Multi-unmanned aerial vehicle cooperative computing task offloading, resource allocation and trajectory planning method and system thereof

Abstract

The application discloses a multi-unmanned aerial vehicle cooperative computing task unloading, resource allocation and trajectory planning method and system, wherein the method specifically comprises the following steps: initializing all state information, and determining a target function; obtaining a multi-unmanned aerial vehicle cooperative computing task unloading result; updating the target function according to the computing task unloading result, and obtaining an unmanned aerial vehicle computing resource allocation scheme according to the updated target function; updating the updated target function again according to the unmanned aerial vehicle computing resource allocation scheme, and obtaining an unmanned aerial vehicle horizontal trajectory; judging whether the maximum iteration number is reached; if the maximum iteration number is not reached, the iteration number is increased by 1, and the above steps are repeatedly executed. The application excavates the cooperation capability among the multi-unmanned aerial vehicles, and achieves the purposes of multi-unmanned aerial vehicle cooperative computing task unloading, computing resource allocation and unmanned aerial vehicle flight trajectory planning.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of computer, in particular, to a multi-unmanned aerial vehicle cooperative computing task offloading, resource allocation and trajectory planning method and system thereof. BACKGROUND

[0002] With the rise of the Internet of Things, the number of different types of ground terminal devices such as cloud sensors, smart phones, wearable devices is increasing; intelligent applications such as face recognition, interactive games, virtual reality are also emerging. However, due to the weak computing power of these terminal devices and the limited battery capacity, it is a difficult problem to effectively compute the large amount of computing data generated by these applications. In this context, mobile edge computing (MEC) as a technology with development potential can help ground terminals to perform data processing and computing at the network edge, thereby solving the above problems. With flexible mobility, unmanned aerial vehicle assisted MEC offloading technology has received more and more research attention in recent years. Taking the emergency communication scene as an example, due to the damage of ground communication link, deploying edge server on unmanned aerial vehicle can more effectively utilize MEC offloading technology to ensure that the computing tasks with high real-time requirements such as image and video data processing can be quickly responded. However, in the existing scheme, single unmanned aerial vehicle assisted MEC network faces the shortcomings of insufficient communication coverage, weak computing power, poor network robustness, etc. Therefore, multi-unmanned aerial vehicle assisted MEC network deployment scheme is more effective. Although the combination of unmanned aerial vehicle and MEC technology provides an effective solution for the computing of ground terminal, through the assistance design of unmanned aerial vehicle, the flexibility and computing power of the system can also be further enhanced. However, there are still many difficult problems: 1) Most current research focuses on the sustainability of unmanned aerial vehicle MEC system, taking energy consumption as the optimization target, while in the application scene such as emergency communication with high demand for time delay, it is more meaningful to design an edge computing offloading strategy that meets the demand of time delay sensitive task. 2) In the current existing research, most of them only stop at optimizing the computing offloading between ground terminal and unmanned aerial vehicle, lack of mining the cooperative ability of multi-unmanned aerial vehicle, resulting in unbalanced computing load of unmanned aerial vehicle, which cannot further reduce the task execution time delay.

[0003] Therefore, how to realize an edge computing offloading method based on multi-unmanned aerial vehicle cooperation for time delay sensitive task has become a problem urgently to be solved in the field. SUMMARY

[0004] The application provides a multi-unmanned aerial vehicle cooperative computing task offloading, resource allocation and trajectory planning method, specifically comprising the following sub-steps: step S1, initializing all state information and determining a target function; step S2, obtaining a multi-unmanned aerial vehicle cooperative computing task offloading result according to the target function; step S3, updating the target function according to the computing task offloading result, and obtaining an unmanned aerial vehicle computing resource allocation scheme according to the updated target function; step S4, updating the updated target function again according to the unmanned aerial vehicle computing resource allocation scheme, and obtaining an unmanned aerial vehicle horizontal trajectory according to the target function updated again; step S5, determining whether the maximum iteration number is reached; if the maximum iteration number is reached, the obtained computing task offloading result, unmanned aerial vehicle computing resource allocation scheme and unmanned aerial vehicle horizontal trajectory are output as the best unmanned aerial vehicle offloading result, the best unmanned aerial vehicle computing resource allocation scheme and the best unmanned aerial vehicle flight trajectory respectively; if the maximum iteration number is not reached, the iteration number is increased by 1, and steps S2-S4 are repeatedly executed.

[0005] The above, wherein the initializing all state information comprises initializing a multi-unmanned aerial vehicle cooperative computing task offloading scheme, initializing an unmanned aerial vehicle computing resource allocation scheme and initializing an unmanned aerial vehicle flight trajectory.

[0006] The above, wherein the initializing all state information and determining a target function comprises determining a signal transmission speed of a ground terminal k and an unmanned aerial vehicle m in a time slot n. is expressed as:

[0007]

[0008] wherein P D represents the signal transmission power of the ground terminal, B D represents the total bandwidth of the air-ground communication, K represents the number of ground terminals, and the ground terminals are evenly distributed in the bandwidth. represents the channel gain between the ground terminals when the transmission distance is 1 m, d km [n] represents the distance between the ground terminal k and the unmanned aerial vehicle m in the time slot n.

[0009] The above, wherein the position vector of the ground terminal k is assumed to be The position vector of the unmanned aerial vehicle m in the time slot n is The distance d km [n] between the ground terminal k and the unmanned aerial vehicle m in the time slot n is expressed as

[0010] The above, wherein the initializing all state information and determining a target function comprises further comprising determining a signal transmission rate of the unmanned aerial vehicle m and the unmanned aerial vehicle l (l≠m) in the time slot n. is expressed as:

[0011]

[0012] wherein M represents the number of UAVs, d ml represents the distance between UAV m and UAV l in time slot n, P A represents the signal transmission power of the ground terminal, B A represents the total bandwidth of the communication between the UAVs, the plurality of UAVs evenly divide the bandwidth, represents the channel gain between the UAVs when the transmission distance is 1 m.

[0013] A multi-UAV cooperative computing task offloading, resource allocation and trajectory planning system, comprising a cloud server, the cloud server specifically comprising an initialization and target function determination unit, a computing task offloading result acquisition unit, a computing resource allocation scheme acquisition unit, a UAV horizontal trajectory acquisition unit, a judgment unit, and an output unit; the target function determination unit is used for initializing all state information and determining a target function. The computing task offloading result acquisition unit is used for acquiring a multi-UAV cooperative computing task offloading result according to the target function. The computing resource allocation scheme acquisition unit is used for updating the target function according to the computing task offloading result, and acquiring a UAV computing resource allocation scheme according to the updated target function. The UAV horizontal trajectory acquisition unit is used for updating the updated target function again according to the UAV computing resource allocation result, and acquiring a UAV horizontal trajectory according to the updated target function. The judgment unit is used for judging whether the maximum iteration number is reached; if the maximum iteration number is reached, the computing task offloading result, the UAV computing resource allocation scheme and the UAV horizontal trajectory obtained above are respectively output as the best UAV offloading result, the best UAV computing resource allocation scheme and the best UAV flight trajectory by the output unit; otherwise, the iteration number is increased by 1, and the process is repeated until the maximum iteration number is reached, and the output unit outputs the best UAV offloading result, the best UAV computing resource allocation scheme and the best UAV flight trajectory.

[0014] As above, wherein the initialization and target function determination unit initializes all state information, including a multi-UAV cooperative initialization computing task offloading scheme, a UAV initialization computing resource allocation scheme, and a UAV initialization flight trajectory.

[0015] As above, wherein the initialization and target function determination unit initializes all state information and determines a target function, including determining the signal transmission speed between the ground terminal k and the UAV m in time slot n is represented as:

[0016]

[0017] wherein P D represents the signal transmission power of the ground terminal, BD represents the total bandwidth of the air-ground communication, K represents the number of ground terminals, and the ground terminals are allocated bandwidth on average, represents the channel gain between the ground terminals when the transmission distance is 1 m, d km [n] represents the distance between the ground terminal k and the unmanned aerial vehicle m in the time slot n.

[0018] As above, in the initialization and target function determination unit, it is assumed that the position vector of the ground terminal k is The position vector of the unmanned aerial vehicle m in the time slot n is The distance d between the ground terminal k and the unmanned aerial vehicle m in the time slot n is km [n] represents

[0019] As above, in the initialization and target function determination unit, all state information is initialized, and the target function is determined, which includes determining the signal transmission rate between the unmanned aerial vehicle m and the unmanned aerial vehicle l (l≠m) in the time slot n. represents:

[0020]

[0021] where M represents the number of unmanned aerial vehicles, d ml represents the distance between the unmanned aerial vehicle m and the unmanned aerial vehicle l in the time slot n, P A represents the signal transmission power of the unmanned aerial vehicle, B A represents the total bandwidth of the communication between the unmanned aerial vehicles, and the bandwidth is divided among the plurality of unmanned aerial vehicles, represents the channel gain between the unmanned aerial vehicles when the transmission distance is 1 m. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiment or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art based on these drawings.

[0023] Figure 1 is a flowchart of a multi-unmanned aerial vehicle cooperative computing task offloading, resource allocation and trajectory planning method provided according to an embodiment of the present application;

[0024] Figure 2 is an internal structure schematic diagram of a multi-unmanned aerial vehicle cooperative computing task offloading, resource allocation and trajectory planning system provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0025] With reference to the accompanying drawings, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only some of the embodiments of the present application, but not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts are within the scope of the present application.

[0026] The present application provides a multi-unmanned aerial vehicle cooperative computing task offloading, resource allocation and trajectory planning method, which reduces the system computing task execution delay by jointly optimizing the multi-unmanned aerial vehicle cooperative computing task offloading, computing resource allocation and flight trajectory planning.

[0027] Embodiment one

[0028] It is assumed that there are K ground terminals in each cycle of the system, and the set is represented as The ground terminal needs to perform the corresponding task on the ground, but its computing power is weak, so M unmanned aerial vehicles are used to assist in offloading computation in the air, and the set is represented as M={1,......M}, wherein the trajectory set of the unmanned aerial vehicle can be represented as N={1,......N}, that is, the total execution time of the task includes N time slots. The task can be selected to be executed locally or cooperatively in the unmanned aerial vehicle network according to the task information, and the secondary offloading between the unmanned aerial vehicles can be performed by using the cooperative computing capability to ensure the balance of the work load. In addition, x km [n]∈{0,1} represents the offloading variable between the ground terminal and the unmanned aerial vehicle; λ lm [n]∈[0,1] represents the secondary offloading variable between the unmanned aerial vehicles; f ml [n] represents the computing resource allocation variable of the unmanned aerial vehicle; and represents the flight trajectory variable of the unmanned aerial vehicle.

[0029] As shown in FIG. 1, a multi-unmanned aerial vehicle cooperative computing task offloading, resource allocation and trajectory planning method provided by the present embodiment is specifically as follows: Figure 1

[0030] Step S110: initialize all state information and determine the objective function.

[0031] Before the system is running, the system initialization parameters are designed first, including the multi-unmanned aerial vehicle cooperative initialization computing task offloading scheme, the unmanned aerial vehicle initialization computing resource allocation scheme, and the unmanned aerial vehicle initialization flight trajectory.

[0032] ​Specifically, based on the location distribution of ground terminals, the initialization computational task offloading scheme, the UAV initialization computational resource allocation scheme, and the UAV flight trajectory are initialized for multi-UAV collaboration. For the computational task offloading scheme, an initialization method based on communication distance can be used; for the UAV computational resource allocation scheme, an equal distribution method can be used; for the UAV flight trajectory, a regular trajectory initialization method can be used to initialize the UAV trajectory, such as using a straight, uniform speed flight method, flying from a given starting point to an ending point within a total time T.

[0033] Assume there are K ground terminals in each cycle of the system, and the set is represented as follows: The ground terminal needs to perform corresponding tasks on the ground, but its computing power is relatively weak. Therefore, M drones assist in unloading the computation in the air, and the set is represented as M = {1, ..., M}. The set of drone trajectories can be represented as N = {1, ..., N}, that is, the total execution time of the task includes N time slots.

[0034] In the 3D coordinate system, assume the position vector of the ground terminal k is... Where x k y k Let x and y represent the horizontal and vertical coordinates of the ground terminal in the coordinate system, respectively, and let m be the position vector of the UAV in time slot n. Where x m [n]、y m [n]、h m [n] represents the horizontal and vertical coordinates and altitude in time slot n, respectively. The distance between the ground terminal k and the UAV m in time slot n can be expressed as: Let the signal transmission power of the ground terminal be P. D The total bandwidth for air-to-ground communication is B. D With average bandwidth allocation at ground terminals, the channel gain at a transmission distance of 1m is: Since the air-to-ground communication channel is mainly a log-normal fading channel, the signal transmission speed between the ground terminal k and the UAV m in time slot n is... It can be represented as:

[0035]

[0036] Where σ 2 Let P represent the Gaussian noise power spectral density of the channel. Similarly, let the signal transmission power of the UAV be P. A The total bandwidth for communication between drones is B. A With multiple drones sharing the bandwidth, the channel gain at a transmission distance of 1m is: The distance between UAV m and UAV l in time slot n is The signal transmission rate of the UAV m to the UAV l (l≠m) in the time slot n may be expressed as:

[0037]

[0038] With the cooperative ability among multiple UAVs, the secondary offloading among UAVs is considered after the computing task is offloaded from the ground terminal to the UAV. Let the task data size generated by the ground terminal k in the time slot n be D k [n] and assume that the task offloading variable between the ground terminal k and the UAV m is x km [n] and x km [n]∈{0,1} indicates whether the task of the ground terminal k is offloaded for computing on the UAV m in the time slot n. That is, x km [n] = 1, the UAV m assists the ground terminal k to complete the task computing, x km [n] = 0, on the contrary. In particular, let x k,0 [n] indicate whether the ground terminal k performs local computing in the time slot n, x k,0 [n] = 1 indicates that the local computing is performed, otherwise x k,0 [n] = 0. In the secondary offloading, let the task offloading variable of the UAV m assisting the l to perform edge computing be λ lm [n] (m, l ∈ M), and λ lm [n] ∈ [0, 1] indicates the proportion of the task offloaded from the UAV l to the UAV m in the time slot n. In particular, when l = m, that is, λ mm [n] indicates the proportion of the task remaining on the UAV m after the secondary offloading, and satisfies

[0039] Let the CPU computing amount required by each ground terminal for 1 bit of data be κ, and the CPU computing rate of the ground terminal k be f k , then the delay of the local computing of the ground terminal k in the time slot n

[0040] Therefore, in the time slot n, the offloading transmission delay of the ground terminal k is Further, after the primary offloading, the task data amount received by the UAV m in the time slot n The transmission delay of the UAV l offloading to the m in the time slot n wherein indicates the task data amount received by the UAV m in the time slot n. According to the offloading model set above, the offloading transmission of the task needs to be first transmitted from the ground terminal to the air, and then transmitted among the UAVs, so the edge offloading transmission delay of the task via the UAV m after the secondary offloading in the time slot n is which is specifically expressed as:

[0041]

[0042] The CPU of the UAV m assigned to the I in the time slot n calculates the rate as f ml [n], assuming that the task obtained by the current time slot is stored in the buffer in the UAV, and the calculation is performed in the next time slot, the edge task calculation delay of the UAV m in the time slot n where λ lm [n-1] represents the task offloading variable in the time slot n-1.

[0043] Since the local execution of the task, the offloading transmission of the task, and the edge execution of the task in the model can be performed in parallel. Therefore, the length T[n] of the time slot n satisfies:

[0044]

[0045] where that is, the objective function described in the step.

[0046] Step S120: obtaining the calculation task offloading result of the multi-UAV cooperation according to the objective function.

[0047] By using the variable relaxation and the penalty function, the optimized calculation task offloading scheme between the ground terminal-UAV and the UAV-UAV in each time slot can be obtained according to the given UAV computing resource allocation scheme and the UAV flight trajectory. Specifically, the task offloading variable x km [n] is relaxed to a continuous variable between 0 and 1, the non-convex constraint is relaxed to a convex constraint by using the method of Successive Convex Approximation (SCA), and the calculation task offloading scheme between the ground terminal-UAV and the UAV-UAV in each time slot is output.

[0048] Specifically, the calculation task offloading result of the multi-UAV cooperation includes that the formula 4 is solved by using the penalty function and the SCA method, and the optimized calculation task offloading scheme is obtained, that is, the calculation task offloading scheme between the ground terminal-UAV and the UAV-UAV in each time slot. By using the convex optimization toolbox, the formula 4 is solved according to the given initialization information, and the solution of the calculation task offloading variable x km [n] and λ lm [n] is obtained.

[0049] Step S130: updating the objective function according to the calculation task offloading result, and obtaining the UAV computing resource allocation scheme according to the updated objective function.

[0050] Wherein the updating of the objective function according to the calculation task offloading result is specifically updating the variable f km [n] and λ lm [n] is updated, so that the objective function is also updated as a whole.

[0051] The calculation task offloading result of the multiple unmanned aerial vehicles obtained through step S120, the initial flight trajectory of the unmanned aerial vehicle, and the constraint of the computing resource of the unmanned aerial vehicle are used to update the objective function. For the updated objective function, the variable f ml [n] is optimized by using the KKT (Karush-Kuhn-Tucker) condition, and the computing resource allocation scheme of the unmanned aerial vehicle is output.

[0052] Step S140: The updated objective function is updated again according to the computing resource allocation scheme of the unmanned aerial vehicle, and the horizontal trajectory of the unmanned aerial vehicle is obtained according to the objective function updated again.

[0053] Wherein the updating of the updated objective function according to the computing resource allocation result of the unmanned aerial vehicle is specifically updating the variable f ml [n] is updated, so that the objective function is updated again.

[0054] The calculation task offloading result of the multiple unmanned aerial vehicles obtained through step S120 and the computing resource allocation scheme of the unmanned aerial vehicle obtained through step S130 are also used to relax the non-convex constraint into a convex constraint by using the SCA method, and the solution of the updated objective function is obtained again, and the optimized flight trajectory of the unmanned aerial vehicle is output.

[0055] Step S150: It is judged whether the maximum iteration number is reached.

[0056] If the maximum iteration number is reached, the calculation task offloading result obtained through step S120 is the best, the computing resource allocation scheme of the unmanned aerial vehicle obtained through step S130 is the best, and the flight trajectory of the unmanned aerial vehicle obtained through step S140 is the best, and step S160 is executed. Otherwise, the iteration number is increased by 1, and steps S120-140 are repeatedly executed until the maximum iteration number is reached, and step S160 is executed.

[0057] Step S160: The best calculation task offloading result, the best computing resource allocation scheme of the unmanned aerial vehicle, and the best flight trajectory of the unmanned aerial vehicle are output.

[0058] Wherein the output result is broadcast to each ground terminal and unmanned aerial vehicle in the system, and the corresponding task is offloaded and executed.

[0059] Embodiment two

[0060] The embodiment provides a multi-unmanned aerial vehicle cooperative computing task offloading, resource allocation and trajectory planning system.

[0061] The cloud server specifically comprises an initialization and target function determination unit 210, a computing task offloading result acquisition unit 220, a computing resource allocation scheme acquisition unit 230, an unmanned aerial vehicle horizontal trajectory acquisition unit 240, a judgment unit 250 and an output unit 260.

[0062] The target function determination unit 210 is used for initializing all state information and determining a target function.

[0063] The computing task offloading result acquisition unit 220 is used for acquiring a multi-unmanned aerial vehicle cooperative computing task offloading result according to the target function.

[0064] The computing resource allocation scheme acquisition unit 230 is used for updating the target function according to the computing task offloading result and acquiring an unmanned aerial vehicle computing resource allocation scheme according to the updated target function.

[0065] The unmanned aerial vehicle horizontal trajectory acquisition unit 240 is used for updating the target function again according to the unmanned aerial vehicle computing resource allocation result and acquiring an unmanned aerial vehicle horizontal trajectory according to the updated target function.

[0066] The judgment unit 250 is used for judging whether a maximum iteration number is reached.

[0067] If the maximum iteration number is reached, the output unit 260 outputs the best unmanned aerial vehicle offloading result, the best unmanned aerial vehicle computing resource allocation scheme and the best unmanned aerial vehicle flight trajectory.

[0068] The application has the following beneficial effects:

[0069] The application is directed to a multi-unmanned aerial vehicle mobile edge computing system in a high latency demand scene such as emergency communication, excavates the cooperative ability among the multi-unmanned aerial vehicles, achieves the purposes of multi-unmanned aerial vehicle cooperative computing task offloading, computing resource allocation and unmanned aerial vehicle flight trajectory planning, and proposes a complete joint design idea. Meanwhile, the design goal of minimizing system task execution latency can be achieved, the work load among the multi-unmanned aerial vehicle formations is balanced, and the execution of the computing task in the high latency demand scene such as emergency communication is effectively optimized.

[0070] While the examples of the present application are described with reference to the drawings, they are merely examples and are not intended to limit the present application. Changes, additions and / or deletions can be made to the embodiments without departing from the scope of the present application.

[0071] The above description is only specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present application, which should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for multi-UAV cooperative computing of task offloading, resource allocation and trajectory planning, characterized in that, Specifically comprising the following sub-steps: Step S1, initializing all state information, determining the objective function; Step S2, obtaining the multi-UAV cooperative computing task offloading result according to the objective function; Step S3, updating the objective function according to the computing task offloading result, and obtaining the UAV computing resource allocation scheme according to the updated objective function; Step S4, updating the updated objective function again according to the UAV computing resource allocation scheme, and obtaining the UAV horizontal trajectory according to the updated objective function again; Step S5, determining whether the maximum iteration number is reached; If the maximum iteration number is reached, the obtained computing task offloading result, UAV computing resource allocation scheme, and UAV horizontal trajectory are output as the best UAV offloading result, the best UAV computing resource allocation scheme, and the best UAV flight trajectory respectively; If the maximum iteration number is not reached, the iteration number is increased by 1, and steps S2-S4 are repeatedly executed; performing initialization of all state information, determining the objective function includes determining the signal transmission rate of the ground terminal k and the unmanned aerial vehicle m in the time slot n is represented as: where P D denotes the signal transmission power of the ground terminal, B D denotes the total bandwidth of the air-ground communication, K denotes the number of ground terminals, and the ground terminals are evenly allocated bandwidth, denotes the channel gain between the ground terminals when the transmission distance is 1 m, d km [n] denotes the distance between the ground terminal k and the unmanned aerial vehicle m in time slot n, σ 2 denotes the Gaussian noise power spectral density of the channel; Let the position vector of ground terminal k be The position vector of UAV m at time slot n is The distance d between ground terminal k and UAV m at time slot n is km [n] is represented as performing initialization of all state information, determining the objective function includes determining the signal transmission rate of the unmanned aerial vehicle m and the unmanned aerial vehicle l (l≠m) in the time slot n is represented as: where M represents the number of UAVs, d ml represents the distance between UAV m and UAV l in time slot n, P A represents the signal transmission power of the UAV, B A represents the total bandwidth of the communication between the UAVs, and the plurality of UAVs divides the bandwidth, represents the inter-UAV channel gain when the transmission distance is 1 m, σ 2 represents the Gaussian noise power spectral density of the channel; With the cooperation ability among multiple UAVs, the secondary offloading among UAVs is considered after the computing tasks are offloaded from the ground terminal to the UAVs. Let the task data size generated by the ground terminal k at time slot n be D k [n] and assume that the task offloading variable between the ground terminal k and the UAV m is x km [n] and assume that the task offloading variable between the ground terminal k and the UAV m is x km [n]∈{0,1}, which represents whether the task of the ground terminal k is offloaded to the UAV m for computation at time slot n. That is, when x km [n] = 1, the UAV m assists the ground terminal k to complete the task computation, and when x km [n] = 0, the opposite is true. In the secondary offloading, let the task offloading variable of the UAV m assisting the UAV l in edge computing be λ lm [n](m, l ∈ M), and λ lm [n] ∈ [0, 1] represents the proportion of tasks offloaded from the UAV l to the UAV m in the time slot n. In particular, when l = m, i.e. λ mm [n] represents the proportion of tasks remaining in the UAV m after the secondary offloading, and satisfies Let the CPU computation amount required for each ground terminal per 1 bit of data be κ, and the CPU computation rate of the ground terminal k be f k Then, the time delay of the local computation of the ground terminal k in the time slot n is Thus, in time slot n, the offloading transmission latency of ground terminal k Further, in time slot n, the amount of task data received by drone m after the initial offloading In time slot n, the transmission latency of drone l offloading to m wherein denotes the amount of task data received by drone m in time slot n; According to the unloading model set above, the task unloading transmission needs to be transmitted from the ground terminal to the air first, and then transmitted by the unmanned aerial vehicle, so the edge unloading transmission delay of the task of the unmanned aerial vehicle m after the secondary unloading in the time slot n Specifically represented as: Let the CPU assigned to drone m in time slot n compute at rate f ml [n] assuming that the task assigned to the drone in the current time slot is stored in the buffer and computation is performed in the next time slot, the edge task computation latency of drone m in time slot n λ lm [n-1] denotes the task offloading variable in time slot n-1. Since the local execution of the task, the offloading transmission of the task, and the edge execution of the task in the model can be performed in parallel, the length T[n] of the time slot n satisfies: wherein i.e. the objective function. 2.The method of claim 1, wherein, The initialization of all state information includes the initialization of the multi-UAV cooperative computing task offloading scheme, the initialization of the UAV computing resource allocation scheme, and the initialization of the UAV flight trajectory.

3. A multi-UAV cooperative computing task offloading, resource allocation and trajectory planning system, characterized in that, The cloud server specifically includes an initialization and objective function determination unit, a computing task offloading result acquisition unit, a computing resource allocation scheme acquisition unit, a UAV horizontal trajectory acquisition unit, a judgment unit, and an output unit. The objective function determination unit is configured to initialize all state information and determine the objective function. The computing task offloading result acquisition unit is configured to obtain the multi-UAV cooperative computing task offloading result according to the objective function. The computing resource allocation scheme acquisition unit is configured to update the objective function according to the computing task offloading result, and obtain the UAV computing resource allocation scheme according to the updated objective function. The UAV horizontal trajectory acquisition unit is configured to update the updated objective function again according to the UAV computing resource allocation result, and obtain the UAV horizontal trajectory according to the updated objective function. The judgment unit is configured to determine whether the maximum iteration number is reached. If the maximum iteration number is reached, the obtained computing task offloading result, UAV computing resource allocation scheme, and UAV horizontal trajectory are output as the best UAV offloading result, the best UAV computing resource allocation scheme, and the best UAV flight trajectory respectively by the output unit; otherwise, the iteration number is increased by 1, and the output unit outputs the best UAV offloading result, the best UAV computing resource allocation scheme, and the best UAV flight trajectory until the maximum iteration number is reached. performing initialization of all state information, determining the objective function includes determining the signal transmission rate of the ground terminal k and the unmanned aerial vehicle m in the time slot n is represented as: where P D denotes the signal transmission power of the ground terminal, B D denotes the total bandwidth of the air-ground communication, K denotes the number of ground terminals, and the ground terminals are evenly allocated bandwidth, denotes the channel gain between the ground terminals when the transmission distance is 1 m, d km [n] denotes the distance between the ground terminal k and the unmanned aerial vehicle m in the time slot n, σ 2 denotes the Gaussian noise power spectral density of the channel; Let the position vector of ground terminal k be The position vector of UAV m at time slot n is The distance d between ground terminal k and UAV m at time slot n is km [n] is represented as performing initialization of all state information, determining the objective function includes determining the signal transmission rate of the unmanned aerial vehicle m with the unmanned aerial vehicle l (l≠m) in the time slot n is represented as: where M represents the number of UAVs, d ml represents the distance between UAV m and UAV l in time slot n, P A represents the signal transmission power of the UAV, B A represents the total bandwidth of the communication between the UAVs, and the plurality of UAVs divides the bandwidth, represents the inter-UAV channel gain when the transmission distance is 1 m, σ 2 represents the Gaussian noise power spectral density of the channel; With the cooperation ability among multiple UAVs, the secondary offloading among UAVs is considered after the computing tasks are offloaded from the ground terminal to the UAVs. Let the data size of the tasks generated by the ground terminal k at time slot n be D k [n] and assume that the task offloading variable between the ground terminal k and the UAV m is x km [n] and assume that the task offloading variable between the ground terminal k and the UAV m is x km [n]∈{0,1}, which indicates whether the task of the ground terminal k is offloaded to the UAV m for computation at time slot n. That is, when x km [n] = 1, the UAV m assists the ground terminal k to complete the task computation, and when x km [n] = 0, the opposite is true. In the secondary offloading, let the task offloading variable of the UAV m assisting the UAV l in edge computing be λ lm [n] (m, l ∈ M), and λ lm [n] ∈ [0, 1] represents the proportion of tasks offloaded from the UAV l to the UAV m in the time slot n. In particular, when l = m, i.e. λ mm [n] represents the proportion of tasks remaining in the UAV m after the secondary offloading, and satisfies Let the CPU computation amount required for each ground terminal per 1 bit of data be κ, and the CPU computation rate of the ground terminal k be f k Then, the time delay of the local computation of the ground terminal k in the time slot n is Thus, in time slot n, the offloading transmission latency of the ground terminal k Further, in time slot n, the amount of task data received by the UAV m after the initial offloading In time slot n, the transmission latency of the UAV l offloading to m wherein denotes the amount of task data received by the UAV m in time slot n; According to the unloading model set above, the task unloading transmission needs to be transmitted from the ground terminal to the air first, and then transmitted by the unmanned aerial vehicle, so the edge unloading transmission delay of the task of the unmanned aerial vehicle m after the secondary unloading in the time slot n Specifically represented as: Let the CPU assigned to drone m in time slot n compute at rate f ml [n] assuming that the task assigned to the drone in the current time slot is stored in the buffer and computation is performed in the next time slot, the edge task computation latency of drone m in time slot n λ lm [n-1] denotes the task offloading variable in time slot n-1. Since the local execution of the task, the offloading transmission of the task, and the edge execution of the task in the model can be performed in parallel, the length T[n] of the time slot n satisfies: wherein i.e. the objective function. 4.The multi-UAV cooperative computing’s task offloading, resource allocation and trajectory planning system of claim 3, wherein, The initialization of all state information in the initialization and objective function determination unit includes the initialization of the multi-UAV cooperative computing task offloading scheme, the initialization of the UAV computing resource allocation scheme, and the initialization of the UAV flight trajectory.

Images