Task offloading method and apparatus, unmanned aerial vehicle, medium and product

Abstract

The application relates to a task offloading method and device, a UAV, a medium and a product. The method comprises the following steps: after detecting a new task, the positions and calculation time delays of UAVs in the same group are acquired; a target decision, a target segmentation strategy and a target transmission power are determined according to the positions and calculation time delays of the UAVs and a pre-established time delay optimization function; the target decision is used to represent a target UAV for processing a current task, and the target segmentation strategy comprises a segmentation point of the current task; and the current task is processed according to the target decision, the target segmentation strategy and the target transmission power. By using the method, the calculation resources of other UAVs can be fully called to perform collaborative reasoning of a neural network, the completion time delay of the current task is ensured to be low, and the waiting time delay of the new task is reduced.

Description

Technical Field

[0001] This application relates to the field of unmanned aerial vehicle (UAV) technology, and in particular to a task unloading method, apparatus, UAV, medium, and product. Background Technology

[0002] As a type of intelligent flight device, drones have the advantages of convenient deployment, flexible mobility, and line-of-sight communication, and can participate in mobile edge computing in different roles.

[0003] In traditional technologies, images are typically captured by drones, which then unload the captured images onto powerful ground base stations or edge servers for processing, thereby achieving target detection.

[0004] However, in some scenarios, drones may be unable to offload captured images to a ground server and must perform target detection locally. Due to the limitations of a single drone's computing power, when a new detection task is generated, it needs to wait for the current detection task to complete, and in some cases, the waiting time is quite long. Summary of the Invention

[0005] Therefore, it is necessary to provide a task unloading method, apparatus, drone, medium, and product that can solve the problem of long waiting time for new detection tasks, in order to address the above-mentioned technical problems.

[0006] Firstly, this application provides a task unloading method. The method includes:

[0007] After a new task is detected, the position and computational latency of each drone in the same group are obtained;

[0008] Based on the location and computational latency of each UAV, as well as the pre-established latency optimization function, the target decision, target segmentation strategy, and target transmission power are determined. The target decision is used to characterize the target UAV that is processing the current task, and the target segmentation strategy includes the segmentation point of the current task.

[0009] The current task is processed based on target decision, target segmentation strategy, and target transmission power.

[0010] In one embodiment, the determination of target decision, target segmentation strategy, and target transmission power based on the location and computational latency of each UAV, as well as a pre-established latency optimization function, includes:

[0011] Obtain initial decision, initial segmentation strategy, and initial transmit power;

[0012] Based on the initial decision, initial segmentation strategy, initial transmission power, and the position and computational delay of each UAV, the delay optimization function is iteratively calculated to obtain the target decision, target segmentation strategy, and target transmission power.

[0013] In one embodiment, the delay optimization function is iteratively calculated alternately based on the initial decision, initial segmentation strategy, initial transmission power, and the position and computational delay of each UAV, including:

[0014] The initial segmentation strategy, initial transmission power, and the position and computation delay of each UAV are substituted into the delay optimization function, and intermediate decisions are calculated based on preset constraints.

[0015] Substitute the intermediate decision, initial transmission power, and the position and computation delay of each UAV into the delay optimization function, and calculate the intermediate segmentation strategy according to the preset constraints.

[0016] The intermediate decision-making and intermediate segmentation strategies, as well as the positions and computational delays of each UAV, are substituted into the delay optimization function, and the intermediate transmission power is calculated according to the preset constraints.

[0017] If the intermediate decision, intermediate segmentation strategy, and intermediate transmission power meet the convergence conditions, then the intermediate decision is determined as the target decision, the intermediate segmentation strategy is taken as the target segmentation strategy, and the intermediate transmission power is taken as the target transmission power.

[0018] In one embodiment, the aforementioned preset constraints include delay constraints and energy consumption constraints. An intermediate decision is calculated based on these preset constraints, including:

[0019] Intermediate decisions are calculated based on time delay constraints and energy consumption constraints.

[0020] In one embodiment, the aforementioned preset constraints include delay constraints, energy consumption constraints, and segmentation constraints. An intermediate segmentation strategy is calculated based on these preset constraints, including:

[0021] The intermediate segmentation strategy is calculated based on the time delay constraint, energy consumption constraint, and segmentation constraint.

[0022] In one embodiment, the aforementioned preset constraints include energy consumption constraints and power constraints. The intermediate transmit power is calculated based on the preset constraints, including:

[0023] The intermediate transmit power is calculated based on energy consumption constraints and power constraints.

[0024] In one embodiment, the above-mentioned iterative calculation of the delay optimization function based on the initial decision, initial segmentation strategy, initial transmission power, and the position and computational delay of each UAV further includes:

[0025] If the intermediate decision, intermediate segmentation strategy, and intermediate transmission power do not meet the convergence conditions, then the intermediate decision, intermediate segmentation strategy, and intermediate transmission power are recalculated until they meet the convergence conditions.

[0026] In one embodiment, the above-mentioned iterative calculation of the delay optimization function based on the initial decision, initial segmentation strategy, initial transmission power, and the position and computational delay of each UAV further includes:

[0027] Establish an energy consumption model;

[0028] Determine energy consumption constraints based on the energy consumption model.

[0029] In one embodiment, the task unloading method described above further includes:

[0030] Establish computational delay models and transmission delay models, and determine the total delay model based on the computational delay models and transmission delay models;

[0031] The delay optimization function and delay constraints are determined based on the total delay model.

[0032] Secondly, this application also provides a task unloading device. The device includes:

[0033] The acquisition module is used to acquire the position and computational latency of each drone in the same group after a new task is detected;

[0034] The determination module is used to determine the target decision, target segmentation strategy, and target transmission power based on the position and computational delay of each UAV and a pre-established delay optimization function; the target decision is used to characterize the target UAV for processing the current task, and the target segmentation strategy includes the segmentation point of the current task;

[0035] The processing module is used to process the current task based on the target decision, target segmentation strategy, and target transmission power.

[0036] In one embodiment, the determining module includes:

[0037] The acquisition submodule is used to acquire the initial decision, initial segmentation strategy, and initial transmit power;

[0038] The calculation submodule performs alternating iterative calculations on the delay optimization function based on the initial decision, initial segmentation strategy, initial transmission power, and the position and calculation delay of each UAV, to obtain the target decision, target segmentation strategy, and target transmission power respectively.

[0039] In one embodiment, the computing submodule includes:

[0040] The decision calculation unit is used to substitute the initial segmentation strategy, initial transmission power, and the position and calculation delay of each UAV into the delay optimization function, and calculate the intermediate decision according to the preset constraints.

[0041] The segmentation strategy calculation unit is used to substitute intermediate decisions, initial transmission power, and the positions and calculation delays of each UAV into the delay optimization function, and calculate the intermediate segmentation strategy according to preset constraints.

[0042] The power calculation unit is used to substitute the intermediate decision and intermediate segmentation strategy, as well as the position and calculation delay of each UAV, into the delay optimization function, and calculate the intermediate transmission power according to the preset constraints.

[0043] The determining unit is used to determine the intermediate decision as the target decision, the intermediate segmentation strategy as the target segmentation strategy, and the intermediate transmission power as the target transmission power if the intermediate decision, intermediate segmentation strategy, and intermediate transmission power meet the convergence conditions.

[0044] In one embodiment, the segmentation strategy calculation unit is specifically used to calculate an intermediate segmentation strategy based on the time delay constraint, energy consumption constraint and segmentation constraint.

[0045] In one embodiment, the power calculation unit is specifically used to calculate the intermediate transmit power based on energy consumption constraints and power constraints.

[0046] In one embodiment, the calculation submodule is further configured to recalculate the intermediate decision, intermediate segmentation strategy, and intermediate transmission power if the intermediate decision, intermediate segmentation strategy, and intermediate transmission power do not meet the convergence conditions, until the intermediate decision, intermediate segmentation strategy, and intermediate transmission power meet the convergence conditions.

[0047] In one embodiment, the task unloading device further includes:

[0048] The first model building module is used to build the energy consumption model;

[0049] The constraint determination module is used to determine energy consumption constraints based on the energy consumption model.

[0050] In one embodiment, the task unloading device further includes:

[0051] The second model building module is used to build the computation delay model and the transmission delay model, and to determine the total delay model based on the computation delay model and the transmission delay model.

[0052] The function determination module is used to determine the delay optimization function and delay constraints based on the total delay model.

[0053] Thirdly, this application also provides an unmanned aerial vehicle (UAV). The UAV includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to perform the steps described in the first aspect.

[0054] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, performs the steps described in the first aspect.

[0055] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, performs the steps described in the first aspect.

[0056] The aforementioned task unloading method, apparatus, UAV, medium, and product, upon detecting a new task, acquire the position and computational latency of each UAV in the same group; determine the target decision, target segmentation strategy, and target transmission power based on the position and computational latency of each UAV and a pre-established latency optimization function; the target decision characterizes the target UAV for processing the current task, and the target segmentation strategy includes the segmentation point of the current task; the current task is processed based on the target decision, target segmentation strategy, and target transmission power. Using this embodiment, the computational resources of other UAVs can be fully utilized to collaboratively process tasks, ensuring a low latency for completing the current task while reducing the waiting latency for new tasks. Attached Figure Description

[0057] Figure 1 This is an application environment diagram of the task unloading method in one embodiment;

[0058] Figure 2 This is one of the flowcharts illustrating a task unloading method in one embodiment;

[0059] Figure 3 This is a second flowchart illustrating a task unloading method in one embodiment;

[0060] Figure 4 This is the third flowchart of a task unloading method in one embodiment;

[0061] Figure 5 This is a flowchart of the task unloading method in one embodiment;

[0062] Figure 6 This is the fifth flowchart of a task unloading method in one embodiment;

[0063] Figure 7 This is a flowchart of the task unloading method in one embodiment, number six.

[0064] Figure 8This is one of the structural block diagrams of the task unloading device in one embodiment;

[0065] Figure 9 This is a second structural block diagram of the task unloading device in one embodiment;

[0066] Figure 10 This is the third structural block diagram of the task unloading device in one embodiment;

[0067] Figure 11 This is the fourth structural block diagram of the task unloading device in one embodiment;

[0068] Figure 12 This is a schematic diagram of the internal structure of a computer device in one embodiment. Detailed Implementation

[0069] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0070] The task unloading method provided in this application embodiment can be applied to, for example, Figure 1 The application environment shown includes multiple drones that can communicate with each other via a network. These drones may include, but are not limited to, inspection drones and aerial photography drones.

[0071] In one embodiment, such as Figure 2 As shown, a task unloading method is provided. This embodiment illustrates the application of this method to a drone. In this embodiment, the method includes the following steps:

[0072] S101, after detecting a new task, obtains the position and computational latency of each UAV in the same group.

[0073] In this context, each drone in the same unit includes the local drone U0 and other drones, and the set of other drones is represented as K = {U1, U2, U3, ..., U...}. K The positions of each UAV are represented as (x...). i ,y i ,h i ), where i = 0, 1, 2, ..., K.

[0074] Current task unloaded to drone U K It may be necessary to wait for the computation latency of other drones already performing their tasks, therefore it is necessary to consider the computation latency of other drones. k computation latency of existing tasks

[0075] When the local drone U0 detects a new task, it obtains the positions (x) of all drones in the same group. i ,y i ,h i ) and computation delay

[0076] S102, based on the position and computational delay of each UAV, as well as the pre-established delay optimization function, the target decision, target segmentation strategy, and target transmission power are determined; the target decision is used to characterize the target UAV for processing the current task, and the target segmentation strategy includes the segmentation point of the current task.

[0077] Wherein, the delay optimization function is

[0078] Decision A = {a k ,0≤k≤K,k∈Z}, where a k The unloading selection function indicates whether the current task R should be completed on the local drone U0. old , can be represented as:

[0079]

[0080] Segmentation strategy M = {l m ,n≤m≤N,m∈Z}, where,l m This is the split point for the current task.

[0081] Transmit power Where p is the signal transmission power of the UAV U0. This refers to the maximum value of the drone's transmission power. However, this application embodiment does not impose a limit on the maximum value of the drone's transmission power, and it can be set according to the actual situation.

[0082] Based on the location (x) of each drone i ,y i ,h i ) and computation delay and the pre-established delay optimization function Determine the target decision A (1) Target segmentation strategy M (1) and target transmit power P (1) .

[0083] S103 processes the current task based on target decision, target segmentation strategy, and target transmission power.

[0084] When the objective decision A (1) When the value is 0, the current task will continue to be completed on the local drone U0.

[0085] When the objective decision A (1)When the value is 1, the current task is segmented according to the target segmentation strategy M, and the current task is offloaded to the target decision A with the target transmission power P. (1) The drone with the specified value.

[0086] In the above embodiments, after a new task is detected, the positions and computational latency of each UAV in the same group are obtained. Based on the positions and computational latency of each UAV, and a pre-established latency optimization function, a target decision, a target segmentation strategy, and a target transmission power are determined. The target decision characterizes the target UAV for processing the current task, and the target segmentation strategy includes the segmentation point of the current task. Processing the current task according to the target decision, target segmentation strategy, and target transmission power can fully utilize the computational resources of other UAVs to coordinate task processing, ensuring a low latency for completing the current task, while reducing the waiting latency for new tasks.

[0087] In one embodiment, such as Figure 3 As shown, the steps described above for determining the target decision, target segmentation strategy, and target transmission power based on the location and computational delay of each UAV, as well as a pre-established delay optimization function, may include:

[0088] S201, obtain initial decision, initial segmentation strategy and initial transmit power.

[0089] Obtain user input regarding the settings for decision-making, segmentation strategy, and transmit power to obtain initial decision A. (0) Initial Segmentation Strategy M (0) and initial transmit power P (0) Alternatively, obtain the default decision, segmentation strategy, and transmit power to get the initial decision A. (0) Initial Segmentation Strategy M (0) and initial transmit power P (0) Initial decision A (0) Initial Segmentation Strategy M (0) and initial transmit power P (0) The method of obtaining the data can be specifically set according to the actual situation, and this application embodiment does not limit it.

[0090] S202, based on the initial decision, initial segmentation strategy, initial transmission power, and the position and computational delay of each UAV, the delay optimization function is iteratively calculated to obtain the target decision, target segmentation strategy, and target transmission power respectively.

[0091] Obtain the initial decision A (0) Initial Segmentation Strategy M (0) and initial transmit power P (0) Then, set the fault tolerance parameters, number of iterations, and maximum number of iterations for the alternating iteration algorithm.

[0092] Based on initial decision A (0) Initial Segmentation Strategy M (0) and initial transmit power P (0) and the location of each drone (x i ,y i ,h i ) and computation delay Calculate the initial completion delay of the current task. .

[0093] The initial completion delay of the current task can be obtained using the following formula:

[0094]

[0095] In the formula, a k For the unloading selection function, T k Unload the current task to another drone. k The completion delay.

[0096] The current task is unloaded to another drone. k Completion delay T k It can be represented as:

[0097]

[0098] In the formula, Indicates local computation latency. For transmission delay, For other drones U k Processing delay of existing tasks, For the current mission, other UAVs k The computational latency.

[0099] Transmission delay It is determined by the position (x) of each drone i ,y i ,h i Obtain the channel gain h k According to the channel gain h k The data transmission rate r is obtained by combining relevant data. k Combined with formula Calculations show that D lm Indicate l m The size of the output data of the node.

[0100] Specifically, the distance d between the two drones k It can be represented as:

[0101]

[0102] According to the channel gain h kThe data transmission rate r is obtained by combining relevant data. k :

[0103]

[0104] In the formula, B represents the signal bandwidth, p represents the signal transmission power of the local UAV U0, β represents the channel gain coefficient, and σ 2 h represents the variance of the Gaussian white noise in the channel. k Channel gain can be expressed by the formula get.

[0105] Finally, according to the formula The transmission delay was calculated. In the formula, Indicate l m The size of the output data of the node.

[0106] Since the three optimization variables in the delay optimization function are uncoupled, this embodiment uses an alternating iteration method to calculate the three optimization variables in the delay optimization function. When calculating any optimization variable in the delay optimization function, the other two optimization variables are fixed to obtain the target decision A. (1) Target segmentation strategy M (1) and target transmit power P (1) .

[0107] In the above embodiments, the initial decision, initial segmentation strategy, and initial transmission power are obtained. Based on the initial decision, initial segmentation strategy, initial transmission power, and the positions and computational delays of each UAV, the delay optimization function is iteratively calculated alternately to obtain the target decision, target segmentation strategy, and target transmission power. By obtaining the optimal solution of the delay optimization function through the alternating iterative algorithm, the current task can be offloaded to other UAVs in the cluster for processing at the most appropriate number of segmentation layers with a small delay cost, and new tasks can begin to be processed locally, significantly reducing the waiting delay for new tasks.

[0108] In one embodiment, such as Figure 4 As shown, the above-mentioned iterative calculation of the delay optimization function based on the initial decision, initial segmentation strategy, initial transmission power, and the position and computational delay of each UAV can include:

[0109] S301, substitute the initial segmentation strategy, initial transmission power, and the position and computational delay of each UAV into the delay optimization function, and calculate the intermediate decision according to the preset constraints.

[0110] Calculate intermediate decision A (x) At that time, the initial segmentation strategy M (0) and initial transmit power P (0)For a fixed quantity, the initial segmentation strategy M is... (0) Initial transmit power P (0) and the location of each drone (x i ,y i ,h i ) and computation delay Substitute into the delay optimization function In the middle, and based on the preset constraints, the intermediate decision A is calculated. (x) .

[0111] Optionally, in embodiments of this application, the intermediate segmentation strategy M may also be calculated first. (x) Initial decision A (0) and initial transmit power P (0) As fixed quantities, the initial decision, initial transmission power, and the position (x) of each UAV are included. i ,y i ,h i ) and calculation delay T k comp Substitute the values ​​into the delay optimization function and calculate the intermediate splitting strategy M based on the preset constraints. (x) .

[0112] Optionally, embodiments of this application may also first calculate the intermediate transmit power P. (x) , the initial decision A (0) Initial Segmentation Strategy M (0) and the location of each drone (x i ,y i ,h i ) and computation delay Substituting the values ​​into the time delay optimization function and calculating the intermediate transmit power P based on preset constraints, (x) .

[0113] S302 substitutes the intermediate decision, initial transmission power, and the position and computational delay of each UAV into the delay optimization function, and calculates the intermediate segmentation strategy according to the preset constraints.

[0114] Calculate the intermediate splitting strategy M (x) At that time, intermediate decision A (x) and initial transmit power P (0) For a fixed quantity, the intermediate decision A is... (x) Initial transmit power P (0) and the location of each drone (x i ,y i ,h i ) and computation delay Substitute into the delay optimization function In the middle, the intermediate segmentation strategy M is calculated based on preset constraints. (x) .

[0115] S303 substitutes the intermediate decision and intermediate segmentation strategies, as well as the positions and computational delays of each UAV, into the delay optimization function, and calculates the intermediate transmit power according to preset constraints.

[0116] Calculate the intermediate transmit power P (x) At that time, intermediate decision A (x) And the intermediate segmentation strategy M (x) For a fixed quantity, the intermediate decision A is... (x) And the intermediate segmentation strategy M (x) and the location of each drone (x i ,y i ,h i ) and computation delay Substitute into the delay optimization function The intermediate transmission power P is calculated based on preset constraints. (x) .

[0117] S304. If the intermediate decision, intermediate segmentation strategy, and intermediate transmission power meet the convergence conditions, then the intermediate decision is determined as the target decision, the intermediate segmentation strategy is taken as the target segmentation strategy, and the intermediate transmission power is taken as the target transmission power.

[0118] The convergence condition is that the number of iterations of the alternating iteration algorithm reaches the maximum number of iterations or the completion delay of the current task at the current number of iterations. Completion delay of the current task compared to the previous iteration number The difference is less than the fault tolerance parameter.

[0119] If the intermediate decision, intermediate segmentation strategy, and intermediate transmission power meet the above convergence conditions, then intermediate decision A will be... (x) Determined as target decision A (1) The middle segmentation strategy M (x) As the target segmentation strategy M (1) The intermediate transmit power P (x) As the target transmission power P (1) .

[0120] In the above embodiments, the initial segmentation strategy, initial transmission power, and the position and computation delay of each UAV are substituted into the delay optimization function, and an intermediate decision is calculated according to preset constraints; the intermediate decision, initial transmission power, and the position and computation delay of each UAV are substituted into the delay optimization function, and an intermediate segmentation strategy is calculated according to preset constraints; the intermediate decision, intermediate segmentation strategy, and the position and computation delay of each UAV are substituted into the delay optimization function, and an intermediate transmission power is calculated according to preset constraints; if the intermediate decision, intermediate segmentation strategy, and intermediate transmission power meet the convergence conditions, the intermediate decision is determined as the target decision, the intermediate segmentation strategy is taken as the target segmentation strategy, and the intermediate transmission power is taken as the target transmission power. Since the three variables to be optimized in the delay optimization function are uncoupled, the intermediate decision, intermediate splitting strategy, and intermediate transmit power are calculated given the other two variables and related parameters, which reduces the complexity of solving the delay optimization function. Furthermore, when the convergence condition is met, the target decision, target splitting strategy, and target transmit power are converged, so that the task can be offloaded according to the target decision, target splitting strategy, and target transmit power.

[0121] In one embodiment, the preset constraints include time delay constraints and energy consumption constraints. The step of calculating intermediate decisions based on the preset constraints may include:

[0122] Intermediate decisions are calculated based on time delay constraints and energy consumption constraints.

[0123] The time delay constraint can be expressed as:

[0124] T new ≤T th (6)

[0125] In the formula, T new The waiting delay for a new task is T, which is the time delay for waiting for a new task when the local drone U0 completes the current task. new The completion delay T0 of the current task is equal to the processing delay of the local drone U0 for the current task, that is: When the current task is not completed by the local drone U0, the waiting delay for the new task is the time it takes for the local drone U0 to process the current task and move it to the offloading node l. m The required computation time and the waiting time for new tasks can be expressed as: T th The maximum allowable waiting time for a new task can be set according to the actual situation, and this application embodiment does not impose any restrictions on it.

[0126] The energy consumption constraint can be expressed as:

[0127] E≤Eth (7)

[0128] In the formula, E represents the total energy consumption of the current task, which can be calculated using the following formula (8):

[0129]

[0130] in, The energy consumption for mission transmission is the energy consumption for mission transmission when the local drone U0 completes the current mission. When the drone U0 is not completing its current task locally, the task transmission energy consumption is 0; φ0(l n ,l m ) represents the locally calculated energy consumption, φ k (l m Unload the current task to another drone. k The computational energy consumption.

[0131] The optimization problem of intermediate decision-making can be expressed as:

[0132]

[0133] The optimization problem of intermediate decisions can be handled using the relaxation method employed in convex optimization, where the objective decision A is... (1) If we relax the binary variables to continuous variables, the optimization problem of the above intermediate decision can be expressed as:

[0134]

[0135] Due to goal decision A (1) After relaxation, it becomes a continuous variable, so its value may not be an integer. In this application's embodiment, the target decision A is... (1) The largest value of a in the middle k Set to 1, and the rest of a k By setting the value to 0, an integer solution is obtained as the intermediate decision A. (x) .

[0136] Intermediate decision A is calculated based on time delay constraints and energy consumption constraints. (x) .

[0137] In the above embodiments, intermediate decisions are calculated based on time delay and energy consumption constraints. The optimization problem of intermediate decisions is transformed into a standard linear programming problem using relaxation methods in convex optimization. This avoids the situation where the complexity of the optimization problem increases dramatically with the number of drones when using the direct traversal method, thus reducing the solution complexity of the intermediate decision optimization problem and making it convenient and efficient.

[0138] In one embodiment, the aforementioned preset constraints include delay constraints, energy consumption constraints, and segmentation constraints. The step of calculating the intermediate segmentation strategy based on the preset constraints may include:

[0139] The intermediate segmentation strategy is calculated based on the time delay constraint, energy consumption constraint, and segmentation constraint.

[0140] The partitioning constraint can be expressed as:

[0141] n≤m≤N,m∈Z + (11)

[0142] The optimization problem of the intermediate segmentation strategy can be expressed as:

[0143]

[0144] Based on the partitioning constraint, solution sets C1 and C2 that satisfy the time delay constraint and energy consumption constraint are found through a traversal method. Then, the feasible solution set C that satisfies the time delay constraint, energy consumption constraint, and partitioning constraint is obtained. * =C1∩C2, then from the feasible solution set C * The middle segmentation strategy is obtained by selecting from the middle segments:

[0145]

[0146] In the above embodiments, the intermediate segmentation strategy is calculated based on the time delay constraint, energy consumption constraint and segmentation constraint. The intermediate segmentation strategy is solved by the traversal method. As the number of network layers increases, the complexity of the solution does not increase significantly, and the intermediate segmentation strategy can be obtained relatively quickly.

[0147] In one embodiment, the aforementioned preset constraints include energy consumption constraints and power constraints. The step of calculating the intermediate transmission power based on the preset constraints may include:

[0148] The intermediate transmit power is calculated based on energy consumption constraints and power constraints.

[0149] The power constraint can be expressed as:

[0150]

[0151] In the formula, p represents the signal transmission power of the local drone U0. The maximum value of the transmission power is not limited in this embodiment of the application, and can be set according to the actual situation.

[0152] The optimization problem of intermediate transmit power can be expressed as:

[0153]

[0154] When solving for the intermediate transmit power, it is necessary to classify the algorithms for processing the optimization problem of the intermediate transmit power according to the values ​​of A and M.

[0155] When the local drone U0 completes its current task, the optimization problem of the intermediate transmit power mentioned above... At this point, since the current task is completed on the local drone U0, no unloading is required. Therefore, the intermediate transmission power P... (x) The value is 0; when the current task is not completed locally by drone U0, if At this point, T(A,M,P) in the above optimization problem of intermediate transmission power can be expressed as:

[0156]

[0157] If the current task is not completed by the local drone U0, then The optimization problem of intermediate transmission power can be solved using convex optimization methods.

[0158] If the current task is not completed by the local drone U0, then The optimization problem of intermediate transmission power is divided into two different optimization problems based on the value of the signal transmission power p of the local UAV U0. According to formula (17), the solution of the two problems is... and The optimal solution is selected to obtain the optimal transmission power that minimizes the delay in completing the current task, i.e., the intermediate transmission power:

[0159]

[0160] In the above embodiments, the intermediate transmission power is calculated based on energy consumption constraints and power constraints. By comparing the computation latency of existing tasks of other UAVs with the computation latency of the local UAV, the optimization problem of the intermediate transmission power is segmented, making the obtained intermediate transmission power more accurate.

[0161] In one embodiment, the step of iteratively calculating the delay optimization function based on the initial decision, initial segmentation strategy, initial transmission power, and the position and computation delay of each UAV further includes: if the intermediate decision, intermediate segmentation strategy, and intermediate transmission power do not meet the convergence conditions, then recalculate the intermediate decision, intermediate segmentation strategy, and intermediate transmission power until the intermediate decision, intermediate segmentation strategy, and intermediate transmission power meet the convergence conditions.

[0162] The convergence condition is that the number of iterations of the alternating iteration algorithm reaches the maximum number of iterations, or the completion delay of the current task at the current number of iterations. Completion delay of the current task compared to the previous iteration number The difference is less than the fault tolerance parameter.

[0163] If intermediate decision A (x) Intermediate Segmentation Strategy M (x) and intermediate transmit power P (x) If the above convergence conditions are not met, then the intermediate decision A should be recalculated. (x) Intermediate Segmentation Strategy M (x) and intermediate transmit power P (x) Until intermediate decision A (x) Intermediate Segmentation Strategy M (x) and intermediate transmit power P (x) Continue until the convergence condition is met.

[0164] In the above embodiments, if the intermediate decision, intermediate segmentation strategy, and intermediate transmission power do not meet the convergence conditions, the intermediate decision, intermediate segmentation strategy, and intermediate transmission power are recalculated until they meet the convergence conditions, so as to obtain a suitable intermediate decision, intermediate segmentation strategy, and intermediate transmission power.

[0165] In one embodiment, such as Figure 5 As shown, embodiments of this disclosure may further include the following steps:

[0166] S401, Establish an energy consumption model.

[0167] Unlike ground-based servers, drones need to perform computational tasks in the air and cannot continuously receive power from the ground. Therefore, they rely on their own batteries for energy, and the constraints of transmission power and energy must be considered when performing task processing. This application's embodiments establish the computational energy consumption E required for task processing. comp Energy consumption for transmission between the model and the drone Model.

[0168] Task transmission power consumption The model is:

[0169]

[0170] UAV i From l j The computational energy required for a node to start deep neural network inference and output the result is Where, α i / 2 is for drones U i The CPU calculates the effective capacitance coefficient of the chipset. Indicate l jThe size of the node's output data determines the computational energy E of the current task. comp The model can be represented as:

[0171]

[0172] According to formulas (18) and (19), the total energy consumption E for completing the current task can be expressed as:

[0173]

[0174] Establish the above transmission energy consumption Model, computational energy consumption E comp Model and total energy consumption E model.

[0175] S402, Determine energy consumption constraints based on the energy consumption model.

[0176] Based on the energy consumption model, the energy consumption constraint condition E≤E is determined. th E th This indicates the maximum allowable energy consumption for processing the task. The maximum allowable energy consumption for processing the task can be set according to the actual situation, and this application embodiment does not impose any restrictions on it.

[0177] In the above embodiments, an energy consumption model is established, and energy consumption constraints are determined based on the energy consumption model, so that a better offloading decision can be obtained under the energy consumption constraints.

[0178] In one embodiment, such as Figure 6 As shown, the embodiments of this disclosure also include the following steps:

[0179] S501, establish the computational delay model and the transmission delay model, and determine the total delay model based on the computational delay model and the transmission delay model.

[0180] The computational latency model takes into account various scenarios. When the local drone U0 completes the current task, the computational latency of the current task is... When the current task is not completed on the local drone U0, the current task needs to be loaded from the current node l on the local drone U0 before the task is unloaded. n Execution to split node l m ,use Local computation latency can be represented as:

[0181]

[0182] After unloading the task, the current task is performed on other drones. k The computation latency is

[0183] In addition to considering the computational latency of the current task, this application embodiment also considers that other drones in the cluster may also be currently performing tasks. Therefore, the current task is offloaded to other drones. k It still needs to wait for it to complete its local task before starting execution, so other drones need to be considered. k Processing delay of existing tasks Other drones U k The xth task currently being processed has been inferred to l. xnode From the nodes, we can obtain:

[0184]

[0185] The transmission delay model also considers various situations, such as the transmission delay when the local drone U0 completes its current task. When the current task is not completed on the local drone U0, the current task is offloaded to another drone U0. k , This indicates that the local drone U0 is different from other drones U0. k To address the task transmission delay between tasks, this application establishes a communication transmission model in free space.

[0186] Given that the local drone U0 has the position coordinates (x0, y0, h0), and other drones U0... k The position coordinates are (x k ,y k ,h k The distance d between the two drones k It can be represented as:

[0187]

[0188] Channel gain h k It can be represented as:

[0189]

[0190] In the formula, β represents the channel gain coefficient. According to Shannon's formula, the data transmission rate r k It can be represented as:

[0191]

[0192] In the formula, B represents the signal bandwidth, p represents the signal transmission power of the local UAV U0, and σ 2 This represents the variance of Gaussian white noise in the channel.

[0193] According to the data transmission rate r k The transmission delay can be expressed as:

[0194]

[0195] In the formula, Indicate l m The size of the output data of the node.

[0196] Based on the above calculation delay model and transmission delay model, the delay T from the current time to the output detection result of the current task can be obtained. old And the waiting time T between the current moment and the start of processing of a new task. new .

[0197] When the local drone U0 completes its current task, the waiting delay T for the new task... new The completion delay T0 of the current task is equal to the processing delay of the local drone U0 for the current task, that is:

[0198] When a drone U0 is not completing its current task locally, the completion delay of the current task is the time elapsed until other drones U0 complete their tasks. k Latency before processing starts and other UAVs k The sum of computational latency during operation, including the latency before processing begins, includes two scenarios: one is the transmission of data from the current task to other UAVs. k Afterwards, other drones U k If the system is already in an idle state, the latency before processing begins is the local computation latency. With transmission delay The sum; secondly, the current task is transmitted to other UAVs. k Afterwards, other drones U k The processing of existing tasks has not yet been completed, therefore the delay before processing begins is... Therefore, the current task is offloaded to another drone. k Completion delay T k It can be represented as:

[0199]

[0200] The waiting time for a new task is the time it takes for the local drone U0 to process the current task and move it to the offload node l. m Required computation latency, waiting latency T for new tasks new It can be represented as:

[0201]

[0202] remember The completion delay T of the current task old for:

[0203]

[0204] The waiting time for a new task is:

[0205]

[0206] Establish computational delay model and transmission delay model, and determine total delay model based on computational delay model and transmission delay model.

[0207] S502, determine the delay optimization function and delay constraints based on the total delay model.

[0208] The time delay optimization function can be expressed as:

[0209]

[0210] In the formula, T k Unload the current task to another drone. k The completion delay is obtained from the total delay model.

[0211] The total delay model also yields the delay constraints:

[0212] T new ≤T th (32)

[0213] In the formula, T new The waiting delay for a new task is T, which is the time delay for waiting for a new task when the local drone U0 completes the current task. new The completion delay T0 of the current task is equal to the processing delay of the local drone U0 for the current task, that is: When the current task is not completed by the local drone U0, the waiting delay for the new task is the time it takes for the local drone U0 to process the current task and move it to the offloading node l. m The required computation time and the waiting time for new tasks can be expressed as: T th The maximum allowable waiting time for a new task can be set according to the actual situation, and this application embodiment does not impose any restrictions on it.

[0214] In the above embodiments, a computational delay model and a transmission delay model are established, and a total delay model is determined based on the computational delay model and the transmission delay model. The delay optimization function and delay constraints are then determined based on the total delay model. Since the delay optimization function and delay constraints are obtained, it is beneficial to solve the delay optimization function under these constraints to obtain a UAV unloading decision with lower delay.

[0215] In one embodiment, such as Figure 7 The diagram illustrates the task unloading process, using an example of its application to a drone. The process includes the following steps:

[0216] S601, after detecting a new task, obtains the position and computational latency of each UAV in the same group.

[0217] S602, obtain initial decision, initial segmentation strategy and initial transmit power.

[0218] S603 substitutes the initial segmentation strategy, initial transmission power, and the position and computational delay of each UAV into the delay optimization function, and calculates the intermediate decision based on preset constraints.

[0219] S604 substitutes the intermediate decision, initial transmission power, and the position and computational delay of each UAV into the delay optimization function, and calculates the intermediate segmentation strategy according to the preset constraints.

[0220] S605 substitutes the intermediate decision and intermediate segmentation strategies, as well as the positions and computational delays of each UAV, into the delay optimization function, and calculates the intermediate transmit power according to preset constraints.

[0221] S606, if the intermediate decision, intermediate segmentation strategy, and intermediate transmission power meet the convergence conditions, then the intermediate decision is determined as the target decision, the intermediate segmentation strategy is taken as the target segmentation strategy, and the intermediate transmission power is taken as the target transmission power.

[0222] S607 If the intermediate decision, intermediate segmentation strategy, and intermediate transmission power do not meet the convergence conditions, then the intermediate decision, intermediate segmentation strategy, and intermediate transmission power are recalculated until they meet the convergence conditions.

[0223] S608 processes the current task based on target decision, target segmentation strategy, and target transmit power.

[0224] In the above embodiments, after detecting a new task, the positions and computational delays of each UAV in the same group are obtained; the initial decision, initial segmentation strategy, and initial transmit power are obtained; the initial segmentation strategy, initial transmit power, and the positions and computational delays of each UAV are substituted into the delay optimization function, and an intermediate decision is calculated according to preset constraints; the intermediate decision, initial transmit power, and the positions and computational delays of each UAV are substituted into the delay optimization function, and an intermediate segmentation strategy is calculated according to preset constraints; the intermediate decision, intermediate segmentation strategy, and the positions and computational delays of each UAV are substituted into the delay optimization function. The intermediate transmission power is calculated based on preset constraints. If the intermediate decision, intermediate segmentation strategy, and intermediate transmission power meet the convergence conditions, the intermediate decision is determined as the target decision, the intermediate segmentation strategy is taken as the target segmentation strategy, and the intermediate transmission power is taken as the target transmission power. If the intermediate decision, intermediate segmentation strategy, and intermediate transmission power do not meet the convergence conditions, they are recalculated until they meet the convergence conditions. The current task is then processed based on the target decision, target segmentation strategy, and target transmission power. Processing the current task based on the target decision, target segmentation strategy, and target transmission power can fully utilize the computing resources of other UAVs to coordinate task processing, ensuring low latency for completing the current task while reducing the waiting latency for new tasks.

[0225] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0226] Based on the same inventive concept, this application also provides a task unloading device for implementing the task unloading method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more task unloading device embodiments provided below can be found in the limitations of the task unloading method described above, and will not be repeated here.

[0227] In one embodiment, such as Figure 8 As shown, a task unloading device is provided, comprising:

[0228] The acquisition module 701 is used to acquire the position and computational latency of each UAV in the same group after a new task is detected;

[0229] The determination module 702 is used to determine the target decision, target segmentation strategy and target transmission power based on the position and computational delay of each UAV and a pre-established delay optimization function; the target decision is used to characterize the target UAV for processing the current task, and the target segmentation strategy includes the segmentation point of the current task;

[0230] The processing module 703 is used to process the current task based on the target decision, target segmentation strategy and target transmission power.

[0231] In one embodiment, such as Figure 9 As shown, the determining module 702 includes:

[0232] The acquisition submodule 7021 is used to acquire the initial decision, initial segmentation strategy, and initial transmit power;

[0233] The calculation submodule 7022 performs alternating iterative calculations on the delay optimization function based on the initial decision, initial segmentation strategy, initial transmission power, and the position and calculation delay of each UAV, to obtain the target decision, target segmentation strategy, and target transmission power respectively.

[0234] In one embodiment, the computing submodule includes:

[0235] The decision calculation unit is used to substitute the initial segmentation strategy, initial transmission power, and the position and calculation delay of each UAV into the delay optimization function, and calculate the intermediate decision according to the preset constraints.

[0236] The segmentation strategy calculation unit is used to substitute intermediate decisions, initial transmission power, and the positions and calculation delays of each UAV into the delay optimization function, and calculate the intermediate segmentation strategy according to preset constraints.

[0237] The power calculation unit is used to substitute the intermediate decision and intermediate segmentation strategy, as well as the position and calculation delay of each UAV, into the delay optimization function, and calculate the intermediate transmission power according to the preset constraints.

[0238] The determining unit is used to determine the intermediate decision as the target decision, the intermediate segmentation strategy as the target segmentation strategy, and the intermediate transmission power as the target transmission power if the intermediate decision, intermediate segmentation strategy, and intermediate transmission power meet the convergence conditions.

[0239] In one embodiment, the segmentation strategy calculation unit is specifically used to calculate an intermediate segmentation strategy based on the time delay constraint, energy consumption constraint and segmentation constraint.

[0240] In one embodiment, the power calculation unit is specifically used to calculate the intermediate transmit power based on energy consumption constraints and power constraints.

[0241] In one embodiment, the calculation submodule is further configured to recalculate the intermediate decision, intermediate segmentation strategy, and intermediate transmission power if the intermediate decision, intermediate segmentation strategy, and intermediate transmission power do not meet the convergence conditions, until the intermediate decision, intermediate segmentation strategy, and intermediate transmission power meet the convergence conditions.

[0242] In one embodiment, such as Figure 10 As shown, the task unloading device also includes:

[0243] The first model building module 704 is used to build the energy consumption model;

[0244] The constraint determination module 705 is used to determine energy consumption constraints based on the energy consumption model.

[0245] In one embodiment, such as Figure 11 As shown, the task unloading device also includes:

[0246] The second model establishment module 706 is used to establish the computation delay model and the transmission delay model, and to determine the total delay model based on the computation delay model and the transmission delay model.

[0247] The function determination module 707 is used to determine the delay optimization function and delay constraints based on the total delay model.

[0248] Each module in the aforementioned task unloading device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0249] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 12 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores task offloading data. The network interface communicates with external terminals via a network connection. When executed by the processor, the computer program implements a task offloading method.

[0250] Those skilled in the art will understand that Figure 12 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0251] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0252] After a new task is detected, the position and computational latency of each drone in the same group are obtained;

[0253] Based on the location and computational latency of each UAV, as well as the pre-established latency optimization function, the target decision, target segmentation strategy, and target transmission power are determined. The target decision is used to characterize the target UAV that is processing the current task, and the target segmentation strategy includes the segmentation point of the current task.

[0254] The current task is processed based on target decision, target segmentation strategy, and target transmission power.

[0255] In one embodiment, the processor further performs the following steps when executing the computer program:

[0256] Obtain initial decision, initial segmentation strategy, and initial transmit power;

[0257] Based on the initial decision, initial segmentation strategy, initial transmission power, and the position and computational delay of each UAV, the delay optimization function is iteratively calculated to obtain the target decision, target segmentation strategy, and target transmission power.

[0258] In one embodiment, the processor further performs the following steps when executing the computer program:

[0259] The initial segmentation strategy, initial transmission power, and the position and computation delay of each UAV are substituted into the delay optimization function, and intermediate decisions are calculated based on preset constraints.

[0260] Substitute the intermediate decision, initial transmission power, and the position and computation delay of each UAV into the delay optimization function, and calculate the intermediate segmentation strategy according to the preset constraints.

[0261] The intermediate decision-making and intermediate segmentation strategies, as well as the positions and computational delays of each UAV, are substituted into the delay optimization function, and the intermediate transmission power is calculated according to the preset constraints.

[0262] If the intermediate decision, intermediate segmentation strategy, and intermediate transmission power meet the convergence conditions, then the intermediate decision is determined as the target decision, the intermediate segmentation strategy is taken as the target segmentation strategy, and the intermediate transmission power is taken as the target transmission power.

[0263] In one embodiment, the processor further performs the following steps when executing the computer program:

[0264] Intermediate decisions are calculated based on time delay constraints and energy consumption constraints.

[0265] In one embodiment, the processor further performs the following steps when executing the computer program:

[0266] The intermediate segmentation strategy is calculated based on the time delay constraint, energy consumption constraint, and segmentation constraint.

[0267] In one embodiment, the processor further performs the following steps when executing the computer program:

[0268] The intermediate transmit power is calculated based on energy consumption constraints and power constraints.

[0269] In one embodiment, the processor further performs the following steps when executing the computer program:

[0270] If the intermediate decision, intermediate segmentation strategy, and intermediate transmission power do not meet the convergence conditions, then the intermediate decision, intermediate segmentation strategy, and intermediate transmission power are recalculated until they meet the convergence conditions.

[0271] In one embodiment, the processor further performs the following steps when executing the computer program:

[0272] Establish an energy consumption model;

[0273] Determine energy consumption constraints based on the energy consumption model.

[0274] In one embodiment, the processor further performs the following steps when executing the computer program:

[0275] The delay optimization function and delay constraints are determined based on the total delay model.

[0276] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0277] After a new task is detected, the position and computational latency of each drone in the same group are obtained;

[0278] Based on the location and computational latency of each UAV, as well as the pre-established latency optimization function, the target decision, target segmentation strategy, and target transmission power are determined. The target decision is used to characterize the target UAV that is processing the current task, and the target segmentation strategy includes the segmentation point of the current task.

[0279] The current task is processed based on target decision, target segmentation strategy, and target transmission power.

[0280] In one embodiment, when the computer program is executed by the processor, it further performs the following steps:

[0281] Obtain initial decision, initial segmentation strategy, and initial transmit power;

[0282] Based on the initial decision, initial segmentation strategy, initial transmission power, and the position and computational delay of each UAV, the delay optimization function is iteratively calculated to obtain the target decision, target segmentation strategy, and target transmission power.

[0283] In one embodiment, when the computer program is executed by the processor, it further performs the following steps:

[0284] The initial segmentation strategy, initial transmission power, and the position and computation delay of each UAV are substituted into the delay optimization function, and intermediate decisions are calculated based on preset constraints.

[0285] Substitute the intermediate decision, initial transmission power, and the position and computation delay of each UAV into the delay optimization function, and calculate the intermediate segmentation strategy according to the preset constraints.

[0286] The intermediate decision-making and intermediate segmentation strategies, as well as the positions and computational delays of each UAV, are substituted into the delay optimization function, and the intermediate transmission power is calculated according to the preset constraints.

[0287] If the intermediate decision, intermediate segmentation strategy, and intermediate transmission power meet the convergence conditions, then the intermediate decision is determined as the target decision, the intermediate segmentation strategy is taken as the target segmentation strategy, and the intermediate transmission power is taken as the target transmission power.

[0288] In one embodiment, when the computer program is executed by the processor, it further performs the following steps:

[0289] Intermediate decisions are calculated based on time delay constraints and energy consumption constraints.

[0290] In one embodiment, when the computer program is executed by the processor, it further performs the following steps:

[0291] The intermediate segmentation strategy is calculated based on the time delay constraint, energy consumption constraint, and segmentation constraint.

[0292] In one embodiment, when the computer program is executed by the processor, it further performs the following steps:

[0293] The intermediate transmit power is calculated based on energy consumption constraints and power constraints.

[0294] In one embodiment, when the computer program is executed by the processor, it further performs the following steps:

[0295] If the intermediate decision, intermediate segmentation strategy, and intermediate transmission power do not meet the convergence conditions, then the intermediate decision, intermediate segmentation strategy, and intermediate transmission power are recalculated until they meet the convergence conditions.

[0296] In one embodiment, when the computer program is executed by the processor, it further performs the following steps:

[0297] Establish an energy consumption model;

[0298] Determine energy consumption constraints based on the energy consumption model.

[0299] In one embodiment, when the computer program is executed by the processor, it further performs the following steps:

[0300] The delay optimization function and delay constraints are determined based on the total delay model.

[0301] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the following steps:

[0302] After a new task is detected, the position and computational latency of each drone in the same group are obtained;

[0303] Based on the location and computational latency of each UAV, as well as the pre-established latency optimization function, the target decision, target segmentation strategy, and target transmission power are determined. The target decision is used to characterize the target UAV that is processing the current task, and the target segmentation strategy includes the segmentation point of the current task.

[0304] The current task is processed based on target decision, target segmentation strategy, and target transmission power.

[0305] In one embodiment, when the computer program is executed by the processor, it further performs the following steps:

[0306] Obtain initial decision, initial segmentation strategy, and initial transmit power;

[0307] Based on the initial decision, initial segmentation strategy, initial transmission power, and the position and computational delay of each UAV, the delay optimization function is iteratively calculated to obtain the target decision, target segmentation strategy, and target transmission power.

[0308] In one embodiment, when the computer program is executed by the processor, it further performs the following steps:

[0309] The initial segmentation strategy, initial transmission power, and the position and computation delay of each UAV are substituted into the delay optimization function, and intermediate decisions are calculated based on preset constraints.

[0310] Substitute the intermediate decision, initial transmission power, and the position and computation delay of each UAV into the delay optimization function, and calculate the intermediate segmentation strategy according to the preset constraints.

[0311] The intermediate decision-making and intermediate segmentation strategies, as well as the positions and computational delays of each UAV, are substituted into the delay optimization function, and the intermediate transmission power is calculated according to the preset constraints.

[0312] If the intermediate decision, intermediate segmentation strategy, and intermediate transmission power meet the convergence conditions, then the intermediate decision is determined as the target decision, the intermediate segmentation strategy is taken as the target segmentation strategy, and the intermediate transmission power is taken as the target transmission power.

[0313] In one embodiment, when the computer program is executed by the processor, it further performs the following steps:

[0314] Intermediate decisions are calculated based on time delay constraints and energy consumption constraints.

[0315] In one embodiment, when the computer program is executed by the processor, it further performs the following steps:

[0316] The intermediate segmentation strategy is calculated based on the time delay constraint, energy consumption constraint, and segmentation constraint.

[0317] In one embodiment, when the computer program is executed by the processor, it further performs the following steps:

[0318] The intermediate transmit power is calculated based on energy consumption constraints and power constraints.

[0319] In one embodiment, when the computer program is executed by the processor, it further performs the following steps:

[0320] If the intermediate decision, intermediate segmentation strategy, and intermediate transmission power do not meet the convergence conditions, then the intermediate decision, intermediate segmentation strategy, and intermediate transmission power are recalculated until they meet the convergence conditions.

[0321] In one embodiment, when the computer program is executed by the processor, it further performs the following steps:

[0322] Establish an energy consumption model;

[0323] Determine energy consumption constraints based on the energy consumption model.

[0324] In one embodiment, when the computer program is executed by the processor, it further performs the following steps:

[0325] The delay optimization function and delay constraints are determined based on the total delay model.

[0326] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0327] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0328] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A task unloading method, characterized in that, The method includes: After a new task is detected, the position and computational latency of each drone in the same group are obtained; Based on the location and computational delay of each UAV, and a pre-established delay optimization function, the target decision, target segmentation strategy, and target transmission power are determined; the target decision is used to characterize the target UAV processing the current task, and the target segmentation strategy includes the segmentation point of the current task; The current task is processed based on the target decision, the target segmentation strategy, and the target transmission power. The process of determining the target decision, target segmentation strategy, and target transmission power based on the location and computational delay of each UAV, and a pre-established delay optimization function, includes: Obtain initial decision, initial segmentation strategy, and initial transmit power; Based on the initial decision, the initial segmentation strategy, the initial transmission power, and the position and computational delay of each UAV, the delay optimization function is iteratively calculated alternately to obtain the target decision, the target segmentation strategy, and the target transmission power, respectively. The step of iteratively calculating the delay optimization function based on the initial decision, the initial segmentation strategy, the initial transmission power, and the position and computational delay of each UAV includes: The initial segmentation strategy, the initial transmission power, and the position and computational delay of each UAV are substituted into the delay optimization function, and intermediate decisions are calculated according to preset constraints. The intermediate decision, the initial transmission power, and the positions and computational delays of each UAV are substituted into the delay optimization function, and the intermediate segmentation strategy is calculated according to the preset constraints. The intermediate decision, the intermediate segmentation strategy, and the positions and computational delays of each UAV are substituted into the delay optimization function, and the intermediate transmission power is calculated according to the preset constraints. If the intermediate decision, the intermediate segmentation strategy, and the intermediate transmit power meet the convergence condition, then the intermediate decision is determined as the target decision, the intermediate segmentation strategy is taken as the target segmentation strategy, and the intermediate transmit power is taken as the target transmit power.

2. The method according to claim 1, characterized in that, The preset constraints include time delay constraints and energy consumption constraints. The intermediate decision calculated based on the preset constraints includes: The intermediate decision is calculated based on the time delay constraint and the energy consumption constraint.

3. The method according to claim 1, characterized in that, The preset constraints include time delay constraints, energy consumption constraints, and segmentation constraints. The calculation of the intermediate segmentation strategy based on the preset constraints includes: The intermediate segmentation strategy is calculated based on the time delay constraint, the energy consumption constraint, and the segmentation constraint.

4. The method according to claim 1, characterized in that, The preset constraints include energy consumption constraints and power constraints. The calculation of the intermediate transmission power based on the preset constraints includes: The intermediate transmission power is calculated based on the energy consumption constraint and the power constraint.

5. The method according to claim 1, characterized in that, The step of iteratively calculating the delay optimization function based on the initial decision, the initial segmentation strategy, the initial transmission power, and the position and computational delay of each UAV further includes: If the intermediate decision, the intermediate segmentation strategy, and the intermediate transmit power do not meet the convergence condition, then the intermediate decision, the intermediate segmentation strategy, and the intermediate transmit power are recalculated until they meet the convergence condition.

6. The method according to claim 1, characterized in that, The method further includes: Establish an energy consumption model; The energy consumption constraints are determined based on the energy consumption model.

7. The method according to any one of claims 1-6, characterized in that, The method further includes: Establish a computational delay model and a transmission delay model, and determine the total delay model based on the computational delay model and the transmission delay model; The delay optimization function and delay constraints are determined based on the total delay model.

8. A task unloading device, characterized in that, The device includes: The acquisition module is used to acquire the position and computational latency of each drone in the same group after a new task is detected; The determination module is used to determine the target decision, target segmentation strategy, and target transmission power based on the position and computational delay of each UAV and a pre-established delay optimization function; the target decision is used to characterize the target UAV processing the current task, and the target segmentation strategy includes the segmentation point of the current task; The processing module is used to process the current task based on the target decision, the target segmentation strategy, and the target transmission power; The determining module includes: The acquisition submodule is used to acquire the initial decision, initial segmentation strategy, and initial transmit power; The calculation submodule performs alternating iterative calculations on the delay optimization function based on the initial decision, initial segmentation strategy, initial transmission power, and the position and calculation delay of each UAV, to obtain the target decision, target segmentation strategy, and target transmission power respectively. The computing submodule includes: The decision calculation unit is used to substitute the initial segmentation strategy, initial transmission power, and the position and calculation delay of each UAV into the delay optimization function, and calculate the intermediate decision according to the preset constraints. The segmentation strategy calculation unit is used to substitute intermediate decisions, initial transmission power, and the positions and calculation delays of each UAV into the delay optimization function, and calculate the intermediate segmentation strategy according to preset constraints. The power calculation unit is used to substitute the intermediate decision and intermediate segmentation strategy, as well as the position and calculation delay of each UAV, into the delay optimization function, and calculate the intermediate transmission power according to the preset constraints. The determining unit is used to determine the intermediate decision as the target decision, the intermediate segmentation strategy as the target segmentation strategy, and the intermediate transmission power as the target transmission power if the intermediate decision, intermediate segmentation strategy, and intermediate transmission power meet the convergence conditions.

9. A drone, comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

11. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images