A rate optimization method for uplink rate division multiple access system

Abstract

The present invention belongs to the field of wireless communication technology and discloses a rate optimization method for an uplink rate division multiple access system. The present invention includes grouping users according to their transmission rate requirements using a hybrid hierarchical clustering algorithm; modeling the uplink rate division multiple access system in conjunction with drone deployment, decomposing the rate optimization problem of the uplink rate division multiple access system into drone position parameter optimization subproblems, user group parameter optimization subproblems, and RSMA parameter optimization subproblems; solving each subproblem based on augmented Lagrange multipliers and an artificial fish swarm algorithm; and solving the rate optimization problem of the uplink rate division multiple access system using a ternary crossover iterative algorithm to complete system rate optimization. The present invention optimizes the communication system parameters to improve the system throughput, achieve a larger capacity area, and better facilitate fairness among users.

Description

Technical Field

[0001] The present invention belongs to the technical field of wireless communications, and more particularly, relates to a rate optimization method for an uplink rate division multiple access system. Background Art

[0002] Compared to orthogonal access, non-orthogonal access systems can provide greater system throughput and user access. Rate Splitting Multiple Access (RSMA) can be classified as a non-orthogonal multiple access (NOMA) protocol. Unlike Spatial Division Multiple Access (SDMA), which completely treats interference as noise, and NOMA, which completely decodes interference, RSMA can partially decode interference and treat it as noise, allowing users in the same group to share the same time-frequency resources.

[0003] A key issue in RSMA systems is user grouping. Existing user grouping methods are primarily based on the channel gain or distance between users and the base station. However, when the base station location is uncertain, this strategy becomes difficult to implement, necessitating the design of a new grouping strategy. In uplink RSMA systems, achieving a larger capacity region to enhance fairness among users is a technical challenge that needs to be addressed. Summary of the Invention

[0004] The purpose of the present invention is to provide a rate optimization method for an uplink rate division multiple access system, which improves the system throughput and achieves a larger capacity area by optimizing the design of communication system parameters, which is more conducive to achieving fairness among users.

[0005] The present invention provides a rate optimization method for an uplink rate division multiple access system, comprising the following steps:

[0006] Step 1: Group users according to their transmission rate requirements using a hybrid hierarchical clustering algorithm.

[0007] Step 2: Model the uplink rate division multiple access system with UAV deployment, and decompose the rate optimization problem of the uplink rate division multiple access system into the UAV location parameter optimization sub-problem, the user group parameter optimization sub-problem, and the RSMA parameter optimization sub-problem;

[0008] Step 3: Solve each subproblem based on the augmented Lagrange multiplier and artificial fish school algorithm; use the ternary crossover iterative algorithm to solve the rate optimization problem of the uplink rate division multiple access system to complete the system rate optimization.

[0009] Preferably, the step 1 includes the following sub-steps:

[0010] Step 1.1: Sort multiple users according to their transmission rate requirements from low to high. The sorted user set U is as follows:

[0011] U={u1,u2,......,u N-1 ,u N}

[0012] Where N represents the number of users;

[0013] Evenly distribute the sorted N users into K sub-user sets, where the value of K is as follows:

[0014] K=2 k

[0015]

[0016] Among them, the operator Indicates rounding x upwards, and M indicates the upper limit of the capacity of each set;

[0017] User SU assigned to the i-th sub-user set i as follows:

[0018] SU i ={u i ,u i+K ,u i+2K ,...,u i+μK}

[0019] The values ​​of μ are as follows:

[0020]

[0021] Among them, the operator Indicates rounding down x;

[0022] Step 1.2: For each sub-user set, calculate the average inter-group similarity based on the similarity between users, and obtain the user group set corresponding to the sub-user set with the goal of the average inter-group similarity being less than a first threshold;

[0023] Define the similarity s between user i and user j i，j for:

[0024]

[0025] Among them, v e (i) v e (j) are the transmission rate requirements of user i and user j respectively, v e (ii) v e(jj) are the transmission rate requirements of a user in the sub-user set, max ii，jj |v e (ii)-v e (jj)| is the maximum absolute value of the difference in transmission rate requirements of all users;

[0026] Step 1.3: With the goal of ensuring that the average similarity between groups is less than a second threshold, update the groups by merging multiple user group sets to obtain a final user grouping result;

[0027] Define user group g p , User Group g q The similarity between p，q for:

[0028]

[0029] Among them, N p 、N q User group g p , User Group g q The number of users, p i ,q j Represents user group g p The i-th user and user group g q The jth user.

[0030] Preferably, in step 1.2, the user group set corresponding to a single sub-user set is obtained in the following manner:

[0031] Step 1.2.1: Initialize the i-th user group set GS i , the i-th sub-user set SU i Each element is added to the GS as a separate user group i ;

[0032] Step 1.2.2: Calculate GS i The similarity between user groups in the group is calculated, and the two user groups for which similarity calculation is performed are called intra-group user group pairs. The intra-group user group pairs are sorted from high to low according to the similarity, and the sorting results are put into the intra-group user group pair set SoG j Where j is the number of iterations at this time;

[0033] Step 1.2.3: Traverse the SoG j If the number of elements of the two user groups in the user group pair is less than 3, and the two user groups are not the two user groups obtained by splitting in step 1.2.5, then merge the two user groups into a new user group G. new, withdraw from SoG j If user group merging occurs, it is called SoG j Updated;

[0034] Step 1.2.4: If step 1.2.3 is correct for SoG j If the update is made, continue with step 1.2.5; otherwise, exit the iteration and terminate the algorithm;

[0035] Step 1.2.5: If G new If the number of elements is greater than 3, select G new The two elements with the smallest similarity in G are used as two group anchor nodes, and then an exhaustive strategy is used to select new Split into two groups so that the similarity between the groups is minimized;

[0036] Step 1.2.6: According to G new , update GS i ;

[0037] Step 1.2.7: Calculate GS i Average similarity between groups. If the average similarity between groups is less than the set threshold, the iteration is exited and the algorithm is terminated;

[0038] Step 1.2.8: If the number of iterations j at this time reaches the set maximum number of iterations M, then exit the iteration; otherwise, increase the number of iterations by one and return to step 1.2.2.

[0039] Preferably, in step 1.3, the merged update of multiple user group sets is obtained in the following manner:

[0040] Step 1.3.1: Initialize the group merge set GS i，j is an empty set;

[0041] Step 1.3.2: Calculate the i-th user group set GS i The user group in and the j-th user group set GS j The inter-group similarity between user groups in the component is calculated. The two user groups for which component similarity is calculated are called inter-group user group pairs. The inter-group user group pairs are sorted from high to low according to the inter-group similarity, and the sorting results are put into the inter-group user group pair set SoG i，j middle;

[0042] Step 1.3.3: Traverse the SoG i，j For a user grouping pair between groups, if the number of elements of the two user groups in the component user grouping pair is less than 3, and the two user groups are not the two user groups obtained by splitting in step 1.3.5, and if the two user groups are merged, GS i，jIf the average similarity between groups decreases, the two user groups are merged into a new user group G new , withdraw from SoG i，j If user group merging occurs, it is called SoG i，i Updated;

[0043] Step 1.3.4: If step 1.3.3 is correct for SoG i，j If an update is made, continue with step 1.3.5, otherwise exit the iteration;

[0044] Step 1.3.5: If G new If the number of elements is greater than 3, continue with step 1.3.6; otherwise, skip to step 1.3.9;

[0045] Step 1.3.6: Select G new The two elements with the smallest similarity in G are used as two group anchor nodes, and then an exhaustive strategy is used to select new Split into two groups GS p1 GS p2 , so that the similarity between groups is minimized;

[0046] Step 1.3.7: From GS i and GS j Delete the user groups merged in step 1.3.3;

[0047] Step 1.3.8: If GS p1 All from GS i or GS j , then GS will be p1 Add to GS i or GS j ; Otherwise, GS p1 Add to GS i，j ; for GS p2 Repeat the same process as above and skip to step 1.3.10.

[0048] Step 1.3.9: G new Add to GS i，j , while from GS i and GS j Delete the user groups involved in the merge in step 1.3.3;

[0049] Step 1.3.10: If GS i，j GS i and GS j If the average similarity between groups is less than the set threshold, the iteration is exited;

[0050] Step 1.3.11: If the number of iterations k reaches the maximum number of iterations M, then exit the iteration; otherwise, increment the number of iterations by one and return to step 1.3.2.

[0051] Step 1.3.12: GS i and GS j The remaining groups are added to GS i，j , and get the final merge result GS i，j .

[0052] Preferably, in step 2, the objective function corresponding to modeling the uplink rate division multiple access system by the joint UAV deployment is:

[0053]

[0054] The constraints are:

[0055] x min ≤pos u .x≤x max

[0056] y min ≤pos u .y≤y max

[0057] z min ≤pos u .z≤z max

[0058]

[0059] B up (i)≥0,1≤i≤nc up

[0060]

[0061] P up_aloc (i, j, k)≥0, 1≤i≤nc up , 1≤j≤nu up (i), 1≤k≤2

[0062] v up (i, j)≥ve up (i, j), 1≤i≤nc up , 1≤j≤nu up (i)

[0063] Among them, pos u =(pos u .x,pos u .y,pos u.z), is the location of the drone base station; x min 、y min 、z min 、x max 、y max 、z max Both are location restrictions for drones; B up ={B up (1), B up (2), ..., B up (nc up )}, bandwidth allocation for each user group; B ua is the total system bandwidth; Π={od(1),od(2),...,od(nc up )}, the decoding order for each user group;

[0064] For each user group, the user with the lowest transmission rate requirement in the group is not rate-split, and the remaining users are split into two virtual users;

[0065] P up_aloc (i, j, k) is the power allocated to the corresponding virtual user; if the user is not split, then P up_aloc (i, j, 2) = 0, P up_aloc (i, j, 1) is the uplink transmission power of the user; P(i, j) is the power limit of the corresponding user;

[0066] nc up is the number of user groups, nu up (i) is the number of users in the i-th user group, v up (i, j) is the uplink transmission rate of the jth user in the i-th user group; ve up (i, j) is the uplink transmission rate requirement of the corresponding user.

[0067] Preferably, in step 2, the sub-problem of optimizing the position parameters of the UAV is described as follows:

[0068]

[0069] The constraints are:

[0070] uav lc

[0071] v rc

[0072] Among them, V target is the inverse of the sum of the uplink transmission rates, uav lc is the position constraint of the UAV, v rc Constraints on the user's uplink transmission rate;

[0073] For the optimization of user group parameters and RSMA parameters, the parameter to be optimized is denoted as X, and the user group parameter optimization sub-problem and the RSMA parameter optimization sub-problem are uniformly described as:

[0074]

[0075] The constraints are:

[0076]

[0077]

[0078] Where G(X) represents the inverse of the objective function, which is the sum of the uplink transmission rate and m c is the number of equality constraint functions, he1(X) is the optimal uplink bandwidth allocation ratio, he2(X) is the optimal user power split ratio; b=1 k*1 , is the proportional constraint; n c is the number of inequality constraint functions, ge i (X) corresponds to the rate transmission requirement of the i-th user;

[0079] When solving a parameter optimization subproblem, the parameters of the remaining subproblems are treated as constants.

[0080] Preferably, in step 3, for the UAV position parameter optimization subproblem, the UAV position parameter optimization subproblem is converted into an optimization problem containing only UAV position deployment constraints by using augmented Lagrange multipliers, and is solved by using a memory-based dynamic fish school algorithm MD-AFSA;

[0081] The memory-based dynamic fish swarm algorithm MD-AFSA adds memory behavior operations to the basic artificial fish swarm algorithm, executes three search paths in parallel to find the next state point, and updates the step size and field of view of the artificial fish after the state transition is completed.

[0082] Preferably, the artificial fish can perform the memory behavior operation under the condition that the artificial fish has found a better position in the previous iteration. When performing the memory behavior operation, the position update strategy of the artificial fish is as follows:

[0083]

[0084] in, represents the position of artificial fish i after the t+1th iteration, represents the position of artificial fish i after the tth iteration, rand[0,1] represents a random number between [0,1], is the step size after the tth iteration, β1+β2=1, β1 and β2 are constant coefficient factors, dirc with dir p Represents the current and historical directions respectively, and are defined as follows:

[0085]

[0086]

[0087] in, represents the optimal solution among all solutions after the tth iteration, represents the position of artificial fish i after the t-1th iteration, represents the optimal solution among all solutions after the t-1th iteration, Represents the current and historical position validity respectively, and is calculated as follows:

[0088]

[0089]

[0090] in, represents the value of the objective function corresponding to artificial fish i after the t-th iteration, Represents the value of the objective function corresponding to the optimal solution among all solutions after the tth iteration, Represents the value of the objective function corresponding to artificial fish i after the t-1th iteration.

[0091] Preferably, the field of view and step length of the artificial fish are updated after each iteration in the following manner:

[0092]

[0093]

[0094] in, is the step size after the t+1th iteration, t m is the maximum number of iterations of the algorithm, is the step size after the tth iteration, is the lower limit of the step length of artificial fish i, is the field of view after the t+1th iteration, is the field of view after the tth iteration, is the lower limit of the visual field of artificial fish i;

[0095] as well as The update strategy is as follows:

[0096] (a) If continuous t mIn the iteration, if the current artificial fish cannot capture the optimal solution after the previous iteration of all artificial fishes within its field of view, and the current artificial fish has not found a better position, then:

[0097]

[0098]

[0099] in, They represent the updated lower limit of the step length of artificial fish i and the updated lower limit of the field of view of artificial fish i respectively;

[0100] (b) If continuous t th In the iteration, the current artificial fish can capture the optimal solution after all artificial fish iterations are completed within its field of view, then:

[0101]

[0102]

[0103] When the above two conditions are not met, as well as Remain unchanged.

[0104] Preferably, for the user group parameter optimization subproblem and the RSMA parameter optimization subproblem, the user group parameter optimization subproblem and the RSMA parameter optimization subproblem are converted into linear equality constrained optimization problems by using augmented Lagrange multipliers, and are solved by using a linear equality constrained dynamic fish school optimization algorithm LEC-MD-AFSA;

[0105] The linear equality constrained dynamic fish swarm optimization algorithm LEC-MD-AFSA is optimized on the basis of the memory-based dynamic fish swarm algorithm MD-AFSA, and the foraging behavior and random behavior of the artificial fish are modified so that the new solution still satisfies the linear equality constraint;

[0106] For foraging behavior, the following corrections are made:

[0107]

[0108] For random behavior, fix as follows:

[0109]

[0110] Among them, X n Indicates the nth iteration variable, P means P = IA T (AA T ) -1 A, where AX=b is a linear constraint and Rand[-1, 1] represents the generation of a random vector.

[0111] One or more technical solutions provided in the present invention have at least the following technical effects or advantages:

[0112] The present invention first designs a user grouping strategy based on the user's transmission rate requirements for scenarios with variable base station positions and uncertain user channel gains, and designs a hybrid hierarchical clustering user grouping algorithm based on the hierarchical clustering algorithm. The algorithm supports local parallel operation, can quickly implement user grouping, and solve the user grouping problem under uncertain channel gain; then the uplink RSMA system is modeled and analyzed, and the original problem is decomposed into multiple sub-problems (including drone position parameter optimization sub-problem, user group parameter optimization sub-problem and RSMA parameter optimization sub-problem) according to the optimization parameter type, and iteratively solves them; finally, a method based on augmented Lagrange multiplier (ALM) and artificial fish swarm algorithm (AFSA) is designed for solving the sub-problems. This paper proposes an optimization solution algorithm that combines an artificial fish swarm algorithm (AFSA) with an artificial fish swarm algorithm. To improve the convergence speed and optimization ability of the artificial fish swarm algorithm, the optimization algorithm adds a dynamic step size and field of view strategy, and designs the memory behavior of the artificial fish. To solve the optimization problem under linear equality constraints, the optimization algorithm constrains the behavior of the artificial fish by adding a correction factor. A ternary crossover iterative algorithm is used to solve the rate optimization problem of the uplink rate division multiple access system, ultimately completing the system rate optimization. By optimizing the design of communication system parameters, the present invention improves the system throughput, achieves a larger capacity area, and is more conducive to achieving fairness among users. BRIEF DESCRIPTION OF THE DRAWINGS

[0113] Figure 1 This is a framework diagram of a rate optimization method for an uplink rate division multiple access system provided by an embodiment of the present invention;

[0114] Figure 2 This is an example diagram of an uplink communication system with two users in one group;

[0115] Figure 3 It is a flow chart of the user grouping algorithm;

[0116] Figure 4 This is an example diagram of an uplink RSMA system;

[0117] Figure 5 It is a flowchart of the dynamic memory artificial fish school optimization algorithm. DETAILED DESCRIPTION

[0118] Compared to ground-based equipment (such as communication vehicles and fixed base stations), drones offer a better communication signal transmission environment, with less interference from the external environment when signals are transmitted between drone base stations and ground-based end users. They are also easier to deploy and can be deployed to specific locations based on actual application needs, providing services to more users and overcoming the difficulty of communication signals reaching a large number of users in complex environments.

[0119] In the uplink RSMA system, the messages of several users are rate-split, thereby creating virtual users for data transmission. When rate splitting is used, the power allocation parameters can be adjusted to achieve a larger capacity area, which is more conducive to achieving fairness among users.

[0120] Based on the above considerations, the present invention proposes a rate optimization method for an uplink RSMA system deployed in conjunction with a UAV base station, which mainly includes the following aspects:

[0121] 1. Using the user's transmission rate requirement as the grouping basis, a user grouping algorithm that supports local parallel operation is designed based on the divisive hierarchical clustering algorithm and the convergent hierarchical clustering algorithm;

[0122] 2. Combined with UAV deployment, RSMA system rate optimization is performed. Based on the parameter types, the original problem is decomposed into UAV position parameter optimization sub-problems, user group parameter optimization sub-problems, and RSMA parameter optimization sub-problems.

[0123] 3. A subproblem-solving algorithm was designed based on the Augmented Lagrange Method (ALM) and the Artificial Fish Swarm Algorithm (AFSA). To improve the convergence speed and optimization capability of AFSA, the optimization algorithm provided by this invention incorporates a dynamic step size and field of view strategy, and designs the memory behavior of the artificial fish. To solve optimization problems under linear equality constraints, the optimization algorithm provided by this invention constrains the behavior of the artificial fish by adding a correction factor.

[0124] 4. A ternary crossover iterative algorithm is used to solve the uplink RSMA system rate optimization problem for joint UAV deployment. The three subproblems are denoted as: P(1), P(2), and P(3). P(1) represents the UAV position parameter optimization subproblem, P(2) represents the user group parameter optimization subproblem, and P(3) represents the RSMA parameter optimization subproblem. When solving a parameter optimization subproblem, the parameters of the remaining subproblems are treated as constants.

[0125] The present invention provides a rate optimization method for an uplink rate division multiple access system, see Figure 1 , including the following steps:

[0126] Step 1: Group users according to their transmission rate requirements using a hybrid hierarchical clustering algorithm.

[0127] Step 2: Model the uplink rate division multiple access system with UAV deployment, and decompose the rate optimization problem of the uplink rate division multiple access system into the UAV location parameter optimization sub-problem, the user group parameter optimization sub-problem, and the RSMA parameter optimization sub-problem;

[0128] Step 3: Solve each subproblem based on the augmented Lagrange multiplier and artificial fish school algorithm; use the ternary crossover iterative algorithm to solve the rate optimization problem of the uplink rate division multiple access system to complete the system rate optimization.

[0129] In order to better understand the above technical solution, the above technical solution will be described in detail below with reference to the accompanying drawings and specific implementation methods.

[0130] like Figure 2 As shown in the figure, an uplink communication system with two users in a group is used. First, consider an uplink non-orthogonal system with unit transmission bandwidth (B = 1 Hz). According to Shannon's formula, we have (not considering the decoding order issue for now):

[0131]

[0132]

[0133] Among them, v ie↑ (i=1, 2) is the expected uplink rate of user i, P i↑ (i=1, 2) is the uplink transmission power of user i, σ 2 is the power spectral density of Gaussian noise, For the transmission power P i↑ In the case of v ie↑ The channel gain required for the transmission rate is When there is a difference, P i↑ The value of achieves the desired transfer rate.

[0134] If it is an uplink orthogonal system (allocating 1 / 2 Hz bandwidth to each user), the achievable transmission bandwidth is:

[0135]

[0136]

[0137] Among them, v i↑ (i=1, 2) is the channel gain The transmission power is Pi↑ The uplink data transmission rate that user i can achieve under the condition of .

[0138] Compared with the uplink orthogonal system, the rate increment of the non-orthogonal uplink system in the case of two users in a group is Δ 2↑ =v 1e↑ +v 2e↑ -v 1↑ -v 2↑ for:

[0139]

[0140] remember And m>1, n>1, then Let the real part of the logarithm be:

[0141]

[0142] It can be seen that when m is fixed, f(m, n) is an increasing function about n (n>1) and has an upper limit When n is fixed, f(m,n) is about m in Increases upward and gradually approaches the curve Analysis shows that when m is fixed and n is increased, the achievable gain is relatively limited, while when n is fixed and m is increased, a more significant gain can be achieved.

[0143] Preferably, when pairing two users in the uplink system, users with smaller n and larger m can be selected for pairing, which is more conducive to improving the total system rate. When decoding signals, users with higher decoding rate requirements can be selected first.

[0144] For a three-user group, similarly, we can get:

[0145]

[0146] in, When m and n are fixed, Δ 3↑ The real part of Δ tends to be constant; when m and r are fixed, 3↑ The real part of Δ tends to λ1n, where λ1 is a constant. When n and r are fixed, Δ 3↑ The real part of the number tends to λ2m 2 , λ2 is a constant. Analysis shows that increasing m is the most effective, followed by increasing n, and finally r. Decoding is performed from high to low according to the rate requirement.

[0147] As a preference, for a scenario where three users are grouped together, in order to optimize the rate of the entire system, a matching mode similar to the two-user grouping mode can be selected, where users with large differences in required rates are selected for matching and grouping.

[0148] According to the above, the user's transmission rate requirement can be selected as the basis for user grouping. First, the similarity s between user i and user j is defined. i，j for:

[0149]

[0150] Among them, v e (i) v e (j) are the transmission rate requirements of user i and user j, respectively, v e (ii) v e (jj) are the transmission rate requirements of a user in the sub-user set, max ii，jj |v e (ii)-v e (jj)| is the maximum absolute value of the difference in transmission rate requirements of all users.

[0151] Define two user groups g p 、g q The similarity between p，q for:

[0152]

[0153] Among them, N p 、N q For user group g p , User Group g q The number of users, p i ,q j Represents user group g p The i-th user and user group g q The jth user.

[0154] Depends on the user's transmission rate requirements, the present invention designs a hybrid hierarchical clustering algorithm to achieve user grouping. And mainly depends on s i，j and sc p，q Perform relevant operations.

[0155] Specifically, the algorithm of the present invention adopts a local parallel strategy. First, users are sorted (from low to high) according to their rate requirements (uplink), and the sorted user set is denoted as U:

[0156] U={u1,u2,......,u N-1 ,u N}

[0157] Where N is the number of users.

[0158] The sorted users are divided into K sub-user sets, where the value of K is as follows:

[0159] K=2 k

[0160] The value of k is related to the upper capacity limit M (M>0) of each set and the number of users N (N>0), and is determined by the following formula.

[0161]

[0162] Operators Represents rounding x upwards.

[0163] Evenly distribute N users to K sub-user sets, and the user SU assigned to the i-th sub-user set is i (i=1, 2, ..., K) is:

[0164] SU i ={u i ,u i+K ,u i+2K ,...,u i+μK}

[0165] Collection SU i The number of elements is μ+1, and the value of μ is as follows:

[0166]

[0167] in, Represents rounding x down.

[0168] Figure 3 The user grouping algorithm flow of the present invention is shown as follows. The grouping algorithm within a single sub-user set (one-stage algorithm) is as follows:

[0169]

[0170]

[0171] The one-stage algorithm for user grouping based on hybrid hierarchical clustering, that is, the user group set corresponding to a single sub-user set is obtained in the following way:

[0172] Step 1: Initialize the i-th user group set GS i , the i-th sub-user set SU i Each element is added to the GS as a separate user group i ;

[0173] Step 2: Calculate GS i The similarity between user groups in the group is calculated, and the two user groups for which similarity calculation is performed are called intra-group user group pairs. The intra-group user group pairs are sorted from high to low according to the similarity, and the sorting results are put into the intra-group user group pair set SoGj Where j is the number of iterations at this time;

[0174] Step 3: Traverse the SoG j If the number of elements of the two user groups in the user group pair is less than 3, and the two user groups are not the two user groups obtained by splitting in step 5, then merge the two user groups into a new user group G. new , withdraw from SoG j If user group merging occurs, it is called SoG j Updated;

[0175] Step 4: If Step 3 is correct for SoG j If the update is made, continue to step 5; otherwise, exit the iteration and terminate the algorithm;

[0176] Step 5: If G new If the number of elements is greater than 3, select G new The two elements with the smallest similarity in G are used as two group anchor nodes, and then an exhaustive strategy is used to select new Split into two groups so that the similarity between the groups is minimized;

[0177] Step 6: According to G new , update GS i ;

[0178] Step 7: Calculate GS i The average similarity between groups, if the average similarity between groups is less than the set threshold θ s , then exit the iteration and terminate the algorithm;

[0179] Step 8: If the number of iterations j at this time reaches the set maximum number of iterations M, then exit the iteration; otherwise, increase the number of iterations by one and return to step 2.

[0180] The two-stage user grouping algorithm based on hybrid hierarchical clustering mainly realizes the merging of each grouping set and completes the grouping update. After k rounds of merging, the number of grouping sets can be reduced to 1, and the final user grouping result is obtained. Assume that after multiple fusions, there are currently K c user sets are grouped, then in this round of merging, the i-th (i=1, 2, ..., 0.5K c ) grouping set and the i+0.5Kth grouping set c The grouping sets are merged and participate in the next round of merging as the new i-th grouping set.

[0181] The merging algorithm for different user group sets is as follows:

[0182]

[0183]

[0184] The two-stage algorithm for user grouping based on hybrid hierarchical clustering, that is, the merge update of multiple user group sets is obtained in the following way:

[0185] Step 1: Initialize the group merge set GS i，j is an empty set;

[0186] Step 2: Calculate the i-th user group set GS i The user group in and the j-th user group set GS j The inter-group similarity between user groups in the component is calculated. The two user groups for which component similarity is calculated are called inter-group user group pairs. The inter-group user group pairs are sorted from high to low according to the inter-group similarity, and the sorting results are put into the inter-group user group pair set SoG i，j middle;

[0187] Step 3: Traverse the SoG i，j For a user grouping pair between groups, if the number of elements of the two user groups in the component user grouping pair is less than 3, and the two user groups are not the two user groups obtained by splitting in step 5, and if the two user groups are merged, GS i，j If the average similarity between groups decreases, the two user groups are merged into a new user group G new , withdraw from SoG i，j If user group merging occurs, it is called SoG i，j Updated;

[0188] Step 4: If Step 3 is correct for SoG i，j If an update is made, continue to step 5, otherwise exit the iteration;

[0189] Step 5: If G new If the number of elements is greater than 3, continue to step 6; otherwise, skip to step 9;

[0190] Step 6: Select G new The two elements with the smallest similarity in G are used as two group anchor nodes, and then an exhaustive strategy is used to select new Split into two groups GS p1 GS p2 , so that the similarity between groups is minimized;

[0191] Step 7: From GS i and GS j Delete the user groups merged in step 3;

[0192] Step 8: If GSp1 All from GS i or GS j , then GS will be p1 Add to GS i or GS j ; Otherwise, GS p1 Add to GS i，j ; for GS p2 Perform the same process as above; jump to step 10;

[0193] Step 9: G new Add to GS i，j , while from GS i and GS j Delete the user groups involved in the merge in step 3;

[0194] Step 10: If GS i，j GS i and GS j The average similarity between groups is less than the set threshold θ s2 , then exit the iteration;

[0195] Step 11: If the number of iterations k reaches the set maximum number of iterations M, then exit the iteration; otherwise, increase the number of iterations by one and return to step 2;

[0196] Step 12: GS i and GS j The remaining groups are added to GS i，j , and get the final merge result GS i，j .

[0197] Figure 4 The figure shows an example of an uplink RSMA system. Users marked with the same shape in the figure belong to the same user group. The user grouping algorithm is implemented using the aforementioned strategy.

[0198] Consider having nc up For a system with multiple uplink user groups, the total transmission rate of the uplink RSMA system is v up for:

[0199]

[0200] Among them, v up (i) is the sum of the uplink rates of the i-th user group, which is calculated as follows:

[0201]

[0202] Where nu up (i) is the number of users in group i, v up(i, j) is the uplink transmission rate of the jth user in the i-th group. Based on the aforementioned user grouping algorithm, the number of users in each group ranges from {1, 2, 3}. In practical scenarios, users with lower rate requirements typically have lower transmit power. Further splitting these users may cause the received signal to fall below the detection limit.

[0203] Preferably, rate splitting is not performed on the user with the lowest rate transmission requirement in each group, and the remaining users are split into two virtual users.

[0204] Therefore, we can get v up The calculation expression of (i, j) is:

[0205]

[0206] Among them, B up (i) The bandwidth allocated to group i; p up (i, j, k) (k = 1, 2) is the uplink transmission power allocated to each virtual user after user j in group i is split into two virtual users. If the user is not split, then p up (i, j, 2) = 0, p up (i, j, 1) is the uplink transmission power of the user; up(i, j) is the channel gain from the jth user in the i-th group to the base station; σ 2 is the Gaussian noise power spectral density; S(i, j, k) is the inter-signal interference, which is calculated as follows:

[0207]

[0208] Wherein, π(i, j, k) represents the set of users (virtual users) decoded after virtual user k split from user j in group i.

[0209] Considering that the base station is equipped with a single receiving antenna and the end user is equipped with a single transmitting antenna, the user groups are orthogonal in the frequency domain, and users in the same user group share time-frequency resources. Since there is good line-of-sight transmission condition between the drone base station and the end user, it can be assumed that the transmission path between the drone base station and the user is mainly a line-of-sight (LoS) link, then h up (i, j) is:

[0210]

[0211] Where β0 is the channel gain at a distance of 1m from the transmitting antenna, d u，i，j represents the distance between the jth user in the i-th group and the drone base station, which is calculated as follows:

[0212]

[0213] Among them, (x u ,y u , z u ) is the three-dimensional space coordinate of the drone base station, (x i，j ,y i，j ) is the two-dimensional plane coordinate of the j-th user in the i-th group (assuming that all users are distributed on the ground and the vertical coordinate is 0).

[0214] Then the rate optimization problem of the uplink RSMA system (i.e., the objective function corresponding to modeling the uplink rate division multiple access system with joint UAV deployment) can be described as:

[0215]

[0216] St

[0217] x min ≤pos u .x≤x max

[0218] y min ≤pos u .y≤y max

[0219] z min ≤pos u .z≤z max

[0220]

[0221] B up (i)≥0,1≤i≤nc up

[0222]

[0223] P up_aloc (i, j, k)≥0, 1≤i≤nc up , 1≤j≤nu up (i), 1≤k≤2

[0224] v up (i, j)≥ve up (i, j), 1≤i≤nc up , 1≤j≤nu up (i)

[0225] Among them, pos u =(pos u .x,pos u .y,pos u .z) is the location of the drone base station, xmin 、y min 、z min 、x max 、y max 、z max B is the location limit of the drone; up ={B up (1), B up (2), ..., B up (nc up )}Bandwidth allocation for each user group, B ua is the total system bandwidth; Π={od(1),od(2),...,od(nc up )} is the decoding order for each user group; P up_aloc (i, j, k) is the power allocated to the corresponding virtual user. If the user does not split, then P up_aloc (i, j, 2) = 0, P up_aloc (i, j, 1) is the uplink transmission power of the user; P(i, j) is the power limit of the corresponding user; nc up is the number of user groups, nu up (i) is the number of users in the i-th user group, v up (i, j) is the uplink transmission rate of the jth user in the i-th user group; ve up (i, j) is the uplink transmission rate requirement of the corresponding user.

[0226] The present invention decomposes the original optimization problem into three sub-problems according to the parameter type and solves them iteratively, namely the UAV position parameter optimization sub-problem, the user group parameter optimization sub-problem, and the RSMA parameter optimization sub-problem.

[0227] The present invention abstracts the sub-problem of optimizing the position parameters of the UAV into the following form, that is, the sub-problem of optimizing the position parameters of the UAV is described as follows:

[0228]

[0229] st

[0230] uav lc

[0231] v rc

[0232] Among them, V target is the inverse of the sum of the uplink transmission rates, uav lc is the position constraint of the UAV, v rc The uplink transmission rate constraint for the user. When considering the optimization of the UAV's position parameters, the user group parameters and RSMA parameters are treated as constants.

[0233] The present invention converts the above-mentioned UAV position parameter optimization sub-problem into an optimization problem containing only UAV position deployment constraints through ALM, and then adopts an improved artificial fish swarm optimization algorithm to solve it.

[0234] Figure 5 Figure 2 shows the improved artificial fish swarm algorithm model designed by the present invention (memory-based dynamic fish swarm algorithm, MD-AFSA). This invention further incorporates a memory behavior operation based on the basic fish swarm algorithm, and can execute three search paths in parallel to find the next state point. After the artificial fish state transition is completed, the step size and field of view are updated.

[0235] Preferably, the artificial fish can perform the memory behavior operation under the condition that the artificial fish has found a better position in the previous iteration. When performing the memory behavior operation, the position update strategy of the artificial fish is as follows:

[0236]

[0237] in, represents the position of artificial fish i after the t+1th iteration, represents the position of artificial fish i after the tth iteration, rand[0,1] takes a random number between [0,1], is the step size after the tth iteration, β1+β2=1, β1 and β2 are constant coefficient factors, dir c with dir p Represents the current and historical directions respectively, and are defined as follows:

[0238]

[0239]

[0240] in, represents the optimal solution among all solutions after the tth iteration, represents the position of artificial fish i after the t-1th iteration, represents the optimal solution among all solutions after the t-1th iteration, Represents the current and historical position validity respectively, and is calculated as follows:

[0241]

[0242]

[0243] in, represents the value of the objective function corresponding to artificial fish i after the t-th iteration, Represents the value of the objective function corresponding to the optimal solution among all solutions after the tth iteration, Represents the value of the objective function corresponding to artificial fish i after the t-1th iteration.

[0244] The algorithm proposed in this invention designs a separate field of view and step size for each artificial fish. The field of view and step size of the artificial fish are updated after each iteration. The update method is as follows:

[0245]

[0246]

[0247] in, is the step size after the t+1th iteration, t m is the maximum number of iterations of the algorithm, is the step size after the tth iteration, is the field of view after the tth iteration, is the field of view after the t+1th iteration, is the lower limit of the step length of artificial fish i, is the lower limit of the visual field of artificial fish i. It is easy to prove that as well as They are all non-increasing and will eventually reach a lower limit.

[0248] In addition, based on the behavior of artificial fish, the algorithm proposed in this invention is as well as Added update strategy. The update strategy is as follows:

[0249] 1. If continuous t m In the iteration, if the current artificial fish cannot capture the optimal solution after the previous iteration of all artificial fishes within its field of view, and the current artificial fish has not found a better position, then:

[0250]

[0251]

[0252] Among them, t m is the maximum number of iterations, as well as Represents the updated lower limit of the step size of artificial fish i and the updated lower limit of the field of view of artificial fish i.

[0253] 2. If continuous t th In the iteration, the current artificial fish can capture the optimal solution after all artificial fish iterations are completed within its field of view, then:

[0254]

[0255]

[0256] When the above two conditions are not met, as well as Remain unchanged.

[0257] For the UAV position parameter optimization subproblem, the UAV position parameter optimization subproblem is converted into an optimization problem containing only UAV position deployment constraints through augmented Lagrange multipliers, and is solved by using a memory-based dynamic fish school algorithm (MD-AFSA). The memory-based dynamic fish school algorithm (MD-AFSA) adds memory behavior operations to the operations of the basic artificial fish school algorithm, executes three search paths in parallel to find the next state point, and updates the step size and field of view of the artificial fish after the state transition is completed.

[0258] The complete solution algorithm for the UAV position parameter optimization subproblem is as follows:

[0259]

[0260]

[0261] That is, the UAV position optimization algorithm based on ALM and MD-AFSA provided by the present invention includes the following steps:

[0262] Step 1: Initialize parameters, including setting parameters related to the Lagrange multiplier and the MD-AFSA algorithm;

[0263] Step 2: Set the maximum number of iterations and start iteration;

[0264] Step 3: Calculate the augmented Lagrangian function

[0265] Step 4: Use MD-AFSA to Solve and get the solution

[0266] Step 5: Update the Lagrange multiplier and penalty function;

[0267] Step 6: If the iteration end condition is met or the maximum number of iterations is reached, output the final drone position Otherwise continue iterating.

[0268] For the user group parameters and RSMA parameters, the parameters to be optimized are denoted as X. Then the rate optimization problem for the user group parameters and RSMA parameters can be uniformly described as:

[0269]

[0270] st

[0271]

[0272]

[0273] Among them, G(X) represents the inverse of the objective function (uplink rate sum), m c is the number of equality constraint functions, he i (X) corresponds to the aforementioned parameter allocation ratio, which is a linear constraint. he1(X) is the optimal uplink bandwidth allocation ratio, and he2(X) is the optimal user power split ratio; b=1 k*1 , is the proportional constraint; n c is the number of inequality constraint functions, ge i (X) corresponds to the user's transmission rate requirement. In addition, each dimension of X is non-negative.

[0274] The present invention first uses ALM to transform the user group parameter and RSMA parameter optimization sub-problems into sub-problems without inequality constraints, and the remaining equality constraints are linear equality constraints. The present invention designs an artificial fish swarm optimization solution algorithm under linear equality constraints. The algorithm is implemented by adding constraints to each step of the artificial fish operation.

[0275] First, let P = IA T (AA T ) -1 A, for X1, it satisfies AX1=b. Let X2 be:

[0276] X2=X1+α.*P*Xr

[0277] Where α is a non-negative proportional control factor, .* represents the multiplication of a scalar and a vector, and Xr is an arbitrary vector.

[0278] but:

[0279] AX2=A(X1+α.*P*X r )=b+α.*(A*P)*X r

[0280] And A*P=0, so AX2=b, satisfying the linear equality constraint.

[0281] The aforementioned MD-AFSA algorithm is further optimized to produce the Linear Equality Constrained Dynamic Fish School Optimization Algorithm (LEC-MD-AFSA). LEC-MD-AFSA modifies the artificial fish's operational behavior so that the resulting solution still satisfies the linear equality constraints.

[0282] Since X1 and X2 satisfy the constraint AX=b, it is easy to prove AX3=b, where:

[0283] X3=X1+k*(X2-X1)

[0284] Therefore, for the clustering and tailgating behaviors of MD-AFSA, the new solutions still meet the constraints and no correction is required.

[0285] Similarly, for memory behavior, its direction factor is still a linear combination of feasible solutions, so no correction is required.

[0286] For foraging, the artificial fish needs to try to find a new position within its field of view. In order to ensure that the position found satisfies the linear equality constraint, the operation of the artificial fish trying to find a new position is modified as follows:

[0287]

[0288] For random behavior, we also need to ensure that the new solution satisfies the linear equality constraints, so the correction is as follows:

[0289]

[0290] Among them, X n Represents the nth iteration variable, and Rand[-1, 1] represents the generation of a random vector.

[0291] The complete solution algorithm for the user group parameter and RSMA parameter optimization subproblem is as follows:

[0292]

[0293] That is, the user group and RSMA parameter optimization algorithm based on ALM and LEC-MU-AFSA provided by the present invention includes the following steps:

[0294] Step 1: Initialize parameters, including setting parameters related to the Lagrange multiplier and the LEC-MD-AFSA algorithm;

[0295] Step 2: Set the maximum number of iterations and start iteration;

[0296] Step 3: Calculate the augmented Lagrangian function and transform the problem into a linear constraint problem;

[0297] Step 4: Use the LEC-MD-AFSA algorithm to solve the linear constraint problem and obtain the solution

[0298] Step 5: Update the Lagrange multiplier and penalty function;

[0299] Step 6: If the iteration end condition is met or the maximum number of iterations is reached, output the optimized parameters, otherwise continue the iteration.

[0300] Finally, a ternary crossover iterative algorithm is used to solve the uplink RSMA system rate optimization problem for joint UAV deployment. The three subproblems are denoted as: P(1), P(2), and P(3). Among them, P(1) represents the UAV position parameter optimization subproblem, P(2) represents the user group parameter optimization subproblem, and P(3) represents the RSMA parameter optimization subproblem. When solving a parameter optimization subproblem, the parameters of the remaining subproblems are treated as constants.

[0301]

[0302]

[0303] That is, the ternary crossover iterative optimization algorithm provided by the present invention includes the following steps:

[0304] Step 1: Initialize the parameters, including setting the parameters related to each sub-problem solving algorithm and the original optimization problem constraint parameters;

[0305] Step 2: Set the maximum number of iterations and start iteration;

[0306] Step 3: Fix the parameters of the user group parameter optimization sub-problem and the RSMA parameter optimization sub-problem and optimize the UAV position parameters;

[0307] Step 4: Fix the parameters of the UAV position parameter optimization sub-problem and the RSMA parameter optimization sub-problem, and optimize the user group parameters;

[0308] Step 5: Fix the parameters of the drone position parameter optimization sub-problem and the user group parameter optimization sub-problem, and optimize the RSMA parameters;

[0309] Step 6: Get the current optimal parameter ζ i ;

[0310] Step 7: If the iteration end condition is met or the maximum number of iterations is reached, output the optimized parameters, otherwise continue the iteration.

[0311] Finally, it should be noted that the above specific implementation methods are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to examples, those skilled in the art should understand that the technical solutions of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A rate optimization method for an uplink rate division multiple access system, characterized in that: The following steps are involved: Step 1: Group users according to their transmission rate requirements using a hybrid hierarchical clustering algorithm. Step 2: Model the uplink rate division multiple access system with UAV deployment, and decompose the rate optimization problem of the uplink rate division multiple access system into the UAV location parameter optimization sub-problem, the user group parameter optimization sub-problem, and the RSMA parameter optimization sub-problem; In step 2, the sub-problem of optimizing the UAV position parameters is described as: The constraints are: uav lc v rc Among them, V target is the inverse of the sum of the uplink transmission rates, uav lc is the position constraint of the UAV, v rc Constraints on the user's uplink transmission rate; For the optimization of user group parameters and RSMA parameters, the parameter to be optimized is denoted as X, and the user group parameter optimization sub-problem and the RSMA parameter optimization sub-problem are uniformly described as: The constraints are: Where G(X) represents the inverse of the objective function, which is the sum of the uplink transmission rate and m c is the number of equality constraint functions, he1(X) is the optimal uplink bandwidth allocation ratio, he2(X) is the optimal user power split ratio; b=1 k*1 , is the proportional constraint; n c is the number of inequality constraint functions, ge i (X) corresponds to the rate transmission requirement of the i-th user; When solving a parameter optimization sub-problem, the parameters of the remaining sub-problems are treated as constants; Step 3: Solve each subproblem based on the augmented Lagrange multiplier and artificial fish school algorithm; use a ternary crossover iterative algorithm to solve the rate optimization problem of the uplink rate division multiple access system to complete the system rate optimization; In step 3, for the UAV position parameter optimization subproblem, the UAV position parameter optimization subproblem is converted into an optimization problem containing only UAV position deployment constraints by using augmented Lagrange multipliers, and the memory-based dynamic fish school algorithm MD-AFSA is used to solve it; The memory-based dynamic fish swarm algorithm MD-AFSA adds memory behavior operations to the basic artificial fish swarm algorithm, executes three search paths in parallel to find the next state point, and updates the step size and field of view of the artificial fish after the state transition is completed.

2. The rate optimization method for an uplink rate division multiple access system according to claim 1, wherein: The step 1 includes the following sub-steps: Step 1.1: Sort multiple users according to their transmission rate requirements from low to high. The sorted user set U is as follows: U={u1,u2,......,u N-1 ,u N } Where N represents the number of users; Evenly distribute the sorted N users into K sub-user sets, where the value of K is as follows: K＝2 k Among them, the operator Indicates rounding x upwards, and M indicates the upper limit of the capacity of each set; User SU assigned to the i-th sub-user set i as follows: ARE i ={in i ,in i+K ,in i+2K ,...,in i+μK } The values ​​of μ are as follows: Among them, the operator Indicates rounding down x; Step 1.2: For each sub-user set, calculate the average similarity between groups based on the similarity between users, and obtain the user group set corresponding to the sub-user set with the goal of the average similarity between groups being less than a first threshold; Define the similarity s between user i and user j i，j for: Among them, v e (i) v e (j) are the transmission rate requirements of user i and user j respectively, v e (ii) v e (jj) are the transmission rate requirements of a user in the sub-user set, max ii，jj |v e (ii)-v e (jj)| is the maximum absolute value of the difference in transmission rate requirements of all users; Step 1.3: With the goal of ensuring that the average similarity between groups is less than a second threshold, update the groups by merging multiple user group sets to obtain a final user grouping result; Define user group g p , User Group g q The similarity between p，q for: Among them, N p 、N q User group g p , User Group g q The number of users, p i ,q j Represents user group g p The i-th user and user group g q The jth user.

3. The rate optimization method for uplink rate division multiple access system according to claim 2, characterized in that: In step 1.2, the user group set corresponding to a single sub-user set is obtained in the following manner: Step 1.2.1: Initialize the i-th user group set GS i , the i-th sub-user set SU i Each element is added to the GS as a separate user group i ; Step 1.2.2: Calculate GS i The similarity between user groups in the group is calculated, and the two user groups for which similarity calculation is performed are called intra-group user group pairs. The intra-group user group pairs are sorted from high to low according to the similarity, and the sorting results are put into the intra-group user group pair set SoG j Where j is the number of iterations at this time; Step 1.2.3: Traverse the SoG j If the number of elements of the two user groups in the user group pair is less than 3, and the two user groups are not the two user groups obtained by splitting in step 1.2.5, then merge the two user groups into a new user group G. new , withdraw from SoG j If user group merging occurs, it is called SoG j Updated; Step 1.2.4: If step 1.2.3 is correct for SoG j If the update is made, continue with step 1.2.5; otherwise, exit the iteration and terminate the algorithm; Step 1.2.5: If G new If the number of elements is greater than 3, select G new The two elements with the smallest similarity in G are used as two group anchor nodes, and then an exhaustive strategy is used to select new Split into two groups so that the similarity between the groups is minimized; Step 1.2.6: According to G new , update GS i ; Step 1.2.7: Calculate GS i Average similarity between groups. If the average similarity between groups is less than the set threshold, the iteration is exited and the algorithm is terminated; Step 1.2.8: If the number of iterations j at this time reaches the set maximum number of iterations M, then exit the iteration; otherwise, increase the number of iterations by one and return to step 1.2.

2.

4. The rate optimization method for an uplink rate division multiple access system according to claim 2, wherein: In step 1.3, the merged update of multiple user group sets is obtained in the following manner: Step 1.3.1: Initialize the group merge set GS i，j is an empty set; Step 1.3.2: Calculate the i-th user group set GS i The user group in and the j-th user group set GS j The inter-group similarity between user groups in the component is calculated. The two user groups for which component similarity is calculated are called inter-group user group pairs. The inter-group user group pairs are sorted from high to low according to the inter-group similarity, and the sorting results are put into the inter-group user group pair set SoG i，j middle; Step 1.3.3: Traverse the SoG i，j For a user grouping pair between groups, if the number of elements of the two user groups in the component user grouping pair is less than 3, and the two user groups are not the two user groups obtained by splitting in step 1.3.5, and if the two user groups are merged, GS i，j If the average similarity between groups decreases, the two user groups are merged into a new user group G new , withdraw from SoG i，j If user group merging occurs, it is called SoG i，j Updated; Step 1.3.4: If step 1.3.3 is correct for SoG i，j If an update is made, continue with step 1.3.5, otherwise exit the iteration; Step 1.3.5: If G new If the number of elements is greater than 3, continue with step 1.3.6; otherwise, skip to step 1.3.9; Step 1.3.6: Select G new The two elements with the smallest similarity in G are used as two group anchor nodes, and then an exhaustive strategy is used to select new Split into two groups GS p1 GS p2 , so that the similarity between groups is minimized; Step 1.3.7: From GS i and GS j Delete the user groups merged in step 1.3.3; Step 1.3.8: If GS p1 All from GS i or GS j , then GS will be p1 Add to GS i or GS j ; Otherwise, GS p1 Add to GS i，j ; for GS p2 Repeat the same process as above and skip to step 1.3.

10. Step 1.3.9: G new Add to GS i，j , while from GS i and GS j Delete the user groups involved in the merge in step 1.3.3; Step 1.3.10: If GS i，j GS i and GS j If the average similarity between groups is less than the set threshold, the iteration is exited; Step 1.3.11: If the number of iterations k reaches the maximum number of iterations M, then exit the iteration; otherwise, increment the number of iterations by one and return to step 1.3.

2. Step 1.3.12: GS i and GS j The remaining groups are added to GS i,j , and get the final merge result GS i，j .

5. The rate optimization method for uplink rate division multiple access system according to claim 1, characterized in that: In step 2, the objective function corresponding to the modeling of the uplink rate division multiple access system by the joint UAV deployment is: The constraints are: x min ≤pos u .x≤x max and min ≤pos u .y≤y max With min ≤pos u .z≤z max B up (i)≥0,1≤i≤nc up P up_aloc (i,j,k)≥0,1≤i≤nc up ,1≤j≤nu up (i),1≤K≤2 v up (i,j)≥ve up (i,j),1≤i≤nc up ,1≤j≤nu up (i) Among them, pos u =(pos u x, pos u y,pos u z), is the location of the drone base station; x min 、y min 、z min 、x max 、y max 、z max Both are location restrictions for drones; B up ={B up (1), B up (2), ..., B up (nc up )}, bandwidth allocation for each user group; B ua is the total system bandwidth; Π={od(1),od(2),...,od(nc up )}, the decoding order for each user group; For each user group, the user with the lowest transmission rate requirement in the group is not rate-split, and the remaining users are split into two virtual users; P up_aloc (i, j, k) is the power allocated to the corresponding virtual user; if the user is not split, then P up_aloc (i, j, 2) = 0, p up_aloc (i, j, 1) is the uplink transmission power of the user; P(i, j) is the power limit of the corresponding user; nc up is the number of user groups, nu up (i) is the number of users in the i-th user group, v up (i, j) is the uplink transmission rate of the jth user in the i-th user group; ve up (i, j) is the uplink transmission rate requirement of the corresponding user.

6. The rate optimization method for an uplink rate division multiple access system according to claim 1, wherein: The artificial fish can perform the memory behavior operation only if it has found a better position in the previous iteration. When performing the memory behavior operation, the position update strategy of the artificial fish is as follows: in, represents the position of artificial fish i after the t+1th iteration, represents the position of artificial fish i after the tth iteration, rand[0,1] represents a random number between [0,1], is the step size after the tth iteration, β1+β2=1, β1 and β2 are constant coefficient factors, dir c with dir p Represents the current and historical directions respectively, and are defined as follows: in, represents the optimal solution among all solutions after the tth iteration, represents the position of artificial fish i after the t-1th iteration, represents the optimal solution among all solutions after the t-1th iteration, Represents the current and historical position validity respectively, and is calculated as follows: in, represents the value of the objective function corresponding to artificial fish i after the t-th iteration, Represents the value of the objective function corresponding to the optimal solution among all solutions after the tth iteration, Represents the value of the objective function corresponding to artificial fish i after the t-1th iteration.

7. The rate optimization method for an uplink rate division multiple access system according to claim 1, wherein: After each iteration, the field of view and step length of the artificial fish are updated as follows: in, is the step size after the t+1th iteration, t m is the maximum number of iterations of the algorithm, is the step size after the tth iteration, is the lower limit of the step length of artificial fish i, is the field of view after the t+1th iteration, is the field of view after the tth iteration, is the lower limit of the visual field of artificial fish i; as well as The update strategy is as follows: (a) If continuous t m In the iteration, if the current artificial fish cannot capture the optimal solution after the previous iteration of all artificial fishes within its field of view, and the current artificial fish has not found a better position, then: in, They represent the updated lower limit of the step length of artificial fish i and the updated lower limit of the field of view of artificial fish i respectively; (b) If continuous t th In the iteration, the current artificial fish can capture the optimal solution after all artificial fish iterations are completed within its field of view, then: When the above two conditions are not met, as well as Remain unchanged.

8. The rate optimization method for an uplink rate division multiple access system according to claim 7, characterized in that: For the user group parameter optimization subproblem and the RSMA parameter optimization subproblem, the user group parameter optimization subproblem and the RSMA parameter optimization subproblem are converted into linear equality constrained optimization problems by using augmented Lagrange multipliers, and the linear equality constrained dynamic fish school optimization algorithm LEC-MD-AFSA is used to solve them; The linear equality constrained dynamic fish swarm optimization algorithm LEC-MD-AFSA is optimized on the basis of the memory-based dynamic fish swarm algorithm MD-AFSA, and the foraging behavior and random behavior of the artificial fish are modified so that the new solution still satisfies the linear equality constraint; For foraging behavior, the following corrections are made: For random behavior, fix as follows: Among them, X n Indicates the nth iteration variable, P means P = IA T (AA T ) -1 A, where AX=b is a linear constraint and Rand[-1, 1] represents the generation of a random vector.

Images