Control node selection method for improving connectivity of communication network of unmanned aerial vehicle

Abstract

The embodiment of the invention provides a control node selection method for improving communication network connectivity of an unmanned aerial vehicle. The invention provides a control node selection method for improving connectivity of a communication network of an unmanned aerial vehicle for a scene of frequent change of network topology of an unmanned aerial vehicle cluster. According to the method, a heuristic genetic algorithm is used for selecting control nodes, an optimal result can be obtained in most scenes, and the algorithm can be triggered when the network connectivity is low by setting a preset minimum value Rth, so that automatic networking adjustment is realized.

Description

Technical Field

[0001] The present application relates to the field of complex network technology, and in particular to a control node selection method for improving the connectivity of unmanned aerial vehicle communication networks. Background Art

[0002] With the widespread application of drones in emergency rescue, logistics distribution, environmental monitoring, and agricultural management, how to ensure the connectivity of drone communication networks has become an important issue that needs to be solved urgently. Drone networks are highly dynamic, with frequently changing topology and easily interrupted communication links. In addition, the communication radius and available bandwidth of each drone are limited. At the same time, due to the battery capacity, long-term or high-frequency communication will accelerate energy consumption, making network maintenance a great challenge. Due to these characteristics, when some nodes or links in the network fail, it may lead to interruption of mission execution, delay of information transmission, or network fragmentation, seriously affecting the collaborative work and command and control of drone groups. Therefore, in a complex and changing environment, maintaining stable communication and efficient information exchange between drones is not only a necessary condition for completing key tasks, but also a basic guarantee for dealing with dynamic environments and emergencies.

[0003] In order to solve these problems, selecting appropriate control nodes in the UAV communication network has become an important means to improve network connectivity. By undertaking information aggregation, routing selection and forwarding tasks, control nodes can effectively reduce the delay of data transmission, optimize communication paths and avoid bottleneck effects. The process of selecting control nodes needs to comprehensively consider multiple factors, such as the centrality and location of nodes and the dynamically changing network topology, to ensure that the control nodes can not only work efficiently, but also provide sufficient redundant paths when some nodes fail. Control node selection methods often rely on centrality analysis and distributed control algorithms to enhance the robustness and fault tolerance of the network. Through the analysis of indicators such as betweenness centrality and degree centrality, the nodes that play a key role in the network can be found, and the distributed algorithm can enable the network to achieve adaptive adjustment without centralized control, making the communication and control between UAVs more flexible.

[0004] In order to cope with the high dynamics of drone networks, the selection of control nodes must have real-time adjustment capabilities, and be able to quickly update the selection of nodes and paths when the network structure changes. At the same time, redundancy design must be considered to deal with the failure of certain nodes or network fragmentation. Therefore, the study of control node selection methods for improving the connectivity of drone communication networks has important theoretical significance and application value. It can not only optimize the collaborative work efficiency of drone groups, but also lay a technical foundation for communication support for complex tasks such as emergency rescue. Summary of the invention

[0005] The purpose of the present invention is to provide a control node selection method for improving the connectivity of an unmanned aerial vehicle communication network.

[0006] In order to achieve the above-mentioned object of the invention, an embodiment of the present invention provides a control node selection method for improving the connectivity of a UAV communication network, the method comprising:

[0007] Step S1, obtaining the communication relationship between cluster machines at the current moment;

[0008] Step S2, constructing a network model according to the communication relationship between cluster machines, and assigning values ​​to the edges in the network model according to the communication success probability of each link;

[0009] Step S3, using a heuristic genetic algorithm to select a control node in the communication network, and the ground station establishes a communication connection with the control node;

[0010] Step S4: The ground station monitors the changes in network connectivity in real time. When the network connectivity is lower than a predetermined value R th When , the above control node selection process is repeated.

[0011] In some embodiments, the step S1, obtaining the communication relationship between cluster machines at the current moment, includes:

[0012] Each drone broadcasts to other drones in the cluster to determine which nodes have communication connections with itself, and records the packet loss rate and delay information of the communication link.

[0013] In some embodiments, the step S2, constructing a network model according to the communication relationship between cluster machines, and assigning values ​​to edges in the network model according to the communication success probability of each link, includes:

[0014] The network model of the UAV cluster communication network is equivalent to an undirected graph G(V,E), in which each node corresponds to a UAV. If there is a communication relationship between two nodes, there is a path between the nodes;

[0015] Each edge corresponds to a communication connection and is assigned the probability of success of the corresponding communication connection.

[0016] In some embodiments, the step S3, using a heuristic genetic algorithm to select a control node in the communication network, and the ground station establishing a communication connection with the control node, includes:

[0017] For a UAV communication network with N nodes, K nodes are selected as primary control nodes, which are directly controlled by the ground station, including:

[0018] Step S301, encoding and initializing the population;

[0019] Step S302, calculating individual fitness;

[0020] Step S303, selecting a parent individual;

[0021] Step S304, a strategy for generating offspring based on a greedy algorithm;

[0022] Step S305, calculating the fitness of the new individual.

[0023] In some embodiments, the step S301, encoding and initializing the population, includes:

[0024] The length of individual gene is K, and each individual x i Contains K drone node numbers in [1, 2, 3, ..., N], representing the K control nodes selected by the individual;

[0025] The individuals are generated sequentially, so that the number of times each drone node gene appears in the initial population is approximately and not less than 2 times. Specifically, the frequency of each gene is determined by the following formula:

[0026]

[0027] Among them, N is the number of drone nodes, K is the required control node, is the number of all solutions to the problem, is the proportional coefficient of the initial population solution. This formula ensures that each gene appears at least twice. The size of the population is f·d, including f groups of solutions, and the number of solutions in each group is d;

[0028] After determining the frequency and population size of each gene, the drone node number needs to be assigned to each solution;

[0029] For the first set of solutions, [1,2,…,K] is the first solution, [K+1,K+2,…2K] is the second solution… Repeat this process until d solutions are obtained;

[0030] For the second set of solutions, [1,3,…,2K-1] is the first solution, [2K+1,2K+3,…,4K-1] is the second solution, and so on. Repeat this process until d solutions are obtained.

[0031] In some embodiments, the step S302, calculating individual fitness, includes:

[0032] Construct the following fitness function

[0033] f(x i )=P[Gisconnected]

[0034] That is, select x i The probability of full connectivity of the UAV network after acting as a control node.

[0035] In some embodiments, the step S303, selecting a parent individual, includes:

[0036] Randomly sample 2 non-repeating individuals from the population and use them as two parents to generate new individuals.

[0037] In some embodiments, the step S304, the offspring generation strategy based on the greedy algorithm, includes:

[0038] This strategy merges the two parents selected in the previous step, removes duplicate genes, generates a temporary offspring of length K' (K'>K), and then removes the redundant K'-K genes through a greedy algorithm;

[0039] Gene removal follows two strategies:

[0040] 1) Genes present in both parents are retained;

[0041] 2) Remove the gene that contributes the least to improving network connectivity at each step;

[0042] When calculating the gene with the smallest contribution, each gene needs to be removed in turn and the network connectivity after removal is compared. If the network connectivity after the gene is removed is the largest, it is considered that the gene contributes the least to the improvement of network connectivity.

[0043] Through K'-K steps of removal, the length of the offspring is shortened to K.

[0044] In some embodiments, the step S305, calculating the fitness of the new individual, includes:

[0045] After generating the offspring of this iteration, it is necessary to determine whether the offspring is a duplicate of an existing individual in the population;

[0046] If there is a repetition, it will directly enter the next round of iteration and increase the number of iterations by one;

[0047] If the individual does not overlap with the existing individuals in the population, then determine whether the individual is better than the individual with the worst fitness in the population. If it is better than the individual with the worst fitness, then replace it, update the population and reset the number of iterations to zero. If it is worse than the individual with the worst fitness, then do not update the population and increase the number of iterations by one.

[0048] If the number of iterations is greater than N, the program is terminated and the optimal solution in the current population is output.

[0049] In some embodiments, in step S4, the ground station monitors the changes in network connectivity in real time, and when the network connectivity is lower than a predetermined value R th When , the above control node selection process is repeated, including:

[0050] The ground station monitors the changes in network connectivity in real time. When the network connectivity is lower than the preset value R th When the control node selection process is repeated, the above control node selection process is repeated to meet the demand for real-time adjustment of control node selection under the high dynamics of the UAV network, and the node and path selection can be quickly updated when the network structure changes.

[0051] The beneficial effects of the above embodiments of the present invention include:

[0052] Aiming at the scenario where the topology of the drone cluster network changes frequently, the present invention provides a control node selection method for improving the connectivity of the drone communication network. The method uses a heuristic genetic algorithm to select control nodes, which can obtain the optimal result in most scenarios. By setting a predetermined minimum value R_th, the algorithm can be triggered when the network connectivity is low, thereby realizing automatic networking adjustment. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The drawings illustrate generally, by way of example and not limitation, various embodiments discussed herein.

[0054] Figure 1 Flowchart for selecting control nodes for UAV communication network;

[0055] Figure 2 It is the flow chart of heuristic genetic algorithm;

[0056] Figure 3 This is the network topology example 1 (N=20, K=2);

[0057] Figure 4 This is an example of a network topology diagram (N=10, K=2). DETAILED DESCRIPTION

[0058] In order to enable a more detailed understanding of the features and technical contents of the embodiments of the present application, the implementation of the embodiments of the present application is described in detail below in conjunction with the accompanying drawings. The attached drawings are for reference only and are not used to limit the embodiments of the present application.

[0059] In the embodiments of the present application, it should be noted that, unless otherwise specified and limited, the term "connection" should be understood in a broad sense. For example, it can be an electrical connection or a connection between two components. It can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to the specific circumstances.

[0060] It should be noted that the terms "first\second\third" involved in the embodiments of the present application are only used to distinguish similar objects, and do not represent a specific order for the objects. It is understandable that the specific order or sequence of "first\second\third" can be interchanged where permitted. It should be understood that the objects distinguished by "first\second\third" can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein.

[0061] The embodiment of the present invention provides a control node selection method for improving the connectivity of a UAV communication network. Figure 1 As shown, the method includes:

[0062] Step S1, obtaining the communication relationship between cluster machines at the current moment.

[0063] Step S2, constructing a network model according to the communication relationship between cluster machines, and assigning values ​​to the edges in the network model according to the communication success probability of each link.

[0064] Step S3, using a heuristic genetic algorithm to select a control node in the communication network, and the ground station establishes a communication connection with the control node.

[0065] Step S4: The ground station monitors the changes in network connectivity in real time. When the network connectivity is lower than a predetermined value R th When , the above control node selection process is repeated.

[0066] In some embodiments, step S1, obtaining the communication relationship between cluster machines at the current moment, includes:

[0067] Each drone broadcasts to other drones in the cluster to determine which nodes have communication connections with itself, and records the packet loss rate and delay information of the communication link.

[0068] In some embodiments, step S2, building a network model according to the communication relationship between cluster machines, and assigning values ​​to edges in the network model according to the communication success probability of each link, includes:

[0069] The network model of the drone cluster communication network is equivalent to an undirected graph G(V,E), in which each node corresponds to a drone. If there is a communication relationship between two nodes, there is a link between the nodes.

[0070] Each edge corresponds to a communication connection and is assigned the probability of success of the corresponding communication connection.

[0071] In some embodiments, step S3, using a heuristic genetic algorithm to select a control node in the communication network, and the ground station establishes a communication connection with the control node, includes:

[0072] For a UAV communication network with N nodes, K nodes are selected as primary control nodes and directly controlled by the ground station, such as Figure 2 As shown, specifically including:

[0073] Step S301, encoding and initializing the population.

[0074] Step S302, calculating individual fitness.

[0075] Step S303, selecting a parent individual.

[0076] Step S304, a strategy for generating offspring based on a greedy algorithm.

[0077] Step S305, calculating the fitness of the new individual.

[0078] In some embodiments, step S301, encoding and initializing a population, includes:

[0079] The length of individual gene is K, and each individual x i Contains K drone node numbers in [1, 2, 3, ..., N], representing the K control nodes selected by the individual.

[0080] The individuals are generated sequentially, so that the number of times each drone node gene appears in the initial population is approximately and not less than 2 times. Specifically, the frequency of each gene is determined by the following formula:

[0081]

[0082] Among them, N is the number of drone nodes, K is the required control node, is the number of all solutions to the problem, is the proportional coefficient of the initial population solution. This formula ensures that each gene appears at least twice. The size of the population is f·d, including f groups of solutions, and the number of solutions in each group is d.

[0083] After determining the frequency of occurrence and population size of each gene, the drone node numbers need to be assigned to each solution.

[0084] For the first set of solutions, [1,2,…,K] is the first solution, [K+1,K+2,…2K] is the second solution, and so on. Repeat this process until d solutions are obtained.

[0085] For the second set of solutions, [1,3,…,2K-1] is the first solution, [2K+1,2K+3,…,4K-1] is the second solution, and so on. Repeat this process until d solutions are obtained.

[0086] In some embodiments, step S302, calculating individual fitness, includes:

[0087] Construct the following fitness function

[0088] f(x i )=P[Gisconnected]

[0089] That is, select x i The probability of full connectivity of the UAV network after acting as a control node.

[0090] In some embodiments, step S303, selecting a parent individual, includes:

[0091] Randomly sample 2 non-repeating individuals from the population and use them as two parents to generate new individuals.

[0092] In some embodiments, step S304, the offspring generation strategy based on the greedy algorithm includes:

[0093] This strategy merges the two parents selected in the previous step, removes duplicate genes, generates a temporary offspring of length K' (K'>K), and then removes the redundant K'-K genes through a greedy algorithm.

[0094] Gene removal follows two strategies:

[0095] 1) Genes present in both parents are retained.

[0096] 2) At each step, remove the gene that contributes the least to improving network connectivity.

[0097] When calculating the gene with the smallest contribution, each gene needs to be removed in turn and the network connectivity after removal is compared. If the network connectivity of the gene is the largest after removal, it is considered that the gene contributes the least to the improvement of network connectivity.

[0098] Through K'-K steps of removal, the length of the offspring is shortened to K.

[0099] In some embodiments, step S305, calculating the fitness of the new individual, includes:

[0100] After generating the offspring of this iteration, it is necessary to determine whether the offspring is a duplicate of an existing individual in the population.

[0101] If there is a repetition, it will directly enter the next round of iteration and increase the number of iterations by one.

[0102] If it does not overlap with the existing individuals in the population, determine whether the individual is better than the individual with the worst fitness in the population. If it is better than the individual with the worst fitness, replace it, update the population and reset the number of iterations to zero. If it is worse than the individual with the worst fitness, do not update the population and increase the number of iterations by one.

[0103] If the number of iterations is greater than N, the program is terminated and the optimal solution in the current population is output.

[0104] In some embodiments, step S4, the ground station monitors the network connectivity changes in real time, and when the network connectivity is lower than a predetermined value R th When , the above control node selection process is repeated, including:

[0105] The ground station monitors the changes in network connectivity in real time. When the network connectivity is lower than the preset value R th When the control node selection process is repeated, the above control node selection process is repeated to meet the demand for real-time adjustment of control node selection under the high dynamics of the UAV network, and the node and path selection can be quickly updated when the network structure changes.

[0106] The above embodiments of the present invention are described below with reference to specific cases.

[0107] Assume that a drone cluster consists of 10 drones. Due to communication restrictions, the ground station needs to select two nodes as primary control nodes, and the remaining drones can be directly or indirectly controlled by the control nodes via relays. This example provides a control node selection method for improving the connectivity of drone communication networks.

[0108] like Figure 1 As shown, the method comprises the following steps:

[0109] The communication relationship between cluster machines at the current moment obtained in step S1;

[0110] Each drone broadcasts to other drones in the cluster to determine which nodes have communication connections with itself, and records the packet loss rate, latency and other information of the communication link. Taking a certain moment in the flight as an example, the communication connection status at that moment is as follows:

[0111] Drone No. 1 has a communication connection with Drone No. 2 and Drone No. 10; Drone No. 2 has a communication connection with Drone No. 1 and Drone No. 3; Drone No. 3 has a communication connection with Drone No. 2 and Drone No. 4; Drone No. 4 has a communication connection with Drone No. 3 and Drone No. 5; Drone No. 5 has a communication connection with Drone No. 4 and Drone No. 6; Drone No. 6 has a communication connection with Drone No. 5 and Drone No. 7; Drone No. 7 has a communication connection with Drone No. 6 and Drone No. 8; Drone No. 8 has a communication connection with Drone No. 7 and Drone No. 9; Drone No. 9 has a communication connection with Drone No. 8 and Drone No. 10; Drone No. 10 has a communication connection with Drone No. 9 and Drone No. 1. The 10 drones form a ring network.

[0112] Step S2 constructs a network model based on the communication relationship between cluster machines, and assigns values ​​to the edges in the network model based on the communication success probability of each link. Figure 3As shown in the figure, the network model of the drone cluster communication network is equivalent to an undirected graph G(V,E), in which each node corresponds to a drone. If there is a communication relationship between two nodes, there is an edge between the nodes. Each edge corresponds to a communication connection and is assigned the probability of a successful communication connection. To simplify the calculation demonstration process, in this example, it is assumed that the connection probability between drones is 0.9 and the connection probability between drones and ground stations is 1.

[0113] Step S3 uses a heuristic genetic algorithm to select a control node in the communication network. The ground station establishes a communication connection with the control node. For a UAV communication network with 10 nodes, two of the nodes are selected as primary control nodes, which are directly controlled by the ground station. Figure 2 As shown, including:

[0114] Step S301: Encoding and initializing population

[0115] In this problem, the length of individual genes is 2, and each individual x i Contains 2 drone node numbers in [1, 2, 3..., 10], representing the 2 control nodes selected by the individual. This method uses the method of sequentially generating individuals, so that the number of times each drone node number (gene) appears in the initial population is approximately and not less than 2 times. Specifically, the frequency of each gene is determined by the following formula:

[0116]

[0117] The available population size is

[0118]

[0119] After determining the frequency of occurrence of each gene and the population size, the drone node numbers need to be assigned to each solution. For the first set of solutions, [1,2] is the first solution, [3,4] is the second solution, and so on. Repeat this process until 5 solutions are obtained. For the second set of solutions, [1,3] is the first solution, [5,7] is the second solution, and so on. Repeat this process until 5 solutions are obtained.

[0120] It can be obtained that the initial population is {[1,2],[3,4],[5,6],[7,8],[9,10],[1,3],[5,7],[9,2],[4,6],[8,10]}.

[0121] Step S302: Calculate individual fitness

[0122] Construct the following fitness function

[0123] f(x i )=P[Gisconnected]

[0124] That is, select x i The probability of full connectivity of the UAV network after acting as a control node.

[0125] Calculation shows that the fitness of each individual in the population is {0.775, 0.775, 0.775, 0.775, 0.775, 0.805, 0.805, 0.805, 0.805}.

[0126] Step S303: Select parent individuals

[0127] The individual selection strategy adopts random selection, that is, randomly sampling 2 non-repetitive individuals from the population and making them serve as two parents to generate new individuals.

[0128] Step S304: Offspring generation strategy based on greedy algorithm

[0129] This strategy merges the two parents selected in the previous step, removes duplicate genes, generates a temporary child of length K' (K'>2), and then removes the redundant K'-2 genes through a greedy algorithm. Gene removal follows the following two strategies: 1) retain genes that appear in both parents; 2) remove a gene that contributes the least to improving network connectivity at each step.

[0130] When calculating the gene with the lowest contribution, each gene needs to be removed in turn and the network connectivity after removal is compared. If the network connectivity after the gene is removed is the largest, it is considered that the gene contributes the least to the improvement of network connectivity. By removing K'-2 steps, the offspring length can be shortened to 2.

[0131] Taking [3,4], [1,3] as an example, first merge the two parents and remove duplicate genes to get [1, 3, 4]. The length of this temporary offspring is 3, and one more gene needs to be removed. "3" appears in both parents, so it needs to be retained. Therefore, compare the contribution of "1" and "4" to improving network connectivity, and remove genes with small contributions.

[0132] Assuming that "1" is removed, the offspring obtained is [3, 4], and its fitness is calculated to be 0.775; assuming that "4" is removed, the offspring obtained is [1, 3], and its fitness is calculated to be 0.805. Therefore, if "4" is removed in this round, the offspring obtained from this crossover is [1, 3].

[0133] Taking [1, 2], [9, 10] as an example, first merge the two parents and remove duplicate genes to obtain [1, 2, 9, 10]. The length of this temporary offspring is 4, and two more genes need to be removed.

[0134] In the first round, it is necessary to compare the contributions of "1", "2", "9" and "10" to improving network connectivity, and remove genes with small contributions. Assuming that "1" is removed, the offspring is [2, 9, 10], and its fitness is calculated to be 0.842; assuming that "2" is removed, the offspring is [1, 9, 10], and its fitness is calculated to be 0.813; assuming that "9" is removed, the offspring is [1, 2, 10], and its fitness is calculated to be 0.813; assuming that "10" is removed, the offspring is [1, 2, 9], and its fitness is calculated to be 0.842. Therefore, this round randomly chooses to remove "1" or "10". This round removes "1" and obtains the temporary offspring [2, 9, 10].

[0135] In the second round, we need to compare the contributions of "2", "9" and "10" to improving network connectivity, and remove genes with small contributions. Assuming that "2" is removed, the offspring is [9, 10], and its fitness is calculated to be 0.775; assuming that "9" is removed, the offspring is [2, 10], and its fitness is calculated to be 0.805; assuming that "10" is removed, the offspring is [2, 9], and its fitness is calculated to be 0.826. Therefore, if "10" is removed in this round, the final offspring is [2, 9].

[0136] Step S305: Calculate the fitness of the new individual

[0137] After generating the offspring of this iteration, it is necessary to determine whether the offspring is repeated with the individuals in the population. If it is repeated, it will directly enter the next round of iteration and increase the number of iterations by one; if it is not repeated with the individuals in the population, it will be determined whether the individual is better than the individual with the worst fitness in the population. If it is better than the individual with the worst fitness, it will be replaced, the population will be updated and the number of iterations will be reset to zero. If it is worse than the individual with the worst fitness, the population will not be updated and the number of iterations will be increased by one. If the number of iterations is greater than 10, the program will be terminated and the optimal solution in the current population will be output.

[0138] Taking the offspring [1,3] obtained in the previous step as an example, [1,3] already exists in the initial population, so the population is not updated and the next round of selection is entered directly.

[0139] Take the offspring [2,9] obtained in the previous step as an example. This individual does not exist in the initial population. Compare the fitness of the offspring [2,9], 0.826, with the minimum fitness of 0.775 in the population. Since 0.826>0.775, randomly replace an individual with a fitness of 0.775 in the population with [2,9]. The initial population is {[1,2],[3,4],[5,6],[7,8],[9,10],[1,3],[5,7],[9,2],[4,6],[8,10]}, and the fitness is {0.775, 0.775, 0.775, 0.775, 0.775, 0.805, 0.805, 0.805, 0.805}; The population after replacement is {[1,2],[3,4],[5,6],[7,8],[2,9],[1,3],[5,7],[9,2],[4,6],[8,10]}, and the fitness is {0.775, 0.775, 0.775, 0.775, 0.826, 0.805, 0.805, 0.805, 0.805, 0.805}

[0140] Repeating the above crossover, replacement, and population update process will increase the average fitness of individuals in the population, and individuals will not be repeated, ensuring the diversity of individuals in the population. After the optimal solution value remains unchanged in 10 consecutive iterations, the program is terminated and the optimal solution is output, namely [1,6], [2,7], [3,8], [4,9] and [5,10], with a fitness of 0.844.

[0141] Step S4: The ground station monitors the changes in network connectivity in real time. When the network connectivity is lower than the preset value R th =0.5, the above control node selection process is repeated to meet the demand for real-time adjustment of control node selection under the high dynamics of the UAV network, and to quickly update the selection of nodes and paths when the network structure changes.

[0142] The algorithm is also applicable to networks with more complex topologies (such as Figure 4 , “2” and “19” are the optimal solutions for K=2), which can quickly solve complex network problems.

[0143] The technical solutions described in the embodiments of the present application can be combined arbitrarily without conflict.

[0144] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A control node selection method for improving the connectivity of unmanned aerial vehicle communication networks, characterized in that: The method comprises: Step S1, obtaining the communication relationship between cluster machines at the current moment; Step S2, constructing a network model according to the communication relationship between cluster machines, and assigning values ​​to the edges in the network model according to the communication success probability of each link; Step S3, using a heuristic genetic algorithm to select a control node in the communication network, and the ground station establishes a communication connection with the control node; Step S4: The ground station monitors the changes in network connectivity in real time. When the network connectivity is lower than a predetermined value R th When , the above control node selection process is repeated.

2. The control node selection method for improving the connectivity of UAV communication network according to claim 1 is characterized in that: The step S1, obtaining the communication relationship between cluster machines at the current moment, includes: Each drone broadcasts to other drones in the cluster to determine which nodes have communication connections with itself, and records the packet loss rate and delay information of the communication link.

3. The control node selection method for improving the connectivity of UAV communication network according to claim 2 is characterized in that: The step S2, constructing a network model according to the communication relationship between cluster machines, and assigning values ​​to edges in the network model according to the communication success probability of each link, includes: The network model of the UAV cluster communication network is equivalent to an undirected graph G(V,E), in which each node corresponds to a UAV. If there is a communication relationship between two nodes, there is a path between the nodes; Each edge corresponds to a communication connection and is assigned the probability of success of the corresponding communication connection.

4. The control node selection method for improving the connectivity of UAV communication network according to claim 3 is characterized in that: The step S3, using a heuristic genetic algorithm to select a control node in the communication network, and the ground station establishes a communication connection with the control node, includes: For a UAV communication network with N nodes, K nodes are selected as primary control nodes, which are directly controlled by the ground station, including: Step S301, encoding and initializing the population; Step S302, calculating individual fitness; Step S303, selecting a parent individual; Step S304, a strategy for generating offspring based on a greedy algorithm; Step S305, calculating the fitness of the new individual.

5. The control node selection method for improving the connectivity of UAV communication network according to claim 4 is characterized in that: The step S301, encoding and initializing the population, includes: The length of individual gene is K, and each individual x i Contains K drone node numbers in [1, 2, 3, ..., N], representing the K control nodes selected by the individual; The individuals are generated sequentially, so that the number of times each drone node gene appears in the initial population is approximately and not less than 2 times. Specifically, the frequency of each gene is determined by the following formula: Among them, N is the number of drone nodes, K is the required control node, is the number of all solutions to the problem, is the proportional coefficient of the initial population solution. This formula ensures that each gene appears at least twice. The size of the population is f·d, including f groups of solutions, and the number of solutions in each group is d; After determining the frequency and population size of each gene, the drone node number needs to be assigned to each solution; For the first set of solutions, [1,2,…,K] is the first solution, [K+1,K+2,…2K] is the second solution… Repeat this process until d solutions are obtained; For the second set of solutions, [1,3,…,2K-1] is the first solution, [2K+1,2K+3,…,4K-1] is the second solution, and so on. Repeat this process until d solutions are obtained.

6. The control node selection method for improving the connectivity of UAV communication network according to claim 5 is characterized in that: The step S302, calculating individual fitness, includes: Construct the following fitness function f(x i )=P[Gisconnected] That is, select x i The probability of full connectivity of the UAV network after acting as a control node.

7. The control node selection method for improving the connectivity of UAV communication network according to claim 6, characterized in that: The step S303, selecting a parent individual, includes: Randomly sample 2 non-repeating individuals from the population and use them as two parents to generate new individuals.

8. The control node selection method for improving the connectivity of UAV communication network according to claim 7, characterized in that: The step S304, the offspring generation strategy based on the greedy algorithm, includes: This strategy merges the two parents selected in the previous step, removes duplicate genes, generates a temporary offspring of length K' (K'>K), and then removes the redundant K'-K genes through a greedy algorithm; Gene removal follows two strategies: 1) Genes present in both parents are retained; 2) Remove the gene that contributes the least to improving network connectivity at each step; When calculating the gene with the smallest contribution, each gene needs to be removed in turn and the network connectivity after removal is compared. If the network connectivity after the gene is removed is the largest, it is considered that the gene contributes the least to the improvement of network connectivity. Through K'-K steps of removal, the length of the offspring is shortened to K.

9. The control node selection method for improving the connectivity of UAV communication network according to claim 8, characterized in that: The step S305, calculating the fitness of the new individual, includes: After generating the offspring of this iteration, it is necessary to determine whether the offspring is a duplicate of an existing individual in the population; If there is a repetition, it will directly enter the next round of iteration and increase the number of iterations by one; If the individual does not overlap with the existing individuals in the population, then determine whether the individual is better than the individual with the worst fitness in the population. If it is better than the individual with the worst fitness, then replace it, update the population and reset the number of iterations to zero. If it is worse than the individual with the worst fitness, then do not update the population and increase the number of iterations by one. If the number of iterations is greater than N, the program is terminated and the optimal solution in the current population is output.

10. The control node selection method for improving the connectivity of UAV communication network according to claim 4, characterized in that: In step S4, the ground station monitors the changes in network connectivity in real time, and when the network connectivity is lower than a predetermined value R th When , the above control node selection process is repeated, including: The ground station monitors the changes in network connectivity in real time. When the network connectivity is lower than the preset value R th When the control node selection process is repeated, the above control node selection process is repeated to meet the demand for real-time adjustment of control node selection under the high dynamics of the UAV network, and the node and path selection can be quickly updated when the network structure changes.

Images