Unmanned aerial vehicle base station deployment method based on dynamic multi-population particle swarm optimization

Abstract

The invention discloses an unmanned aerial vehicle base station deployment method based on dynamic multi-population particle swarm optimization, and aims to solve the problems that a traditional optimization algorithm is easy to fall into local optimum and slow in convergence speed. According to the method, a dominant group, a common group and an inferior group are dynamically divided, and a differential learning strategy is adopted for different groups: the dominant group performs local fine search by using dynamic multi-scale Gaussian variation, the common group keeps balanced exploration and development of a standard particle swarm algorithm, and the inferior group enhances global search ability through pure social learning. Meanwhile, system utility normalization processing is introduced, the coverage rate and the throughput are weighted to solve the problem of multi-target dimension difference, and collaborative optimization is ensured. The deployment efficiency and the communication performance of the unmanned aerial vehicle base station are remarkably improved, the method is suitable for emergency communication and other complex scenes, and the coverage rate and the throughput capacity are higher.

Description

Technical Field

[0001] This invention belongs to the field of UAV base station deployment optimization technology, and in particular relates to a UAV base station deployment method based on dynamic multi-swarm particle swarm optimization. Background Technology

[0002] In the field of UAV base station deployment optimization, existing technologies mainly focus on the application of traditional particle swarm optimization (PSO) algorithms and their improved versions. Traditional PSO algorithms simulate the cooperative and competitive behavior of particles in the search space to find the optimal deployment location for UAV base stations, thereby improving the coverage and throughput of communication systems. However, traditional PSO algorithms are prone to getting trapped in local optima when dealing with complex optimization problems and lack adaptability in dynamic environments.

[0003] Chinese patent document CN110108050B proposes a method for deploying nodes in wireless sensor networks based on particle swarm optimization algorithm. The method optimizes node deployment by adjusting the velocity and position of particles, but the convergence speed and accuracy of this method still need to be improved in complex environments.

[0004] In addition, some improved PSO algorithms, such as adaptive PSO, may improve the performance of the algorithm to a certain extent, but when adjusting the inertia weight, the algorithm may search too wide in the early iterations and converge too slowly in the later iterations. Summary of the Invention

[0005] The purpose of this invention is to provide a method for deploying UAV base stations based on dynamic multi-population particle swarm optimization, which addresses the problems of slow convergence speed, easy getting trapped in local optima, and insufficient dynamic adaptability in existing technologies.

[0006] To achieve the objective of this invention, a method for deploying unmanned aerial vehicle (UAV) base stations based on dynamic multi-population particle swarm optimization is provided, comprising the following steps:

[0007] S1. Collect information from the fixed base station side, the UAV base station side and the user side to obtain the spatial location information of the fixed base station, the UAV base station and the user in three-dimensional Cartesian coordinates, as well as the straight-line distance between the user and the fixed base station and between the user and the UAV base station.

[0008] S2. Using the spatial location information and the straight-line distance, construct a probabilistic link model between the user and the drone base station and a path loss model between the user and the fixed base station in hotspot scenarios. By combining the above models, a spatial propagation model is obtained.

[0009] S3. The spatial propagation model is processed by the signal-to-interference-plus-noise ratio model to obtain the interference model of the user during communication with the associated service node;

[0010] S4. By introducing binary decision variables, the decision variables between each service node and the user are set in the interference model to obtain the connection relationship between the service node and the user.

[0011] S5. By initializing the drone base station deployment scheme population, obtain the initial flight speed and initial position of the drone base station, and define the individual learning and group learning intensity, inertia weight and number of iterations;

[0012] S6. Utilize the connection relationship between the service node and the user to define the optimization target of the system utility; when the optimization target is met, the coverage and throughput of the communication system are introduced and normalized, and the normalized coverage and throughput are added together to obtain the system utility processing method.

[0013] S7. The system utility of the initial unmanned aerial vehicle (UAV) base station deployment scheme population is obtained through the system utility processing method;

[0014] S8. Divide the system utility of each deployment scheme into population groups: if the system utility ranks in the top 20%, it is judged as the dominant group; if the system utility ranks in the bottom 20%, it is judged as the disadvantaged group; the remaining deployment schemes are judged as the ordinary group.

[0015] S9. Using the defined number of iterations, the Gaussian density function is dynamically adjusted through a dynamic multi-scale Gaussian mutation mechanism to perform a local fine search for the dominant group, thereby obtaining the learning strategy corresponding to the dominant group.

[0016] S10. Using the defined individual and group learning strengths, inertia weights, and iteration counts, the disadvantaged group is given a learning strategy corresponding to the disadvantaged group by adopting social learning alone.

[0017] S11. Using the defined individual and group learning strengths, inertia weights, and iteration counts, the ordinary group obtains the corresponding learning strategy through standard PSO.

[0018] The significant advancement of this invention compared to existing technologies lies in:

[0019] (1) This invention solves the problem of the difference in the dimensions of coverage and throughput by adopting the normalization of system utility and introducing a proportional factor to adjust the weights of coverage and throughput; specifically, it adopts a weighted summation method and combines the dynamic balancing of multiple objectives by preference factors to ensure that coverage and throughput affect system utility on the same order of magnitude, and avoids a single objective dominating the optimization process.

[0020] (2) This invention achieves a significant improvement in convergence speed and solution accuracy by dynamically dividing the dominant group, the ordinary group and the disadvantaged group, and by specifically adopting mutation learning, social individual hybrid learning and pure social learning strategies.

[0021] (3) Compared with standard PSO, genetic algorithm (GA) and differential evolution (DE), the present invention utilizes dynamic multi-population particle swarm optimization (DMPSO) to achieve better performance in coverage, throughput and system utility. In particular, it can jump out of local optima faster in complex scenarios and achieve more stable global optimization results. Its adaptive learning mechanism and multi-objective collaborative optimization capability give it significant advantages in engineering problems such as UAV base station deployment.

[0022] To more clearly illustrate the functional characteristics and structural parameters of the present invention, further explanation is provided below in conjunction with the accompanying drawings and specific embodiments. Attached Figure Description

[0023] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0024] Figure 1 This is a flowchart of the steps of the present invention;

[0025] Figure 2 This is a schematic diagram of particle grouping according to system utility in the dynamic multi-group update strategy of this invention;

[0026] Figure 3 These are optimized effect diagrams produced by embodiments of the present invention, wherein... Figure 3 (a) shows the user coverage effect when only fixed base stations are deployed. Figure 3 (b) is the user coverage effect when fixed base stations and drones are deployed in a coordinated manner;

[0027] Figure 4 This is a graph showing the effect of the weighting factor ratio of the present invention on coverage and the ratio of system throughput to total performance. Detailed Implementation

[0028] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] This invention provides a method for deploying unmanned aerial vehicle (UAV) base stations based on dynamic multi-population particle swarm optimization, combined with... Figure 1 This includes the following steps:

[0030] S1. Collect information from the fixed base station side, the UAV base station side and the user side to obtain the spatial location information of the fixed base station, the UAV base station and the user in three-dimensional Cartesian coordinates, as well as the straight-line distance between the user and the fixed base station and between the user and the UAV base station.

[0031] Let the set of the number of fixed base stations be F = {1, 2, ..., F...} Max}, F Max ≥1, where F Max Represents the maximum value in set F;

[0032] Define the spatial position of a fixed base station f in a three-dimensional Cartesian coordinate system: s f =(x f y f h f ), Its position is fixed; where x f y f h f They are s f Position on the x, y, and z axes;

[0033] Define the set of unmanned aerial vehicle (UAV) base stations: M = {1, 2, ..., M} Max}, M Max ≥1, where M Max This represents the maximum value in set M;

[0034] Define the spatial position of the UAV base station m in a three-dimensional Cartesian coordinate system:

[0035] s m =(x m y m h m ), Where, x m y m h m They are s m Position on the x, y, and z axes;

[0036] By taking the union of sets F and M, we obtain the set of all service nodes:

[0037] J = {1, 2, ..., J} Max}, J Max ≥1, where J=F∪M;

[0038] The set of UAV base stations, U = {1, 2, ..., U}, is obtained through initialization settings. Max}, U Max ≥1, where U Max Represents the maximum value in set U;

[0039] Assuming the ground user is relatively stationary within the square area, the spatial position of user u in the three-dimensional Cartesian coordinate system is obtained through initialization settings. u =(x u y u h u ), Where, x u y u h u They are s u Position on the x, y, and z axes;

[0040] Determine the straight-line distance d between user u and drone base station m. u，m :

[0041]

[0042] Determine the straight-line distance d between user u and fixed base station f. u，f :

[0043]

[0044] S2. Using the spatial location information and the straight-line distance, construct a probabilistic link model between the user and the drone base station and a path loss model between the user and the fixed base station in hotspot scenarios. By combining the above models, a spatial propagation model is obtained.

[0045] 2-1. The straight-line distance between the user and the drone base station is calculated using the Fries free-space propagation path loss model to obtain the free-space propagation path loss.

[0046]

[0047] Among them, f c The carrier frequency is expressed in Hz; c is the speed of light.

[0048] 2-2. By integrating the free space propagation path loss with the additional path loss, a LoS and NLoS path loss model is obtained between the UAV base station communication node and the ground terminal user.

[0049] LoS path loss model As shown in the following formula:

[0050]

[0051] NLoS path loss model As shown in the following formula:

[0052]

[0053] Where, η LOS Additional path loss for line-of-sight links, in dB; η NLoS Additional path loss for non-line-of-sight links, in dB;

[0054] 2-3. Using the line-of-sight transmission probability calculation formula that correlates the elevation angle parameters between the UAV platform and the ground terminal, the LoS and NLoS path loss models are used to determine the probabilistic link model between the user and the UAV base station.

[0055] The formula for calculating the line-of-sight transmission probability associated with the elevation angle parameter between the UAV platform and the ground terminal is as follows:

[0056]

[0057] P(NLoS, θ) u，m )=1-P(LoS,θ u，m );

[0058] Probabilistic link model L between users and drone base stations u,m As shown in the following formula:

[0059]

[0060] Where P(LoS,θ) u，m ) represents the probability of line-of-sight connection; P(NLoS, θ) u，m ) represents the probability of line-of-sight connectivity; a represents the ratio between the construction land area and the total area; b represents the average number of buildings per unit area; θ represents the probability of line-of-sight connectivity. u，m Let m be the elevation angle between the drone base station and user u.

[0061] 2-4. By defining the additional loss caused by the NLoS link and the path loss index when the user communicates with the fixed base station, the path loss model between the user and the fixed base station in hotspot scenarios (with the influence of obstruction) is obtained.

[0062] The path loss model L between the user and the fixed base station in the hotspot scenario (with the influence of obstruction) u，f As shown in the following formula:

[0063] L u，f =η G +10μlog 10 (d u,f );

[0064] Where, η G The additional loss caused by the NLoS link when user u communicates with fixed base station f is expressed in dB; μ is the path loss exponent; du，f The straight-line distance between the user and the fixed base station, in km.

[0065] 2-5. The probabilistic link model between the user and the drone base station is combined with the path loss model between the user and the fixed base station in the hotspot scenario to form a spatial propagation model.

[0066] S3. The spatial propagation model is processed by the signal-to-interference-plus-noise ratio model to obtain the interference model of the user during communication with the associated service node;

[0067] By applying a signal-to-interference-plus-noise ratio (SINR) model to the communication process between user u and service node j (i.e., when the user receives data from its associated service node, in addition to system noise, it is also subject to interference signals from other non-associated service nodes), an interference model γ is obtained for user u during its communication process with associated service node j. u，j for:

[0068]

[0069] Where, p u，j The transmit power of service node j when transmitting data to user equipment u; g u，j The channel gain between service node j and user equipment u. N0 is the noise power, N0 = -174 + 10lg B, where B is the base station bandwidth.

[0070] S4. By introducing binary decision variables, the decision variables between each service node and the user are set in the interference model to obtain the connection relationship between the service node and the user.

[0071] By introducing a binary decision variable α, a method for describing the connection relationship between the service node and the user is obtained: when the user communicates with the service node, α = 1; when the service node and the user do not exchange information, α = 0.

[0072] Using the above-described method and the interference model, decision variables between each service node and the user are set to obtain the connection relationship between the service node and the user:

[0073]

[0074] in, N represents the service node with the highest signal-to-interference-plus-noise ratio. j，max Indicates the maximum number of users that can access the service.

[0075] S5. By initializing the drone base station deployment scheme population, obtain the initial flight speed and initial position of the drone base station, and define the individual learning and group learning intensity, inertia weight and number of iterations;

[0076] S6. Utilize the connection relationship between the service node and the user to define the optimization target of the system utility; when the optimization target is met, the coverage and throughput of the communication system are introduced and normalized, and the normalized coverage and throughput are added together to obtain the system utility processing method.

[0077] The optimization objective is shown in the following formula:

[0078]

[0079] Among them, P cov C represents the coverage of the communication system; C represents the throughput of the communication system; C′ is the normalization factor to ensure the two optimization objectives P cov It remains on the same order of magnitude as C;

[0080] C1 represents the location restriction of the drone base station, which limits the research target to a certain range.

[0081] C2 represents the (0, 1) distribution of the association between any user and the service node;

[0082] C3 allows users to select and associate with the service node that has the highest signal-to-interference-plus-noise ratio (SINR) from either fixed base stations or drone base stations.

[0083] C4 allows each user to have at most one node serving them.

[0084] C5 has a capacity limit for service nodes;

[0085] C6 is the minimum threshold value that the signal-to-interference-plus-noise ratio (SIR) must meet for normal communication.

[0086] S7. The system utility of the initial unmanned aerial vehicle (UAV) base station deployment scheme population is obtained through the system utility processing method;

[0087] S8. Divide the system utility of each deployment scheme into population groups: those ranking in the top 20% are considered the dominant group; those ranking in the bottom 20% are considered the suboptimal group; the remaining deployment schemes are considered the ordinary group; combined with... Figure 2 This is a schematic diagram of the particle grouping method based on system utility in the dynamic multi-group update strategy provided by the present invention.

[0088] S9. Using the defined number of iterations, the Gaussian density function is dynamically adjusted through a dynamic multi-scale Gaussian mutation mechanism to perform a local fine search for the dominant group, thereby obtaining the learning strategy corresponding to the dominant group.

[0089] The Gaussian density function is shown in the following equation:

[0090]

[0091] Where μ is the mean (set to 0, indicating variation around the current position), and σ is the standard deviation (controlling the magnitude of variation);

[0092] Furthermore, the dynamic multi-scale Gaussian mutation mechanism is shown in the following equation:

[0093]

[0094] Where t is the current iteration number, T is the total number of iterations, σ(t) is the standard deviation at the current iteration number, δ is the decay rate, and σ start It is the initial scale, σ end It is the ultimate measure.

[0095] S10. Using the defined individual and group learning strengths, inertia weights, and iteration counts, the disadvantaged group is given a learning strategy corresponding to the disadvantaged group by adopting social learning alone.

[0096] The method involves using social learning alone to obtain the velocity and position update formulas for the dominant group particles, which serve as the learning strategy for the dominant group.

[0097]

[0098] in, Let be the velocity of the i-th particle in the j-th dimension at iteration step t; w is the inertia weight, controlling the decay of the particle's velocity; c1 and c2 are learning factors (acceleration constants), representing the intensity of individual learning and group learning, respectively; r1 and r2 are random numbers, following a uniform distribution [0, 1], to increase randomness; gbest j Let be the coordinates of the global optimal solution in the j-th dimension; Let t be the coordinates of particle i in the j-th dimension when the iteration step is t.

[0099] S11. Using the defined individual and group learning strengths, inertia weights, and iteration counts, the ordinary group is explored and developed by following the standard PSO's social and individual learning balance to obtain the learning strategy corresponding to the ordinary group.

[0100] By exploring and developing the social and individual learning balance of the standard PSO, we obtained the velocity update and position update model of ordinary swarm particles, which serves as the learning strategy for the ordinary swarm.

[0101] The velocity update model for the particles is shown in the following equation:

[0102]

[0103] Among them, pbest ij The historical best position of particle i in the j-th dimension

[0104] Obtain a position update model using the current velocity and current position:

[0105]

[0106] Example

[0107] The specific parameter values ​​used in implementing the above steps in this embodiment are shown in the table below:

[0108]

[0109] like Figure 3 of Figure 3 (a) and Figure 3 Figure (b) shows the optimized effect of the embodiment of the present invention under the above parameters. It can be observed that, through the implementation of the above method, compared with the scheme using only fixed base stations, the joint deployment scheme significantly improves user coverage. By introducing drone base stations, the system can serve more users who were previously unable to be served, and the user connectivity rate is greatly improved.

[0110] like Figure 4 The diagram illustrates the impact of the tradeoff factor ratio provided by this invention on coverage and the ratio of system throughput to overall performance. By varying β1:β2 within the range of 3:1 to 1:3, coverage decreased from 97% to 91%, while throughput increased from 7850 to 8200. In practical deployments, the β1:β2 ratio can be adjusted to adapt to different task scenarios.

[0111] For example, if the objective is to provide communication services to as many user devices as possible, maximizing coverage is more important than maximizing system throughput; if economic efficiency is a consideration or to ensure network quality for a portion of the population, maximizing system throughput is more important.

[0112] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0113] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for deploying unmanned aerial vehicle (UAV) base stations based on dynamic multi-population particle swarm optimization, characterized in that, Includes the following steps: S1. Collect information from the fixed base station side, the UAV base station side and the user side to obtain the spatial location information of the fixed base station, the UAV base station and the user in three-dimensional Cartesian coordinates, as well as the straight-line distance between the user and the fixed base station and between the user and the UAV base station. S2. Using the spatial location information and the straight-line distance, construct a probabilistic link model between the user and the drone base station and a path loss model between the user and the fixed base station in hotspot scenarios. By combining the above models, a spatial propagation model is obtained. S3. The spatial propagation model is processed by the signal-to-interference-plus-noise ratio model to obtain the interference model of the user during communication with the associated service node; S4. By introducing binary decision variables, the decision variables between each service node and the user are set in the interference model to obtain the connection relationship between the service node and the user. S5. By initializing the drone base station deployment scheme population, obtain the initial flight speed and initial position of the drone base station, and define the individual learning and group learning intensity, inertia weight and number of iterations; S6. Utilize the connection relationship between the service node and the user to define the optimization target of the system utility; when the optimization target is met, the coverage and throughput of the communication system are introduced and normalized, and the normalized coverage and throughput are added together to obtain the system utility processing method. S7. The system utility of the initial unmanned aerial vehicle (UAV) base station deployment scheme population is obtained through the system utility processing method; S8. Divide the system utility of each deployment scheme into population groups: if the system utility ranks in the top 20%, it is judged as the dominant group; if the system utility ranks in the bottom 20%, it is judged as the disadvantaged group; the remaining deployment schemes are judged as the ordinary group. S9. Using the defined number of iterations, the Gaussian density function is dynamically adjusted through a dynamic multi-scale Gaussian mutation mechanism to perform a local fine search for the dominant group, thereby obtaining the learning strategy corresponding to the dominant group. S10. Using the defined individual and group learning strengths, inertia weights, and iteration counts, the disadvantaged group is given a learning strategy corresponding to the disadvantaged group by adopting social learning alone. S11. Using the defined individual and group learning strengths, inertia weights, and iteration counts, the ordinary group obtains the corresponding learning strategy through standard PSO.

2. The method for deploying UAV base stations based on dynamic multi-population particle swarm optimization according to claim 1, characterized in that, The information collection in step 1 is as follows: Let the set of the number of fixed base stations be F = {1, 2, ..., F...} Max }, F Max ≥1, where F Max Represents the maximum value in set F; Define the spatial position of the fixed base station f in a three-dimensional Cartesian coordinate system: Where, x f y f h f They are s f Position on the x, y, and z axes; Define the set of unmanned aerial vehicle (UAV) base stations: M = {1, 2, ..., M} Max }, M Max ≥1, where M Max This represents the maximum value in set M; Define the spatial position of the UAV base station m in a three-dimensional Cartesian coordinate system: Where, x m y m h m They are s m Position on the x, y, and z axes; By taking the union of sets F and M, we obtain the set of all service nodes: J = {1, 2, ..., J} Max }, J Max ≥1, where J=F∪M; The set of UAV base stations, U = {1, 2, ..., U}, is obtained through initialization settings. Max }, U Max ≥1, where U Max Represents the maximum value in set U; Assuming the ground user is relatively stationary within the square area, the spatial position of user u in the three-dimensional Cartesian coordinate system is obtained through initialization settings. Where, x u y u h u They are s u Position on the x, y, and z axes; Determine the straight-line distance d between user u and drone base station m. u，m : Determine the straight-line distance d between user u and fixed base station f. u，f :

3. The method for deploying UAV base stations based on dynamic multi-population particle swarm optimization according to claim 2, characterized in that, Step 2 includes the following steps: 2-1. The straight-line distance between the user and the drone base station is used to obtain the free-space propagation path loss by adopting the Fries free-space propagation path loss model; 2-2. By integrating the free space propagation path loss with the additional path loss, a LoS and NLoS path loss model is obtained between the UAV base station communication node and the ground terminal user. 2-3. Using the line-of-sight transmission probability calculation formula that correlates the elevation angle parameters between the UAV platform and the ground terminal, the LoS and NLoS path loss models are used to determine the probabilistic link model between the user and the UAV base station. 2-4. The distance between the user and the fixed base station is determined by defining the additional loss caused by the NLoS link and the path loss index when the user communicates with the fixed base station, thus obtaining the path loss model between the user and the fixed base station in the hot spot scenario. 2-5. The probabilistic link model between the user and the drone base station is combined with the path loss model between the user and the fixed base station in the hotspot scenario to form a spatial propagation model.

4. The method for deploying UAV base stations based on dynamic multi-population particle swarm optimization according to claim 3, characterized in that, The formula for calculating the line-of-sight transmission probability associated with the elevation angle parameter between the UAV platform and the ground terminal in steps 2-3 is as follows: P(NLoS,θ u，m )=1-P(LoS,θ u，m ); Probabilistic link model L between users and drone base stations u，m As shown in the following formula: Where P(LoS,θ) u，m P(NLoS, θ) represents the probability of line-of-sight connection; u，m ) represents the probability of line-of-sight connectivity; a represents the ratio between the construction land area and the total area; b represents the average number of buildings per unit area; θ represents the probability of line-of-sight connectivity. u，m Let m be the elevation angle between the drone base station and user u. The path loss model L between users and fixed base stations in the hotspot scenario described in steps 2-4 u，f As shown in the following formula: L u，f =the G +10μlog 10 (d u,f ); Where, η G d is the additional loss caused by the NLoS link when user u communicates with fixed base station f; μ is the path loss exponent; d u，f This refers to the straight-line distance between the user and the fixed base station.

5. The method for deploying unmanned aerial vehicle (UAV) base stations based on dynamic multi-population particle swarm optimization according to claim 4, characterized in that, Step 3 is described in detail below: By applying a signal-to-interference-plus-noise ratio (SINR) model to the communication process between user u and associated service node j, an interference model γ is obtained. u，j for: Where, p u，j The transmit power of service node j when transmitting data to user equipment u; g u,j The channel gain between service node j and user equipment u. N0 is the noise power, N0 = -174 + 10lg B, where B is the base station bandwidth.

6. The method for deploying unmanned aerial vehicle (UAV) base stations based on dynamic multi-population particle swarm optimization according to claim 5, characterized in that, Step 4 is described in detail below: By introducing the decision variable α, we obtain a representation of the connection relationship between the service node and the user: when the user communicates with the service node, α = 1; when the service node and the user do not exchange information, α = 0. By setting decision variables between each service node and the user, and using the aforementioned interference model, the connection relationship between the service node and the user is obtained: in, N represents the service node with the highest signal-to-interference-plus-noise ratio. j，max Indicates the maximum number of users that can access the service.

7. The method for deploying UAV base stations based on dynamic multi-population particle swarm optimization according to claim 6, characterized in that, The optimization objective of step 6 is shown in the following formula: Among them, P cov C represents the coverage of the communication system; C represents the throughput of the communication system; C′ is the normalization factor to ensure the two optimization objectives P cov It remains on the same order of magnitude as C; C1 represents the location restriction of the drone base station, which limits the research target to a certain range. C2 represents the (0, 1) distribution of the association between any user and the service node; C3 allows users to select and associate with the service node that has the highest signal-to-interference-plus-noise ratio (SINR) from either fixed base stations or drone base stations. C4 allows each user to have at most one node serving them. C5 has a capacity limit for service nodes; C6 is the minimum threshold value that the signal-to-interference-plus-noise ratio (SIR) must meet for normal communication.

8. The method for deploying unmanned aerial vehicle (UAV) base stations based on dynamic multi-population particle swarm optimization according to claim 7, characterized in that, The Gaussian density function in step 9 is shown in the following equation: Where μ is the mean and σ is the standard deviation; Furthermore, the dynamic multi-scale Gaussian mutation mechanism is shown in the following equation: Where t is the current iteration number, T is the total number of iterations, σ(t) is the standard deviation at the current iteration number, δ is the decay rate, and σ start It is the initial scale, σ end It is the ultimate measure.

9. The method for deploying unmanned aerial vehicle (UAV) base stations based on dynamic multi-population particle swarm optimization according to claim 8, characterized in that, In step 10, by employing social learning alone, the velocity and position update formulas for the dominant group particles are obtained, serving as the learning strategy corresponding to the dominant group. in, Let be the velocity of the i-th particle in the j-th dimension at iteration step t; w is the inertia weight, controlling the decay of the particle velocity; c1 and c2 are learning factors, representing the intensity of individual learning and group learning, respectively; r1 and r2 are random numbers, following a uniform distribution [0, 1], to increase randomness; gbest j Let be the coordinates of the global optimal solution in the j-th dimension; Let t be the coordinates of particle i in the j-th dimension when the iteration step is t.

10. The method for deploying unmanned aerial vehicle (UAV) base stations based on dynamic multi-population particle swarm optimization according to claim 9, characterized in that, In step 11, the velocity update and position update model of ordinary swarm particles are obtained through standard PSO, which serves as the learning strategy for the ordinary swarm. The velocity update model for the particles is shown in the following equation: Among them, pbest ij The historical best position of particle i in the j-th dimension Obtain a position update model using the current velocity and current position:

Images