An AoI-sensitive data collection method and system in multiple uavs

Abstract

The application provides an AoI-sensitive data collection method and system in multiple unmanned aerial vehicles, relates to the technical field of mobile communication, and comprises the following steps: collecting the number and position information of ground sensors and initializing the number and position of the ground sensors; determining the number of data collection points and the position coordinates of each data collection point, and establishing SN-CP association; clustering the data collection points according to the position coordinates of the data collection points and the SN-CP association, and then establishing CP-UAV association; optimizing the unmanned aerial vehicle trajectory under each data collection point according to the coordinates of the data collection points and the CP-UAV association, so as to minimize the average AoI and the maximum AoI of each sensor data, and complete data collection; and the data collection task is completed while the freshness of the sensor information is ensured.

Description

Technical Field

[0001] This invention relates to the field of mobile communication technology, and specifically to an AoI-sensitive data collection method and system for multiple unmanned aerial vehicles (UAVs). Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] With the development of artificial intelligence, big data, cloud computing, and mobile edge computing, today's mobile applications are becoming more sensitive to latency. Drones are playing an increasingly important role in fields such as military, disaster relief, and healthcare.

[0004] In recent years, with the increasing demand for global wireless communication network coverage, the combination of unmanned aerial vehicles (UAVs) and mobile networks can support UAV communication in a low-cost and highly mobile manner, and also provide the possibility of establishing new dedicated air-to-ground communication links. When natural disasters such as epidemics, floods and earthquakes occur, UAVs can be deployed to data collection scenarios to improve the freshness of information and reduce disaster losses.

[0005] Unmanned aerial vehicles (UAVs), deployed as mobile communication base stations in wireless communication networks, are widely used in military, disaster relief, medical, and other fields due to their advantages such as low cost, high flexibility, and flexible deployment. For example, ground base stations in disaster areas are often destroyed, failing to provide timely communication services, which hinders rescue efforts. In urban traffic hotspots such as parties and concerts, traditional ground base station coverage often cannot meet demand, leaving user equipment without signal. Given these issues, UAV-assisted data collection in wireless sensor networks has attracted widespread attention. UAVs are more flexible and mobile than ground base stations, providing reliable communication for small and energy-constrained ground user equipment and enabling more timely data collection for ground sensor nodes (SNs) to ensure data freshness.

[0006] Drones have a wide range of applications and many advantages in mobile communications. However, due to their small size, limited payload and energy storage capacity, drones cannot collect data for extended periods. The sequential uploading of sensor data and the cumbersome paths the drone takes between data collection points (CPs) result in low data freshness. Furthermore, in scenarios with high timeliness requirements, the limited endurance of drones makes it difficult for a single drone to complete data collection while maintaining sensor information freshness, especially in large areas with densely packed sensors. Summary of the Invention

[0007] To address the aforementioned issues, this invention proposes an AoI-sensitive data collection method and system for multi-UAVs. The method uses AoI to measure data freshness, minimizes the maximum and average AoI of the data in the sensors, and introduces an association and planning strategy from start to finish. Through an iterative three-step process, the two AoIs of the data in the sensors are optimized.

[0008] According to some embodiments, the present invention adopts the following technical solution:

[0009] A method for AoI-sensitive data collection in multi-UAV systems includes:

[0010] Collect information on the number and location of ground sensors, and initialize the number and location of ground sensors;

[0011] Determine the number of data collection points and the location coordinates of each data collection point, and establish the association between the sensor and the data collection points;

[0012] Based on the location coordinates of the data collection points and the association between the sensors and the data collection points, the data collection points are clustered, and then the association between the data collection points and the drones is established.

[0013] Based on the coordinates of the data collection points and the correlation between the data collection points and the drone, the drone trajectory at each data collection point is optimized to minimize the average AoI and maximum AoI of each sensor data, thus completing the data collection.

[0014] According to some embodiments, the present invention adopts the following technical solution:

[0015] An AoI-sensitive data collection system for multiple unmanned aerial vehicles includes:

[0016] The data acquisition module is used to collect information on the number and location of ground sensors, and to initialize the number and location of the ground sensors.

[0017] The data association module is used to determine the number of data collection points and the location coordinates of each data collection point, and to establish the association between the sensor and the data collection point. Based on the location coordinates of the data collection point and the association between the sensor and the data collection point, the data collection point is clustered, and then the association between the data collection point and the drone is established.

[0018] The optimization module is used to optimize the drone trajectory at each data acquisition point based on the coordinates of the data acquisition point and the correlation between the data acquisition point and the drone, in order to minimize the average AoI and maximum AoI of each sensor data, thus completing the data collection.

[0019] According to some embodiments, the present invention adopts the following technical solution:

[0020] A computer-readable storage medium storing a plurality of instructions adapted for loading and execution by a processor of a terminal device of the aforementioned AoI-sensitive data collection method in multiple unmanned aerial vehicles (UAVs).

[0021] According to some embodiments, the present invention adopts the following technical solution:

[0022] A terminal device includes a processor and a computer-readable storage medium, the processor being configured to implement various instructions; the computer-readable storage medium being configured to store multiple instructions adapted to be loaded and executed by the processor as described in the AoI-sensitive data collection method for multiple unmanned aerial vehicles.

[0023] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0024] This invention utilizes multiple drones to collaboratively collect data, improving data freshness. AoI (Aspect-Oriented Intelligence) is used to measure data freshness, primarily including the data upload time from sensors, drone flight time, and data unloading time. An AoI-sensitive data collection problem is established within a multi-drone-assisted wireless sensor network to simultaneously minimize both the maximum and average AoI of the data in the sensors. A start-to-end association and planning strategy is introduced, optimizing both AoIs of the sensor data through an iterative three-step process. Under limited endurance, an improved ant colony algorithm is employed to optimize the drone flight trajectories. This strategy optimizes both the maximum and average AoI of the sensor networks (SNs), thus improving data freshness. Attached Figure Description

[0025] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0026] Figure 1 This is a schematic diagram of a drone data collection scenario in an embodiment of the present invention;

[0027] Figure 2 This is a flowchart illustrating the implementation of the method in an embodiment of the present invention. Detailed implementation method:

[0028] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0029] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0030] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0031] AOI: AoI is an important indicator for measuring data freshness, reflecting the timeliness of information. In this invention, AoI is defined as the time it takes for data to travel from the sensor to the data center.

[0032] Example 1

[0033] One embodiment of the present invention provides a method for collecting AoI-sensitive data in multiple unmanned aerial vehicles, comprising:

[0034] Step 1: Collect the number and location information of the ground sensors, and initialize the number and location of the ground sensors;

[0035] Step 2: Determine the number of data acquisition points and the location coordinates of each data acquisition point, and establish the association between the sensor and the data acquisition points;

[0036] Step 3: Based on the location coordinates of the data collection points and the association between the sensors and the data collection points, cluster the data collection points and then establish the association between the data collection points and the drone;

[0037] Step 4: Based on the coordinates of the data acquisition points and the correlation between the data acquisition points and the drone, optimize the drone trajectory at each data acquisition point to minimize the average AoI and maximum AoI of each sensor data, thus completing the data collection.

[0038] As one embodiment, the specific method implementation process is as follows:

[0039] S10: Initialize the number and location of ground sensors;

[0040] The position of the ground sensor changes over time. A three-dimensional coordinate system is established, with the origin as the data center. Initially, the number of UAVs is 1. AoI (Aspect-Oriented Identification) is introduced to evaluate the data collection effectiveness. A multi-UAV assisted data collection problem in wireless sensor networks is proposed to minimize the maximum and average AoI of the data from the sensors. Let... w l =(x l ,y l ,h) represent the number of drones, the number of sensors, the number of CPs, and the location of each CP, respectively.

[0041] like Figure 1 As shown, the drone starts from data center v0, passes through CPcl. Data information from the scheduled sensors in CPcl is generated and uploaded. Drone a1 flies from CP c1 to C4 until it passes through all CPs within its trajectory, completing the data collection task, and then returns to the data center to unload the data. Therefore, the drone's trajectory only includes CPs and the data center. The flight trajectory of drone ak is represented as follows. This represents the j-th CP in the trajectory. The data center is the starting and ending point of the drone.

[0042] S11: Considering clustering in the time dimension, we propose the SCADC algorithm that combines density clustering to determine the location of CP and establish SN-CP association;

[0043] Based on practical considerations, the location of sensor nodes often changes over time. Considering clustering in the time dimension, a SCADC algorithm combining density clustering and SN-CP associations is proposed to determine the location of the CP (Collision Point) and establish SN-CP associations. Each sensor belongs to only one CP. To improve the efficiency of UAV data collection, the UAV needs to select the optimal data collection point, denoted as CP, and divide M sensors into the range of L CPs. Within the UAV's signal radius, the ranges of CPs obtained by density clustering do not overlap, avoiding collisions during UAV data collection. Density clustering can form clusters of arbitrary shapes and can handle relatively large datasets. Incorporating threshold constraints during clustering improves the clustering effect. Considering the actual situation of ground users, areas with high density of ground users have a large number of physical devices (i.e., a large number of sensors). Therefore, a sensor clustering algorithm combining density clustering is used to find the location of CPs and establish associations between sensors and CPs.

[0044] Assuming a relatively uniformly distributed wireless communication network, let the longitude of the sensor location be x. t Latitude is y t Sensor dwell time z t Then the spatiotemporal coordinates of the sensor are (x t ,y t ,z t Let a length be t(z) t The duration of ∈t) allows the time-varying density of the sensor to be λ. α (t).

[0045] S110: Spatial distance between sensors

[0046] sensor v m The position can be represented by spacetime coordinates, denoted as vector V. m =(x m,y m ,z m ), sensor v k The position is represented by vector V k =(x k ,y k ,z k Then vector V m With vector V k The distance between them is

[0047]

[0048] sensor v m The number of sensors in the spatiotemporal domain can be considered as v m The number of sensors in a region with radius ε centered at ε, and sensor v m With sensor v k The spatiotemporal distance d(V) between them m V k If ε is less than or equal to ε, it is represented as:

[0049] N ε (v m )={v k |d(V m V k )≤ε,V k ≠V m ,v k ∈M} (2)

[0050] Where M is the number of sensors, ρ(v m )=|N ε (v m )| is the sensor v m The number of sensors in the neighborhood.

[0051] Core object: In sensor v m Within the ε-neighborhood of , if the number of sensors is greater than or equal to MinPts, then the number of sensors v m The core object is represented as ρ(v) m )≥MinPts.

[0052] Density clustering algorithm is used to find the set of clustering points (CPs), optimize the drone's flight time, and establish the association between sensors and CPs. Fewer CPs result in shorter drone flight time but longer sensor data upload time; conversely, more CPs result in longer sensor data upload time but longer drone flight time. Therefore, it is necessary to find an appropriate number of CPs to minimize both sensor upload time and the number of CPs, thereby uniformly optimizing both sensor data upload time and drone flight time. m,k =1 indicates that sensor vm Select sensor v k As the cluster head, ζ m,k (m,k=1,...,M) is the sensor v m The upload time. Further simplifying, ρ(m) is the sensor v. m The set of neighboring nodes, whose distances are less than or equal to the signal coverage range of the drone, ρ + (m) represents the sensor v m and its neighboring nodes, i.e., ρ + (m)=m∪ρ(m). The problem involves optimizing density clustering to form a sensor set, i.e., the number of CP points.

[0053]

[0054]

[0055] Where θ is a positive weighting factor, representing the ability of a sensor node to become a cluster center, i.e., a CP point. The first constraint ensures that each sensor set has a cluster head, the second constraint ensures that some sensors in each sensor set can serve as cluster heads, the third constraint ensures that the core object is as accurate as possible, and the fourth constraint ensures that the sensor-CP point association parameter has only two values: 0 and 1, i.e., whether or not it is associated.

[0056] ζ′ m,k These are parameters related to optimizing sensor data upload time, expressed as follows:

[0057]

[0058] In density clustering algorithms, two types of messages are transmitted between adjacent sensors: (1) α m,k , by v m Send to its neighboring node v k , indicates v k How capable is it of becoming its cluster head; (2)γ m,k , by v k Send to its neighboring node v m This indicates how much ability one has to become a V. m The cluster head. The iterative message update is:

[0059]

[0060]

[0061] When convergence is achieved, if γ k,k +α k,k If the value is greater than 0, then the sensor v k They become cluster heads. Therefore, the set of cluster heads is represented as:

[0062] L={l|γ k,k +α k,k >0} (7)

[0063] There are |L| cluster heads, each sensor v m and its cluster head v k Within a cluster, and satisfying k = argmink∈L ζm,k′ The M sensors are divided into L clusters, where l is the cluster head index. Clustering yields a set {c} of CPs. l} and their positions {w l},

[0064] Step S12: Based on the location coordinates of the CP and the SN-CP association, design the CUKK-means algorithm combined with K-means to cluster the CPs, forming a CP cluster containing several CPs, and establish the CP-UAV association.

[0065] Each CP is associated with only one UAV, and each UAV accesses only CPs in one CP set. The main idea of ​​the CUKK-means algorithm in this paper is as follows: First, CPs in the spatial region are mapped to a high-dimensional kernel space through a nonlinear mapping, highlighting the feature differences between CPs. Then, clustering is performed in this kernel space. Here, a new kernel function is proposed for the clustering algorithm to improve its performance. The kernel function of CUKK-means is:

[0066]

[0067] Among them, o k Represents the k-th cluster center, ||Φ(w) l )-o k || is CP point c l and cluster center o k The distance between them, ||Φ(w0)-o k || This refers to data center v0 and cluster center o k The distance between them. According to the kernel function formula, point C is CP. l With cluster center o k The square of the distance between them can be expressed as w = [w1, w2, ..., w L ] represents the position vector of point CP. If the CP point is horizontal, then the cluster center o k It can be represented as:

[0068]

[0069] Where i represents the iteration index. The optimization objective of CUKK-means is to obtain the CP-UAV associations and the locations of the cluster centers. Based on the proximity principle, each CP is assigned to a cluster.

[0070]

[0071] Initially, K=1. During the iteration process, the cluster center o k (i) Continuously update; at the end of the iteration, obtain the association between CP and UAV, and also obtain the location of the cluster center. k (i) and the set of CPs within each cluster.

[0072] Let T be the maximum flight time that the drone can sustain on its battery. max Initially, K=1, and during the first iteration, i=0. All CPs are in one cluster. Based on a heuristic algorithm, a Hamiltonian path u containing all CPs is found. k,hp Calculate the time T required to complete the collection and unloading of all sensor data along this path. k (u k,hp If the longest working time of the drone cannot complete the collection and unloading of all sensor data within the cluster, a second iteration is performed (i=1, K=K+1). All CP points are then divided into two clusters. The maximum value of the time for collecting sensor data in these two clusters plus the time for the drone to return to the data center for data unloading is combined with T. max The comparison is performed. If the former is greater than the latter, the iteration continues; otherwise, the clustering continues and the number of drones is incremented by one.

[0073] S13: Based on the coordinates of CPs in the CP set and the association between CP and UAV, and based on the improved ant colony optimization algorithm (AOTPACO), a strategy for optimizing the UAV trajectory in each CP set is established;

[0074] First, the drone's endurance and speed are initialized. Then, the pheromone concentration on each sub-path of the drone is initialized using the Analytic Hierarchy Process (AHP) to determine the system's fitness function. Based on the improved Ant Colony Optimization (AOTPACO) algorithm, a strategy for optimizing the drone trajectory in each CP set is established. The CPs in each cluster are represented as follows: To address the trajectory optimization problem for each UAV, the AOTPACO algorithm was designed. Since the previous CP clustering determines the CP-UAV association, during the AOTPACO iteration process, the weight of each CP is selected by the UAV. The migration probability of each UAV to the next CP is...

[0075]

[0076] Where i and j are the starting point and the ending point, respectively, τ ijΦ represents the concentration of pheromones between CP points i and j over a time period t. ij Let c represent the heuristic function and the CP point. l The expectation of being selected by the drone as the next data collection point, where a and b are parameters used to adjust the pheromone concentration and the magnitude of the heuristic function, respectively.

[0077] The present invention aims to minimize the average AoI and maximum AoI of each sensor data. The first objective is to minimize the average AoI of the data in the sensor.

[0078]

[0079] in, This indicates the total time it takes for data from the sensors to be uploaded to the drone. This indicates the total flight time of the drone. This represents the total time it takes for data to be unloaded from the sensor to the data center. Note: The AoI of the data in the sensor is calculated from the moment the sensor is scheduled.

[0080] The second objective is to minimize the maximum AoI of the data in the sensor.

[0081]

[0082] in, This represents the total time it takes for data from the scheduled sensors to be uploaded to the UAV. Therefore, the optimization objective becomes a multi-objective optimization problem, namely, minimizing the average AoI of the data in the sensors while simultaneously minimizing the maximum AoI of the data in the sensors.

[0083]

[0084]

[0085]

[0086] Among them, constraint (a) is that the time consumed by each UAV in the trajectory to and from the data center is less than or equal to the UAV's maximum endurance time; constraint (b) indicates that a sensor is associated with only one CP point; constraint (c) indicates that a CP point is associated with only one UAV; constraint (d) indicates the number of sensors in the CP points visited by all UAVs; and constraint (e) indicates that the association parameters λ and η can only take the values ​​of 0 and 1.

[0087] Since the goal of this invention is to minimize the average AoI and maximum AoI of each sensor data, that is, for the UAV to traverse the most CP points in the least amount of time and complete the maximum amount of data collection, the heuristic function is expressed as:

[0088]

[0089] The parameter k is used to adjust the value of the heuristic function. rt(u k,tsp ) represents the remaining flight time of the drone on the current path, and dis(i,j) represents the path from CP point i to CP point j. This represents the average AoI of the sensor data as the drone travels along this path.

[0090] As shown in the above formula, the transition probability is determined by the pheromone concentration and the heuristic function. To improve the performance in the first iteration, the pheromone concentration is initialized. The two determining factors are the distance between CP points and the average data volume of the sensors within the CP cluster. The pheromone matrix is ​​initialized using the Analytic Hierarchy Process (AHP). The designed AHP model consists of three layers: the first layer is the target layer, which selects a suitable CP point; the second layer is the criterion layer, where the CP point is selected based on the distance between the current CP point and the previous CP point, as well as the average data volume of the CP points; and the third layer is the alternative layer, from CPc1 to CPc. L The selection is made from the data. For each drone, the judgment matrix obtained by comparing the pairwise determinants of the trajectory is represented as follows:

[0091]

[0092] Where 'a' represents the importance of distance relative to the average data volume within a given CP point when the UAV selects the next CP point. In the scheme layer, each scheme requires a corresponding criterion layer matrix. Because the criterion layer has two factors, two matrices, B1 and B2, need to be defined, each of size L×L. For the distance matrix B1, B1(v,w) represents the importance of CP point w relative to CP point v in terms of distance. The distance between the current CP point i and CP points v and w is calculated, resulting in B1(v,w).

[0093]

[0094] Then, based on the average data volume within CP points v and w, B2(v,w) is obtained as follows:

[0095]

[0096] Check the consistency of matrices C, B1, and B2. CI is used to determine the degree of deviation of the matrices. The smaller the value of CI, the closer the matrices are to being consistent.

[0097]

[0098] Where, λ maxis the largest eigenvalue of each matrix, and dim is the dimension of the matrix. Then, to measure the magnitude of CI, a random consistency index RI is introduced, and CR is calculated based on RI and CI.

[0099]

[0100] Generally, if CR < 0.1, the judgment matrix is ​​considered to have passed the consistency test; otherwise, it lacks satisfactory consistency. The weights of the relative importance of all factors in the criterion layer to the target layer are calculated, resulting in weight matrices A, B1, and B2. Matrices w1, w2, and w3 have sizes of 2×1, m×1, and m×1, respectively. The weight of each CP point, i.e., the initial pheromone concentration, is:

[0101] τ(i,j)=w1(1)w2(j)+w1(2)w3(j) (23)

[0102] The pheromone concentration on each sub-path is a constant, but not equal, which makes the feasible solution obtained in the first iteration closer to the optimal solution.

[0103] When all drones select the next CP point c l Or, when the data center is v0, the local pheromone concentration is updated to τ. ij =(1-ρ)τ ij Where ρ is the pheromone evaporation coefficient. The pheromone concentration after all drones have completed pathfinding is updated as follows:

[0104]

[0105] in, It is the pheromone update value obtained based on feedback. It is the pheromone left by the k-th drone along the path from i to j, denoted as

[0106]

[0107] Here, λ1 and λ2 represent the influence of the current solution set and the non-dominated solution set on the increase of pheromone intensity during the (t+1)th iteration, respectively. Additionally, w1 and w2 represent the importance of the average AoI and maximum AoI of the data in the sensor, respectively, set as w1, w2 ∈ [0, 1] and w1 + w2 = 1. Adjusting the average and maximum AoI of the data in the sensor using μ1 and μ2 has different effects on pheromone intensity.

[0108] S14: If a drone cannot complete the data collection task, repeat Step S12 and Step S13, recording the highest fitness value in each iteration. The iteration terminates if the required number of iterations is reached or the optimal solution remains unchanged for a period of time.

[0109] S15: Store the local optima corresponding to the fitness values ​​of the two optimization objectives obtained in Step S14 into the Pareto optimal solution set; set the maximum capacity of the Pareto optimal solution set. If the number of optimal solutions obtained after iteration exceeds the maximum capacity of the Pareto optimal solution set, then eliminate the currently dominant solution in the Pareto optimal solution set using the fitness bias ratio. The current solution is also a local optimum. Let χ(t) represent the optimal solution set affected by the average AoI and maximum AoI of the data in the sensor.

[0110]

[0111]

[0112] The global optimal solution is contained in the Pareto optimal solution set. Let δ(t) denote the global optimal solution set. If the current solution set also belongs to the global optimal solution set, it is added to the Pareto optimal solution set.

[0113]

[0114]

[0115] Where sn represents the number of solutions in the non-dominated solution set, and path(i,j) represents the path between two adjacent CPs.

[0116] To achieve the objectives of minimizing the average AoI and maximizing the AoI, a fitness function is designed for each feasible solution. Dominant solutions are eliminated based on the fitness evaluation function, while non-dominated solutions are retained in the outer set. Path loss is used to estimate the fitness functions for optimizing the average AoI and maximizing the AoI, respectively, as follows:

[0117]

[0118]

[0119] in, This represents the upload time of all sensor data at the l-th CP point.

[0120] For a solution u, It is a fitness evaluation function for the average AoI of the data in the sensor. It is a fitness evaluation function for the maximum AoI of the data in the sensor. For example, given two solutions u i and u j If u i <u j , then u i Dominate u j Otherwise, uj Dominate u i If u i and u j If they do not dominate each other, they are added to the Pareto solution set. Define u i <u j as follows:

[0121]

[0122] Multi-objective optimization uses a Pareto optimal solution set to store the global optimum and coordinate different objective optimizations. After each iteration, a local optimum is selected based on the evaluation function of the comparison solution in formula (32). If multiple local optima exist, one is randomly selected as the current solution in that iteration and added to the Pareto optimal solution set. Then, the newly added solution is compared with the existing solutions in the Pareto optimal solution set, and the dominant solution affected by the added solution is eliminated.

[0123] As the number of iterations increases, the number of solutions in the Pareto solution set also increases, which slows down the convergence speed of the algorithm and reduces computational efficiency. Therefore, the maximum capacity Num of the Pareto solution set Pos is set. Pos If the number of solutions exceeds Num Pos Some solutions will be eliminated. All solutions are sorted according to the defined fitness bias ratio (FDR) as follows:

[0124]

[0125]

[0126]

[0127]

[0128]

[0129] For a solution u, and These are used to represent the average AoI deviation ratio and the maximum AoI deviation ratio, respectively. and These are used to represent the deviation of the evaluation value of solution u from the average fitness evaluation value of all non-dominated solutions in Pos. and Let each represent the non-dominated solutions in Pos. and The average value. To simultaneously reduce both the average AoI and maximum AoI of the data in the sensor, y(u) and z(u) should be controlled to be as small as possible. Therefore, when the number of solutions exceeds Num PosWhen the number of solutions does not exceed Num, the algorithm will discard the solution with the largest FDR value until the number of solutions does not exceed Num. Pos .

[0130] Example 2

[0131] One embodiment of the present invention provides an AoI-sensitive data collection system for multiple unmanned aerial vehicles (UAVs), comprising:

[0132] The data acquisition module is used to collect information on the number and location of ground sensors, and to initialize the number and location of the ground sensors.

[0133] The data association module is used to determine the number of data collection points and the location coordinates of each data collection point, and to establish the association between the sensor and the data collection point. Based on the location coordinates of the data collection point and the association between the sensor and the data collection point, the data collection point is clustered, and then the association between the data collection point and the drone is established.

[0134] The optimization module is used to optimize the drone trajectory at each data acquisition point based on the coordinates of the data acquisition point and the correlation between the data acquisition point and the drone, in order to minimize the average AoI and maximum AoI of each sensor data, thus completing the data collection.

[0135] Example 3

[0136] One embodiment of the present invention provides a computer-readable storage medium storing a plurality of instructions adapted for loading and execution by a processor of a terminal device of steps in an AoI-sensitive data collection method for multiple unmanned aerial vehicles.

[0137] Example 4

[0138] One embodiment of the present invention provides a terminal device, including a processor and a computer-readable storage medium. The processor is used to implement various instructions; the computer-readable storage medium is used to store multiple instructions, which are adapted to be loaded by the processor and executed as steps of the AoI-sensitive data collection method in a multi-UAV.

[0139] The specific implementations of Examples 2, 3, and 4 above are the various steps in the method of Example 1.

[0140] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0141] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0142] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for collecting AoI-sensitive data in multiple unmanned aerial vehicles (UAVs), characterized in that, include: Collect information on the number and location of ground sensors, and initialize the number and location of ground sensors; The number of data acquisition points and the location coordinates of each data acquisition point are determined, and the association between the sensor and the data acquisition point is established. The process is as follows: considering the time dimension, the density clustering SCADC algorithm is used to determine the location of the data acquisition point and establish the association between the sensor and the data acquisition point. Each sensor corresponds to only one data acquisition point. The steps of the SCADC algorithm using density clustering are as follows: Calculate the spatiotemporal distance between sensors; sensors The number of sensors in the spatiotemporal neighborhood is considered as The number of sensors within a region centered at ε, in the sensor... Within the ε-neighborhood of , if the number of sensors is greater than or equal to MinPts, then the number of sensors... As the core object, it serves as a candidate cluster head for subsequent iterations; In density clustering algorithms, two types of messages are passed between adjacent sensors during iterative updates: (1) ,Depend on Send to its neighboring nodes ,express How capable is it of becoming its cluster head; (2) ,Depend on Send to its neighboring nodes To indicate one's ability to become The cluster head; When convergence is reached, if If the value is greater than 0, then the sensor Become the cluster head; Its location is the location of the data acquisition point, and the location association between each sensor and its corresponding cluster head is established; in, Represents self-responsibility, which measures the sensor's self-responsibility level. The degree to which it is suitable as its own cluster head. Represents self-availability, which measures the sensor's self-availability. As the external support rate of the cluster head, Used to determine whether a sensor can become a cluster head, when A value greater than 0 indicates that the sensor The sensor's self-responsibility and self-availability meet the conditions to become a cluster leader. Selected as the cluster head; Based on the location coordinates of the data collection points and the association between the sensors and the data collection points, the data collection points are clustered, and then the association between the data collection points and the drones is established. Based on the coordinates of the data collection points and the correlation between the data collection points and the drone, the drone trajectory at each data collection point is optimized to minimize the average AoI and maximum AoI of each sensor data, thus completing the data collection.

2. The method for collecting AoI-sensitive data in multiple unmanned aerial vehicles as described in claim 1, characterized in that, The specific steps for initializing the number and location of ground sensors are as follows: establish a three-dimensional coordinate system, collect the position coordinates of the ground sensors with the origin as the center, initialize the number and location of the ground sensors, and the position of the ground sensors will change over time. The data center is located at the origin, and the number of drones is 1 initially.

3. The method for collecting AoI-sensitive data in multiple unmanned aerial vehicles as described in claim 1, characterized in that, The specific steps to establish the association between data collection points and drones are as follows: use the CUKK-means algorithm combined with K-means to cluster the data collection points to form a data collection point cluster containing multiple data collection points, establish the association between the data collection points and drones, each data collection point is associated with only one drone, and each drone only accesses the data collection points in one data collection point cluster. The CUKK-means algorithm specifically involves mapping data collection points in the space to a high-dimensional kernel space through a nonlinear mapping, and then performing clustering in this kernel space. When performing clustering in the kernel space, a new kernel function is specifically proposed, which is: in, Represents the center of the k-th cluster. It is a CP point and cluster center The distance between them It is a data center and cluster center The distance between them This indicates the association status between the drone and the data collection point (cp). Indicates the number of sensor clusters, This represents a nonlinear function that transforms the original coordinates to a higher-dimensional space, where L represents the number of data collection points. Indicates data collection point Location coordinates, Indicates data center The location coordinates; the CP point refers to the optimal point for the drone to collect data.

4. The method for collecting AoI-sensitive data in multiple unmanned aerial vehicles as described in claim 1, characterized in that, Based on the coordinates of the data acquisition points and the correlation between the data acquisition points and the UAV, the process of optimizing the UAV trajectory at each data acquisition point to minimize the average AoI and maximum AoI of each sensor data is as follows: First, initialize the UAV's endurance and speed, then use the analytic hierarchy process (AHP) to initialize the pheromone concentration on each sub-path of the UAV, determine the fitness function, and optimize the UAV's trajectory.

5. An AoI-sensitive data collection system for multiple unmanned aerial vehicles (UAVs), characterized in that, include: The data acquisition module is used to collect information on the number and location of ground sensors, and to initialize the number and location of the ground sensors. The data association module is used to determine the number of data collection points and the location coordinates of each data collection point, and to establish the association between the sensor and the data collection point; based on the location coordinates of the data collection point and the association between the sensor and the data collection point, the data collection point is clustered, and then the association between the data collection point and the drone is established. The optimization module is used to optimize the drone trajectory at each data acquisition point based on the coordinates of the data acquisition point and the correlation between the data acquisition point and the drone, in order to minimize the average AoI and maximum AoI of each sensor data, thus completing the data collection. The process of determining the number of data acquisition points and the location coordinates of each data acquisition point, and establishing the association between the sensor and the data acquisition point is as follows: Considering the time dimension, the SCADC algorithm of density clustering is used to determine the location of the data acquisition point and establish the association between the sensor and the data acquisition point. Each sensor corresponds to only one data acquisition point. The steps of the SCADC algorithm using density clustering are as follows: Calculate the spatiotemporal distance between sensors; sensors The number of sensors in the spatiotemporal neighborhood is considered as The number of sensors within a region centered at ε, in the sensor... Within the ε-neighborhood of , if the number of sensors is greater than or equal to MinPts, then the number of sensors... As the core object, it serves as a candidate cluster head for subsequent iterations; In density clustering algorithms, two types of messages are passed between adjacent sensors during iterative updates: (1) ,Depend on Send to its neighboring nodes ,express How capable is it of becoming its cluster head; (2) ,Depend on Send to its neighboring nodes To indicate one's ability to become The cluster head; When convergence is reached, if If the value is greater than 0, then the sensor Become the cluster head; Its location is the location of the data acquisition point, establishing a positional association between each sensor and its corresponding cluster head; among which, Represents self-responsibility, which measures the sensor's self-responsibility level. The degree to which it is suitable as its own cluster head. Represents self-availability, which measures the sensor's self-availability. As the external support rate of the cluster head, Used to determine whether a sensor can become a cluster head, when A value greater than 0 indicates that the sensor The sensor's self-responsibility and self-availability meet the conditions to become a cluster leader. It was selected as the cluster head.

6. A computer-readable storage medium, characterized in that, It stores multiple instructions, which are adapted to be loaded and executed by the processor of the terminal device as any one of claims 1-4, for an AoI-sensitive data collection method in multiple unmanned aerial vehicles.

7. A terminal device, characterized in that, The device includes a processor and a computer-readable storage medium, the processor being used to implement various instructions; the computer-readable storage medium being used to store multiple instructions adapted to be loaded by the processor and executed as described in any one of claims 1-4, a method for AoI-sensitive data collection in multiple unmanned aerial vehicles.

Images