Continuous communication coverage method of drone swarm in emergency communication

Abstract

The present invention discloses a method for continuous communication coverage of a drone cluster under emergency communication, and belongs to the technical field of drone cluster planning. The method proposed in the present application completes the emergency communication task with the minimum total task energy consumption in a scenario with multiple fixed emergency task points; further, the emergency communication task can also be completed with the minimum average energy consumption. In the emergency communication task, the present application ensures the seamlessness and continuity of the drone network at the emergency task point by reasonably planning the working path of the drone, while minimizing the average energy consumption of the drone cluster. Compared with the existing work, the present application takes into account the link continuity and seamlessness of the drone cluster under dynamic deployment, and realizes the replacement and planning of the drone cluster through multiple differentiated periodic interval rotation strategies.

Description

Technical Field

[0001] The present application relates to the technical field of drone cluster planning, and in particular to a method for continuous communication coverage of drone clusters in emergency communications. Background Art

[0002] UAV-based aerial base stations have become an important means of providing wireless network coverage due to their advantages such as flexible deployment on demand and wide service coverage. Every year, the material losses caused by natural disasters such as earthquakes and floods continue to rise around the world. When natural disasters damage ground base stations, drone clusters can quickly reconstruct emergency communication networks and provide continuous coverage, providing communication guarantee for emergency points, which is of great significance to emergency rescue and communication services. However, the limited battery capacity limits the cruising and hovering time of drones, which may make it difficult for drones to provide seamless and continuous information services, but network coverage, emergency communications and other services have high requirements for the seamlessness and continuity of services; and the number of drones required to provide the above information services in natural disaster scenarios is large, while the number of sites where drones can land and recharge is very limited; therefore, how to plan multiple drones that perform continuous tasks in disaster emergency scenarios is a key issue.

[0003] In existing research, the problem of UAV communication network deployment in emergency scenarios is abstracted into a static deployment optimization problem within multiple time slots. For example, some studies have solved the problem of large variance of user resource demand between adjacent time slots and discontinuous coverage in the spatiotemporal dimension based on the resource allocation method of deep reinforcement learning (RDRL); some studies have formulated the problem of joint backhaul link and coverage-aware UAV deployment as integer linear programming, and implemented a low-complexity algorithm with provable performance guarantee to effectively solve the problem of UAV communication network deployment; some studies have modeled the multi-UAV trajectory planning problem as two coupled multi-agent cooperative random game problems, corresponding to path planning and charging scheduling respectively, and the equilibrium of the game is the optimal trajectory of the UAV to achieve long-term and energy-efficient content coverage; some studies have considered the continuously changing energy cost, time cost and social relations between users, and implemented a timely perception information separation scheme based on social relations; some studies have also improved the overall social benefits by optimizing UAV scheduling and trajectory planning, such as modeling the multi-UAV base station assisted emergency communication scenario as a Markov game model, and implementing a UAV cluster emergency communication method based on collaborative intelligence. Other studies have taken into account the limited cruising time of drones and that multiple rechargeable drones can form a closed chain. Based on this, they have studied high-efficiency collaborative strategies for rechargeable multiple drones to provide seamless coverage and long-term information services for IoT nodes.

[0004] It can be seen that in the field of emergency communications, the problem of continuous network coverage has not been well solved, and the energy efficiency problem of how to provide continuous network coverage in the context of emergency communications has not been fully studied. Summary of the invention

[0005] The invention purpose of this application is to provide a method for continuous communication coverage of drone clusters under emergency communication, which is a drone cluster planning strategy based on the continuous communication coverage requirements of fixed points, and aims to provide continuous and seamless communication services for multiple disaster relief points in disaster scenarios, while minimizing the energy consumption of drone clusters.

[0006] The technical solution adopted in this application is: a method for continuous communication coverage of drone clusters under emergency communication, in an emergency rescue mission with multiple fixed emergency mission points and a remote charging base station, performing the following steps:

[0007] Construct a multi-drone tree-like backhaul network to ensure connection with the charging base station: each drone in the unmanned cluster performing emergency rescue tasks is used as a node, and K drones provide one-to-one communication access services for each emergency task point by hovering, and the remaining drones provide relay services or serve as redundant drones; the value of K is the same as the number of emergency task points;

[0008] Based on the departure time of the drones from the charging base station, the multi-drone tree-like backhaul network is divided into multiple rotation paths, and the time intervals between adjacent drones on the same rotation path are the same;

[0009] For each drone in the unmanned swarm performing emergency rescue missions, if it stays at the charging base station, it is defined as a dormant state, otherwise it is defined as a working state; and the state value of the drone in the dormant state is set to 0, and the state value of the drone in the working state is set to 1;

[0010] For any drone in the drone cluster, the instantaneous energy consumption of the drone is obtained based on the product of its drone state value and the drone working power; the total energy consumption of the task is obtained based on the instantaneous energy consumption of all drones within the task duration T. ;

[0011] Under given constraints, the total energy consumption of the task is minimized. To optimize the target, the access service allocation of drones corresponding to the optimization solution is , UAV tracks and network topology Obtain the emergency communication deployment result, that is, plan the optimal rotation path;

[0012] The given constraints include: single UAV energy consumption constraints, one-to-one access service constraints, time interval constraints of rotation paths, and network connection constraints of multi-UAV tree-like backhaul networks.

[0013] Furthermore, the network connection constraints of the multi-drone tree-like backhaul network include: the nodes in the network are accessible (i.e., used to determine that the available network topology is built on the actual communication link), the charging base station and all drones providing access services are included in the topology of the multi-drone tree-like backhaul network at any time, and the network's acyclic tree constraints (i.e., building an acyclic tree in the available network topology to achieve a rotation path).

[0014] Furthermore, the energy consumption constraint of a single UAV is specifically: the current remaining storage energy of the UAV is not less than the current working energy threshold; usually, the maximum storage energy of the UAV can be The difference between the current energy consumption and the current remaining energy consumption is taken as the current remaining energy storage of the drone, where the current energy consumption is the time between the drone and the last departure from the charging base station at the current moment. Working power with drone The current working energy threshold can be set as: the ratio of the distance between the current position of the drone and the charging base station to the flight speed of the drone and the working power of the drone The product of ; UAVs can usually be considered to fly at a constant speed; and all UAVs in a UAV cluster are homogeneous.

[0015] Furthermore, the one-to-one access service constraints are as follows:

[0016] definition Indicates drone exist The time is The binary decision variable of providing communication access service for each emergency task point. If communication access service is provided, then ;otherwise ;

[0017] In solving the minimum total energy consumption hour, satisfy:

[0018]

[0019]

[0020] in, Represents an unmanned swarm that performs emergency rescue missions.

[0021] Furthermore, the time interval constraint of the rotation path is specifically: the time interval of each rotation path Greater than or equal to the preset minimum time interval ,in, is the round-robin path identifier.

[0022] Furthermore, the nodes in the network in the network connection constraint of the multi-machine tree-like backhaul network can be specifically:

[0023] definition express Two nodes in the network at this moment and A binary decision variable for whether to build a network connection between them. If so, ;otherwise ;

[0024] In solving the minimum total energy consumption hour, satisfy:

[0025]

[0026] in, represents a node set, , Indicates charging base station; auxiliary quantity The value depends on the node , The location and maximum relay distance of the drone , which is expressed as:

[0027]

[0028] in, , Respectively represent nodes exist The position at the moment.

[0029] Furthermore, the charging base station and all drones providing access services are included in the multi-drone tree-like backhaul network topology at any time:

[0030] definition Indicates drone exist Time and charging base station A binary decision variable for whether to build a network connection between them. If so, ;otherwise ;

[0031] In solving the minimum total energy consumption When , the topological constraints of the multi-UAV tree-like backhaul network satisfy:

[0032]

[0033]

[0034] in, express Drones that provide access services at all times and A binary decision variable for whether to build a network connection between them. If so, ;otherwise .

[0035] Furthermore, the acyclic tree constraints of the network are as follows:

[0036]

[0037]

[0038]

[0039] in, Represents any subset of network nodes in a multi-UAV tree-like backhaul network.

[0040] Furthermore, under given constraints, the total energy consumption of the task can be minimized. The optimization objective is transformed into:

[0041]

[0042] in, represents the set of rotation paths. The new optimization goal is to find a set of path configurations to minimize the sum of the inverses of the time intervals of each path, that is, to minimize the total energy consumption of the task The optimization objective is transformed into: Under given constraints, determine the optimal set of rotation paths so that the sum of the inverses of the time intervals of each rotation path reaches the minimum value.

[0043] Furthermore, under given constraints, on a given rotation path, the time interval on the rotation path is maximized. To achieve the minimum total energy consumption of the current rotation path, where the time interval on the rotation path is maximized Specifically:

[0044] ;

[0045] in, , Respectively represent the hovering time of all fixed nodes and non-fixed nodes on the same rotation path; , , Respectively represent the number of fixed nodes and non-fixed nodes; Indicates the maximum working time of the drone. , Indicates the total flight time of the drone along the current rotation path from the charging base station to the return process. Indicates that the drone is at the maximum relay distance The time required for the flight is .

[0046] Furthermore, under given constraints, the Max-Min Ant System (MMAS) algorithm is used to optimize the optimization target to achieve the optimal rotation path planning. Furthermore, when the Max-Min Ant System (MMAS) algorithm is used to optimize the optimization target, the dynamic tree return strategy can also be used to update the return link of the UAV at the emergency mission point and the role of the relay node; among them, the role of the relay node includes two types: fixed node and non-fixed node.

[0047] The technical solution provided by this application brings at least the following beneficial effects:

[0048] The method for continuous communication coverage of drone clusters under emergency communications provided in this application is a drone cluster planning scheme that completes emergency communication tasks with the minimum total mission energy consumption in a scenario with multiple fixed emergency mission points (such as disaster relief points); further, the emergency communication task can also be completed with the minimum average energy consumption. In emergency communication tasks, this application ensures the seamlessness and continuity of the drone network at the emergency mission point by reasonably planning the working path of the drone (drone trajectory and network topology), while minimizing the average energy consumption of the drone cluster. Compared with existing work, this application takes into account the link continuity and seamlessness of drone clusters under dynamic deployment, and adopts a multi-differentiated periodic interval rotation strategy (converting the minimization of total mission energy consumption into maximization of the time interval on all paths). The derivative sum, dynamic tree backhaul strategy and cluster rotation path planning strategy based on the maximum-minimum ant colony algorithm) are used to realize the replacement and planning of drone clusters. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0050] Figure 1 A schematic diagram of an emergency communication task scenario shown in an embodiment;

[0051] Figure 2 This is a schematic diagram of a multi-UAV tree-like backhaul network based on a linear distribution scenario;

[0052] Figure 3Schematic diagram of a multi-UAV tree-like backhaul network based on a randomly distributed scenario;

[0053] Figure 4 for Figure 2 A schematic diagram of a planned path obtained by using the method proposed in an embodiment of the present application in a linear distribution scenario is shown;

[0054] Figure 5 for Figure 3 A schematic diagram of a planned path obtained by using the method proposed in an embodiment of the present application in a randomly distributed scenario shown;

[0055] Figure 6 for Figure 3 Schematic diagram of access point drone allocation in a randomly distributed scenario.

[0056] In the figure, station represents a disaster relief point, Access represents an access point, Relay represents a relay point, and Path represents a rotation path, that is, a planned path. DETAILED DESCRIPTION

[0057] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions of the embodiments of the present application will be described in detail and completely in conjunction with the drawings in the embodiments of the present application. Obviously, the embodiments described with reference to the drawings are exemplary and are intended to be used to explain the present application, and cannot be understood as limiting the present application.

[0058] In emergency communication tasks, it is crucial to always maintain the backhaul link of the emergency task point (such as the disaster relief point). Therefore, the goal of the method for continuous communication coverage of drone clusters under emergency communication proposed in the embodiment of the present application is to reasonably plan the working path of the drones, to ensure the seamlessness and continuity of the drone network at the disaster relief point, and to minimize the average energy consumption of the drone cluster.

[0059] In such Figure 1 In the emergency communication mission scenario shown, this application considers a long-distance multi-hop tree network constructed by drones. The unmanned cluster performing emergency rescue missions includes a total of M drones, and each drone can be regarded as an aerial node. K of the drones provide one-to-one communication access services for K ground disaster relief points by hovering, and the remaining drones (MK) are used to build a complete tree-like backhaul network to ensure connection with the back-end charging station, that is, the drones provide relay services or serve as redundant drones (to take over the work of other drones), thereby constructing the multi-drone tree-like backhaul network of this application. In the embodiment of this application, a collection of represents all drones, i.e. Represents an unmanned cluster that performs emergency rescue tasks, and can be obtained In the embodiment of the present application, it is assumed that all drones are homogeneous and the flight altitude Same, maximum working time Due to the limitation of battery capacity, all drones start from the charging station when performing tasks, and then end at the same charging station before the energy is exhausted. In order to simplify the model, we design the charging station and the back-end base station together and use It means charging base station.

[0060] In order to provide seamless and continuous communication services, the method proposed in the embodiment of the present application is used to propose a multi-differentiated periodic rotation path planning method, which divides the multi-UAV tree-like backhaul network of multi-UAV nodes into a rotation path set R with multiple different time interval periods; on any rotation path in the set R, when a UAV stops the current mission (returning to the charging station or going to the next emergency mission point), the next UAV will take turns to replace the previous UAV. Before a UAV stops the service, the next UAV has arrived at a specific location to replace the UAV, so as to ensure the seamlessness of the service, and the same rotation path (such as Figure 1 The time interval between each drone leaving the charging station in the drone trajectory 1 and drone trajectory 2 shown in ,in, is the rotation path identifier, Indicates The time interval between the rotation paths.

[0061] Since all drones need to return to the charging base station for energy replenishment, in order to ensure that the drone returns to the charging base station before the energy is exhausted, the energy consumption of the drone needs to be constrained. The energy consumption of drones during emergency mission execution generally consists of two parts: communication-related energy consumption and motion energy consumption. Among them, communication-related energy consumption includes communication circuit energy, signal processing energy, signal radiation / reception energy, etc.; motion energy consumption is related to the speed, acceleration, and motion state of the drone. In the embodiment of this application, the communication-related power (denoted as ) is considered as a constant with a unit of , and ignore the time it takes for the drone to accelerate from hovering to a constant speed, assuming that the drone moves at a constant speed throughout the emergency mission The drone can hover by circling at a constant speed, so the power of the drone is ,in , They represent the parasitic drag coefficient and the induced drag coefficient respectively, so the UAV working power can be expressed as: Therefore, in the embodiment of the present application, the energy consumption constraint of a single UAV is described as:

[0062]

[0063] in, represents the maximum storage energy of the drone, Indicates drone exist The time from the last departure from the charging base station. Indicates drone exist The location at the moment, represents the location coordinates of the charging base station. In this embodiment, the Cartesian coordinate system is not used. Indicates the current working energy consumption threshold.

[0064] In order to characterize the one-to-one access service constraint between the drone and the emergency mission point, in the embodiment of the present application, it is defined Indicates drone exist The time is The binary decision variable that provides communication access service for emergency mission points takes the value of 0 or 1. exist The time is If communication access services are provided to each emergency task point, ,otherwise In order to ensure seamless and continuous services at all disaster relief sites, there should be a drone at each disaster relief site at any time, and the drone should provide one-to-one services to the disaster relief site. Therefore, this application sets the following two constraints when optimizing the solution:

[0065]

[0066] In addition, the time interval between drones on the rotation path It will affect system performance. If it is too small, the rotation replacement operation of the drones on the rotation path will be too frequent, which will affect the quality of service (QoS) of the disaster relief point. Therefore, it is necessary to add the following constraint to the time interval between drones on the rotation path:

[0067]

[0068] in, It indicates the minimum time interval between drones agreed upon manually, that is, the preset minimum time interval.

[0069] At the same time, in order to realize the network connection constraints of the multi-drone tree-like backhaul network, that is, the nodes in the network are accessible, the charging base station and all drones providing access services are included in the topology of the multi-drone tree-like backhaul network and the network's acyclic tree constraints at any time. In the embodiment of the present application, the network connection constraints of the multi-drone tree-like backhaul network are specifically set as follows:

[0070] As the drone moves forward along a rotation path, the original tree network structure will change, which may cause the drones working at some disaster relief points to be disconnected from the charging base stations at the back end. In order to ensure the seamlessness and continuity of services at all disaster relief points, it is necessary to enable all disaster relief points to establish a tree network through drones and the charging base stations at the back end at any time. To solve this problem, the embodiment of the present application defines a node set , and introduce the following constraints:

[0071]

[0072]

[0073]

[0074]

[0075]

[0076] in, express At this moment, two nodes in the network and A binary decision variable for whether to build a network connection between them. If so, ;otherwise ; Auxiliary amount The value depends on the node , The location and maximum relay distance of the drone , which is expressed as:

[0077]

[0078] in, , Respectively represent nodes exist The location at the moment; Indicates drone exist Time and charging station A binary decision variable for whether to build a network connection between them. If so, ;otherwise ; express Drones that provide access services at all times and A binary decision variable for whether to build a network connection between them. If so, ;otherwise ;

[0079] The set of network nodes with valid topology at time t The expression is:

[0080]

[0081] in, Represents any subset of network nodes in a multi-UAV tree-like backhaul network.

[0082] Among the above constraints, formula (5) restricts the premise that the inter-node link is part of the network topology only if there is a real communication link between them, that is, the network connection constraint of the rotation path; formula (6) is used to ensure that the charging base station is included in the network topology at any time, that is, the charging base station is included in the topology constraint of the multi-UAV tree backhaul network at any time; formula (7) is used to ensure that all access drones at the disaster relief points are included in the network topology at any time; formula (8) ensures Build network topology at all times There is at least connections, for The set of network nodes that participate in building the network topology at all times; the constraints given by formula (9) ensure that any subset of network nodes There is no loop in the topological connection; the two constraints of formula (8) and formula (9) form the backhaul connectivity constraints of the multi-UAV tree-like backhaul network.

[0083] In the embodiment of the present application, the states of the drone during the emergency mission are divided into two types: working state and dormant state, and the binary variable It means that when the drone stays at the charging base station waiting to be replaced by other drones, it is in dormant state. ; Otherwise, the drone is in working state, and . It is assumed that the UAV has no energy consumption in the dormant state, so the energy consumption of the multi-UAV tree-like backhaul network composed of multiple UAVs during the emergency mission can be expressed as:

[0084]

[0085] in, Indicates the duration of the emergency task.

[0086] In summary, the optimization goal of the embodiment of the present application is to optimize the access service allocation of human and machine by joint optimization. , UAV tracks and network topology To minimize the total energy consumption of the task , the optimization objective can be expressed as follows:

[0087]

[0088] in, It represents the nine constraints corresponding to formulas (1) to (9).

[0089] By optimizing and solving the optimization given by formula (13), we can obtain the UAV trajectory, UAV access association, UAV rotation order, network topology, etc., and thus obtain the emergency communication deployment result of the current emergency rescue mission.

[0090] In one embodiment, the optimization objective given by formula (13) may also be transformed to implement a multi-differentiated periodic path rotation strategy.

[0091] Since the straight line is the shortest distance between two points, in order to make full use of the limited storage energy of the drone, this application assumes that the movement of the drone during the rotation process is point-to-point. The movement trajectory of the drone from departure to return to the charging station is a path composed of several straight lines connected end to end. By finding multiple continuous paths that meet the constraints in the optimization problem (13), the optimization problem is decoupled and the total energy consumption of the task is calculated. Make the following transformations:

[0092]

[0093] in, Indicates the rotation path During the emergency mission The total energy consumption of the task can be equalized by the sum of the energy consumption of all paths. .

[0094] On a fixed rotation path, the mission process of the rotating drone is regular, which is reflected in the same time interval between the rotating drones and the unchanged drone mission cycle. The same time interval refers to the time interval between each drone leaving the charging base station on the rotation path. The invariant mission cycle refers to the drones working on the path. Their mission execution process is periodically consistent, that is, they start from the charging base station at different times, perform corresponding tasks at the same location along the same trajectory, and return to the charging base station at different times. It is also assumed that the drone makes maximum use of the limited energy storage of the drone in the mission process, that is, when the drone returns to the charging base station, its remaining energy is 0, so the duration of this cycle can be expressed as: .

[0095] When the period is known, a rotation path is The energy consumption in the cycle is related to the number of repetitions of the cycle, and the number of repetitions of the cycle is related to the time interval between the cycles, so we can get The following definition:

[0096] Due to the duration of the emergency task and the maximum storage energy of the drone is a constant, so the optimization objective of formula (13) can be further transformed into:

[0097]

[0098] There may be multiple relay nodes (relay points) or access nodes (APs) on a rotation path. The drone needs to hover at these mission points (relay points / APs) for a certain period of time to ensure seamless service or connectivity. Number of mission points, cruise energy consumption ratio, and maximum relay distance , the hovering time of different task points is related to the time interval. However, after the specific path is determined, the time interval can only be adjusted by adjusting the hovering time of different task points. There are two types of hovering time for different mission points: one is the hovering time of fixed nodes (referred to as fixed points), and the other is the hovering time of non-fixed nodes (referred to as non-fixed points). The access points are all fixed points. If the distance between the relay point and the two mission points before and after is not greater than , then the relay point is a non-fixed point, otherwise it is a fixed point. This application uses the non-fixed relay point UAV synchronous forward replacement method to efficiently utilize the energy consumption of the UAV, adjust the size of the two hovering times, so that the time interval The essence of this adjustment strategy is that when a replacement drone flies to a non-fixed relay point, the drone at that point does not stay at that point until the replacement drone arrives, but moves forward with a certain distance from the replacement drone, and this distance is equal to the maximum relay distance. The reason why non-fixed relay points have hovering time is that when the replacement drone hovers at the previous mission point to perform the mission, in order to ensure the connection of the communication link, the drone also needs to hover and wait for a period of time at the relay point. And it has been verified that the hovering time of all fixed points and non-fixed points on a rotation path is the same, respectively. , Therefore, the maximum time interval can be obtained ,at this time, It can be solved by the following formula:

[0099]

[0100] in, , Respectively represent the number of fixed nodes and non-fixed nodes; Indicates the maximum working time of the drone. Indicates the total flight time of the drone along the current rotation path from the charging station to the return process. Indicates that the drone is at the maximum relay distance The duration of the flight.

[0101] The optimization goal given by formula (17) is a NLP (Nonlinear Programming) problem, and is a convex problem that can be directly solved by solving convex optimization problem solving tools (such as CVX solver). That is, in the embodiment of the present application, the total energy consumption of the task can be The original optimization problem is transformed into the accumulation of energy consumption of multiple rotation paths, each of which corresponds to a specific time interval, and all rotation paths maintain the same task cycle duration, thereby transforming the original optimization problem. The new optimization goal is to find a set of rotation path configurations to minimize the sum of the inverses of the time intervals of each rotation path. Therefore, the optimization goal is transformed to determine the optimal set of rotation paths under given constraints so that the sum of the inverses of the time intervals of each wheel path reaches the minimum value.

[0102] In one embodiment, the Max-Min Ant System (MMAS) algorithm can be used to optimize and solve the optimization target shown in formula (16) to achieve the planning of the optimal rotation path. In the solution process, the maximum time interval can be obtained based on formula (17): In the embodiment of the present application, when the maximum-minimum ant colony algorithm is used to optimize and solve the optimization target, a dynamic tree backhaul strategy is used to update the backhaul link of the drone at the emergency mission point and the role of the relay node (fixed point / non-fixed point).

[0103] In the embodiment of the present application, the converted optimization target shown in formulas (16) and (17) is essentially analogous to an extended vehicle routing problem (Extended Vehicle Routing Problem, E-VRP), which is a type of path planning problem. In problems such as VRP, the transfer between task points is the core element of path planning. However, since the present application requires the provision of seamless services, path planning must not only consider the optimal transfer between task points, but also the link connectivity problem caused by the transfer between task points. This problem mainly occurs in the process of synchronous forward operation when the drone transfers from one non-fixed relay point to the next task point. Specifically, when the drone is synchronously pushed forward at a non-fixed relay point, other drones at the task points connected to the relay point need to establish a return link through the relay point. This may cause the communication link between these drones to be interrupted, thereby affecting the overall connectivity of the network. In order to ensure the connectivity of the entire network and ensure seamless communication, the embodiment of the present application adopts a dynamic tree return strategy to solve it.

[0104] The dynamic tree backhaul strategy ensures that the network of the entire system maintains a tree structure by dynamically updating the backhaul link of the drone at the mission point and the role of the relay point (fixed point or non-fixed point). The main ideas of the dynamic tree backhaul strategy are as follows:

[0105] 1) When selecting the next task point of the relay point on the rotation path, similar to the traditional vehicle routing problem (VRP), all unplanned task points can be used as candidate points for the current task point to transfer. However, in the embodiment of the present application, there are two situations when the drone transfers from the relay task point to the next task point: when the relay point is a non-fixed point, synchronous advancement transfer can be adopted; when the relay point is a fixed point, it is necessary to wait for the replacement drone to arrive before transferring. This difference comes from the fact that the selection of the next task point will affect the role selection of the current relay point, and thus affect the connectivity of the overall task point.

[0106] 2) Feasibility assessment of changing the backhaul link of the disconnected mission point drone: If some mission points are disconnected due to the synchronous forwarding of non-fixed relay points, it is necessary to determine whether these mission points can restore connectivity by changing the backhaul link. There are two ways to change the backhaul link: one is to establish a connection with the drone on the existing backhaul link, and the other is to establish a connection with an unplanned relay point.

[0107] 3) Relay point task role selection: After determining the next task point, the role selection of the current relay point needs to be evaluated to ensure that the connectivity of other task points is not affected.

[0108] In one embodiment, when determining whether a relay point is fixed or non-fixed, the present application may determine whether the relay point will affect the communication of other nodes on the rotation path if it is not a fixed point. If the impact is too great, it will remain fixed; if the impact is not great or there is no impact, it will not be fixed. The measure of the degree of communication impact can be set as follows: if after setting the current relay node to a non-fixed point, the number of other nodes originally connected to the relay point that are disconnected exceeds a specified value (for example, set to half of the number of all drones), then the impact is considered to be too great.

[0109] Ant colony algorithm is widely used to solve various path planning problems due to its good parallel operation and positive reinforcement mechanism. The maximum minimum ant colony algorithm (MMAS) inherits the advantages of the traditional ant colony algorithm in solving vehicle path problems. By introducing the upper and lower limit mechanism of pheromone concentration, it significantly enhances the global search ability and the quality of the solution, thereby showing better performance and higher robustness. Therefore, in the embodiment of the present application, the maximum minimum ant colony algorithm (Max-Min AntSystem, MMAS) can be used to optimize and solve the optimization objective shown in formula (16). In this embodiment, the solution strategy is described as a maximum minimum ant colony based on the wheel path planning algorithm, which adjusts the design of the heuristic, candidate set, pheromone matrix, node transfer probability, and pheromone update in the MMAS algorithm to make the MMAS algorithm more suitable for the optimization objective constructed in the embodiment of the present application.

[0110] Among them, the heuristic settings are:

[0111]

[0112] in, Indicates the current mission point (relay point / access point), Indicates the next mission point (relay point / access point), Indicates the rotation path Adding mission points The increased equivalent energy consumption, Used to characterize charging base stations; Indicates that a task point has been added The rotation path formed afterwards.

[0113] Given that relay task points can assume two roles (fixed points and non-fixed points), when designing the candidate set of relay task points, corresponding elements should be introduced according to the role of each relay task point to clearly indicate its role status. This design ensures that the specific role of the relay task point can be accurately distinguished and selected during the path planning process, thereby enhancing the flexibility and adaptability of the algorithm. exist Candidate set of relay points It is expressed as follows:

[0114]

[0115]

[0116] Among them, the candidate set of the next task point , represents the set of all mission points, including access mission points (i.e., disaster relief points), relay mission points (i.e., relay points), and charging base stations. Indicates the taboo table, , Represent the relay task points There are two role elements, Relay task point fixed, Relay task point Not fixed; It means that the drone must always meet the energy consumption requirements while heading to the next mission point.

[0117] In the framework of the ant colony algorithm, in order to accurately record the relay task points Pheromones in different roles (fixed and non-fixed), pheromone matrix For each relay task point, two entries are set in both row and column directions: and , to record the pheromone concentration when it is a fixed role and a non-fixed role. Therefore, the pheromone matrix It can also be called the pheromone concentration matrix, and its structure is:

[0118]

[0119] in, represents the field of real numbers, represents the set of relay task points, , Respectively represent sets , The number of elements in the collection.

[0120] In the embodiment of the present application, the next task point can be selected by pseudo-random probability to improve the diversity of solutions, that is, the transition probability of each task point in the candidate set is calculated according to formula (22), and the next task point is selected by roulette selection. :

[0121]

[0122]

[0123]

[0124] in, represents the transfer probability. Formula (24) can be used to clearly distinguish the roles of relay task points to ensure that the pheromone concentration and heuristic factor ( , are the corresponding two inspiration factors), represents the set of access task points, Used to characterize the role of relay task points, , Respectively represent nodes For fixed and non-fixed states, Representation Node is a fixed state, , They represent the corresponding matrix elements in the pheromone matrix, and the subscripts are matrix element indices. This design aims to improve the accuracy of round-robin path selection and enhance the adaptability and solution accuracy of the algorithm to complex task allocation problems.

[0125] In the redesigned pheromone update mechanism of this embodiment, the main adjustment is focused on the index selection of the value to be updated in the pheromone matrix. Specifically, for the relay task point in the planned rotation path and the next mission point , if their roles are fixed, then the corresponding index position in the pheromone matrix is The value will be updated.

[0126] In this embodiment, the designs of heuristics, candidate sets, pheromone matrix, node transfer probability, and pheromone update in the MMAS algorithm are adjusted to make the maximum-minimum ant colony algorithm more suitable for the round-robin path planning defined in this application, that is, in the process of path finding by the ant colony algorithm, the selection of paths and relay point roles is guided or influenced by the strategy of the dynamic tree to ensure that the path planned by the ant colony algorithm can meet the requirements of seamless service.

[0127] In one embodiment, the present application uses the MMAS algorithm to optimize and solve the optimization target, and the specific implementation process of planning the optimal rotation path is as follows:

[0128] Input: environmental information (such as the location coordinates of each fixed emergency task point in the emergency rescue mission scene, the location coordinates of the remote charging station, the drone cluster performing the mission, and the communication connection relationship between the task points, i.e., an adjacency matrix containing the connection relationship of all points, etc.), the number of ants m, and the maximum number of iterations ,parameter (represents the number of elite paths screened out in each round); the number of ants m is an empirical preset value, which can usually be set to 1.5~2.5 times the number of nodes in the multi-UAV tree-like backhaul network.

[0129] Output: The optimal rotation path under the minimum optimization goal ;

[0130] Initialization: Initialize all pheromone matrix elements ( ) is 0, that is ;

[0131] Before reaching the maximum number of iterations The following iterative process is performed:

[0132] (1) Deploy all ants in parallel and define is the ant identifier, where That is, m ants are deployed in parallel, and each ant represents a rotation path, which can be recorded as path ;

[0133] (2) If there are still unvisited mission points, repeat the following steps:

[0134] (2.1) If the current network node k is a relay node, the candidate set of the next task point is determined according to formula (19): ;

[0135] (2.2) Otherwise, if the current network node k is not a relay node, then the candidate set of the next task point is is the set of all remaining task points;

[0136] (2.3) If the current network node is not a relay node, the next task point candidate set The transition probability of each task point in Select the next task node , and update ant The characterized paths and their taboo tables;

[0137] (3) Based on energy consumption assessment ( ) The paths represented by all ants , and select the previous The ants / paths with the lowest energy consumption are regarded as the elite path set;

[0138] (4) Update the pheromone of the elite path and adjust the upper and lower bounds of the pheromone;

[0139] (5) When the maximum number of iterations is reached When , a single path with the lowest energy consumption is selected from the most recently obtained elite path set as the optimal rotation path .

[0140] In this embodiment, the output elite path set is in the form of a list group, and the order of elements in a single list is the order in which a rotation path passes through the task points, for example: {[0, 2, 0], [0, 13, 14, 10, 7, 11, 12, 9,8, 0], [0, 3, 4, 0], [0, 1, 0], [0, 6, 0], [0, 5, 0]}, wherein all elements in each bracket correspond to a rotation path, and the values ​​in the brackets are the numbers of the task points; in addition, a list of the states of all relay points on the rotation path is also output, for example: {[9, "not_fix", 0], [12, "not_fix", 8], [14, "not_fix", 7], [7, "not_fix", 12], [11, "not_fix", 9], [0, "not_fix", 14],[13, "not_fix", 10]}, where the first element of each bracket represents the relay point number, the second element represents the role (fixed point / non-fixed point), and the third element represents the next task point of the relay point.

[0141] In addition, a simulation experiment was conducted to further verify the communication coverage performance of the method proposed in the embodiment of the present application. In the simulation experiment, the area of ​​the emergency rescue mission was 25 km × 25 km, the location coordinates of the charging base station were (0,0), the flight altitude of the UAV was set to 100 meters, and the operating power was 100 watts. The flight speed V of the UAV was 10 meters / second, the communication power was 30 watts, and the maximum relay distance was 6 kilometers.

[0142] In this simulation experiment, two scenarios are considered to correspond to the two main terrain distributions that require emergency rescue: linear distribution and random distribution. The multi-UAV tree-like backhaul network formed by these two distributions is as follows: Figure 2 , Figure 3 shown. Figure 4 , Figure 5 It is a trajectory diagram obtained by solving the maximum-minimum ant colony algorithm adopted in the embodiment of the present application in two scenarios. Figure 6 is Figure 3 The corresponding access point drone allocation in the randomly distributed scenario is Figure 4 , 5It can be seen that three paths are planned for the linear scenario and five paths are planned for the random scenario. In the planned paths, almost all relay drones are not fixed, which makes the maximum use of the drone relay capability. When the task point is transferred, synchronous advancement is performed to minimize energy consumption. In addition, direct round-trip paths are planned for farther access points (such as Path 2 in the linear scenario, Path 4 and Path 5 in the random scenario), which is more conducive to saving energy. Figure 6 It represents the access allocation of access points numbered 1, 2, and 3 on path 1 (Path1) planned in the random scenario. It can be seen that there are drones providing access services for these three mission points at any time.

[0143] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for continuous communication coverage of drone clusters under emergency communication, characterized in that: In an emergency rescue mission with multiple fixed emergency mission points and a remote charging station, perform the following steps: Construct a multi-drone tree-like backhaul network to ensure connection with the charging base station: each drone in the unmanned cluster performing emergency rescue tasks is used as a node, and K drones provide one-to-one communication access services for each emergency task point by hovering, and the remaining drones provide relay services or serve as redundant drones; the value of K is the same as the number of emergency task points; Based on the departure time of the drones from the charging base station, the multi-drone tree-like backhaul network is divided into multiple rotation paths, and the time intervals between adjacent drones on the same rotation path are the same; For each drone in the unmanned swarm performing emergency rescue missions, if it stays at the charging base station, it is defined as a dormant state, otherwise it is defined as a working state; and the state value of the drone in the dormant state is set to 0, and the state value of the drone in the working state is set to 1; For any drone in the drone cluster, the instantaneous energy consumption of the drone is obtained based on the product of its drone state value and the drone working power; the total energy consumption of the task E is obtained based on the instantaneous energy consumption of all drones within the task duration T. all ; Under given constraints, the total energy consumption of the task E is minimized. all For the optimization goal, the emergency communication deployment result is obtained based on the access service allocation A of the UAV, the trajectory Q of the UAV and the network topology N corresponding to the optimization solution result; The given constraints include: single UAV energy consumption constraint, one-to-one access service constraint, rotation path time interval constraint and multi-UAV tree backhaul network network connection constraint; in, The energy consumption constraints of a single UAV are as follows: the current remaining storage energy of the UAV is not less than the current working energy threshold; The current remaining storage energy of the drone is: The maximum storage energy of the drone E max The difference between the current energy consumption and the current energy consumption; the current energy consumption is the time T between the drone and the last time it departed from the charging base station at the current moment i (t) and the UAV working power P ω The product of The current working energy threshold is: the ratio of the distance between the current position of the drone and the charging base station to the flight speed of the drone and the working power P of the drone. ω The product of The specific constraints of one-to-one access service are: Define α k,i (t) represents the binary decision variable that UAV i provides communication access service for the kth emergency task point at time t. If it provides communication access service, then α k,i (t) = 1; otherwise α k,i (t) = 0; In solving the minimum total energy consumption E all When α j,i (t)Satisfy: Among them, I represents the unmanned cluster performing emergency rescue tasks; The time interval constraints of the rotation path are as follows: The time interval Δt of each rotation path r Greater than or equal to the preset minimum time interval T interval , where r is the rotation path identifier; The network connection constraints of the multi-drone tree-like backhaul network include: the nodes in the network are accessible, the charging base station and all drones providing access services are included in the topology of the multi-drone tree-like backhaul network at any time, and the network's acyclic tree constraint; The nodes in the network can be specifically: Define μ p,q (t) is a binary decision variable indicating whether a network connection is established between two nodes p and q in the network at time t. If established, μ p,q (t)=1; otherwise μ p,q (t) = 0; In solving the minimum total energy consumption E all When p,q (t)Satisfy: Where U represents the node set, U = I∪{station}, station represents the charging base station; the auxiliary quantity z p,q The value of (t) depends on the location of nodes p, q and the maximum relay distance D of the drone, which is expressed as: Among them, L p (t), L q (t) represents the position of node p and q at time t respectively; The charging base station and all drones providing access services are included in the multi-drone tree-like backhaul network topology at any time. The specific topology is: Define μ i,station (t) is a binary decision variable indicating whether a network connection is established between drone i and the charging base station at time t. If established, μ i,station (t)=1; otherwise μ i,station (t) = 0; In solving the minimum total energy consumption E all When , the topological constraints of the multi-UAV tree-like backhaul network satisfy: Among them, μ j,i (t) represents the binary decision variable of whether to establish a network connection between UAVs j and i providing access services at time t. If established, then μ j,i (t)=1; otherwise μ j,i (t)=0.

2. The method for continuous communication coverage of drone clusters under emergency communication as claimed in claim 1, characterized in that: The acyclic tree constraints of the network are specifically: Among them, S represents any subset of network nodes in the multi-UAV tree-like backhaul network.

3. The method for continuous communication coverage of drone clusters under emergency communication as claimed in claim 1, characterized in that: Under given constraints, the total energy consumption E of the task will be minimized all The optimization objective is transformed into: Where R represents the set of rotation paths, Δt r Represents the time interval of the rth rotation path.

4. The method for continuous communication coverage of drone clusters under emergency communication as claimed in claim 1, characterized in that: Under given constraints, on a given rotation path, the time interval Δt on the rotation path is maximized. r To achieve the minimum total energy consumption of the current task on the rotation path, where the time interval Δt on the rotation path is maximized r Specifically: Where b represents the hovering time of a non-fixed node on a rotation path, x and y represent the number of fixed nodes and non-fixed nodes on the current rotation path, respectively, and T max represents the maximum working time of the UAV, g represents the total flight time of the UAV from the charging base station to the return along the current rotation path, and l represents the time required for the UAV to fly at the maximum relay distance D.

5. The method for continuous communication coverage of drone clusters in emergency communication according to any one of claims 1 to 4, characterized in that: Under given constraints, the maximum-minimum ant colony algorithm is used to optimize the optimization objective.

Images