss of the optimal solution. The particle position update with the addition of the chaotic perturbation operator can be expressed as follows:

[0093] P i =P i +V i +μP i (1-Pi ) (10)

[0094] Among them, μ is the control variable, which is usually set to 4, at which time the randomness of the chaotic perturbation operator is the best.

[0095] Finally, each particle determines its own position information according to equations (6) and (10) during the iteration, calculates the fitness value and compares the current value with the previous optimal value to update pbest. When all particles determine their positions and update pbest, the group optimal value gbest is updated. This cycle iterates until the iteration ends, and gbest is the optimal solution found.

[0096] S5: Based on the user grouping obtained in S3 and the optimal power allocation matrix obtained in S4, the objective function (4) is substituted into the objective function to obtain the optimal value of the system and rate, thus completing the performance optimization of the uplink NOMA system.

[0097] Starting from S3, the present invention introduces a joint optimization algorithm for user grouping and power allocation. It uses a channel gain algorithm to group users. When the differences in user channel gains are large, a near-far grouping method is selected. For users with similar channel gains, the user with the largest current channel gain is selected to join a designated sub-channel, and then the user with the best performance in the current sub-channel is selected to group with it. This results in a user group that maximizes the overall system and rate. An adaptive chaotic particle swarm power allocation algorithm is used. In this algorithm, the particle position changes are combined with an adaptive algorithm based on the dynamic changes in particle position, so that the inertia factor not only decreases with the number of iterations but also increases with the distance from the optimal solution. In addition, in order to address the possibility of falling into a local optimum, the present invention combines the chaotic sequence of the logistic map with the position changes. The ergodicity and randomness of the chaotic sequence avoid the dilemma of falling into a local optimum.

[0098] Implementation steps: To reduce the complexity of SIC demodulation at the receiver end of the NOMA system, assume that each user can only be assigned one subchannel, and each subchannel is assigned two users. Input initialization parameters include the number of subchannels, number of users, noise power, system bandwidth, cell radius, total user transmit power, particle swarm size, and number of iterations.

[0099] When grouping users, when the number of users is 6, the number of subchannels is 3, and the maximum number of superimposed users in the subchannels is 2, the grouping algorithm of the present invention converges after 11 iterations, and the number of iterations is much lower than that of the exhaustive search algorithm. In addition, the system performance of the grouping algorithm of the present invention is compared with that of the near-far grouping scheme. Assuming that both adopt the fractional power allocation scheme (FTPA) with equal power allocation between subchannels for power allocation, the traditional near-far grouping simply pairs users according to the difference in channel gain. When the difference in user channel gain is not large, it interferes with system performance. In contrast, when the channel gain difference is not large, the present invention performs grouping with the goal of maximizing subchannel performance, which can achieve better system performance. Referring to the changes in system sum rate, under different numbers of users, the system sum rate of the grouping algorithm of the present invention is higher than that of near-far grouping.

[0100] When performing power allocation, NOMA technology is used, superimposing two users on each subchannel. Compared with the Orthogonal Frequency Division Multiple Access (OMA) system, the system throughput is significantly improved. Assuming a cell radius of 1000m, a total user power of 40dBm, and 20 users, the adaptive chaotic particle swarm power allocation algorithm proposed in this paper approaches convergence after about 30 iterations, while the traditional particle swarm power allocation algorithm requires about 40 iterations to approach convergence. The power allocation algorithm proposed in this paper has a faster convergence speed and stronger optimization capability.

[0101] It should be understood that the above description of the preferred embodiment is relatively detailed and cannot be regarded as limiting the scope of protection of the patent of the present invention. Under the guidance of the present invention, ordinary technicians in this field can make substitutions or modifications without departing from the scope of protection of the claims of the present invention, which fall within the scope of protection of the present invention. The scope of protection requested by the present invention shall be based on the attached claims.

Claims

1. A joint optimization method for user grouping and power allocation in uplink NOMA system, characterized by The following steps are involved: S1, building a system model; Consider a single-cell uplink NOMA system. Assume that the system has N subchannels and N×K users, represented by the sets N1={1,2,…,n,…,N} and M1={1,2,…,m,…,M}, where K is the maximum number of users that each subchannel can contain. S2, problem formulation; The objective function of the uplink NOMA system, the joint user grouping and power allocation optimization maximization system, and the rate problem is described as: Among them, constraint C1 is the total allocated power limit on subchannel n, R n is the number of users and data rate of the nth subchannel, α n,k is the user power allocation coefficient, constraint C2 is the minimum value constraint of the user power coefficient, constraint C3 is to ensure that the sum of the power allocated to all users is less than the maximum value of the total power, p n is the transmit power allocated to subchannel n, P tot Indicates the maximum value of the total transmit power of all users; S3, user grouping scheme; Considering the benefits of edge users and the total system throughput, a PF proportional fairness expression is constructed in the uplink NOMA system: Among them, Γ m represents the cumulative average rate of user m, R n,m is the uplink data rate of user m on the nth subchannel, t c is the average time interval; Assume Γ m The initial value is 0, and the gain factor of user m in subchannel n is defined as G m =R n,m Γ m ; Set the baseline value of user channel gain difference based on the actual uplink NOMA system situation judge Gain factor difference between users The size relationship is used to complete the grouping; Taking the number of sub-channel users K = 2 as an example, the grouping algorithm is as follows: (3-1) First, the gain factor G of all users m Arrange in descending order, select the two users m1 and m2 with the largest gain factor difference and calculate the gain factor difference. Indicates that there is a large difference in channel gain between user m1 and user m2. Users m1 and m2 are grouped into a subchannel n and removed from the user set M1. Update Γ m , execute (3-3); if Indicates that the channel gain difference between users m1 and m2 is small. Select the larger one and put it into subchannel n, and remove it from the user set M1. Update Γ m , execute (3-2); (3-2) Select another user pair for subchannel n, randomly select a user from the user set M1, calculate the sum of all users and data rates of subchannel n, select the user with the largest sum and data rate and put it into subchannel n, remove this user from the user set M1, and update Γ m ; (3-3) Let n = n + 1. If n > N, the algorithm ends and outputs the user grouping solution; otherwise, jump to (3-1). S4, the power allocation of the system users in the joint user grouping, the power allocation algorithm based on the chaotic particle swarm is as follows: (4-1) Initialize the particle swarm size L, the maximum number of algorithm iterations T, and the acceleration factor; (4-2) Initialize the position of the particle swarm, that is, the power allocation matrix of the system, and initialize the initial motion step of each particle; (4-3) Calculate the fitness value of each particle according to the fitness function, and update the individual optimal value pbest and the population optimal value gbest according to the fitness value; (4-4) Determine whether the current number of iterations t is less than the maximum number of iterations T. If so, continue execution; if not, execute (4-8); (4-5) Set the inertia factor w; (4-6) Update the motion step size and spatial position of each particle; (4-7) Update the number of iterations t = t + 1 and jump to (4-3); (4-8) The algorithm ends and the power allocation matrix is ​​output; In S5, based on the user grouping obtained in S3 and the optimal power allocation matrix obtained in S4, the objective function (4) is substituted into the objective function to obtain the optimal value of the system and rate, thus completing the performance optimization of the uplink NOMA system.

2. The method for joint optimization of uplink NOMA system user grouping and power allocation according to claim 1, characterized in that: In step 1, assume that the order of subchannel n gain is defined as H n ={h n,1 ≥h n,2 ≥…≥h n,K }, the signal-to-noise ratio of user k on subchannel n is: Among them, p n is the transmit power allocated to subchannel n, α n,k is the user power allocation coefficient, h n,k represents the channel gain of user k on base station subchannel n, σ 2 represents the power of additive white Gaussian noise; According to Shannon's formula, the uplink data rate of user k in the nth subchannel is expressed as: R n,k =Blog2(1+SINR n,k ) (2) Where B represents the fixed bandwidth of each subchannel; Derived from formula (2), the sum of all users and data rates of the nth subchannel is:

3. The method for joint optimization of user grouping and power allocation in an uplink NOMA system according to claim 1, characterized in that: In step 4, the particles self-recognize and self-optimize during the movement, and this process is formulated as: V i =wV i +c1r1(pbest-P i )+c2r2(gbest-P i ) (6) P i =P i +V i (7) Formula (6) is called the step length of particle movement, and Formula (7) represents the position update of the particle during the iteration process. i is the position information of the i-th particle, that is, the power allocation matrix, w is called the inertia factor, which represents the influence of the previous movement step on this time, c1 and c2 are acceleration factors, which respectively represent the degree to which the particle is affected by its own historical trajectory and the historical trajectory of the population during the movement, and parameters r1 and r2 are two random numbers between (0, 1); Formula (4) is defined as the fitness function, and the position of a particle with the largest fitness value in the current iteration process is recorded as the individual optimal value pbest, and the position of the population with the largest fitness value in the current iteration process is recorded as the global optimal value gbest; On the basis of the adaptive weighted particle swarm algorithm, a chaotic perturbation operator is added. The particle position update with the chaotic perturbation operator is expressed as follows: P i =P i +V i +μP i (1-P i ) (10) Among them, μ is the control variable.

4. The method for joint optimization of uplink NOMA system user grouping and power allocation according to claim 3, characterized in that: An adaptive algorithm is constructed to make the inertia factor w change dynamically with the increase of the number of iterations. The inertia factor formula is as follows: Among them, w max and w min Represent the maximum and minimum values ​​of the inertia factor, f avg and f max They represent the average and maximum values ​​of the fitness function of the current population respectively.

5. The method for joint optimization of uplink NOMA system user grouping and power allocation according to claim 3, characterized in that: Formula (7) represents the position update during the particle iteration process. Based on the limitation of the user's total transmission power, each time the particle's position is updated, it is judged whether it exceeds the user's total transmission power. If it exceeds the total transmission power, it is scaled down by the same proportion. The formula is as follows:

6. The method for joint optimization of uplink NOMA system user grouping and power allocation according to claim 3, characterized in that: In formula (10), the control variable μ is set to 4.

Images