A task offloading method based on unmanned aerial vehicle cooperation

Abstract

The present application relates to a kind of task unloading methods based on unmanned aerial vehicle cooperation, comprising: constructing unmanned aerial vehicle cooperative task unloading system;According to the CPU cycle number that ground user can execute in unit time and the CPU cycle number of task, create the local task unloading model of ground user;According to the launch power of ground user, the CPU cycle number that MEC server on original unmanned aerial vehicle and cooperation unmanned aerial vehicle can execute in unit time, the CPU cycle number of task, create the unmanned aerial vehicle task unloading model of ground user;According to the hovering power and hovering delay of unmanned aerial vehicle, create the hovering model of unmanned aerial vehicle;According to the local task unloading model of ground user, the unmanned aerial vehicle task unloading model of ground user and the hovering model of unmanned aerial vehicle, the optimal task unloading allocation scheme is calculated using second price auction algorithm to unload task, the present application uses unmanned aerial vehicle auxiliary mobile edge computing to carry out task unloading, improve the service quality and service experience of user.

Description

Technical Field

[0001] This invention belongs to the field of unmanned aerial vehicles (UAVs), specifically relating to a task offloading method based on UAV collaboration. Background Technology

[0002] Unmanned Aerial Vehicles (UAVs) possess characteristics such as rapid deployment, high flexibility, strong adaptability, powerful functionality, and low cost, making them promising candidates for applications in wireless communication systems and cellular networks. UAVs can act as relay communication devices and flying mobile terminals, greatly contributing to auxiliary communication and improving network capacity. Mobile Edge Computing (MEC) is a technology that performs computation on task resources at the edge, capable of handling resource-intensive and latency-sensitive tasks, improving user service quality (QoS). It is currently widely used in areas such as connected vehicles, artificial intelligence, and the Internet of Things (IoT).

[0003] Drone mobile edge computing is a combination of drones and mobile edge computing, which involves installing mobile edge servers on drones. It combines the advantages of drones and mobile edge computing, enabling the establishment of a good line-of-sight link with ground users (GUs) and the real-time processing of complex and latency-sensitive tasks.

[0004] For task offloading scenarios in remote areas (deserts, plateaus), where populations are sparse and basic communication equipment is lacking, UAV communication plays a more advantageous role compared to ground-based infrastructure. It can not only promptly address the problem of low user offloading efficiency but also fully utilize the low cost and high mobility of UAVs to improve communication service quality and channel gain. However, UAV energy consumption and task offloading latency are critical influencing factors in UAV communication. In recent years, many studies have focused on optimizing UAV energy consumption, task offloading latency, or a weighted sum of these factors. However, they haven't considered how to incentivize ground users to offload tasks to UAVs—whether to assign tasks to UAVs under appropriate circumstances or offload them locally when inappropriate. Furthermore, most studies haven't adequately addressed the task allocation issues among multiple UAVs and multiple ground users, failing to develop reasonable task allocation strategies to maximize the benefits of UAV edge computing systems. Summary of the Invention

[0005] The purpose of this invention is to propose a task offloading method based on UAV collaboration, which improves the task offloading efficiency for users in remote areas, and rationally plans the task offloading allocation between UAVs and ground users. The method includes:

[0006] S1: Construct a UAV collaborative task unloading system, which includes: M = {1,...m...,M} ground users and N = {1,...n...,N} UAVs equipped with MEC servers; when a ground user communicates directly with any UAV, the ground user generates a task that needs to be unloaded;

[0007] S2: Create a local task offloading model for ground users based on the number of CPU cycles that ground users can execute per unit time, the number of CPU cycles for tasks, the data size of tasks, and the tolerable latency of tasks.

[0008] S3: Create a drone task offloading model for the ground user based on the ground user's transmit power, the number of CPU cycles that the MEC servers on the original drone and the cooperative drone can execute per unit time, the number of CPU cycles of the task, the data size of the task, and the tolerable latency of the task; where the original drone represents the drone that communicates directly with the ground user, and the cooperative drone represents the drone that communicates indirectly with the ground user.

[0009] S4: Create a hovering model for the drone based on its hovering power and hovering latency;

[0010] S5: Based on the local task offloading model of the ground user, the UAV task offloading model of the ground user, and the hovering model of the UAV, the optimal task offloading allocation scheme is calculated using the second price auction algorithm to offload the task.

[0011] Furthermore, the local task offloading model for ground users includes:

[0012]

[0013]

[0014]

[0015] in, This represents the task generated at time t when ground user m communicates directly with drone n. For the task Data size, For the task CPU cycle count For the task Tolerable delay; Indicates task Delay in local uninstallation; Indicates task Energy consumption for local unloading; f mk represents the number of CPU cycles that ground user m can execute per unit of time; m This represents the effective capacitance coefficient of ground user m. Indicates the battery capacity of ground users; Indicates task Tolerable delay.

[0016] Furthermore, the drone mission offloading model for creating ground users includes:

[0017] S31: Calculate the uplink transmission rate between the ground user and the original UAV based on the ground user's transmit power.

[0018]

[0019] Where ω is the channel bandwidth, n0 is the noise power, and p m P represents the transmit power of the ground user m. interf This represents the maximum average received interference power. The channel gain between ground user m and the original UAV n; Let m be the uplink transmission rate between ground user m and the original drone n;

[0020] S32: Calculate the latency and energy consumption of offloading the task onto the original drone and the cooperative drone respectively, based on the uplink transmission rate between the ground user and the original drone, the number of CPU cycles that the MEC server on the original drone and the cooperative drone can execute per unit time, the data size of the task, and the tolerable latency of the task.

[0021]

[0022]

[0023]

[0024]

[0025]

[0026] in, Indicates task The delay of unloading from the original drone n; This represents the queuing delay for the unloading task from ground user m to the original drone n; Indicates task Delay of unloading on the collaborative drone h; This represents the transmission rate between the original drone n and the collaborating drone h; F represents the queuing delay for the task to be unloaded from the ground user m onto the collaborative drone h;n F represents the number of CPU cycles that the MEC server on the original drone n can execute per unit time; h This represents the number of CPU cycles that the MEC server on the collaborative drone h can execute per unit of time; k n k represents the effective capacitance coefficient of the original UAV n; h p represents the effective capacitance coefficient of the collaborative drone h; n This represents the transmission power of the original drone n; Indicates task Energy consumption unloaded from the collaborative drone h; Indicates task Energy consumption unloaded from the original drone n; This indicates that ground user m will perform the task. The uplink power consumption of the data uploaded to the drone n.

[0027] S33: Create a drone mission offloading model for ground users based on the latency and energy consumption of offloading the mission from the original drone and the cooperative drone:

[0028]

[0029]

[0030] in, Indicates task Total latency for unloading from the drone; Indicates task Total energy consumption for unloading from the drone; Indicates task The actual processing location parameters, when Time indicates task Unloading is performed on the collaborative drone, while unloading is performed on the original drone.

[0031] Furthermore, the drone hovering model includes:

[0032] E N =P0·T′

[0033] Where P0 represents the hovering power of the drone; T′ is the hovering delay of the drone; E N This indicates the hovering energy consumption of the drone.

[0034] Furthermore, the step of unloading tasks using the optimal task unloading allocation scheme calculated by the second price auction algorithm includes:

[0035] S51: Each drone is designated as a seller, and the ground user directly communicating with the drone is designated as a buyer; the minimum asking price for each ground user is calculated based on the ground user's drone task offloading model, the drone's hovering model, and the number of times the ground user participates in the auction; the bidding content for each ground user is initialized based on the minimum asking price for each ground user; the bidding content for each ground user includes:

[0036]

[0037] in, This represents the bidding content of ground user m. This represents the minimum asking price of drone seller n to ground user m, i.e., the initial bid price of ground user m.

[0038] S52: Create a profit model for ground users and drones based on each buyer's bid content, the drone mission offloading model for ground users, the drone hovering model, and the number of times ground users participate in the auction.

[0039] S53: Each ground user determines whether to participate in the auction based on preset participation conditions. When a ground user participates, it is determined whether the user's original drone is idle. If the user's original drone is idle, the task processing location is determined. If the user's original drone is not idle, it is determined whether the user's collaborating drone is idle. If the user's collaborating drone is idle, the task processing location is determined. Otherwise, the task is unloaded locally. The preset participation conditions include:

[0040]

[0041]

[0042]

[0043]

[0044]

[0045]

[0046]

[0047]

[0048] in, This means that when ground user m and drone n are performing task unloading, the coverage radius of drone n must be greater than or equal to the horizontal distance between ground user m and drone n, X. m,nH represents the horizontal distance between ground user m and its original drone n, H is the height of the drone above the ground, and θ is the fixed beamwidth of the drone and the ground user.

[0049] S54: Based on the task processing location of ground users, the profit model of ground users and drones, and with the optimization objective of maximizing the actual total profit of all ground users and all drones, create a profit optimization model for the drone collaborative task unloading system; solve the profit optimization model of the drone collaborative task unloading system to determine the optimal ground user corresponding to each drone seller as the winner to unload the task, and the remaining ground users as losers; and determine the actual bid of each ground user winner according to the second payment rule.

[0050] S55: The losing ground user updates the bid price for the next round of the auction, while the winning ground user uses the actual bid as the bid price for the next round of the auction, and then repeats steps S53-S55.

[0051] Furthermore, the minimum charge for the drone to each ground user includes:

[0052]

[0053] in, γ represents the minimum asking price of drone seller n to ground user m, i.e., the initial bid price of ground user m; E and γ C These represent the economic costs per unit of energy consumption and per unit of time, respectively; τ' represents the duration of each auction. This indicates the number of times ground user m participated in the auction.

[0054] Furthermore, the profit model for the ground users and drones includes:

[0055]

[0056]

[0057]

[0058]

[0059] in, Represents the task of ground user m The profit of ground user m when unloading is performed on the drone; This indicates the task of the drone to ground user m. Profit from unloading the drone; η m , where is the weight, representing the bid metric for ground user m.

[0060] Furthermore, the profit optimization model of the UAV collaborative task unloading system includes:

[0061]

[0062] stC1-C9

[0063]

[0064]

[0065]

[0066]

[0067]

[0068]

[0069] C7:d n1,n2 ≥d min

[0070]

[0071]

[0072] Where P1 is the objective optimization function, and T represents the total system time slots. This indicates that the drone n has established a direct communication connection with the ground user m, otherwise it has not; C1 represents the energy consumption constraint of the ground user. Indicates task Unloading occurs on drone n, but not vice versa; Indicates task Energy consumption for local unloading; This indicates that ground user m will perform the task. C1 represents the uplink transmission power consumption of the UAV n; C2 represents the latency constraint of the ground user; C3 represents the task offloading decision and task processing location constraint of the ground user; C4 represents the communication establishment constraint between the ground user and the UAV; C5 represents the transmission power constraint of the ground user and the transmission power constraint of the UAV. and C1 represents the maximum transmission power of the ground user and the drone, respectively; C6 represents the coverage constraint between the ground user and the drone; C7 represents the constraint condition to avoid collisions between drones, d n1,n2 d represents the distance between the two drones. min C8 represents the minimum distance between two drones, C8 represents the constraint on the maximum number of unloaded tasks by the drone swarm, Q represents the total number of tasks that the ground user needs to unload, and n represents the maximum distance between two drones. maxThe maximum number of tasks that can be processed by each drone; C9 represents the constraint on the number of drones that provide offloading services in each round of auction.

[0073] Furthermore, determining the actual bid of each ground user winner according to the second payment rule includes:

[0074] For each drone, the bid prices of all ground users are grouped into a set P. The winning ground user's bid and bids higher than the winning ground user's bid are removed from set P. The highest bid among the remaining bid prices in set P is selected as the winning ground user's actual bid.

[0075] Furthermore, the losing ground user's updated bid price for the next round of auction includes:

[0076]

[0077] Where, 1 < α < 2, This indicates the bid price of the losing ground user in the next round of auction.

[0078] The present invention has at least the following beneficial effects

[0079] This invention comprehensively considers the practical application scenarios in remote areas and proposes a solution for offloading tasks for Gateway Units (GUs) in remote areas using mobile edge computing (MEC) from UAVs. Applied in complex and variable environments with low population density, this solution leverages the flexibility, low cost, and compact size of UAVs to provide wider coverage. Compared to base station MEC offloading services, it not only meets the service offloading needs of a wider range of GUs but also improves the QoS of GUs.

[0080] Meanwhile, this invention proposes a multi-UAV auction algorithm based on MEC to incentivize communication between users and UAVs, optimize user task allocation, and comprehensively consider the actual bids, willingness to pay, and personal preferences of UAVs and UAVs under constraints such as transmission power, energy consumption, and latency of UAVs and UAVs, thereby maximizing the social welfare of the system while ensuring the authenticity of the auction system.

[0081] This invention optimizes the efficiency of the entire system by designing a second-price auction algorithm that maximizes total profit, enabling the entire auction process to proceed sustainably and smoothly. Attached Figure Description

[0082] Figure 1 This is a schematic diagram of the unmanned aerial vehicle (UAV) collaborative task unloading system of the present invention;

[0083] Figure 2 This is a diagram illustrating the auction role structure of the present invention.

[0084] Figure 3This is a model diagram of the auction system of the present invention;

[0085] Figure 4 This is a flowchart of the auction process of the present invention. Detailed Implementation

[0086] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0087] For scenarios involving multiple ground users in remote areas requiring task unloading, a task unloading scheme based on UAV collaboration is designed. First, the network model, computational unloading model, UAV hovering model, and profit model of the entire system are designed. Then, an objective function is constructed based on the profit model, and an auction algorithm is used to incentivize ground users to unload tasks on UAVs. Finally, a second-price auction algorithm based on maximizing total profit is used to solve the objective optimization function, thereby maximizing the total profit of the entire system while ensuring the performance characteristics of the entire auction system.

[0088] like Figure 1 As shown, the present invention provides a task offloading method based on UAV collaboration, the method comprising:

[0089] S1: Construct a UAV collaborative task unloading system, which includes: M = {1,...m...,M} ground users and N = {1,...n...,N} UAVs equipped with MEC servers; when a ground user communicates directly with any UAV, the ground user generates a task that needs to be unloaded;

[0090] S2: Create a local task offloading model for ground users based on the number of CPU cycles that ground users can execute per unit time, the number of CPU cycles for tasks, the data size of tasks, and the tolerable latency of tasks.

[0091] Furthermore, the local task offloading model for ground users includes:

[0092]

[0093]

[0094]

[0095] in, This represents the task generated at time t when ground user m communicates directly with drone n. For the task Data size, For the task CPU cycle count For the task Tolerable delay; Indicates task Delay in local uninstallation; Indicates task Energy consumption for local unloading; f m k represents the number of CPU cycles that ground user m can execute per unit of time; m This represents the effective capacitance coefficient of ground user m. Indicates the battery capacity of ground users; Indicates task Tolerable delay.

[0096] S3: Create a drone task offloading model for the ground user based on the ground user's transmit power, the number of CPU cycles that the MEC servers on the original drone and the cooperative drone can execute per unit time, the number of CPU cycles of the task, the data size of the task, and the tolerable latency of the task; where the original drone represents the drone that communicates directly with the ground user, and the cooperative drone represents the drone that communicates indirectly with the ground user.

[0097] Furthermore, the drone mission offloading model for creating ground users includes:

[0098] S31: Calculate the uplink transmission rate between the ground user and the original UAV based on the ground user's transmit power.

[0099]

[0100] Where ω is the channel bandwidth, n0 is the noise power, and p m P represents the transmit power of the ground user m. interf This represents the maximum average received interference power. The channel gain between ground user m and the original UAV n; Let m be the uplink transmission rate between ground user m and the original drone n;

[0101] S32: Calculate the latency and energy consumption of offloading the task onto the original drone and the cooperative drone respectively, based on the uplink transmission rate between the ground user and the original drone, the number of CPU cycles that the MEC server on the original drone and the cooperative drone can execute per unit time, the data size of the task, and the tolerable latency of the task.

[0102]

[0103]

[0104]

[0105]

[0106]

[0107] in, Indicates task The delay of unloading from the original drone n; This represents the queuing delay for the unloading task from ground user m to the original drone n; Indicates task Delay of unloading on the collaborative drone h; This represents the transmission rate between the original drone n and the collaborating drone h; F represents the queuing delay for the task to be unloaded from the ground user m onto the collaborative drone h; n F represents the number of CPU cycles that the MEC server on the original drone n can execute per unit time; h This represents the number of CPU cycles that the MEC server on the collaborative drone h can execute per unit of time; k n k represents the effective capacitance coefficient of the original UAV n; h p represents the effective capacitance coefficient of the collaborative drone h; n This represents the transmission power of the original drone n; Indicates task Energy consumption unloaded from the collaborative drone h; Indicates task Energy consumption unloaded from the original drone n; This indicates that ground user m will perform the task. The uplink power consumption of the data uploaded to the drone n.

[0108] S33: Create a drone mission offloading model for ground users based on the latency and energy consumption of offloading the mission from the original drone and the cooperative drone:

[0109]

[0110]

[0111] in, Indicates task Total latency for unloading from the drone; Indicates task Total energy consumption for unloading from the drone; Indicates task The actual processing location parameters, when Time indicates task Unloading is performed on the collaborative drone, while unloading is performed on the original drone.

[0112] S4: Create a hovering model for the drone based on its hovering power and hovering latency;

[0113] Furthermore, the drone hovering model includes:

[0114] E N =P0·T′

[0115] Where P0 represents the hovering power of the drone; T′ is the hovering delay of the drone; E N This indicates the hovering energy consumption of the drone.

[0116] S5: Based on the local task offloading model of the ground user, the UAV task offloading model of the ground user, and the hovering model of the UAV, the optimal task offloading allocation scheme is calculated using the second price auction algorithm to offload the task.

[0117] Please see Figure 4 Considering the above design goals, we should minimize the energy consumption of UAV nodes to extend the UAV's endurance, increase the network's lifespan, and always pay attention to the latency requirements of ground user tasks. Figure 2 This is a structural diagram of the auction role of the present invention, and based on the above design objectives, it is established as follows: Figure 3 The drone auction system model diagram shown further illustrates that the process of unloading tasks using the optimal task unloading allocation scheme calculated by the second price auction algorithm includes:

[0118] S51: Treat each drone as a seller and the ground user who communicates directly with the drone as a buyer; calculate the minimum bid for each ground user based on the drone's mission offloading model, the drone's hovering model, and the number of times the ground user participates in the auction; initialize the bidding content of each ground user based on the minimum bid for each ground user.

[0119] To ensure that the basic profit of the drone is greater than zero in each auction, the minimum bid from ground users can be set as the minimum asking price for the drone, thus guaranteeing a positive profit for the drone in the first round of auctions.

[0120] Furthermore, the minimum charge for the drone to each ground user includes:

[0121]

[0122] in, γ represents the minimum asking price of drone seller n to ground user m, i.e., the initial bid price of ground user m; E and γ C τ' represents the economic cost per unit of energy consumption and per unit of time, respectively; τ' represents the duration of each auction. This indicates the number of times ground user m participated in the auction.

[0123] Furthermore, the bidding content for each ground user includes:

[0124]

[0125] in, This represents the bidding content of ground user m.

[0126] S52: Create a profit model for ground users and drones based on each buyer's bid content, the drone mission offloading model for ground users, the drone hovering model, and the number of times ground users participate in the auction.

[0127] Preferably, the profit model for the ground user and the drone includes:

[0128]

[0129]

[0130]

[0131]

[0132] in, Represents the task of ground user m The profit of ground user m when unloading is performed on the drone; This indicates the task of the drone to ground user m. Profit from unloading the drone; η m represents the weight, indicating the bid metric for ground user m.

[0133] Before drones and ground users participate in the auction, the asking price of drones and the bids of ground users are initialized. It is also necessary to ensure that the bids of ground users are always greater than the initial asking price of drones in each round. In order to ensure that the basic profit of drones is greater than zero, the unloading cost of drones can be set as the minimum asking price of drones to ensure that the profit of drones is always positive in each round of auction.

[0134] S53: Each ground user determines whether to participate in the auction based on preset participation conditions. When a ground user participates, it is determined whether the user's original drone is idle. If the user's original drone is idle, the task processing location is determined. If the user's original drone is not idle, it is determined whether the user's collaborating drone is idle. If the user's collaborating drone is idle, the task processing location is determined. Otherwise, the task is unloaded locally. The preset participation conditions include:

[0135]

[0136]

[0137]

[0138]

[0139]

[0140]

[0141]

[0142]

[0143] in, This means that when ground user m and drone n are performing task unloading, the coverage radius of drone n must be greater than or equal to the horizontal distance between ground user m and drone n, X. m,n H represents the horizontal distance between ground user m and its original drone n, H is the height of the drone above the ground, and θ is the fixed beamwidth of the drone and the ground user.

[0144] S54: Based on the task processing location of ground users, the profit model of ground users and drones, and with the optimization objective of maximizing the actual total profit of all ground users and all drones, create a profit optimization model for the drone collaborative task unloading system; solve the profit optimization model of the drone collaborative task unloading system to determine the optimal ground user corresponding to each drone seller as the winner to unload the task, and the remaining ground users as losers; and determine the actual bid of each ground user winner according to the second payment rule.

[0145] Furthermore, the profit optimization model of the UAV collaborative task unloading system includes:

[0146]

[0147] stC1-C9

[0148]

[0149]

[0150]

[0151]

[0152]

[0153]

[0154] C7:d n1,n2 ≥d min

[0155]

[0156]

[0157] Where P1 is the objective optimization function, and T represents the total system time slots. This indicates that the drone n has established a direct communication connection with the ground user m, otherwise it has not; C1 represents the energy consumption constraint of the ground user. Indicates task Unloading occurs on drone n, but not vice versa; Indicates task Energy consumption for local unloading; This indicates that ground user m will perform the task. C1 represents the uplink transmission power consumption of the UAV n; C2 represents the latency constraint of the ground user; C3 represents the task offloading decision and task processing location constraint of the ground user; C4 represents the communication establishment constraint between the ground user and the UAV; C5 represents the transmission power constraint of the ground user and the transmission power constraint of the UAV. and C1 represents the maximum transmission power of the ground user and the drone, respectively; C6 represents the coverage constraint between the ground user and the drone; C7 represents the constraint condition to avoid collisions between drones, d n1,n2 d represents the distance between the two drones. min C8 represents the minimum distance between two drones, C8 represents the constraint on the maximum number of unloaded tasks by the drone swarm, Q represents the total number of tasks that the ground user needs to unload, and n represents the maximum distance between two drones. max The maximum number of tasks that can be processed by each drone; C9 represents the constraint on the number of drones that provide offloading services in each round of auction.

[0158] Furthermore, determining the actual bid of each ground user winner according to the second payment rule includes:

[0159] For each drone, the bid prices of all ground users are grouped into a set P. The winning ground user's bid and bids higher than the winning ground user's bid are removed from set P. The highest bid among the remaining bid prices in set P is selected as the winning ground user's actual bid.

[0160] S55: The losing ground user updates the bid price for the next round of the auction, while the winning ground user uses the actual bid as the bid price for the next round of the auction, and then repeats steps S53-S55.

[0161] Furthermore, the losing ground user's updated bid price for the next round of auction includes:

[0162]

[0163] Where, 1 < α < 2, This indicates the bid price of the losing ground user in the next round of auction.

[0164] Please see Figure 3 In this embodiment, a second-price auction algorithm based on maximizing total profit is designed. Figure 3 The model diagram of the entire auction system is shown below. The specific rules for determining the winner and the payment process are as follows: 1) First, drone 1 and drone 2 calculate their minimum bids, and ground users 1-6 submit bids based on the minimum bids of the drones.

[0165] 2) Drone 1 calculates the buyer's profit of Ground User 1 and the seller's profit of Drone 1 based on the bid of Ground User 1, and then sums them to obtain the total bid profit of Ground User 1.

[0166] 3) Drone 2 calculates the buyer's profit of ground user 4 and the seller's profit of drone 2 based on the bid of ground user 4, and then sums them to obtain the total bid profit of ground user 4; Note: Drone 1 and drone 2 are performed in parallel.

[0167] 4) Drone 1 then calculates the total bid profit of ground users 2 and 3, and Drone 2 then calculates the total bid profit of ground users 5 and 6.

[0168] 5) After that, the winner of this round of auction is determined by the total profit calculated by drone 1 and drone 2, that is, the one with the highest total profit in the bidding is the winner of this round of auction.

[0169] 6) Then, UAV 1 and UAV 2 sort the bids from largest to smallest based on the bids of ground users 1-3 and 4-6, denoted as sets A1 and A2, and assume that the winning bids determined by UAV 1 and UAV 2 are a1 and a2.

[0170] 7) Finally, the actual payment of the winning ground user corresponding to drone 1 is: after removing a1 and higher bids from set A1, the highest bid among the remaining bids is the actual bid of the winning ground user a1; similarly, the actual bid of the winning ground user a2 corresponding to drone 2 can be obtained.

[0171] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A task unloading method based on UAV collaboration, characterized in that, include: S1: Construct a drone collaborative task offloading system, which includes: M ground users and N drones with MEC servers; when a ground user communicates directly with any drone, the ground user generates a task that needs to be offloaded. S2: Create a local task offloading model for ground users based on the number of CPU cycles that ground users can execute per unit time, the number of CPU cycles for tasks, the data size of tasks, and the tolerable latency of tasks. S3: Create a drone task offloading model for the ground user based on the ground user's transmit power, the number of CPU cycles that the MEC servers on the original drone and the cooperative drone can execute per unit time, the number of CPU cycles of the task, the data size of the task, and the tolerable latency of the task; where the original drone represents the drone that communicates directly with the ground user, and the cooperative drone represents the drone that communicates indirectly with the ground user. S4: Create a hovering model for the drone based on its hovering power and hovering latency; S5: Based on the local task offloading model of the ground user, the UAV task offloading model of the ground user, and the hovering model of the UAV, the optimal task offloading allocation scheme is calculated using the second price auction algorithm to offload the task. The process of unloading tasks by calculating the optimal task unloading allocation scheme using the second price auction algorithm includes: S51: Each drone is designated as a seller, and the ground user directly communicating with the drone is designated as a buyer; the minimum asking price for each ground user is calculated based on the ground user's drone task offloading model, the drone's hovering model, and the number of times the ground user participates in the auction; the bidding content for each ground user is initialized based on the minimum asking price for each ground user; the bidding content for each ground user includes: in, This represents the bidding content of ground user m. Indicates drone seller For ground users The lowest asking price, i.e., for ground users The initial bid price; S52: Create a profit model for ground users and drones based on each buyer's bid content, the drone mission offloading model for ground users, the drone hovering model, and the number of times ground users participate in the auction. S53: Each ground user determines whether to participate in the auction based on preset participation conditions. When a ground user participates, it is determined whether the user's original drone is idle. If the user's original drone is idle, the task processing location is determined. If the user's original drone is not idle, it is determined whether the user's collaborating drone is idle. If the user's collaborating drone is idle, the task processing location is determined. Otherwise, the task is unloaded locally. The preset participation conditions include: in, This means that when ground user m and drone n are performing task unloading, the coverage radius of drone n must be greater than or equal to the horizontal distance between ground user m and drone n. H represents the horizontal distance between ground user m and its original drone n, where H is the height of the drone above the ground. Fixed beamwidth for drones and ground users; Indicates the duration of each auction. Indicates task Total latency for unloading from the drone; Indicates ground users With drones During communication Tasks generated in real time; For the task Tolerable delay; Indicates task Delay in local uninstallation; Indicates task Energy consumption for local unloading; Indicates the battery capacity of ground users; Indicates ground users The task Upload to drone Uplink transmission power consumption; Represents the task of ground user m The profit of ground user m when unloading is performed on the drone; S54: Based on the task processing location of ground users, the profit model of ground users and drones, and with the optimization objective of maximizing the actual total profit of all ground users and all drones, create a profit optimization model for the drone collaborative task unloading system; solve the profit optimization model of the drone collaborative task unloading system to determine the optimal ground user corresponding to each drone seller as the winner to unload the task, and the remaining ground users as losers; and determine the actual bid of each ground user winner according to the second payment rule. S55: The losing ground user updates the bid price for the next round of the auction, while the winning ground user uses the actual bid as the bid price for the next round of the auction, and then repeats steps S53-S55.

2. The task offloading method based on UAV collaboration according to claim 1, characterized in that, The local task offloading model for ground users includes: ， ， in, Indicates ground users With drones During communication Tasks generated in real time For the task Data size, For the task CPU cycle count For the task Tolerable delay; Indicates task Delay in local uninstallation; Indicates task Energy consumption for local unloading; Indicates ground users The number of CPU cycles that can be executed per unit of time; Indicates ground users The effective capacitance coefficient, This indicates the battery capacity of ground users.

3. A task offloading method based on UAV collaboration according to claim 2, characterized in that, The drone mission offloading model for creating ground users includes: S31: Calculate the uplink transmission rate between the ground user and the original UAV based on the ground user's transmit power. in, For channel bandwidth, For noise power, For ground users The transmission power; This represents the maximum average received interference power. For ground users With the original drone Channel gain between; For ground users With the original drone Uplink transmission rate between; S32: Calculate the latency and energy consumption of offloading the task onto the original drone and the cooperative drone respectively, based on the uplink transmission rate between the ground user and the original drone, the number of CPU cycles that the MEC server on the original drone and the cooperative drone can execute per unit time, the data size of the task, and the tolerable latency of the task. ， in, Indicates task In the original drone Loading and unloading latency; Indicates ground users To the original drone Queuing delay for loading and unloading tasks; Indicates task Collaborative drones Loading and unloading latency; Indicates the original drone and collaborative drones The transmission rate between them; Indicates ground users Queuing delay for unloading tasks onto the collaborative drone h; Indicates the original drone The number of CPU cycles that the MEC server can execute per unit of time; Indicating collaborative drones The number of CPU cycles that an MEC server can execute per unit of time; Indicates the original drone The effective capacitance coefficient; Indicating collaborative drones The effective capacitance coefficient; Indicates the original drone The transmission power; Indicates task Collaborative drones Energy consumption during loading and unloading; Indicates task In the original drone Energy consumption during loading and unloading; Indicates ground users The task Upload to drone Uplink transmission power consumption; S33: Create a drone mission offloading model for ground users based on the latency and energy consumption of offloading the mission from the original drone and the cooperative drone: in, Indicates task Total latency for unloading from the drone; Indicates task Total energy consumption for unloading from the drone; Indicates task The actual processing location parameters, ,when Time indicates task Unloading is performed on the collaborative drone, while unloading is performed on the original drone.

4. A task offloading method based on UAV collaboration according to claim 3, characterized in that, The hovering model of the drone includes: in, Indicates the hovering power of the drone; For the hovering delay of the drone; This indicates the hovering energy consumption of the drone.

5. A task offloading method based on UAV collaboration according to claim 1, characterized in that, The minimum charge for the drone to each ground user includes: in, Indicates drone seller For ground users The lowest asking price, i.e., for ground users The initial bid price; and These represent the economic cost per unit of energy consumption and per unit of time, respectively. Indicates the duration of each auction; Indicates ground users The number of times you participate in an auction.

6. A task offloading method based on UAV collaboration according to claim 5, characterized in that, The profit model for ground users and drones includes: in, Represents the task of ground user m The profit of ground user m when unloading is performed on the drone; Indicates drones to ground users Task The profit from unloading the drone; , where is the weight, representing the bid metric for ground user m.

7. A task offloading method based on UAV collaboration according to claim 6, characterized in that, The profit optimization model of the UAV collaborative task unloading system includes: in, Optimize the function for the objective. Indicates the total number of system time slots. Indicates drone With ground users A direct communication connection is established if one is not established, otherwise it is not. This represents the energy consumption constraints for ground users. Indicates task Unloading occurs on drone n, but not vice versa; Indicates task Energy consumption for local unloading; Indicates ground users The task Upload to drone Uplink transmission power consumption; This represents the latency constraints for ground users; This represents the task offloading decision and task processing location constraints for ground users; This indicates the establishment of communication constraints between ground users and drones; This represents the transmit power constraints for ground users and the transmission power constraints for drones; and These represent the maximum transmission power for ground users and drones, respectively. Indicates the coverage constraints between ground users and drones; This represents the constraints that prevent collisions between drones. Indicates the distance between the two drones. This indicates the minimum distance between the two drones. This indicates the maximum number of unloading tasks that can be performed by a drone swarm. This indicates the total number of tasks that ground users need to uninstall. The maximum number of tasks that can be processed for each drone; This indicates the quantity constraint for drones providing unloading services in each round of the auction.

8. A task offloading method based on UAV collaboration according to claim 1, characterized in that, The determination of the actual bid for each ground user winner according to the second payment rule includes: For each drone, the bid prices of all ground users are grouped into a set P. The winning ground user's bid and bids higher than the winning ground user's bid are removed from set P. The highest bid among the remaining bid prices in set P is selected as the winning ground user's actual bid.

9. A task offloading method based on UAV collaboration according to claim 1, characterized in that, The bid price for the losing ground users to participate in the next round of auction includes: in, , This indicates the bid price of the losing ground user in the next round of auction.

Images