A method and system for scheduling multi-UAV collaborative communication and computing tasks

By optimizing the task upload and unloading strategies between drones on the SDN platform, and using the two-layer game method, the problems of insufficient resources and unbalanced load in drone-assisted edge computing are solved, and the low-latency multi-node Human-computer collaborative communication and computing task scheduling.

CN115243321BActive Publication Date: 2024-11-19NORTHWESTERN POLYTECHNICAL UNIV
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202210939647.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-08-05
Publication Date
2024-11-19
Estimated Expiration
2042-08-05

AI Technical Summary

Technical Problem

The existing drone-assisted edge computing does not consider the collaborative communication and calculation between drones when handling ground tasks, resulting in insufficient drone resources and power in user-intensive areas, long task processing delays, and inability to use user service quality. Guaranteed and unbalanced load.

Method used

Multi-UAV collaborative communication and computing task scheduling methods are adopted, and task uploading strategies and sub-task offloading strategies are determined on the SDN platform, and the two-layer game method is used to optimize the processing delay of tasks on the drone to achieve drone load balancing.

Benefits of technology

On the premise of meeting the energy consumption constraints of drones, the total processing delay of tasks is reduced, the load balancing of multiple drones is realized, and the computing and communication services with low latency are provided, which are suitable for outdoor scenarios without ground infrastructure.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115243321B_ABST
    Figure CN115243321B_ABST
Patent Text Reader

Abstract

The present invention discloses a method and system for scheduling multi-UAV cooperative communication and computing tasks, establishing a multi-UAV aerial computing platform; uploading tasks to UAVs according to an upload strategy with the minimum task upload delay, dividing tasks into associated subtask flows according to the characteristics of a task association model; based on the divided associated subtask flows, each computing UAV determines the computing UAV of a subtask according to the existing computing resources of the UAV, the computing resources required for the subtask, the UAV where the predecessor task of the subtask is located, and the air-to-air communication delay between UAVs, using a double-layer game method; the total processing delay of the task and the task scheduling scheme are obtained according to the minimum upload delay and the minimum unloading delay, so as to realize multi-UAV cooperative communication and computing task scheduling. Under the premise of meeting the energy consumption constraint of the UAV, the total processing delay of the task is minimized, and load balancing of multiple UAVs is realized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of unmanned aerial vehicle (UAV) aerial computing, and in particular relates to a method and system for scheduling multi-UAV collaborative communication and computing tasks. Background Art

[0002] Drone aerial computing, which loads computing servers onto drones, is regarded as an important technology for serving computing-intensive and delay-sensitive applications, and has received extensive attention and research from industry and academia. Due to the characteristics of drones such as fast movement, flexible deployment, and strong maneuverability, drones can be quickly deployed to designated locations, especially in disaster areas, epidemic prevention and control, battlefields, and other areas that lack ground infrastructure or resources. Furthermore, the line-of-sight link between the drone and the ground can reduce the latency of ground users in processing tasks.

[0003] As a mode of drone aerial computing, drone-assisted edge computing has developed from single drone-assisted edge computing to multi-drone-assisted edge computing. Considering the limited resources and power of a single drone, multi-drone-assisted edge computing has become a research hotspot. The more resources and larger service areas of multiple drones enable multi-drone-assisted edge computing to serve more application scenarios.

[0004] However, existing drone-assisted edge computing does not consider the collaborative communication and computing between drones when processing ground tasks. Affected by the coverage of drones, tasks can only be processed on drones that cover their areas. Therefore, in the case of uneven distribution of ground users, once drones in user-dense areas continue to provide services, it will cause problems such as insufficient drone resources and power, long task processing delays, inability to guarantee user service quality, and unbalanced loads on multiple drones. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a method and system for scheduling multi-UAV collaborative communication and computing tasks in response to the above-mentioned deficiencies in the prior art. In a multi-UAV aerial computing platform based on software defined networking (SDN), the task scheduling strategy is optimally selected to minimize the total processing delay of the task while satisfying the energy consumption constraints of the UAVs, thereby achieving multi-UAV load balancing.

[0006] The present invention adopts the following technical solutions:

[0007] A method for scheduling multi-UAV collaborative communication and computing tasks, comprising the following steps:

[0008] S1. Establish a multi-UAV aerial computing platform;

[0009] S2. Determine the upload strategy with the minimum task upload delay based on the communication delay between the ground user and each drone in the multi-drone aerial computing platform established in step S1 and the communication delay between the ground users, and calculate the minimum upload delay of the task;

[0010] S3, uploading the task to the drone according to the uploading strategy of step S2, and dividing the task into associated subtask flows according to the characteristics of the task association model;

[0011] S4. Based on the associated subtask flows divided in step S3, each computing UAV determines the computing UAV of the subtask using a double-layer game method according to the existing computing resources of the UAV, the computing resources required for the subtask, the UAV where the predecessor task of the subtask is located, and the air-to-air communication delay between UAVs, and calculates the minimum unloading delay obtained by processing the subtask on this UAV;

[0012] S5. According to the minimum upload delay obtained in step S2 and the minimum unloading delay obtained in step S4, the total processing delay and task scheduling scheme of the task are obtained to realize multi-UAV collaborative communication and computing task scheduling.

[0013] Specifically, in step S1, the multi-UAV aerial computing platform includes multiple UAVs equipped with communication and computing resources, and multiple ground users; the UAVs are connected via air-to-air communication links, the ground users are connected via ground-to-ground communication links, and the UAVs and ground users are connected via air-to-ground communication links.

[0014] Furthermore, the multi-UAV aerial computing platform is based on the SDN architecture, which divides the UAVs into management nodes and computing nodes. The management node acts as an SDN controller to centrally collect and record information of all computing nodes, including task execution status, data storage information, etc.; the computing node mainly executes tasks offloaded to the node and assists in data transmission. The tasks adopt a partial offloading mode, and each computing node maintains a queue of tasks to be executed.

[0015] Specifically, step S2 is as follows:

[0016] S201. Define binary variables Indicates whether subtask w(i) is sent to ground user k, It indicates that task w(i) is transmitted to ground user k; It means that task w(i) is not transmitted to ground user k;

[0017] S202, define the upload delay T of all tasks upl The mathematical expression of

[0018] S203, calculating the air-to-ground communication delay of user k1 directly transmitting task w(k1) to the UAV as task upload mode 1;

[0019] S204, calculating the ground-to-ground communication delay of user k1 transmitting task w(k1) to other user k2 in the same area, and calculating the air-to-ground communication delay of user k2 uploading task w(k1) to the drone as task upload mode 2;

[0020] S205, obtaining the task upload strategy and the corresponding task upload delay according to the task upload delay obtained by the upload mode in step S203 and step S204;

[0021] S206, repeat the above steps to obtain the upload strategy and total upload delay T of all tasks upl .

[0022] Furthermore, in step S202, the upload delay of all tasks is T upl for:

[0023]

[0024] Among them, I(k∈N v ) represents an indicator function, t G2G is the transmission delay of the task from ground user k to ground user k', t G2A is the time delay of uploading a task from ground user k to UAV v, i is any task, k is any ground user, N v is the set of tasks covered by the UAV, A[i] is the area where task i is located, Y is the scheduling strategy of tasks among ground users, and n is the number of tasks.

[0025] Specifically, in step S3, the task association model features include linear, tree and network.

[0026] Specifically, in step S4, the computing drone for the subtask is determined by using a double-layer game method:

[0027] S401, initializing the uninstallation strategy of all subtasks;

[0028] S402: Record the current time as t, and initialize the subtask set U that needs to be updated j (t) and the task set U i (t);

[0029] S403, calculating the delay t(w(i), s(j), t) of task w(i) at the current moment;

[0030] S404, modify the subtasks of task w(i) The unloading strategy at time t+1 is And calculate the unloading delay t(w(i),s(j),t+1) of the subtask at time t+1;

[0031] S405, determine whether t(w(i), s(j), t+1) is less than the unloading delay t(w(i), s(j), t) at time t;

[0032] S406. If t(w(i), s(j), t+1) is less than t(w(i), s(j), t), calculate the energy consumption of UAV v Determine whether the drone load e0 is exceeded;

[0033] S407, if the energy consumption of UAV v Less than or equal to the drone load e0, the subtask Add to collection j (t), and record the unloading strategy that minimizes its unloading delay;

[0034] S408, repeat steps S403 to S407 to process all subtasks of task w(i);

[0035] S409, judging the set U j (t) Whether it is empty;

[0036] S410, if the set U j (t) is not empty, randomly select a subtask from the set Update the subtask strategy to minimize the current subtask offloading delay;

[0037] S411, calculate the unloading delay t(w(i),t+1) of task w(i) at time t+1;

[0038] S412: If t(w(i), t+1) is less than the unloading delay t(w(i), t) at time t, add task w(i) to the set U. i (t) medium;

[0039] S413, repeat steps S403 to S412 to process all tasks;

[0040] S414, judging set U i (t) Whether it is empty;

[0041] S415, if it is not empty, randomly select a task w(i) in the set and update its strategy to minimize the unloading delay of its current task;

[0042] S416, update time to t+1, loop through the above steps to obtain the minimum task unloading delay T off And the corresponding task offloading scheme.

[0043] Furthermore, in step S406, the energy consumption of the drone v for:

[0044]

[0045] Among them, e h represents the minimum power of the drone hovering, η represents the power efficiency; ε is the effective switching capacitance of the drone, n is the number of tasks, m is the number of subtasks contained in the task, V is the drone set, p v represents the transmission power of UAV v, t A2A represents the delay of subtask scheduling in the air, For subtask Whether to unload to drone v', f v is the computing power of UAV v, c w(i),s(j) For subtask The number of CPU cycles required, For subtask Unload onto drone v.

[0046] Furthermore, in step S416, the minimum task unloading delay T off for:

[0047]

[0048] Among them, X is the offloading strategy of the subtask, and m is the number of subtasks contained in the task.

[0049] In a second aspect, an embodiment of the present invention provides a multi-UAV collaborative communication and computing task scheduling system, including:

[0050] Platform module, building a multi-UAV aerial computing platform;

[0051] The strategy module determines the upload strategy with the minimum task upload delay based on the communication delay between each drone in the multi-drone aerial computing platform established by the ground user and the platform module and the communication delay between ground users, and calculates the minimum upload delay of the task;

[0052] The division module uploads the task to the UAV according to the upload strategy of the strategy module, and divides the task into associated sub-task flows according to the characteristics of the task association model;

[0053] The game module is based on the associated subtask flows divided by the partition module. Each computing UAV uses a double-layer game method to determine the computing UAV of the subtask according to the existing computing resources of the UAV, the computing resources required for the subtask, the UAV where the predecessor task of the subtask is located, and the air-to-air communication delay between UAVs, and calculates the minimum unloading delay obtained by processing the subtask on this UAV.

[0054] The scheduling module obtains the total processing delay and task scheduling plan of the task based on the minimum upload delay obtained by the strategy module and the minimum unloading delay obtained by the game module, thereby realizing multi-UAV collaborative communication and computing task scheduling.

[0055] Compared with the prior art, the present invention has at least the following beneficial effects:

[0056] The present invention discloses a method for scheduling multi-UAV collaborative communication and computing tasks. In outdoor scenes without ground infrastructure, multiple UAVs are used as aerial computing platforms, which have the characteristics of rapid deployment and low cost, and can provide timely computing and communication services for sudden ground resource needs. Users upload tasks to UAVs for computing. Limited by the service range of a single UAV, users can only upload tasks to UAVs that cover them. At the same time, considering the instability and diversity of air-to-ground communication links, users can forward tasks to users with better air-to-ground communication links through D2D communication, and then upload tasks. Therefore, there are two upload strategies for tasks: direct upload and forwarding first and then uploading. After the task is uploaded to the UAV, in order to ensure the load balance of the UAV, the task is scheduled between the UAVs, and the calculation is selected on the appropriate UAV to finally complete the processing of the task.

[0057] Furthermore, the actual scenario is modeled, where the UAV is loaded with communication and computing resources to provide services to ground users, the UAV and ground users communicate via air-to-ground communication links, and the UAVs communicate with each other via air-to-air communication links.

[0058] Furthermore, to ensure efficient coordination of multiple drones, the roles of drones are divided. The management drone and the computing drone work together to facilitate the scheduling of tasks between drones. In addition, tasks are inherently related. Compared with the full offloading mode, the partial offloading mode is used to offload various parts of the task to different drones. Tasks will be calculated at a smaller granularity, resource allocation will be more flexible, and task processing latency can be effectively reduced.

[0059] Furthermore, the task upload strategy is divided into two types: one is to upload directly by the user, and the other is to forward to the user with good air-to-ground communication link performance before uploading. The user can choose the strategy with less upload delay to upload the task, thus reducing the total delay of task processing.

[0060] Furthermore, the task upload delay is related to the input data of the task and the air-ground communication rate. Task upload is the first step in the task processing process. By optimizing the task upload strategy, the task upload delay can be effectively reduced, thereby further reducing the total task processing delay.

[0061] Furthermore, typical task association models include linear, tree, and mesh. The characteristic of linear association is that the output data generated by each subtask is used as the input data of the next subtask; the characteristic of tree association is that the node at the top of the tree is the entry node of the computing task, and the remaining subtasks expand from the top downward; the characteristic of mesh association is that unlike tree association, there is no obvious hierarchical relationship between subtasks.

[0062] Furthermore, game theory is widely used to solve decision-making problems between multiple game participants with different goals. The computational drone selection problem of subtasks has a two-level game relationship, so a two-level game method is proposed. In the inner game, all subtasks in the subtask flow are game participants, and the approximate optimal computational offloading strategy of the subtask flow is found after a finite number of games; in the outer game, all subtask flows are game participants, and the approximate optimal computational offloading strategy of all players is found after a finite number of games, and finally the minimum task offloading delay is obtained.

[0063] Furthermore, considering the limited energy of UAVs, in order to ensure that UAVs can provide long-term communication and computing services, the energy consumption of UAVs cannot exceed the available energy load.

[0064] Furthermore, the minimum unloading delay of the task is related to the processing delay of each subtask. By optimizing the unloading strategy of the subtask, the minimum unloading delay of the task can be reduced, thereby effectively reducing the total processing delay of the task.

[0065] It can be understood that the beneficial effects of the second aspect mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here.

[0066] In summary, the present invention proposes a multi-UAV collaborative communication and computing method for outdoor scenarios without ground infrastructure, which ensures the load balancing of UAVs while providing low-latency services for task processing of ground users.

[0067] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 This is a diagram of the multi-UAV cooperative communication and computing model described in the present invention;

[0069] Figure 2 The task flow model diagram of the present invention;

[0070] Figure 3 It is a flow chart of the double-layer game method of the present invention;

[0071] Figure 4 This is a comparison diagram of the total task processing delay in the partial unloading and full unloading modes of the present invention;

[0072] Figure 5 Comparison of upload latency for missions with and without ground user-to-ground communication strategy optimization;

[0073] Figure 6 Comparison of task offloading latency with and without offloading strategy optimization;

[0074] Figure 7 It is a comparison chart of the total task processing delay of the traditional multi-UAV solution (without collaboration) and the multi-UAV collaboration solution proposed by the present invention under different task flow models;

[0075] Figure 8 A comparison chart of the total energy consumption of drones using the traditional multi-drone solution (without collaboration) and the multi-drone collaboration solution proposed in the present invention to process different subtask flow models;

[0076] Fig. 9 is the average convergence times graph of the two-layer game method;

[0077] Fig.10 This is a comparison chart of the total task processing delay between the greedy method and the two-layer game method;

[0078] Fig.11 This is a comparison chart of the processing delay of each task when processing linear task flows by the traditional multi-UAV solution (without collaboration) and the multi-UAV collaboration solution proposed in the present invention;

[0079] Fig.12 The figure is a comparison chart of the energy consumption of each drone when processing linear task flows by the traditional multi-drone solution (without collaboration) and the multi-drone collaboration solution proposed in the present invention. DETAILED DESCRIPTION

[0080] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0081] In the description of the present invention, it should be understood that the terms “include” and “comprises” indicate the presence of described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.

[0082] It should also be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0083] It should be further understood that the term "and / or" used in the present specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes these combinations. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects are in an "or" relationship.

[0084] It should be understood that, although the terms first, second, third, etc. may be used to describe preset ranges, etc. in the embodiments of the present invention, these preset ranges should not be limited to these terms. These terms are only used to distinguish preset ranges from each other. For example, without departing from the scope of the embodiments of the present invention, the first preset range may also be referred to as the second preset range, and similarly, the second preset range may also be referred to as the first preset range.

[0085] The word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to determining" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)", depending on the context.

[0086] Various structural schematic diagrams of the embodiments disclosed in the present invention are shown in the accompanying drawings. These figures are not drawn to scale, and some details are magnified and some details may be omitted for the purpose of clear expression. The shapes of various regions and layers shown in the figures and the relative sizes and positional relationships therebetween are only exemplary, and may deviate in practice due to manufacturing tolerances or technical limitations, and those skilled in the art may additionally design regions / layers with different shapes, sizes, and relative positions according to actual needs.

[0087] The present invention provides a method for scheduling multi-UAV cooperative communication and computing tasks, establishing a multi-UAV aerial computing platform; determining an upload strategy with the minimum task upload delay according to the communication delay between a ground user and a UAV and the communication delay between ground users; uploading tasks to UAVs according to the upload strategy, dividing tasks into associated subtask flows according to the characteristics of a task association model; based on the divided associated subtask flows, each computing UAV determines the computing UAV of a subtask by using a double-layer game method according to the existing computing resources of the UAV, the computing resources required for the subtask, the UAV where the predecessor task of the subtask is located, and the air-to-air communication delay between the UAVs; obtaining the total processing delay of the task and the task scheduling scheme according to the minimum upload delay and the minimum unloading delay, and realizing multi-UAV cooperative communication and computing task scheduling. Under the premise of meeting the energy consumption constraint of the UAV, the total processing delay of the task is minimized, and multi-UAV load balancing is realized.

[0088] The present invention provides a method for scheduling multi-UAV cooperative communication and computing tasks, comprising the following steps:

[0089] S1. Establish a multi-UAV aerial computing platform;

[0090] See also Figure 1 , including multiple drones equipped with communication and computing resources, and multiple ground users. The drones are connected via air-to-air communication links, the ground users are connected via ground-to-ground communication links, and the drones and ground users are connected via air-to-ground communication links.

[0091] The details of the multi-UAV aerial computing platform are as follows:

[0092] The platform is based on the SDN architecture and divides drones into management nodes and computing nodes. The management node acts as an SDN controller to centrally collect and record information from all computing nodes, including task execution status, data storage information, etc. The computing node mainly executes tasks offloaded to the node and assists in data transmission. The task adopts a partial offload mode, and each computing node maintains a queue of tasks to be executed.

[0093] The information of multiple drones is represented as:

[0094] U={V,F,P,L,N}

[0095] Where V = {1, 2, ..., |V|} represents the serial number set of drones, |V| represents the number of drones, and F = {f1, f2, ..., f |V|} represents the computing power of the drone, P = {p1, p2, ..., p |V|} represents the transmission power of the drone, L={(x1,y1),(x2,y2),...,(x |V| ,y|V| )} represents the horizontal position of the drone, N={N1,N2,...,N |V|} represents the set of users covered by the drone. In addition, the drone is hovering in the air at a height of H.

[0096] There are n users on the ground, and the information of each user is expressed as:

[0097] G k ={p k ,(x k ,y k )}

[0098] Among them, p k represents the transmission power of user k, (x k ,y k ) represents the horizontal coordinate of the user. In a certain period of time, each user has only one task to be offloaded to the drone.

[0099] A task that contains multiple associated subtasks is defined as a task flow. The set of task flows in this system is represented as:

[0100] W={w(1),w(2),...,w(n)}

[0101] The subtasks contained in task w(i) are expressed as:

[0102] w(i)={s(1),s(2),...,s(m)}

[0103] The information of the subtask is represented as:

[0104] φ w(i),s(j) =(α w(i),s(j) ,β w(i),s(j) ,c w(i),s(j) )

[0105] Among them, α w(i),s(j) represents the input data size of the subtask, β w(i),s(j) Indicates the output data size of the subtask, c w(i),s(j) Represents the computing power required by the subtask.

[0106] The mathematical expression of the air-to-ground communication rate between the ground user and the UAV is:

[0107]

[0108] Among them, B G represents the bandwidth of air-to-air communication; g k,v represents the channel gain between the ground user k and the UAV v; represents the signal interference power of other users in the coverage area of ​​UAV v; N0 represents the noise power.

[0109] The mathematical expression of the time delay for uploading a task from ground user k to drone v is:

[0110]

[0111] The mathematical expression of the ground-to-ground communication rate between ground users is:

[0112]

[0113] Among them, B U represents the bandwidth of air-to-air communication; g k,k' represents the channel gain between ground user k and k'; A[k] represents the drone coverage area where ground user k is located. represents the signal interference power from other users in the same UAV coverage area as ground user k.

[0114] The mathematical expression of the time delay of task transmission from ground user k to ground user k' is:

[0115]

[0116] The mathematical expression of the air-to-air communication rate between drones is:

[0117]

[0118] Among them, B A represents the bandwidth for air-to-air communications; represents the path loss of communication between UAV v and UAV v'; σ 2 represents the power spectral density of Gaussian noise.

[0119] The mathematical expression of the time delay of the task transmission from UAV v to UAV v' is:

[0120]

[0121] Subtasks The mathematical expression of the calculation delay is:

[0122]

[0123] in, Represents the arrival rate of subtasks on UAV v.

[0124] S2, task upload strategy optimization;

[0125] According to the communication delay between ground users and UAVs and the communication delay between ground users, the upload strategy with the minimum task upload delay is determined.

[0126] The task upload strategy optimization specifically includes the following steps:

[0127] S201. Define binary variables Indicates whether subtask w(i) is sent to ground user k, It indicates that task w(i) is transmitted to ground user k; It means that task w(i) is not transmitted to ground user k;

[0128] S202, define the upload delay T of all tasks upl The mathematical expression is:

[0129]

[0130] Among them, I(k∈N v ) represents an indicator function, I(k∈N v )=1 indicates that the ground user k is in the coverage area of ​​the UAV v; otherwise, it indicates that the ground user k is not in the coverage area of ​​the UAV v.

[0131] S203, task upload mode 1: calculating the air-to-ground communication delay of user k1 directly transmitting task w(k1) to the drone;

[0132] S204, task upload mode 2: calculate the ground-to-ground communication delay of user k1 transmitting task w(k1) to other user k2 in the same area; calculate the air-to-ground communication delay of user k2 uploading task w(k1) to the drone. The total task upload delay is the sum of the ground-to-ground communication delay and the air-to-ground communication delay.

[0133] S205, calculating the task upload delay obtained in task upload mode 1, and in task upload mode 2, calculating the total delay obtained by different users k2, and selecting the user k2 that minimizes the total delay; according to the task upload delays obtained in the two upload modes, obtaining the task upload strategy (selecting upload mode 1 or upload mode 2) and the corresponding task upload delay;

[0134] S206, repeat the above steps to obtain the upload strategy and total upload delay T of all tasks upl .

[0135] S3. After the task is uploaded to the drone, the task is divided into associated subtask flows according to the characteristics of three classic task association models, namely linear, tree and mesh;

[0136] See also Figure 2,The intrinsic association characteristics of tasks are as follows: (a) Linear association: the output data generated by each subtask serves as the input data of the next subtask; (b) Tree association: the node at the top of the tree is the entry node of the computing task, and the remaining subtasks expand from the top to the bottom; (c) Mesh association: different from the tree association, there is no obvious hierarchical relationship between subtasks.

[0137] S4, task offloading strategy optimization;

[0138] Each computing UAV uses a two-layer game method to determine the computing UAV for the subtask based on the existing computing resources of the UAV, the computing resources required for the subtask, the UAV where the predecessor task of the subtask is located, and the air-to-air communication delay between UAVs.

[0139] In the task offloading strategy optimization process, the total delay from uploading the task to the completion of the task execution is defined as the task offloading delay, including the delay of task transmission between drones, queuing delay and calculation delay. In the tree and mesh task models, some subtasks are executed in parallel. Therefore, the task offloading delay depends on the maximum offloading delay of the subtask. For a subtask, the set of all predecessor subtasks of the subtask s(j) of the task is defined as S(w(i),s(j)).

[0140] Define sub-binary variables Indicates the offloading strategy of the subtask. Represents subtask φ w(i),s(j) Offload the calculation to the drone v, Represents subtask φ w(i),s(j) The calculation will not be offloaded to the drone v.

[0141] Subtasks The output results of all predecessor tasks are transmitted to the drone where the subtask is located, and the subtask can be scheduled for execution. The transmission of the predecessor subtask and the unloading process of the subtask are concurrent, so the processing delay of the task depends on the predecessor subtask. Processing latency and subtasks The maximum value of the transmission delay, that is, max{t(w(i),s(j')),t A2A (w(i),s(j),v,v')}.

[0142] Subtasks The mathematical expression of the processing delay is:

[0143]

[0144] The processing time of a task depends on the maximum delay of its subtasks.

[0145] The processing delay of task w(i) is defined as the maximum value of the uplink delay from user i to the UAV and the processing delay of all subtasks in the task. Its mathematical expression is:

[0146]

[0147] The offloading delay of all tasks is T off The mathematical expression is:

[0148]

[0149] The mathematical expression for calculating the energy consumption of a drone is:

[0150]

[0151] Among them, e h represents the minimum power of the drone to hover, η represents the power efficiency; ε is the effective switching capacitance of the drone.

[0152] See also Figure 3 , the two-layer game method is as follows:

[0153] S401, initializing the uninstallation strategy of all subtasks;

[0154] S402: Record the current time as t, and initialize the subtask set U that needs to be updated j (t) and the task set U i (t);

[0155] S403, for task w(i), calculate its current delay t(w(i), s(j), t);

[0156] S404, modify its subtask The unloading strategy at time t+1 is And calculate the unloading delay t(w(i),s(j),t+1) of the subtask at time t+1;

[0157] S405, determine whether t(w(i), s(j), t+1) is less than the unloading delay t(w(i), s(j), t) at time t;

[0158] S406. If t(w(i), s(j), t+1) is less than t(w(i), s(j), t), calculate the energy consumption of UAV v Determine whether the drone load e0 is exceeded;

[0159] S407, if Less than or equal to e0, the subtask Add to collection j(t), and record the unloading strategy that minimizes its unloading delay;

[0160] S408, repeat S403 to S407, and perform the processing from S403 to S407 on all subtasks of task w(i);

[0161] S409, judging the set U j (t) Whether it is empty;

[0162] S410, if not empty, randomly select a subtask in the set Update its strategy to minimize the offloading delay of its current subtask;

[0163] S411, calculate the unloading delay t(w(i),t+1) of task w(i) at time t+1;

[0164] S412: If t(w(i), t+1) is less than the unloading delay t(w(i), t) at time t, add task w(i) to the set U. i (t) medium;

[0165] S413, repeat S403 to S412 to process all tasks;

[0166] S414, judging set U i (t) Whether it is empty;

[0167] S415, if it is not empty, randomly select a task w(i) in the set and update its strategy to minimize the unloading delay of its current task;

[0168] S416, update time to t+1, loop through the above steps to obtain the minimum task unloading delay T off And the corresponding task offloading scheme.

[0169] S5. According to the obtained minimum upload delay and minimum unload delay, the total processing delay of the task and the task scheduling plan are obtained.

[0170] In another embodiment of the present invention, a multi-UAV collaborative communication and computing task scheduling system is provided, which can be used to implement the above-mentioned multi-UAV collaborative communication and computing task scheduling method. Specifically, the multi-UAV collaborative communication and computing task scheduling system includes a platform module, a strategy module, a partitioning module, a game module and a scheduling module.

[0171] Among them, the platform module builds a multi-UAV aerial computing platform;

[0172] The strategy module determines the upload strategy with the minimum task upload delay based on the communication delay between each drone in the multi-drone aerial computing platform established by the ground user and the platform module and the communication delay between ground users, and calculates the minimum upload delay of the task;

[0173] The division module uploads the task to the UAV according to the upload strategy of the strategy module, and divides the task into associated sub-task flows according to the characteristics of the task association model;

[0174] The game module is based on the associated subtask flows divided by the partition module. Each computing UAV uses a double-layer game method to determine the computing UAV of the subtask according to the existing computing resources of the UAV, the computing resources required for the subtask, the UAV where the predecessor task of the subtask is located, and the air-to-air communication delay between UAVs, and calculates the minimum unloading delay obtained by processing the subtask on this UAV.

[0175] The scheduling module obtains the total processing delay and task scheduling plan of the task based on the minimum upload delay obtained by the strategy module and the minimum unloading delay obtained by the game module, thereby realizing multi-UAV collaborative communication and computing task scheduling.

[0176] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention described and shown in the drawings here can usually be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0177] See also Figure 4 , Figure 4 The comparison of the total processing delay of tasks in two offloading modes, partial offloading and full offloading, is given. The results show that under the three task association models, the total processing delay of partial offloading is lower than that of full offloading mode, and is reduced by 21%, 30% and 20% respectively.

[0178] See also Figure 5 , Figure 5The comparison of the minimum uplink delay of tasks without and with the optimization of communication strategies between users is given. The results show that the minimum uplink delay is reduced after optimizing the communication strategy between users. Since the dynamic link between air-ground communication will cause a long uplink delay of the task, ground cooperative communication uses the communication resources of users to help complete the uplink task, so that some remote users can transfer tasks to users with better air-ground communication links. Therefore, by optimizing the communication strategy between users, the uplink delay can be reduced.

[0179] See also Figure 6 , Figure 6 The minimum queuing delay and computation delay of the linear subtask flow without and with offloading strategy optimization are compared. The results show that the queuing and computation delays can be reduced by optimizing the offloading strategy. By sending subtasks to idle drones with abundant resources, low-latency and high-quality services can be provided. Therefore, communication between drones helps to effectively utilize the computing resources of drones.

[0180] See also Figure 7 and Figure 8 , the total minimum delay and energy consumption of the three subtask flow models are compared when the traditional multi-UAV (non-cooperative) solution and the multi-UAV cooperative solution proposed in this invention are used. Figure 7 (a)- Figure 7 (c) It can be seen that the multi-UAV cooperative scheme outperforms the traditional multi-UAV scheme (without cooperation) in processing the subtask flows of linear models, tree models, and grid models, with a reduction of more than 50.0%. However, for the total energy consumption of UAVs, Figure 8 (a)- Figure 8 (c) It can be seen that when processing the three subtask flow models, the UAV cooperation scheme consumes more energy than the traditional scheme, and the increase is less than 5%. This is because the communication between UAVs consumes more energy. These results show that the multi-UAV cooperation scheme greatly reduces the overall processing delay of the task while increasing the energy consumption of fewer UAVs.

[0181] See also Fig. 9 , Fig. 9 This is the average iterative process of the convergence of the double-layer game algorithm used in the present invention. This line shows a linear trend as the number of subtask flows increases, which shows that the algorithm has fast convergence speed, good stability, and good scalability as the number of computing tasks increases.

[0182] See also Fig.10 , Fig.10The performance of the greedy-based method and the two-layer game algorithm proposed in this invention are compared. It can be seen that as the number of subtask flows increases, the task processing delay of the two-layer game algorithm is always smaller than that of the greedy-based method. From the comparison results, the effectiveness and high performance of the proposed method are proved.

[0183] See also Fig.11 , Fig.11 In order to solve the load balancing problem in the multi-drone collaborative solution, the delay performance of each subtask flow was tested under the condition of n=50 subtask flows, m=5 tasks in each subtask flow, and |V|=10 drones. As can be seen from the figure, the delay data of the traditional multi-drone solution (without collaboration) fluctuates greatly, while the data of the multi-drone collaborative solution is stable, and the task processing delay value is smaller than that of the traditional multi-drone solution. The experimental results show that the multi-drone writing solution can provide stable low-latency services.

[0184] See also Fig.12 , Fig.12 The energy consumption of each drone is compared. The comparison results show that in the multi-drone collaboration scheme, when executing 50 subtask flows, the energy consumed by each drone is close to the average energy consumption. However, the energy consumption of drones in the traditional multi-drone scheme (without collaboration) varies greatly. The energy consumption of some drones is much higher than the average energy consumption, causing them to run out of power and unable to provide computing services to users. In short, the multi-drone collaboration scheme proposed in the present invention has good load balancing performance.

[0185] Example

[0186] Without loss of generality, this implementation example considers a ground user randomly distributed in 1000×1000m 2 Multi-UAV scenario in a region. For the user-generated subtasks, it is assumed that the input and output data sizes are set to 5-10MB and 0.1-0.5MB respectively, and the CPU cycles required for the subtasks are between 500 and 1000 megacycles. The drones hover at a fixed altitude of 10m, and the hovering energy consumption of each drone is 220W per unit time, with an energy budget of 500kJ. The drone computing power is set to 6-8 gigacycles per second. In addition, the transmit power of users and drones is set between 30-50mW and 150-200mW respectively. The channel bandwidths for inter-user communication, air-to-air communication, and air-to-ground communication are set to 1MHz, 1MHz, and no more than 1MHz, respectively.

[0187] This embodiment simulates the calculation, communication and other parameters of the drone and the user, which is more in line with engineering practice, and is simulated based on the embodiment.

[0188] In summary, the present invention provides a method and system for scheduling multi-UAV collaborative communication and computing tasks, which can realize reasonable task scheduling according to the type of task and the resources of the multi-UAV computing platform, and has better performance in reducing task processing delay and realizing UAV load balancing compared with traditional methods.

[0189] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0190] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0191] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0192] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0193] The above contents are only for explaining the technical idea of ​​the present invention and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.

Claims

1. A method for scheduling multi-UAV collaborative communication and computing tasks, characterized in that: The following steps are involved: S1. Establish a multi-UAV aerial computing platform; S2. Determine the upload strategy with the minimum task upload delay based on the communication delay between the ground user and each drone in the multi-drone aerial computing platform established in step S1 and the communication delay between the ground users, and calculate the minimum upload delay of the task; S3, uploading the task to the drone according to the uploading strategy of step S2, and dividing the task into associated subtask flows according to the characteristics of the task association model; S4. Based on the associated subtask flows divided in step S3, each computing drone uses a double-layer game method to determine the computing drone for the subtask according to the existing computing resources of the drone, the computing resources required for the subtask, the drone where the predecessor task of the subtask is located, and the air-to-air communication delay between drones, and calculates the minimum unloading delay obtained by processing the subtask on this drone; the specific method of determining the computing drone for the subtask using the double-layer game method is: S401, initializing the uninstallation strategy of all subtasks; S402, record the current time as t , initialize the subtask set that needs to be updated and task collection ; S403, calculation task w ( i ) The delay at the current moment ; S404, modify the task w ( i ) exist t The uninstallation strategy at time +1 is , and calculate t +1 time subtask unloading delay ; S405, judgment Is it less than t Unloading delay at any time ; S406, if Less than , computing drones v Energy consumption , determine whether the drone load is exceeded , drones v Energy consumption for: in, Represents the minimum power of the drone to hover, represents power efficiency; is the effective switching capacitor of the drone, is the number of tasks, is the number of subtasks contained in the task, For drone collection, represents the transmission power of the UAV v, represents the delay of subtask scheduling in the air, For subtask Whether to unload to drone , is the computing power of the drone v, For subtask The number of CPU cycles required, For subtask Offloading to drone v superior; S407, If the drone v Energy consumption Less than or equal to drone payload , the subtask Add to Collection and record the unloading strategy that minimizes the unloading delay; S408, repeat steps S403 to S407, and w ( i ) to process all subtasks; S409, judging the set Is it empty? S410, if the collection Not empty, randomly select a subtask from the collection Update the subtask strategy to minimize the current subtask offloading delay; S411, Computational Tasks w ( i )exist t +1 time unloading delay ; S412, if Less than t Unloading delay at any time , the task w ( i ) Add to collection middle; S413, repeat steps S403 to S412 to process all tasks; S414, judge the set Whether it is empty; S415, if not empty, randomly select a task in the set w ( i ) Update its strategy to minimize the offloading delay of its current task; S416, update time is t +1, loop through the above steps to get the minimum unloading delay of the task And the corresponding task offloading scheme, the minimum task offloading delay for: in, is the offloading strategy of the subtask, The number of subtasks contained in the task; S5. According to the minimum upload delay obtained in step S2 and the minimum unloading delay obtained in step S4, the total processing delay and task scheduling scheme of the task are obtained to realize multi-UAV collaborative communication and computing task scheduling.

2. The method for scheduling multi-UAV cooperative communication and computing tasks according to claim 1, characterized in that: In step S1, the multi-UAV aerial computing platform includes multiple UAVs equipped with communication and computing resources, and multiple ground users; the UAVs are connected via air-to-air communication links, the ground users are connected via ground-to-ground communication links, and the UAVs and ground users are connected via air-to-ground communication links.

3. The method for scheduling multi-UAV cooperative communication and computing tasks according to claim 2, characterized in that: The multi-UAV aerial computing platform is based on the SDN architecture, which divides the UAVs into management nodes and computing nodes. The management node acts as an SDN controller to centrally collect and record the information of all computing nodes, including task execution status, data storage information, etc. The computing nodes mainly execute the tasks offloaded to the nodes and assist in data transmission. The tasks adopt the partial offloading mode, and each computing node maintains a queue of tasks to be executed.

4. The method for scheduling multi-UAV cooperative communication and computing tasks according to claim 1, characterized in that: Step S2 is specifically as follows: S201. Define binary variables Represents a subtask w ( i ) whether to send to ground users k , Representation Task w ( i ) is transmitted to ground users k ; Representation Task w ( i ) is not transmitted to ground users k ; S202. Define the upload delay of all tasks The mathematical expression of S203. Calculate users k1 The task Air-to-ground communication latency transmitted directly to the UAV as mission upload mode 1; S204. Calculate users k1 The task Transmit to other users in the same area k2 The ground-to-ground communication delay is calculated for the user k2 The task The air-to-ground communication latency uploaded to the drone as mission upload mode 2; S205, obtaining the task upload strategy and the corresponding task upload delay according to the task upload delay obtained by the upload mode in step S203 and step S204; S206: Repeat the above steps to obtain the upload strategy and total upload delay of all tasks. .

5. The method for scheduling multi-UAV cooperative communication and computing tasks according to claim 4, characterized in that: In step S202, the upload delay of all tasks for: in, represents an indicator function, For tasks from ground users k Transmission to ground users k’ The delay, For tasks from ground users k Upload to drone v The delay, For any task, For any ground user, The set of tasks covered by the drone, For the task i The area where you are located, is the scheduling strategy for tasks among ground users, is the number of tasks.

6. The method for scheduling multi-UAV cooperative communication and computing tasks according to claim 1, characterized in that: In step S3, the task association model features include linear, tree and network.

7. A multi-UAV collaborative communication and computing task scheduling system, characterized in that: include: Platform module, used to build a multi-UAV aerial computing platform; The strategy module is used to determine the upload strategy with the minimum task upload delay according to the communication delay between each drone in the multi-drone aerial computing platform established by the ground user and the platform module and the communication delay between ground users, and calculate the minimum upload delay of the task; A division module is used to upload tasks to the UAV according to the upload strategy of the strategy module, and divide the tasks into associated sub-task flows according to the characteristics of the task association model; The game module is used for determining the computing drone of the subtask based on the associated subtask flows divided by the division module. Each computing drone uses a double-layer game method to determine the computing drone of the subtask according to the existing computing resources of the drone, the computing resources required for the subtask, the drone where the predecessor task of the subtask is located, and the air-to-air communication delay between drones. The minimum unloading delay obtained by processing the computing subtask on this drone is determined by the double-layer game method. Specifically, the computing drone of the subtask is determined as follows: Initialize the unloading strategy of all subtasks; record the current time as t , initialize the subtask set that needs to be updated and task collection ; Computational tasks w ( i ) The delay at the current moment ; Modify the task w ( i ) exist t The uninstallation strategy at time +1 is , and calculate t +1 time subtask unloading delay ;judge Is it less than t Unloading delay at any time ;if Less than , computing drones v Energy consumption , determine whether the drone load is exceeded If the drone v Energy consumption Less than or equal to drone payload , the subtask Add to Collection and record the unloading strategy that minimizes the unloading delay; repeat the task w ( i ) to process all subtasks; judge the set Is it empty? If the collection Not empty, randomly select a subtask from the collection Update the subtask strategy to minimize the current subtask offloading delay; Computational tasks w ( i )exist t +1 time unloading delay ;if Less than t Unloading delay at any time , the task w ( i )Add to collection Repeat the process for all tasks; judge the set Whether it is empty; If not empty, randomly select a task from the collection w ( i ) Update its strategy to minimize the offloading delay of its current task; Update time is t +1, loop through the above steps to get the minimum unloading delay of the task And the corresponding task offloading scheme; Drones v Energy consumption for: in, Represents the minimum power of the drone to hover, represents power efficiency; is the effective switching capacitor of the drone, is the number of tasks, is the number of subtasks contained in the task, For drone collection, represents the transmission power of UAV v, represents the delay of subtask scheduling in the air, For subtask Whether to unload to drone , is the computing power of the drone v, For subtask The number of CPU cycles required, For subtask Offloading to drone v superior; Minimum task offloading delay for: in, is the offloading strategy of the subtask, The number of subtasks contained in the task; The scheduling module is used to obtain the total processing delay and task scheduling plan of the task according to the minimum upload delay obtained by the strategy module and the minimum unloading delay obtained by the game module, so as to realize multi-UAV collaborative communication and computing task scheduling.

8. The multi-UAV cooperative communication and computing task scheduling system according to claim 7, characterized in that: In the platform module, the multi-UAV aerial computing platform includes multiple UAVs equipped with communication and computing resources, and multiple ground users; UAVs are connected through air-to-air communication links, ground users are connected through ground-to-ground communication links, and UAVs and ground users are connected through air-to-ground communication links. The multi-UAV aerial computing platform is based on the SDN architecture and divides UAVs into management nodes and computing nodes. The management node acts as an SDN controller to centrally collect and record information of all computing nodes, including task execution status, data storage information, etc. The computing nodes mainly execute the tasks offloaded to the nodes and assist in data transmission. The tasks adopt the partial offloading mode, and each computing node maintains a queue of tasks to be executed.

9. The multi-UAV cooperative communication and computing task scheduling system according to claim 7, characterized in that: In the strategy module, define binary variables Represents a subtask w ( i ) whether to send to ground users k , Representation Task w ( i ) is transmitted to ground users k ; Representation Task w ( i ) is not transmitted to ground users k ; Define the upload delay for all tasks The mathematical expression of Counting users k1 The task Air-to-ground communication latency transmitted directly to the drone as task upload mode 1; calculate user k1 The task Transmit to other users in the same area k2 The ground-to-ground communication delay is calculated for the user k2 The task The air-to-ground communication delay uploaded to the UAV is used as task upload mode 2; according to the task upload delay obtained by the upload mode, the task upload strategy and the corresponding task upload delay are obtained; Repeat the above steps to obtain the upload strategy and total upload delay of all tasks. .

10. The multi-UAV cooperative communication and computing task scheduling system according to claim 9, characterized in that: Upload latency of all tasks for: in, represents an indicator function, For tasks from ground users k Transmission to ground users k’ The delay, For tasks from ground users k Upload to drone v The delay, For any task, For any ground user, The set of tasks covered by the drone, For the task i The area where you are located, is the scheduling strategy for tasks among ground users, is the number of tasks.

Citation Information

Cited By

  • Low-altitude multi-unmanned aerial vehicle cooperative prevention and control scheduling optimization method

    CN121900453A