Method for unmanned aerial vehicle position deployment and ground user clustering communication facing data service transmission demand

Abstract

The application provides a method for unmanned aerial vehicle position deployment and ground user clustering communication aiming at data service transmission requirements. First, the related parameters of unmanned aerial vehicles and static ground users are set to construct a communication scenario; then, the cooperation relationship between the unmanned aerial vehicle position deployment and the ground user clustering is established in combination with the data service transmission requirements, and the deployment and clustering communication strategy is distributed to the unmanned aerial vehicles and the ground users in a centralized manner; finally, the air-ground joint task of the unmanned aerial vehicle deployment and the ground user clustering communication is completed. According to the application, more ground users are effectively served by the unmanned aerial vehicles in the form of joining clusters, and the communication overhead of the ground users is reduced. The traditional alliance formation algorithm is improved in view of the problem of limited coverage range of the unmanned aerial vehicles, and the convergence number required for obtaining a stable solution is obviously reduced under the condition of less unmanned aerial vehicles.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wireless communication, in particular to a method for unmanned aerial vehicle position deployment and ground user clustering communication aiming at data service transmission demand. BACKGROUND

[0002] Due to the characteristics of flexible deployment, high efficiency, low cost and the like, unmanned aerial vehicles are widely used as air base stations to meet the communication needs of ground users. Unmanned aerial vehicle position deployment is a basic problem in air-ground communication. Since the coverage of unmanned aerial vehicles is limited, unmanned aerial vehicles need to be deployed at appropriate positions to meet the communication needs of more ground users. In addition, with the development of Internet of Things technology, the communication needs of ground users have increased significantly, and the limited coverage of unmanned aerial vehicles is insufficient to meet the communication needs of all ground users. Ground users cooperate in the form of clusters, which is an effective way to further improve the performance of air-ground networks.

[0003] However, current researches are limited to clustering ground users and unmanned aerial vehicles respectively, and there are few studies on how to combine ground user clustering and unmanned aerial vehicles. SUMMARY

[0004] The present application provides a method for unmanned aerial vehicle position deployment and ground user clustering communication aiming at data service transmission demand, which can be used to solve the technical problem that ground users far away from unmanned aerial vehicles cannot receive signals due to limited transmission power of unmanned aerial vehicles and line-of-sight obstruction of signals.

[0005] The present application provides a method for unmanned aerial vehicle position deployment and ground user clustering communication aiming at data service transmission demand, which includes the following steps:

[0006] Step 1, set the communication related parameters of unmanned aerial vehicles and static ground users in air-ground networks, construct a communication scenario, and determine the downlink data transmission rate of unmanned aerial vehicles to ground users and the data transmission rate between ground users;

[0007] Step 2, establish the cooperation relationship between unmanned aerial vehicle deployment and ground user clustering communication in combination with data service demand;

[0008] Step 3, the centralized controller respectively issues the unmanned aerial vehicle deployment strategy and the ground user clustering strategy obtained by the optimization algorithm to the unmanned aerial vehicle group and the ground users, and the unmanned aerial vehicles and the ground users execute the issued strategies to finally complete the joint task of unmanned aerial vehicle position deployment and ground user clustering communication under data service demand.

[0009] Further, set the communication related parameters of the unmanned aerial vehicles and static ground users in the air-ground network, and build a communication scenario: the air-ground network includes M unmanned aerial vehicles and N ground users; the unmanned aerial vehicles unicast the required service data to the ground users according to the service data requirements of the ground users, the default channel resource is sufficient in the air-ground communication scenario, the co-frequency interference in wireless communication is not considered, and the deployment heights of the unmanned aerial vehicles are all the same;

[0010] Before the deployment of the unmanned aerial vehicles, the centralized controller collects the communication related parameters in the air-ground network, including the positions of the ground users, the transmission powers of the unmanned aerial vehicles and the ground users, and the electromagnetic environment parameters, thereby building the air-ground communication scenario and obtaining the service data transmission rates between the air and the ground and between the ground and the ground.

[0011] Further, in step 2, the cooperation relationship between the unmanned aerial vehicle deployment and the ground user clustering communication is established in combination with the data service requirements, including:

[0012] Step 2-1, determining the set of ground users effectively covered by the unmanned aerial vehicles, designing a ground user clustering mechanism, and establishing a ground user utility model;

[0013] Step 2-2, establishing the relationship between the unmanned aerial vehicle position deployment strategy and the ground user clustering strategy as a hierarchical game model; the hierarchical game model is divided into an outer layer position unmanned aerial vehicle deployment game and an inner layer alliance formation game;

[0014] Step 2-3, the centralized controller executes a joint optimization algorithm for the unmanned aerial vehicle position deployment and the ground user clustering; the unmanned aerial vehicle Nash equilibrium position deployment strategy and the ground user Nash equilibrium clustering strategy are obtained respectively.

[0015] Further, the unmanned aerial vehicle to ground user downlink data transmission rate is determined by the following method:

[0016] Suppose that the air-ground network includes M unmanned aerial vehicles and N ground users; the sets of the unmanned aerial vehicles and the ground users are respectively denoted as and Suppose that the deployment heights of the unmanned aerial vehicles are all h, the task area is discretized in two dimensions, and the coordinates of the unmanned aerial vehicle m and the ground user n are respectively denoted as and

[0017] In the communication scenario, due to the complex environment terrain, there are line of sight (LOS) and non line of sight (NLOS) in the air-ground channel transmission link; in the widely used air-ground link channel model, the path loss between the unmanned aerial vehicle and the ground user is modeled as a probability weighted average value of the line of sight link loss and the non line of sight link loss; the line of sight link probability between the unmanned aerial vehicle m and the ground user n is defined as follows:

[0018]

[0019] Among them, ζ1 and ζ2 are two environmental impact factors, the magnitude of which depends on building density and topography. The elevation angle (in degrees) of the drone m as seen from ground user n is determined as follows:

[0020]

[0021] in, Let m be the straight-line distance from drone m to ground user n. Based on the line-of-sight link probability, the probability of NLoS transmission between drone m and ground user n is:

[0022] The transmission path loss from drone m to ground user n is:

[0023]

[0024] Among them, f m Let m be the carrier frequency for transmitting data from the drone, c0 be the speed of light, and μ be the carrier frequency. LoS and μ NLoS These are the attenuation factors for line-of-sight links and non-line-of-sight links, respectively.

[0025] Assuming all drones have identical hardware performance, the antenna gain of drone m is:

[0026]

[0027] Where, θ t θ is the beamwidth of the main lobe of the transmitter antenna for each UAV. m,n Angle of elevation The complementary angle, G t N0 represents the main lobe gain of each drone transmitter and the number of drone antennas.

[0028] Taking into account the impact of Los and NLos links on the signal received by ground users, the signal-to-noise ratio of the signal received by ground user n from UAV m is determined as follows:

[0029]

[0030] Where P0 is the transmit power of a single UAV transmitter, σ 2 The variance of the additive white Gaussian noise;

[0031] The downlink data transmission rate from drone m to ground user n is:

[0032]

[0033] where the channel bandwidth of the UAV is fixed, denoted as B.

[0034] Further, the data transmission rate between ground users is determined by the following method:

[0035] The data distribution mechanism within the cluster relies on the data forwarding between ground users, so the communication relationship between ground users needs to be considered; the communication between ground users considers the NLoS link, assuming that the hardware properties of all ground users are the same, the signal-to-noise ratio of the communication between ground users n1 and n2 is obtained as:

[0036]

[0037] where β0 is the channel gain of a unit energy signal at a distance of 1 meter from the source, P1 is the transmission power of the ground user, is the communication distance between ground users n1 and n2, and α0 is the loss factor of the communication between ground users;

[0038] The data transmission rate between ground users n1 and n2 is:

[0039]

[0040] The intra-cluster users relay and forward data in the form of decode-and-forward, so the data transmission rate of the ground user receiving data from other ground users within the same cluster is the minimum value of the data transmission rate of the UAV-intra-cluster user and the intra-cluster user-intra-cluster user link; since the data transmission rate of the UAV-intra-cluster user is greater than R th , as long as the data transmission rate of the intra-cluster user when receiving data is greater than R th , it is considered that the user successfully receives the required data; for the relationship between any two ground users n1 and n2, denoted as:

[0041]

[0042] Only when , it is indicated that the ground users n1 and n2 can transmit data, that is, a neighbor relationship is established between the two ground users.

[0043] Further, the set of ground users effectively covered by the UAV is determined by the following method:

[0044] The information rate threshold R th is used as a standard to judge whether a user is effectively served, if the information rate of the ground user receiving the signal of a UAV is greater than the threshold value R thIf a ground user is covered by multiple UAVs, the ground user will receive the UAV signal with the highest signal-to-noise ratio; define the set of ground users effectively covered by UAVs as:

[0045]

[0046] The set of ground users not effectively covered by UAVs is:

[0047]

[0048] Further, the ground user clustering mechanism determines by the following method:

[0049] The way ground users obtain data is related to whether they are covered by UAVs; ground users in set can be effectively served, so ground users in set can directly download required data from UAVs, or obtain required data by acting as a cluster head or joining a cluster from the perspective of reducing overhead; ground users in set can only obtain required data by joining a cluster through a cluster head;

[0050] To enhance the willingness of ground users in set to initiate a cluster, a group purchase mechanism is introduced, i.e., cluster members entrust the cluster head to download data and need to share the overhead generated by downloading required data of the cluster head; when cluster members have similar data requirements as the cluster head, since other cluster members share part of the overhead required by the cluster head, the cluster head obtains required data with smaller overhead, so ground users in set act as cluster heads and can also reduce their own data acquisition overhead when helping other cluster members download data;

[0051] The overhead required by cluster members under the group purchase mechanism is as follows:

[0052] Let ground user n1 be a cluster head, and the user cluster managed by n1 be For any member n2 in cluster , the similarity between its data requirements and those of other members in the cluster is evaluated; specifically, by comparing the data requirements of ground user n2 with those of all other members in the cluster one by one, the similarity of each data requirement item is quantitatively calculated, and the quantitatively expressed way is as follows:

[0053]

[0054] wherein represents the length of data required by the ground user, represents the i-th data requirement of ground user n2 in cluster The degree of similarity between the data requirements of the data in the cluster and those of other members within the cluster;

[0055] Given a certain data similarity, the download cost is calculated. From a fairness perspective, the download cost incurred by cluster head n1 downloading certain data should be shared equally among all ground users within the cluster who need that data. For any member n2 within the cluster, the download cost is:

[0056]

[0057] Where, ν d l(R) represents the download cost incurred by the cluster head when downloading certain data. n1 ) represents a cluster Total data requirement length, L max This indicates the maximum data length that the cluster head can receive; if l(R) n1 )≤L max This indicates that cluster head n1 can download the data required by all users in the cluster; otherwise, the current cluster is invalid, and users in the cluster covered by the drone download their required data individually, while users in the cluster not covered by the drone cannot obtain their required data.

[0058] Based on the neighbor relationships and similar data needs among ground users, and combined with the deployment of UAVs, ground users are divided into multiple clusters; the cluster head node is responsible for acquiring data from UAVs and enabling all users in the cluster to receive the data through intra-cluster forwarding; to improve the reliability of intra-cluster data transmission, intra-cluster users use a flooding mechanism to determine the data transmission path.

[0059] Given a defined intra-cluster data transmission path, the forwarding overhead borne by users within the cluster is calculated; the forwarding overhead incurred by a single data forwarding action by a ground user is denoted as... Assume that ground users n2 are clusters Based on the members of the cluster and the position of the user within the cluster, the calculation method for forwarding overhead is divided into three cases: In the first case, when the ground user n2 is the source node (i.e., the cluster head), it is only responsible for forwarding data, and the forwarding overhead is calculated in three ways. The data transmission cost is shared equally between the next-level member receiving the data and ground user n2, depending on the data transmission path. In the second case, when ground user n2 is a leaf node, only the required data needs to be received, and the forwarding overhead is minimal. According to the data transmission path, the cost is shared by the ground user sending the data, other ground users receiving data from the same ground user at the same time, and ground user n2. In the third case, when ground user n2 does not belong to the above two cases, according to the data transmission path, it needs to receive data from the superior ground user and forward the received data to the subordinate ground user. The forwarding cost is the sum of the first two cases.

[0060] The forwarding overhead of the ground user n2 is calculated as follows:

[0061]

[0062] wherein the next level ground user set receiving data from the ground user n2 is denoted as LowNode, the upper level ground user sending data to the ground user n2 and other ground user set receiving data from the upper level ground user are denoted as OtherNode according to the data transmission path.

[0063] Further, the ground user utility model is determined by the following method:

[0064] Due to the limitation of communication example and cluster scale, some users in the set cannot obtain the required data, and the ground user is divided into two parts according to whether the required data is received, and

[0065]

[0066] Only when λ n = 1, the ground user n successfully receives the required data;

[0067] For the ground user, two aspects are considered, one is whether the required data can be received, and the other is how much overhead is spent when the required data is received, and the utility function of the ground user is:

[0068] η n = λ n [1-α(D n +F n )].

[0069] Further, in step 2-2, the hierarchical game model is established by the following method:

[0070] The outer layer unmanned aerial vehicle deployment game is modeled as follows:

[0071]

[0072] wherein denotes the set of unmanned aerial vehicles, J m denotes the strategy space of the unmanned aerial vehicle m, denotes the airspace with a height of h corresponding to the task area, the unmanned aerial vehicle m will explore its deployment strategy in the airspace, u m denotes the utility of the unmanned aerial vehicle m; the marginal utility is used to construct the utility u m of the unmanned aerial vehicle m, which is defined as follows:

[0073]

[0074] where J -m = (J1, J2,... J m-1 , J m+1 ,..., J M ) represents the deployment strategy set of other UAVs except UAV m, A(J) represents the coalition formation strategy set of ground users, U(J m , J -m , A(J)) represents the total utility of all UAVs;

[0075] The optimization objective of the outer-layer UAV deployment game is modeled as:

[0076]

[0077] The inner-layer coalition formation game is modeled as:

[0078]

[0079] where a n represents the coalition formation strategy of ground user n, a-n = (a1, a2,..., an-1, an +1 ,..., a N ) represents the coalition formation strategy set of other users except ground user n, represents the coalition set in the current task area, u n represents the utility of ground user n;

[0080] The optimization objective of the inner-layer coalition formation game is modeled as:

[0081]

[0082] The criterion adopted in the ground user coalition formation is the Pareto criterion, i.e., assuming that ground user n is currently in coalition CO j , if user wants to join coalition CO i , the following relationship must be satisfied:

[0083]

[0084] According to the Pareto criterion, ground users can choose to join the coalition that maximizes their utility, however, only ground users that are effectively covered by UAVs can initiate a coalition and improve their utility. For ground users that cannot be effectively served by UAVs, they cannot initiate a coalition and belong to no coalition in the initial stage, thus they cannot join other coalitions to improve their utility through the Pareto criterion. To solve this problem, the Pareto criterion is supplemented and a resource-constrained Pareto criterion is proposed, which is defined as follows:

[0085] Due to resource constraints, only the set The ground users in the set V can form coalitions, if there exists a ground user n that is neither in the set V nor belongs to any coalition, when the user wants to join a coalition CO , the following relationship is satisfied: i

[0086]

[0087] By combining the resource-constrained Pareto criterion with the Pareto criterion, the formation of the ground user coalition under the given position of the UAV will eventually converge to a Nash equilibrium.

[0088] Further, in step 2-3, the centralized controller executes a joint optimization algorithm for the UAV position deployment and the ground user clustering; and respectively obtains a UAV Nash equilibrium position deployment strategy and a ground user Nash equilibrium clustering strategy, including:

[0089] The joint optimization algorithm is implemented by using a hierarchical structure, the outer layer is a local space adaptive algorithm, which is used to optimize the position deployment of the UAV, and the inner layer is a coalition formation game algorithm, which is used to optimize the formation of the ground user coalition; when the UAV explores the position in the outer layer, the ground user in the inner layer forms the coalition according to the position of the UAV, and the steps are as follows:

[0090] Step A1, a UAV deployment strategy J = {J1, J2,..., J M} is given;

[0091] Step A2, a UAV m is randomly selected, and s deployment strategies are randomly selected from the strategy space , the current strategy is recorded as J m (q1), and the newly selected deployment strategy is recorded as {J1, J2,..., J s};

[0092] Step A3, the inner layer algorithm is run to calculate the total utility of the ground user corresponding to different UAV deployment strategies, so as to obtain the utility of the UAV m corresponding to different deployment strategies, and the UAV m updates its deployment position according to the utility under different deployment strategies, the utility of the UAV is in the form of:

[0093]

[0094] The probability that the UAV m updates the deployment strategy to d m is:

[0095]

[0096] Step A4, repeat steps A1-A3 until convergence;

[0097] ​The inner layer algorithm considers two aspects in the process of alliance formation. Firstly, ground users attempt to join other alliances. In order to enable more ground users to join the alliance, two principles are followed in the processing order. On the one hand, ground users who have not joined the alliance are given priority to attempt to join the alliance. On the other hand, ground users with fewer neighboring alliances are given priority to attempt to join the alliance. Secondly, the fusion and exchange between alliances are considered. In order to reduce the communication overhead of ground users as much as possible, the data service requirements of ground users in the alliance need to be as similar as possible. Therefore, before fusion and exchange, the alliances are matched first, and alliances with high data similarity are matched in pairs. Then, the alliance members between alliances are exchanged according to the data similarity. The steps are as follows:

[0098] Firstly, the algorithm for ground users attempting to join other alliances is as follows:

[0099] Step B1, determine the user set in the alliance and the alliance neighbor user set not in any alliance

[0100] Step B2, generate a ground user processing order. The ground users in set are processed first, and then the ground users in set are processed. In addition, the ground users in set are sorted according to the principle that ground users with fewer neighboring alliances are processed first, so as to finally obtain the ground user processing order.

[0101] Step B3, attempt to let ground users join other alliances according to the processing order. The set of neighboring alliances of ground user n is P n (CO)={CO1,CO 2, ,...,CO k}, respectively calculate the utility obtained after the user joins different neighboring alliances u n (CO)={η 2n (CO1),η 2n (CO2),...,η 2n (CO k )}, and finally the ground user determines which alliance to join according to the Pareto criterion.

[0102] Step B5, after the user is processed according to the order, run the fusion and exchange algorithm between alliances to obtain a new alliance structure.

[0103] Step B6, repeat steps B1-B5 until convergence.

[0104] Secondly, the fusion and exchange algorithm between alliances is as follows:

[0105] Step C1, each alliance generates a preferred matching list of neighboring alliances according to the data similarity.

[0106] Step C2, randomly select a coalition, match with neighbor coalition according to preference matching list, if neighbor coalition has no matching pair, then match successfully, otherwise, neighbor coalition matches with more preferred coalition according to preference matching list, and rejects another coalition;

[0107] Step C3, update matching list of rejected and rejecting coalition, and delete each other from matching list;

[0108] Step C4, repeat step C1-step C3 until convergence, and finally obtain several coalition matching pairs;

[0109] Step C5, sequentially perform fusion and exchange operation on several coalition matching pairs, first try to fuse coalition matching pairs into a coalition, if it meets Pareto criterion, directly output the result, otherwise, coalition matching pairs try to exchange ground users in equal quantity according to business data similarity, and judge whether the exchange is established through Pareto criterion.

[0110] Compared with the prior art, the significant progress of the present application is that: 1) focusing on the leading application scenario of air-ground communication, the joint decision method of multi-unmanned aerial vehicle position deployment and ground user clustering communication under data service transmission demand can effectively increase the number of ground user services and reduce the communication overhead of ground users; 2) the traditional coalition formation algorithm is improved, compared with the traditional coalition formation algorithm, the performance of the proposed coalition formation algorithm is improved by about 10% under the premise of ensuring the effectiveness of the result, and in the case of fewer unmanned aerial vehicles, the convergence number required for the proposed algorithm to obtain a stable solution is significantly reduced. BRIEF DESCRIPTION OF DRAWINGS

[0111] Figure 1 is the air-ground network model diagram based on ground user clustering communication of the present application;

[0112] Figure 2 is the convergence performance schematic diagram of the method in embodiment 1 of the present application under different network scales;

[0113] Figure 3 is the performance comparison schematic diagram of the method and the comparative algorithm in embodiment 2 of the present application;

[0114] Figure 4 is the convergence comparison schematic diagram of the method and the comparative algorithm in embodiment 2 of the present application. DETAILED DESCRIPTION

[0115] In order to make the purpose, technical scheme and advantages of the present application more clear, the embodiments of the present application will be further described in detail below with reference to the drawings.

[0116] Firstly, the embodiments of the present application will be introduced below with reference to the drawings.

[0117] This invention proposes a method for UAV location deployment and ground user clustering communication oriented towards data service transmission needs, specifically a hierarchical game model and decision-making method for UAV location deployment and ground user clustering communication in an air-to-ground communication scenario. In air-to-ground communication networks, due to the limited transmission power of UAVs and signal line-of-sight obstruction, ground users far from the UAVs have difficulty receiving signals. Clustering ground users to receive signals from UAVs can effectively solve this problem. Furthermore, compared to directly forwarding signals, rationally clustering ground users can effectively reduce network communication overhead and improve spectrum utilization efficiency. The proposed game model uses a hierarchical game framework to model UAV location deployment and ground user clustering. The outer layer is modeled as a UAV deployment game, and the inner layer as a coalition formation game. A Pareto criterion under resource constraints is proposed in the inner layer game. The proposed decision-making method is a hierarchical algorithm based on a partial spatial adaptive algorithm and an optimal response algorithm, respectively addressing the problems of UAV location deployment and ground user clustering communication. Finally, simulation results under different air-to-ground network parameters demonstrate the effectiveness of user clustering communication and the proposed algorithm. This invention includes the following steps:

[0118] Step 1: Set the relevant parameters for drones and static ground users in the air-to-ground network to construct a communication scenario;

[0119] Step 2: Establish a collaborative relationship between UAV deployment and ground user cluster communication based on data service requirements;

[0120] Step 3: Centrally distribute the joint strategy to the drones and ground users. Ground users receive the required service data from the drones in clusters according to the joint strategy; finally, the joint task of drone location deployment and ground user cluster communication under the data service requirements is completed.

[0121] like Figure 1 As shown, Figure 1 This is a diagram of an air-to-ground network model based on ground-based clustered communication. The model consists of M UAVs and N ground users, with the sets of UAVs and ground users represented as follows: and Assuming all drones are deployed at altitude h, and the mission area is discretized in two dimensions, the coordinates of drone m and ground user n can be represented as follows: and

[0122] Due to the complex environmental terrain, there are Line of Sight (LOS) and Non Line of Sight (NLOS) in air-ground channel transmission link. In the channel model of air-ground link which is widely used, the path loss between UAV and ground user is modeled as the probability weighted average of LOS link loss and NLOS link loss. The probability of LOS link between UAV m and ground user n is defined as follows:

[0123]

[0124] where ζ1 and ζ2 are environmental factors, whose size depends on the building density, topography and other factors, is the elevation angle (in degrees) from the ground user n to the UAV m, and the calculation formula is as follows:

[0125]

[0126] where, is the straight-line distance from the UAV m to the ground user n, and the probability of NLoS transmission between the UAV m and the ground user n is:

[0127] The transmission path loss of the UAV m to the ground user n is:

[0128]

[0129] where, f m is the carrier frequency of the UAV m transmission data, c0 is the speed of light, μ LoS and μ NLoS are the attenuation factors of LOS link and NLOS link respectively;

[0130] Assuming that the hardware performance of all UAVs is the same, the antenna gain of the UAV m is:

[0131]

[0132] where, θ t is the beam width of each UAV transmitter antenna main lobe, θ m,n is the complementary angle of the elevation angle , G t is the main lobe gain of each UAV transmitter, and N0 is the number of UAV antennas;

[0133] Considering the influence of Los and NLos link on the received signal of the ground user, the signal-to-noise ratio of the ground user n receiving the signal from the UAV m is designed as:

[0134]

[0135] where P0is the transmission power of the single unmanned aerial vehicle transmitter, σ 2 is the variance of the additive white Gaussian noise;

[0136] Based on the above analysis, the downlink data transmission rate of unmanned aerial vehicle m to ground user n is obtained as:

[0137]

[0138] where the channel bandwidth of the unmanned aerial vehicle is fixed, denoted as B.

[0139] The data distribution mechanism within the cluster relies on the data forwarding between ground users, so the communication relationship between ground users also needs to be considered. The communication between ground users mainly considers the NLoS link, and assuming that all ground users have the same hardware properties, the signal-to-noise ratio of the communication between ground users n1 and n2 is obtained as:

[0140]

[0141] where β0is the channel gain of a unit energy signal at a distance of 1 meter from the source, P1is the ground user transmission power, is the communication distance between ground users n1 and n2, and α0is the loss factor of the communication between ground users;

[0142] The data transmission rate between ground users n1 and n2 is:

[0143]

[0144] Considering that the users within the cluster relay and forward data in the form of decoding forwarding, the data transmission rate received by the ground user from other ground users within the same cluster is the minimum value of the data transmission rates of the unmanned aerial vehicle-cluster user and the cluster user-cluster user link. Since the data transmission rate of the unmanned aerial vehicle-cluster user is greater than R th , as long as the data transmission rate when the cluster user receives data is greater than R th , it can be considered that the user successfully receives the required data. For the relationship between any two ground users n1 and n2, it can be expressed as:

[0145]

[0146] Only when , it means that the ground users n1 and n2 can transmit data, that is, a neighbor relationship is established between the two ground users.

[0147] The information rate threshold R th is used as a standard to judge whether a user has been effectively served. If the information rate of the ground user receiving a signal from a certain unmanned aerial vehicle is greater than the threshold value Rth If a ground user is effectively covered by multiple UAVs, the ground user will receive the UAV signal with the highest signal-to-noise ratio. Therefore, the set of ground users effectively covered by UAVs is defined as:

[0148]

[0149] The set of ground users not effectively covered by UAVs is:

[0150]

[0151] The way a ground user acquires data is related to whether it is covered by a UAV or not. Since the ground users in set are effectively served, the ground users in this set can either directly download the required data from the UAV or acquire the required data as a cluster head or by joining a cluster from the perspective of reducing overhead. The ground users in set can only acquire the required data through a cluster head by joining a cluster. To enhance the willingness of ground users in set to initiate a cluster, this section introduces a group purchase mechanism, i.e., cluster members entrust the cluster head to download data and need to share the overhead generated by downloading the required data of the cluster head. When cluster members have similar data requirements as the cluster head, the cluster head can acquire the required data with smaller overhead since other cluster members share a portion of the overhead for the required data of the cluster head. Therefore, the ground users in set as cluster heads can also reduce their own data acquisition overhead when helping other cluster members download data;

[0152] The required overhead of cluster members under the group purchase mechanism is described. Let ground user n1 be the cluster head, and the user cluster managed by n1 be denoted as For any member n2 in cluster , the similarity between its data requirements and those of other members in the cluster needs to be evaluated first. Specifically, by comparing the data requirements of ground user n2 with those of all other members in the cluster one by one, the similarity of each data requirement item can be quantitatively calculated, and the quantified expression is as follows:

[0153]

[0154] wherein denotes the length of the data required by the ground user, denotes the similarity of the i-th data requirement of ground user n2 to the data requirements of other members in cluster R n1

[0155] In order to quantify the similarity ​To better understand, with Figure 1 Taking the clustering results in the figure as an example, ground user 1, ground user 3, ground user 4 and ground user 5 form a cluster on the right side of the figure. Their data requirements are (5,6,7,8), (4,6,8,9), (4,6,7,9) and (6,7,8,9) respectively. Taking ground user 5 as an example, by comparing its required data with the data requirements of other users in the cluster, the similarity of each data requirement item can be quantified as (3,2,2,2).

[0156] Given a certain data similarity, the download cost to be borne is calculated. From a fairness perspective, the download cost incurred by cluster head n1 downloading certain data should be shared equally among all ground users within the cluster who need that data. For any member n2 within the cluster, the download cost is calculated as follows:

[0157]

[0158] Where, ν d l(R) represents the download cost incurred by the cluster head when downloading certain data. n1 ) represents a cluster Total data requirement length, L max This indicates the maximum data length that the cluster head can receive. If l(R) n1 )≤L max This indicates that cluster head n1 can download the data required by all users in the cluster; otherwise, the current cluster is invalid, and users in the cluster covered by the drone download their required data individually, while users in the cluster not covered by the drone cannot obtain their required data.

[0159] pass Figure 1 The clustering results on the right provide an example of calculating the download cost for users within a cluster. After analyzing the data similarity of each user within the cluster, the download cost is calculated for each user. Taking ground user 5 as an example, its data similarity is (3,2,2,2), and the download cost it incurs for acquiring data 6 is (1 / 4)ν. d The download cost for data 7, 8, and 9 is (1 / 3)ν. d ;

[0160] Based on the neighbor relationships and similarity of data needs among ground users, and combined with the deployment of drones, ground users are divided into multiple clusters. The cluster head node is responsible for acquiring data from drones and ensuring that all users within the cluster receive the data through intra-cluster forwarding. To improve the reliability of intra-cluster data transmission, users within the cluster utilize a flooding mechanism to determine the data transmission path.

[0161] Given a fixed intra-cluster data transmission path, the forwarding overhead borne by users within the cluster is calculated. The forwarding overhead incurred by a single data forwarding action by a ground user is denoted as... Assume that ground users n2 are clusters The calculation of forwarding overhead for members within a cluster, based on their position within the cluster, is divided into three cases: The first case is when ground user n2 is the source node (cluster head), it is only responsible for forwarding data, and the forwarding overhead is... The data transmission cost is shared equally between the next-level member receiving the data and ground user n2, depending on the data transmission path. In the second case, when ground user n2 is a leaf node, it only needs to receive the required data, and the forwarding overhead is minimal. The forwarding overhead is shared by the ground user sending the data, other ground users simultaneously receiving data from the same ground user, and ground user n2, according to the data transmission path. In the third case, where ground user n2 does not fall into either of the above two categories, it needs to receive data from the superior ground user and forward the received data to the subordinate ground user; the forwarding overhead is the sum of the first two cases. In summary, the calculation method for the forwarding overhead of ground user n2 is as follows:

[0162]

[0163] In this context, the set of next-level ground users receiving data from ground user n2 is represented as LowNode according to the data transmission path, and the set of upper-level ground users sending data to ground user n2 and other ground users that jointly receive data from upper-level ground users is represented as OtherNode.

[0164] pass Figure 1 The clustering results on the right illustrate the calculation of user forwarding overhead within a cluster. Based on the data transmission path, three neighbor relationships are established within the cluster: ground users 1 and 3, ground users 1 and 4, and ground users 3 and 5. Ground user 1, as the source node, bears a forwarding overhead of 2ν. s Ground users 4 and 5, acting as leaf nodes, bear forwarding overheads of 2ν respectively. s and 3ν s The forwarding overhead borne by ground user 3 is (2+3)ν s =5ν s ;

[0165] Due to limitations in communication examples and cluster size, the set Some users may not be able to obtain the required data. We can categorize ground users based on whether they receive the required data as follows:

[0166]

[0167] Only when λ n When = 1, ground user n successfully received the required data;

[0168] Two aspects are mainly considered for the ground users, one is whether they can receive the required data, and the other is how much cost they spend when receiving the required data. Combining the above two factors, the utility function of the ground users can be designed as:

[0169] η n =λ n [1-α(D n +F n )].

[0170] The joint optimization problem is modeled as follows:

[0171]

[0172] Wherein, wherein C1 and C2 define the task area according to the position of the ground user, C3 represents that the cost of the ground user in the set is less than the cost of the ground user independently downloading data after joining the cluster.

[0173] The relationship between the unmanned aerial vehicle position deployment strategy and the ground user clustering strategy is established as a hierarchical game model, and the hierarchical game model is divided into an outer layer unmanned aerial vehicle deployment game and an inner layer alliance formation game.

[0174] The outer layer unmanned aerial vehicle deployment game is modeled as follows:

[0175]

[0176] Wherein, represents a set of unmanned aerial vehicles, J m represents the strategy space of unmanned aerial vehicle m, represents a space with a height of h corresponding to the task area, and unmanned aerial vehicle m will explore its deployment strategy in the space, u m represents the utility of unmanned aerial vehicle m; the utility u m of unmanned aerial vehicle m is constructed by using marginal utility, and the definition is as follows:

[0177]

[0178] Wherein, J -m =(J1,J2,...J m-1 ,J m+1 ,...,J M ) represents a set of deployment strategies of other unmanned aerial vehicles except unmanned aerial vehicle m, A(J) represents a set of alliance formation strategies of ground users, U(J m ,J -m ,A(J)) represents the total utility of all unmanned aerial vehicles.

[0179] The optimization objective of the outer layer unmanned aerial vehicle deployment game is modeled as:

[0180]

[0181] The inner coalition formation game is modeled as follows:

[0182]

[0183] where a n represents the coalition formation strategy of ground user n, a-n = (a1, a2,..., an-1, an +1 ,...,a N represents the coalition formation strategy set of other users except ground user n, represents the coalition set in the current task area, u n represents the utility of ground user n;

[0184] The optimization objective of the inner coalition formation game is modeled as:

[0185]

[0186] The method adopts the criterion as the Pareto criterion in the ground user coalition formation, that is, assuming that the ground user n is currently in the coalition CO j , if the user wants to join the coalition CO i , the following relationship should be met:

[0187]

[0188] According to the Pareto criterion, the ground user can select the coalition to join to maximize the utility, however, only the ground user who is not effectively covered by the UAV can initiate the coalition and improve the utility, for the ground user who cannot be effectively served by the UAV, it cannot initiate the coalition and does not belong to any coalition in the initial stage, so it cannot join other coalitions to improve the utility through the Pareto criterion. To solve this problem, the Pareto criterion is supplemented, and a resource-constrained Pareto criterion is proposed, which is defined as follows:

[0189] Due to resource constraints, only the ground users in the set can form a coalition, if there is a ground user n who is neither in the set nor belongs to any coalition, when the user wants to join the coalition CO i , the following relationship should be met:

[0190]

[0191] By combining the resource-constrained Pareto criterion with the Pareto criterion, the ground user coalition formation will eventually converge to the Nash equilibrium under the given UAV position.

[0192] The joint optimization algorithm is implemented by using a hierarchical structure, the outer layer is a local space adaptive algorithm, which is used to optimize the position deployment of the unmanned aerial vehicle, and the inner layer is an alliance formation game algorithm, which is used to optimize the alliance formation of the ground users; when the unmanned aerial vehicle explores the position in the outer layer, the ground users in the inner layer will form alliances according to the position of the unmanned aerial vehicle, and the steps are as follows:

[0193] Step A1, a given unmanned aerial vehicle deployment strategy J={J1, J2,...,J M};

[0194] Step A2, randomly select an unmanned aerial vehicle m, and randomly select s deployment strategies from the strategy space , the current strategy is recorded as J m (q1), and the newly selected deployment strategy is recorded as {J1, J2,...,J s};

[0195] Step A3, run the inner layer algorithm to calculate the total utility of the ground users corresponding to different unmanned aerial vehicle deployment strategies, so as to obtain the utility of the unmanned aerial vehicle m corresponding to different deployment strategies, and the unmanned aerial vehicle m updates the deployment position according to the utility under different deployment strategies with a probability, and the utility of the unmanned aerial vehicle is in the form of:

[0196]

[0197] The probability that the unmanned aerial vehicle m updates the deployment strategy to d m is:

[0198]

[0199] Step A4, repeat steps A1-A3 until convergence;

[0200] The inner layer algorithm considers two aspects in the process of alliance formation, one is that the ground users try to join other alliances, in order to enable more ground users to join the alliance, two principles are followed in the processing order, one is to preferentially let the ground users who have not joined the alliance try to join the alliance, and the other is to preferentially let the ground users whose neighbor alliances are less try to join the alliance; the other is the fusion and exchange between alliances, in order to reduce the communication overhead of the ground users as much as possible, it is necessary to let the data service demand of the ground users in the alliance be as similar as possible, therefore, before fusion and exchange, the alliances are matched first, and the alliances with high data similarity are matched in pairs, and then the alliance members between the alliances are exchanged according to the data similarity. The specific steps are as follows:

[0201] First is the ground user trying to join other alliance algorithm:

[0202] Step B1, determine the user set in the alliance and the alliance neighbor user set not in any alliance

[0203] Step B2, generate the ground user processing order, first process the ground users in set , then process the ground users in set , and further sort the ground users in set according to the principle of giving priority to processing ground users with fewer neighbor alliances, so as to finally obtain the ground user processing order;

[0204] Step B3, try to make the ground user join other alliances according to the processing order, and the set of neighbor alliances of the ground user n is P n (CO) = {CO1, CO 2, ,...,CO k}, respectively calculate the utility of the user after joining different neighbor alliances u n (CO) = {η 2n (CO1), η 2n (CO2),..., η 2n (CO k )}, and finally the ground user determines which alliance to join according to the Pareto criterion;

[0205] Step B5, run the fusion and exchange algorithm between alliances after the user completes processing according to the order, and obtain a new alliance structure;

[0206] Step B6, repeat steps B1-B5 until convergence;

[0207] Then the fusion and exchange algorithm between alliances:

[0208] Step C1, each alliance generates a preferred matching list of neighbor alliances according to data similarity;

[0209] Step C2, randomly select a union, match with neighbor alliances according to the preferred matching list, if the neighbor alliance has no current matching pair, the matching is successful, otherwise the neighbor alliance matches with the alliance it prefers more according to its preferred matching list, and rejects the other alliance;

[0210] Step C3, the rejected and rejected alliances update the matching list and delete each other from the matching list;

[0211] Step C4, repeat steps C1-C3 until convergence, and finally obtain several alliance matching pairs;

[0212] Step C5, perform fusion and exchange operations on several alliance matching pairs in turn, first try to fuse the alliance matching pairs into one alliance, if it meets the Pareto criterion, directly output the result, otherwise the alliance matching pairs try to exchange the ground users in equal quantities according to the similarity of business data, and judge whether the exchange is established by the Pareto criterion.

[0213] Example 1

[0214] The first example of this invention is described in detail below. The system simulation is performed using MATLAB, and the parameter design does not affect the generality. The ground mission area is set as a 1km × 1km square area. Figure 2 As shown, this example verifies the convergence of the proposed algorithm under different air-to-ground network sizes, where M is the number of drones and N is the number of ground users. The parameter settings are as follows:

[0215]

[0216]

[0217] Example 2

[0218] The second example of this invention is described in detail below. The system simulation was performed using MATLAB, and the parameter settings do not affect the generality. This embodiment compares different algorithms to verify the superiority of the proposed method.

[0219] like Figure 3 As shown, Figure 3 The performance of the proposed coalition formation algorithm and the traditional coalition formation algorithm was verified in an air-to-ground network scenario, with 3 drones and the number of ground users as shown in the figure. As the number of ground users increases, the proposed algorithm significantly improves the total utility of ground users. This improvement is due to the proposed algorithm's focus on getting more ground users to join the coalition. Furthermore, the more ground users there are, the more coalitions can be formed. During the fusion and exchange process, the proposed algorithm utilizes data similarity to create more coalition matches, further amplifying its advantages. Simulation results show that the proposed joint optimization algorithm improves the total utility of ground users by approximately 10%.

[0220] like Figure 4 As shown, Figure 4 The convergence of the proposed coalition formation algorithm and the traditional coalition formation algorithm was compared in an air-to-ground network with a small number of drones (2 drones and 20 ground users). The simulation results clearly demonstrate the superior convergence of the proposed algorithm when the number of drones is small. This improvement is attributed to the proposed algorithm's optimization of the coalition formation mechanism for ground users, thereby expanding the set space of optimal drone deployment strategies and enabling drones to more efficiently explore the final deployment scheme that maximizes the total utility of ground users.

[0221] The application focuses on the air-ground integrated communication, a leading application scene, studies a joint decision method of multi-unmanned aerial vehicle position deployment and ground user clustering communication for data service transmission needs, and lets more ground users be effectively served by unmanned aerial vehicles in the form of joining clusters and reduce the communication overhead of ground users. In addition, the traditional coalition formation algorithm is improved for the problem of limited coverage of unmanned aerial vehicles. Compared with the traditional coalition formation algorithm, the proposed coalition formation algorithm improves the performance by about 10% under the premise of ensuring the effectiveness of the result. In the case of fewer unmanned aerial vehicles, the convergence number required for the stable solution of the proposed algorithm is significantly reduced.

[0222] It should be noted that, in this document, relational terms such as first and second and the like can be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0223] The above-described embodiments of the present application are not intended to limit the scope of the present application.

Claims

1. A method for drone location deployment and ground user clustering communication oriented to data service transmission requirement, characterized in that, The method comprises the following steps: Step 1, setting the communication related parameters of the unmanned aerial vehicle and the static ground user in the air-ground network, constructing the communication scene, and determining the data transmission rate of the unmanned aerial vehicle to the ground user downlink and the data transmission rate between the ground users; Step 2, combined with the data service demand, the cooperation relationship between the unmanned aerial vehicle deployment and the ground user clustering communication is established; Step 3, the centralized controller respectively issues the unmanned aerial vehicle deployment strategy and the ground user clustering strategy obtained by the optimization algorithm to the unmanned aerial vehicle group and the ground user, and the unmanned aerial vehicle and the ground user execute according to the issued strategy, and finally complete the joint task of unmanned aerial vehicle position deployment and ground user clustering communication under the data service demand.

2. The method for drone location deployment and ground user clustering communication according to data service transmission requirement of claim 1, wherein, The communication related parameters of the unmanned aerial vehicle and the static ground user in the air-ground network are set, and the communication scene is constructed: the air-ground network comprises M unmanned aerial vehicles and N ground users; the unmanned aerial vehicle unicasts the required service data to the ground user according to the service data demand of the ground user, the default channel resource is sufficient in the air-ground communication scene, the same frequency interference in wireless communication is not considered, and the deployment height of the unmanned aerial vehicle is the same; Before the deployment of the unmanned aerial vehicle, the centralized controller collects the communication related parameters in the air-ground network, including the position of the ground user, the transmission power of the unmanned aerial vehicle and the ground user and the electromagnetic environment parameters, so as to construct the air-ground communication scene and obtain the service data transmission rate between the air and the ground and between the ground. 3.The method of claim 2, wherein, Step 2, combined with the data service demand, the cooperation relationship between the unmanned aerial vehicle deployment and the ground user clustering communication is established, including: Step 2-1, determining the ground user set effectively covered by the unmanned aerial vehicle, designing the ground user clustering mechanism, and establishing the ground user utility model; Step 2-2, the relationship between the unmanned aerial vehicle position deployment strategy and the ground user clustering strategy is established as a hierarchical game model; the hierarchical game model is divided into outer layer position unmanned aerial vehicle deployment game and inner layer alliance formation game; Step 2-3, the centralized controller executes the joint optimization algorithm of the unmanned aerial vehicle position deployment and the ground user clustering; the unmanned aerial vehicle Nash equilibrium position deployment strategy and the ground user Nash equilibrium clustering strategy are obtained respectively.

4. The method for drone location deployment and ground user clustering communication according to data service transmission demand of claim 3, wherein, The data transmission rate of the unmanned aerial vehicle to the ground user downlink is determined by the following method: The empty network includes M unmanned aerial vehicles and N ground users; the sets of unmanned aerial vehicles and ground users are respectively denoted as and Suppose the deployment height of the unmanned aerial vehicle is h, the task area is discretized in two dimensions, and the coordinates of the unmanned aerial vehicle m and the ground user n are respectively denoted as m = {x m ,y m ,h}, and n = {x n ,y n ,0}, In the communication scene, there are line of sight (LOS) and non line of sight (NLOS) in the air-ground channel transmission link; the path loss between the unmanned aerial vehicle and the ground user is modeled as a probability weighted average value of the line of sight link loss and the non line of sight link loss; the line of sight link probability between the unmanned aerial vehicle m and the ground user n is defined as follows: where ζ1 and ζ2 are two environmental impact factors, whose sizes depend on the building density, topography, The determination method for the elevation angle of the unmanned aerial vehicle m from the ground user n is as follows: wherein, is the straight-line distance from the UAV m to the ground user n, and the probability of NLoS transmission between the UAV m and the ground user n based on the line-of-sight link probability is: The transmission path loss of the unmanned aerial vehicle m to the ground user n is: Wherein, f m is the carrier frequency of the unmanned aerial vehicle m transmitting data, c0is the speed of light, μ LoS and μ NLoS are the attenuation factors of the line-of-sight link and the non-line-of-sight link, respectively; Assuming that the hardware performance of all unmanned aerial vehicles is the same, the antenna gain of the unmanned aerial vehicle m is: Wherein, θ t is the beam width of the main lobe of the transmitter antenna of each UAV, θ m,n is the elevation angle is the complementary angle of θ t is the main lobe gain of each UAV transmitter, and N0 is the number of UAV antennas. Considering the influence of Los and NLos link on the received signal of the ground user, the signal to noise ratio of the ground user n receiving the signal from the unmanned aerial vehicle m is determined as: where P0 is the transmitting power of the single unmanned aerial vehicle transmitter, σ 2 is the variance of the additive white Gaussian noise; The data transmission rate of the unmanned aerial vehicle m to the ground user n downlink is: Where the channel bandwidth of the unmanned aerial vehicle is fixed, represented as B.

5. The method for drone location deployment and ground user clustering communication according to data service transmission requirement of claim 4, wherein, The data transmission rate between the ground users is determined by the following method: The data distribution mechanism in the cluster relies on the data forwarding among the ground users, so the communication relationship among the ground users needs to be considered; the communication among the ground users considers the NLoS link, assuming that the hardware properties of all ground users are the same, the signal-to-noise ratio of the communication between ground users n1 and n2 is obtained as follows: wherein β0 is the channel gain of a unit energy signal at a distance of 1 meter from the source, P1 is the transmission power of the ground user, is the communication distance between the ground users n1 and n2, and α0 is the loss factor for the communication between the ground users. The data transmission rate between ground users n1 and n2 is: The intra-cluster users relay the data in the form of decode-and-forward, so the ground users receive the data transmission rate from other ground users in the same cluster is the minimum of the data transmission rate of UAV-intra-cluster user and intra-cluster user-intra-cluster user link; since the data transmission rate of UAV-intra-cluster user is greater than R th , as long as the data transmission rate of the intra-cluster user when receiving data is greater than R th , it is considered that the user successfully receives the required data; the relationship between any two ground users n1 and n2 is represented as: Only when n1 and n2, i.e. the two terrestrial users have established a neighborhood relationship.

6. The method for drone location deployment and ground user clustering communication according to data service transmission demand of claim 5, wherein, The set of ground users effectively covered by the unmanned aerial vehicle is determined by the following method: Adopting information rate threshold R th As a criterion to judge whether a user is effectively served, if the information rate of a ground user receiving a certain unmanned aerial vehicle (UAV) signal is greater than a threshold value R th , it is considered that the ground user can be effectively served, i.e., effectively covered by the UAV. If a ground user is effectively covered by multiple UAVs, the ground user will receive the UAV signal with the highest signal-to-noise ratio; and a set of ground users effectively covered by the UAV is defined as: The set of ground users not effectively covered by the unmanned aerial vehicle is:

7. The method for drone location deployment and ground user clustering communication according to data service transmission requirement of claim 6, wherein, The ground user clustering mechanism is determined by the following method: The manner in which the ground users obtain data is related to whether the ground users are covered by the UAVs, and the set in which the ground users can be effectively served, so the ground users in the set can obtain the required data either directly from the UAVs or in a manner of being cluster heads or joining a cluster from the perspective of reducing overhead; and the set in which the ground users can only obtain the required data through the cluster heads in a manner of joining a cluster. A group-purchasing mechanism is introduced, i.e. in-cluster users entrust cluster head to download data, and the cluster head shares the overhead of downloading its required data with in-cluster users; when in-cluster users have similar data requirements with the cluster head, the cluster head acquires the required data with smaller overhead because in-cluster users share part of the overhead of the data required by the cluster head, thus the cluster head can reduce its data acquisition overhead when helping in-cluster users to download data. The overheads required by the cluster members under the group purchase mechanism are as follows: Let ground user n1 be the cluster head, and the cluster of users managed by it be For any member n2 in the cluster For any member n2 in the cluster For any member n2 in the cluster, evaluate the similarity between its data demand and other members in the cluster; specifically, by comparing the data demand of ground user n2 with the data demand of all other members in the cluster one by one, the similarity of each data demand item is quantitatively calculated, and the quantified expression is as follows: wherein, represents the length of the data required by the ground user, represents the degree of similarity of the i-th data requirement of the ground user n2 to the data requirements of other members within the cluster; and represents the degree of similarity of the i-th data requirement of the ground user n2 to the data requirements of other members within the cluster. In the case of data similarity determination, the download overheads to be borne are calculated, and from the fairness point of view, the download overheads generated by the cluster head n1 downloading a certain data should be borne by all ground users in the cluster that need the data, and for any member n2 in the cluster, the download overheads borne by it are: wherein, v d represents the download overhead generated by the cluster head downloading a certain data, l(R n1 ) represents the data demand total length of the cluster , L max represents the maximum data demand length that the cluster head can receive; if l(R n1 )≤L max , it indicates that the cluster head n1 can download all the data required by the users in the cluster, otherwise, the current clustering is invalid, the users in the cluster covered by the unmanned aerial vehicle download the data required by them alone, and the users in the cluster not covered by the unmanned aerial vehicle cannot obtain the data required by them; Based on the neighbor relationship and data demand similarity among the ground users, and combined with the deployment of the unmanned aerial vehicle, the ground users are divided into multiple clusters; the cluster head node is responsible for obtaining data from the unmanned aerial vehicle, and through the cluster forwarding manner, all users in the cluster can receive the data; the cluster users use the flooding mechanism to determine the data transmission path; In the case of cluster data transmission path determination, the forwarding overhead borne by the cluster users is calculated; the forwarding overhead generated by the data forwarding action of the ground user is recorded as Assuming that the ground user n2 is a member of the cluster , combined with the position of the cluster user in the cluster, the calculation method of the forwarding overhead is divided into three cases: the first case when the ground user n2 is the source node, i.e. the cluster head, only responsible for forwarding data, and the forwarding overhead According to the data transmission path, it is borne by the next level member receiving data and the ground user n2; the second case when the ground user n2 is a leaf node, only needs to receive the required data, and the forwarding overhead According to the data transmission path, it is borne by the ground user sending data to it, other ground users receiving data from the same ground user, and the ground user n2; the third case when the ground user n2 does not belong to the above two cases, according to the data transmission path, it needs to receive data from the upper ground user and forward the received data to the lower ground user, and the forwarding overhead is the sum of the first two cases; The forwarding overheads of the ground user n2 are calculated as follows: Wherein, according to the data transmission path, the next level ground user set receiving data from the ground user n2 is represented as LowNode, and the upper level ground user sending data to the ground user n2 and other ground users receiving data from the upper level ground user are represented as OtherNode.

8. The method for drone location deployment and ground user clustering communication according to data service transmission demand of claim 7, wherein, The ground user utility model is determined by the following method: Some users in the set may not be able to obtain the desired data due to communication example and clustering scale limitations, dividing whether the ground user receives the desired data, there are: Only when λ n = 1, the ground user n successfully receives the desired data; For the ground user, two aspects are considered, one is whether it can receive the required data, and the other is how much overhead it spends when receiving the required data, and the utility function of the ground user is expressed as: η n = λ n [1 - α(D n + F n )].

9. The method for drone location deployment and ground user clustering communication according to data service transmission demand of claim 8, wherein, In step 2-2, the hierarchical game model is established by the following method: The outer unmanned aerial vehicle deployment game is modeled as follows: wherein, denotes a set of UAVs, J m denotes the strategy space of UAV m, denotes the strategy space in which UAV m will explore its deployment strategy, u m denotes the utility of UAV m; the utility of UAV m, u m is constructed and defined as follows: wherein J -m = (J1, J2,... J m-1 , J m+1 ,..., J M ) represents a set of deployment strategies of other unmanned aerial vehicles except unmanned aerial vehicle m, A(J) represents a set of ground user coalition formation strategies, U(J m , J -m , A(J)) represents the total utility of all unmanned aerial vehicles; The optimization objective of the outer unmanned aerial vehicle deployment game is modeled as: The inner alliance formation game is modeled as follows: where a n represents the coalition formation strategy of the ground user n, a -n n-1 n+1 N represents the coalition formation strategy set of other users except the ground user n, represents the coalition set in the current task area, u n represents the utility of the ground user n;​​​ The optimization objective of the inner alliance formation game is modeled as: The criterion used in the formation of the ground user coalition is the Pareto criterion, i.e. it is assumed that a ground user n is currently in a coalition CO j If the user wants to join the coalition CO i , the following relation must be satisfied: η 2w (CO i )≥η 2w (CO i \n),w∈CO i \n∧ η 2w (CO j \n)≥η 2w (CO j ),w∈CO j \n The Pareto criterion is supplemented, and a resource-constrained Pareto criterion is proposed, which is defined as follows: Due to resource constraints, only ground users in set are able to form coalitions, and if there exists a ground user n that is neither in set nor belongs to any coalition, the following relation is satisfied when the user wants to join a coalition CO i ​ By combining the resource-constrained Pareto criterion with the Pareto criterion, the ground user alliance formation will finally converge to the Nash equilibrium under the given unmanned aerial vehicle position.

10. The method for drone location deployment and ground user clustering communication according to data service transmission demand of claim 9, wherein, In step 2-3, the centralized controller executes the joint optimization algorithm of unmanned aerial vehicle position deployment and ground user clustering; The unmanned aerial vehicle Nash equilibrium position deployment strategy and the ground user Nash equilibrium clustering strategy are obtained, including: The hierarchical structure is adopted to realize the joint optimization algorithm, the outer layer is a local space adaptive algorithm, which is used to optimize the position deployment of the unmanned aerial vehicle, and the inner layer is an alliance formation game algorithm, which is used to optimize the alliance formation of the ground user; when the unmanned aerial vehicle explores the position in the outer layer, the ground user forms the alliance according to the position of the unmanned aerial vehicle in the inner layer, and the steps are as follows: Step A1, given a drone deployment policy J = {J1, J2,..., J M}; Step A2, randomly pick one drone m and randomly pick s deployment strategies from the strategy space , denote the current strategy as J m (q1), denote the newly picked deployment strategies as {J1, J2,..., J s s}; Step A3, run the inner layer algorithm to calculate the total utility of the ground users corresponding to different UAV deployment strategies, so as to obtain the utility of the UAV m corresponding to different deployment strategies, and the UAV m updates its deployment position according to the utility under different deployment strategies, and the utility of the UAV is in the form of: The drone m updates the deployment policy for d m The probability that Step A4, repeat steps A1-A3 until convergence; The inner layer algorithm considers two aspects in the process of alliance formation. Firstly, ground users try to join other alliances. In order to enable more ground users to join the alliance, two principles are followed in the processing order. On the one hand, ground users who have not joined the alliance are given priority to try to join the alliance. On the other hand, ground users with fewer neighbor alliances are given priority to try to join the alliance. Secondly, the fusion and exchange between alliances are considered. In order to reduce the communication overhead of ground users as much as possible, the data service requirements of ground users in the alliance need to be as similar as possible. Therefore, before fusion and exchange, the alliances are matched first, and alliances with high data similarity are matched in pairs. Then, the members of the alliances are exchanged according to the data similarity. The steps are as follows: Firstly, the algorithm for ground users trying to join other alliances is as follows: Step B1, determining the set of users within the coalition and the set of coalition neighbor users not in any coalition Step B2, generating the ground user processing order, first processing the ground users in set , then processing the ground users in set , and further sorting the ground users in set according to the principle of less-neighborhood-union ground user first processing, thus obtaining the final ground user processing order; Step B3, try to let the ground user join other alliances according to the processing sequence, the set of neighbor alliances of the ground user n is P n (CO) = {CO1, CO 2, ,...,CO k}, respectively calculate the utility of the user after joining different neighbor alliances u n (CO) = {η 2n (CO1), η 2n (CO2),...,η 2n (CO k )}, finally the ground user determines which alliance to join according to the Pareto criterion; Step B5, after the user completes the processing according to the order, run the fusion and exchange algorithm between alliances to obtain a new alliance structure; Step B6, repeat steps B1-B5 until convergence; Secondly, the fusion and exchange algorithm between alliances is as follows: Step C1, each alliance generates a preference matching list of neighbor alliances according to data similarity; Step C2, randomly select a union, match with neighbor alliances according to the preference matching list, if the neighbor alliance has no matching pair at present, the matching is successful, otherwise the neighbor alliance matches with the alliance it prefers more according to its preference matching list, and rejects the other alliance; Step C3, the rejected and rejected alliances update the matching list and delete each other from the matching list; Step C4, repeat steps C1-C3 until convergence, and finally obtain several alliance matching pairs; Step C5, perform fusion and exchange operations on several alliance matching pairs in turn. Firstly, try to fuse the alliance matching pairs into one alliance. If it meets the Pareto criterion, the result can be directly outputted, otherwise the alliance matching pairs try to exchange the ground users in equal quantities according to the business data similarity, and judge whether the exchange is established by the Pareto criterion.

Images