A hierarchical scheduling method for multi-terminal and multi-UAV edge computing resource allocation

Abstract

This invention discloses a hierarchical scheduling method for multi-terminal, multi-UAV assisted edge computing resource allocation. Aiming to minimize the number of UAVs and optimize UAV trajectories and task offloading strategies, a hierarchical UAV scheduling model is established. It comprehensively considers task transmission latency, task offloading decisions, and the minimum energy consumption required to complete tasks. Parameters such as trajectory allocation, flight speed, data transmission rate, and flight path of each UAV node are jointly optimized to reduce the energy consumption of the UAV-assisted edge computing system in completing tasks, thereby obtaining the optimal system resource allocation strategy. The position of each UAV is incorporated into a unit, and the variable-length optimization problem is transformed into a two-dimensional optimization problem. The weighting factor for UAV hovering energy consumption is adjusted to reduce the complexity of the optimization problem. This invention further proposes a hierarchical optimization scheduling method for UAVs to address the problem of inconsistent variable types, effectively reducing algorithm complexity.

Description

Technical Field

[0001] This invention relates to the field of computer wireless communication technology, and in particular to a multi-terminal, multi-UAV hierarchical scheduling-assisted edge computing resource allocation method. Background Technology

[0002] Currently, the number of IoT devices is exploding, with wearable devices, mobile phones, tablets, home appliances, sensors, and other terminal devices increasing dramatically. It is estimated that by 2023, the number of connected devices worldwide will reach 29.3 billion, and by 2025, the global data volume will reach 163 ZB. Computationally intensive applications are causing a surge in cloud computing load and network traffic, resulting in high computing costs, severe network congestion, and an inability to reliably guarantee users' computing needs. At the same time, new technologies are constantly emerging, with applications such as autonomous driving, virtual / augmented reality, and multimedia video streaming becoming increasingly widespread. These applications all have high requirements for computing power, latency, energy consumption, and security, but cloud computing cannot provide low-energy, low-latency, efficient, and secure computing services, compromising user experience quality. To address the challenges of limited cloud computing resources and demanding computing task latency, Mobile Edge Computing (MEC) technology has emerged as an effective solution and has rapidly developed in both academia and industry. Mobile Edge Computing is a novel distributed computing approach based on mobile networks, utilizing cloud servers running at the edge of mobile networks to perform specific tasks. MEC (Multi-access Edge Computing) can bring cloud computing and cloud storage closer to the network edge, creating a high-performance, low-latency, and high-bandwidth service environment. This accelerates the distribution and download of content, services, and applications within the network, providing users with a higher quality network experience. Addressing the issues of high deployment costs, poor mobility, and insufficient network coverage associated with traditional MEC servers, drone-enabled MEC systems offer unique advantages by combining the computing power of drones with the mobility of MEC. Drones combined with cellular networks can support mobile communication in a low-cost and highly mobile manner. Furthermore, when used as base stations, drone base stations are more adaptable to environmental changes compared to ground-based base stations. However, as an emerging technology, drone-assisted MEC also faces challenges that need to be addressed. One significant challenge is the extremely limited endurance of drones. The high energy consumption of drone flight leads to a short network lifespan, limiting computing power. Simultaneously, traditional 2D trajectory optimization does not fully leverage the freedom of multi-dimensional trajectory design for drones, resulting in underdeveloped communication performance. Therefore, researching resource allocation methods for mobile edge computing systems based on hierarchical scheduling of multiple drones in the presence of large-scale ground terminals is of significant importance. Summary of the Invention

[0003] The purpose of this section is to outline some aspects of the embodiments of the present invention and to briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section, as well as in the abstract and title of the present application, to avoid obscuring the purpose of this section, the abstract and title of the invention. Such simplifications or omissions shall not be used to limit the scope of the present invention.

[0004] In view of the problems existing in the above and / or prior art, the present invention is proposed.

[0005] Therefore, the technical problem to be solved by this invention is the extremely limited endurance of drones. The high energy consumption of drone flight leads to a short network lifespan, resulting in limited computing power. At the same time, traditional 2D trajectory optimization does not fully utilize the freedom of drone multi-dimensional trajectory design, and communication performance needs to be improved.

[0006] To address the aforementioned technical problems, this invention provides the following technical solution: a multi-terminal, multi-UAV hierarchical scheduling-assisted edge computing resource allocation method, comprising the following steps: Establish an edge offloading scenario with a large number of ground terminals and multiple drones; Establish a model of the energy consumption required for local execution of tasks; Establish a model of the energy consumption required for offloading tasks to the drone for execution; Establish a model of the energy consumption required for drone hovering; The optimal solution is obtained with the overall goal of minimizing system energy consumption and considering the constraints of UAV trajectory deployment and task offloading strategies. The system is hierarchically scheduled, and the upper-level system uses an improved differential evolution algorithm, i.e., a hierarchical optimization algorithm, to obtain the optimal solution for the deployment of UAV trajectories. After obtaining the optimal solution for the upper-level system, the optimal solution for the task unloading strategy in the lower-level system is obtained.

[0007] As a preferred embodiment of the multi-terminal and multi-UAV hierarchical scheduling assisted edge computing resource allocation method of the present invention, wherein: a mobile edge computing system with hierarchical scheduling of UAVs is considered, comprising M ground terminals and N UAVs; No. The location of each user is represented as The drone flies at a fixed altitude H, and its coordinates are represented as follows: ;definition For the first A task to be executed by a ground terminal; definition As the execution mode for each task, , This indicates that the task will be executed locally. This indicates that the task has been unloaded onto the drone. Execute above, at this time , Indicates task exist Execution mode; define the energy consumption required for the task to be executed locally on the ground terminal device as:

[0008] in, Indicates the effective switching capacitor. It is a constant greater than 0. This indicates the number of CPU revolutions required to execute the task; The energy required for the task to be offloaded to the drone for execution is defined as:

[0009] in, This indicates the transmission power of each mobile device. Indicates user Uploaded data scale Indicates the data upload rate. Indicates the effective switching capacitor; Task The data upload rate is:

[0010] in For channel bandwidth, The transmission power for each ground terminal device, This represents the channel power gain at the reference distance. It is a constant. It represents the noise power spectral density; θ represents the fixed beamwidth directional antenna of the UAV. The energy consumption required for drone hovering is defined as:

[0011] in, Indicates hovering power. Indicates the hovering time.

[0012] As a preferred embodiment of the multi-terminal, multi-UAV hierarchical scheduling assisted edge computing resource allocation method of the present invention, a system model is established, which specifically includes an objective function and constraints; wherein the objective function is to minimize system energy consumption, and the main constraints are UAV deployment strategy and task planning.

[0013] The system energy consumption is minimized in the system model as follows:

[0014] The constraints are as follows:

[0015]

[0016]

[0017]

[0018]

[0019]

[0020]

[0021]

[0022] Wherein, the matrix is ​​defined. For the decision of unloading, This indicates that the task is executed locally. This indicates that the task was offloaded to be executed on an edge server deployed on the drone. Indicates ground terminal With drones Maximum distance for drones The coverage radius, Indicates drone With ground terminal Spacing; This represents the minimum distance constraint to prevent collisions between two drones; This represents the constraint that allows each drone to perform a maximum number of missions. Indicates the task execution completion constraint; and constraint In mode ,matrix express Tasks are assigned in the mode The allocation of computing resources; and Transmission latency constraints for each task.

[0023] As a preferred embodiment of the multi-terminal, multi-UAV hierarchical scheduling-assisted edge computing resource allocation method of the present invention, the system is hierarchically scheduled, and the upper-layer system uses an improved differential evolution algorithm, i.e., a hierarchical optimization algorithm, to obtain the optimal solution for UAV trajectory deployment. The specific steps are as follows: First, an initialization operator is performed to randomly generate and store a location for the first drone. Then, generate a location for the second drone, if the distance between them satisfies the constraints. If the two drones do not collide, then the location of the second drone will be stored. ,at this time Otherwise, the location of the second drone is invalid. Count the number of failures, when When that happens, restart the operator initialization process. If the number does not exceed 200, then a second drone location will be generated. Generate the third, fourth... drone positions until all drone positions are successfully generated, resulting in the initial deployment. ; Perform upper-level optimization to obtain the optimal deployment of drones, i.e., the optimal number and location of drones; set up The initial value is , Then The number is decreased by 1 until at least one task can no longer be completed within the transmission delay constraint. for The first in Unit The mutation and crossover operators are expressed as follows:

[0024]

[0025] in and From Three independent units are randomly selected from the data. , These are the mutation vector and the experimental vector, respectively. It is a measurement vector. It is an integer randomly selected between 1 and 2 to guarantee At least with Differing by at least one dimension Represents a uniformly distributed random number between 0 and 1. This represents the cross-control parameters.

[0026] As a preferred embodiment of the multi-terminal, multi-UAV hierarchical scheduling-assisted edge computing resource allocation method of the present invention, wherein: the hierarchical optimization algorithm obtains the optimal solution for UAV trajectory deployment by including: By encoding the location of each drone into a unit, and then integrating the units into a population, The deployment of drones is represented by a unit of length 2, so the deployment of drones can be optimized in two-dimensional space during the algorithm's evolution. Hierarchical optimization algorithms first give the following: population of units Substitute the initial values ​​into the solution to solve the unloading decision. and resource allocation If all tasks can be completed within the transmission delay limit, the elimination operator continuously reduces the number of units until the tasks can no longer be completed within the transmission delay limit. At this point, a subpopulation is generated using the differential evolution algorithm. Used as an intermediate quantity Used to update the population If the updated task cannot complete within the transmission latency limit, check the variables. , express{ The number of consecutive non-executable items in} The set threshold of 1000 has been reached. It cannot be reduced any further, it has been obtained. The optimal value is found by returning to the most recent state that can be completed within the transmission delay limit to optimize {P,a,f}. If the updated task can be completed within the transmission latency limit, then continue executing the elimination operator, and repeat the above process until the maximum fitness score is reached. End the process.

[0027] As a preferred embodiment of the multi-terminal, multi-UAV hierarchical scheduling-assisted edge computing resource allocation method of the present invention, the optimal solution of the layer system is obtained and then used to obtain the optimal solution of the task offloading strategy of the lower layer system. The specific processing flow is as follows: The lower-level optimization objective is to optimize the task offloading strategy after the drone deployment is known, and the drone deployment satisfies the constraints. , will be established , Substituting the values, we obtain the following model for the lower-level optimization problem:

[0028]

[0029] Under constraints and Down, It cannot be less than the minimum value. and Substitution constraints and get and This leads to the optimal resource allocation. for:

[0030] As a preferred embodiment of the multi-terminal, multi-UAV hierarchical scheduling-assisted edge computing resource allocation method of the present invention, wherein: optimal resource allocation is obtained. Afterwards, constraints Once satisfied, the lower-level optimization problem model can be rewritten as follows:

[0031]

[0032] in This represents the minimum energy consumption under local computation conditions for the task. This represents the minimum energy consumption required to offload the task to the drone's computation; at this point, the lower-level optimization problem only needs to optimize... ,and ; All tasks are divided into three categories: local computing mode, fully unloaded computing mode, and partially unloaded computing mode. The uninstallation decision is obtained by determining the number of tasks included in each of the three task categories. The matrix is ​​as follows:

[0033] As a preferred embodiment of the multi-terminal, multi-UAV hierarchical scheduling-assisted edge computing resource allocation method of the present invention, the following steps are taken: Three types of tasks are prioritized, with the first type having the highest priority. Then, the second and third types of tasks are optimized for unloading decisions and deployment. For the second type of tasks, the task with the fewest unloaded tasks is executed first, thus increasing the likelihood of completing all tasks. Subsequently, a participation mode is selected from these to calculate the minimum energy consumption. For the third type of tasks, both participation mode and task execution energy consumption are considered, with the task having the fewest unloaded tasks and the lowest energy consumption being executed first. This ensures that all tasks can be executed with a low probability of system energy depletion.

[0034] The beneficial effects of this invention are as follows: By introducing hierarchical scheduling optimization technology, the upper layer optimizes UAV trajectory deployment, while the lower layer optimizes task offloading strategies. This allows for finding the optimal solution that minimizes system energy consumption. The hierarchical scheduling model is used to jointly optimize UAV trajectory deployment and task offloading strategies under conditions with a large number of ground terminals, thereby obtaining the lowest overall energy consumption required by the system. This invention solves the high complexity of evolutionary algorithms caused by variable decision variable lengths by incorporating the position of each UAV into a unit, ensuring that the length of each unit is fixed during the evolutionary algorithm execution. Simultaneously, the complexity of the optimization problem is reduced by adjusting the weighting factor for UAV hovering energy consumption. Furthermore, the hierarchical optimization algorithm involves continuous decision variables in the upper layer optimization problem. and The lower-level optimization problem contains binary decision variables. This cleverly solves the problem of inconsistent decision variable types in optimization problems. Attached Figure Description

[0035] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 This is a system flowchart of the present invention; Figure 2 This is a system model diagram of the present invention; Figure 3 This is a diagram illustrating the encoding of the present invention. Detailed Implementation

[0036] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0037] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0038] Secondly, the present invention will be described in detail with reference to the schematic diagrams. When detailing the embodiments of the present invention, for ease of explanation, the cross-sectional views illustrating the device structure will be partially enlarged, not according to the usual scale. Furthermore, the schematic diagrams are merely examples and should not limit the scope of protection of the present invention. In addition, actual fabrication should include three-dimensional spatial dimensions of length, width, and depth.

[0039] Furthermore, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places throughout this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that mutually excludes other embodiments.

[0040] Example Reference Figures 1-3 This embodiment provides a multi-terminal, multi-UAV hierarchical scheduling-assisted edge computing resource allocation method, including the following steps: S1: Establish an edge offloading scenario with a large number of ground terminals and multiple drones; S2: Establish a model of the energy consumption required for local execution of the task; S3: Establish a model of the energy consumption required for offloading tasks to the UAV for execution; S4: Establish a model of the energy consumption required for drone hovering; S5: Based on the energy consumption models required for local execution of tasks, the energy consumption models required for task offloading to UAV execution, and the energy consumption models required for UAV hovering in S2, S3, and S4, the optimal solution is obtained with the goal of minimizing system energy consumption and considering the constraints of UAV trajectory deployment and task offloading strategies. S6: The system is hierarchically scheduled, and the upper-level system uses the improved differential evolution algorithm, i.e., the hierarchical optimization algorithm, to obtain the optimal solution for the deployment of UAV trajectories; S7: After obtaining the optimal solution of the upper-level system, it is used to find the optimal solution for the task unloading strategy of the lower-level system.

[0041] like Figure 1 Regarding step S1, consider a mobile edge computing system with hierarchical scheduling of UAVs, comprising M ground terminals and N UAVs. Each ground terminal possesses certain computing power to meet the needs of performing simple tasks locally. The UAVs equipped with edge servers have strong computing capabilities. Ground terminals can offload data to the UAVs. The sets of ground terminals and UAVs are defined as follows: and .

[0042] No. The location of each user is represented as The drone flies at a fixed altitude H, and its coordinates are represented as follows: ;definition For the first A task to be executed by a ground terminal; definition As the execution mode for each task, , This indicates that the task will be executed locally. This indicates that the task has been unloaded onto the drone. Execute above, at this time , Indicates task exist Execution mode; due to limited local computing power and latency requirements, users can process their tasks locally or offload parts of the tasks to a legitimate drone for processing.

[0043] Furthermore, for steps S2-4, we model them as formula optimization problems using a mathematical model. Based on the above description, the following mathematical model can be established: Task When executed locally on the ground terminal device, the task takes the following time:

[0044] In S2, the energy consumption required for a task to be executed locally on the ground terminal device is defined as:

[0045] in, Indicates the effective switching capacitor. It is a constant greater than 0. This indicates the number of CPU revolutions required to execute the task; In S3, when a task needs to be offloaded to a drone, the task is first transmitted to the drone, then executed by the mobile edge server on the drone, and finally the result is returned to the ground terminal. With drones Distance between:

[0046] At the same time, if In drones Executed on the ground terminal It must be in the drone Within its coverage area, therefore there are constraints. , This indicates the coverage radius of each drone. , This represents the fixed beamwidth of the directional antenna for each drone, and the distance between two drones is expressed as:

[0047] Because the two drones need to maintain a minimum distance to avoid collision, there are constraints. Due to the limited computing power of each mobile edge server, each drone can perform a maximum of [number] tasks. Therefore, there are constraints. .

[0048] Task The data upload rate is:

[0049] in For channel bandwidth, The transmission power for each ground terminal device, This represents the channel power gain at the reference distance. It is a constant. It represents the noise power spectral density; θ represents the fixed beamwidth directional antenna of the UAV; The completion time is obtained by adding the transmission time and the calculation time. Total time :

[0050] The energy required to offload a task onto a drone for execution is defined as: the energy required for transmission and the energy required for computation are added together to obtain the energy required to complete the task. Total energy :

[0051] in, As a constant, This indicates the transmission power of each mobile device. Indicates user Uploaded data scale Indicates the data upload rate. Indicates the effective switching capacitor; Furthermore, the energy consumption required for drone hovering is defined as the energy consumed by the drone at a certain altitude. The energy consumed during hovering is :

[0052] in, Indicates hovering power. Indicates the hovering time.

[0053] Based on the above analysis, in S5, a system model is established, which specifically includes the objective function and constraints. That is, the optimization problem in this invention can be modeled as follows: The system model is as follows:

[0054]

[0055]

[0056]

[0057]

[0058]

[0059]

[0060]

[0061]

[0062] Wherein, the matrix is ​​defined. For the decision of unloading, This indicates that the task is executed locally. This indicates that the task was offloaded to be executed on an edge server deployed on the drone. Indicates ground terminal With drones Maximum distance for drones The coverage radius, Indicates drone With ground terminal Spacing; This represents the minimum distance constraint to prevent collisions between two drones; This represents the constraint that allows each drone to perform a maximum number of missions. Indicates the task execution completion constraint; and constraint In mode ,matrix express Tasks are assigned in the mode The allocation of computing resources; and Transmission latency constraints for each task.

[0063] It should be noted that the optimization problem of the formula for the energy consumption required for local execution of ground terminal equipment is a non-convex nonlinear optimization problem. General optimization methods cannot solve this optimization problem. However, since the traditional differential evolution algorithm is a population-based heuristic search method that does not require gradient information, the traditional differential evolution algorithm provides the possibility of solving this optimization problem.

[0064] Because the location of drone j needs to be determined. Unloading decision Resource allocation To perform joint optimization, the number of decision variables that need to be optimized is... The number of decision variables increases with the increase of M or N. Given the large-scale ground terminal scenario considered in this invention, the number of decision variables is very complex. Furthermore, due to... It is a binary decision variable, where N is an integer decision variable. Since the variables are continuous and of inconsistent types, the deployment and task offloading strategies of UAVs influence each other. Therefore, the hierarchical optimization algorithm of this invention is proposed.

[0065] Furthermore, in S6, the system is hierarchically scheduled. The upper-level system uses an improved differential evolution algorithm, i.e., a hierarchical optimization algorithm, to obtain the optimal solution for UAV trajectory deployment. The specific steps are as follows: First, an initialization operator is performed to randomly generate a position for the first UAV and store it. Then, generate a location for the second drone, if the distance between them satisfies the constraints. If the two drones do not collide, then the location of the second drone will be stored. ,at this time Otherwise, the location of the second drone is invalid. Count the number of failures, when When that happens, restart the operator initialization process. If the number does not exceed 200, then a second drone location will be generated. Generate the third, fourth... drone positions until all drone positions are successfully generated, resulting in the initial deployment. ; Perform upper-level optimization to obtain the optimal deployment of drones, i.e., the optimal number and location of drones; set up The initial value is , Then The number is decreased by 1 until at least one task can no longer be completed within the transmission delay constraint. for The first in Unit The mutation and crossover operators are expressed as follows:

[0066]

[0067] in and From Three independent units are randomly selected from the data. , These are the mutation vector and the experimental vector, respectively. It is a measurement vector. It is an integer randomly selected between 1 and 2 to guarantee At least with Differing by at least one dimension Represents a uniformly distributed random number between 0 and 1. This represents the cross-control parameters.

[0068] In the process of evolution, the differential evolution algorithm is used in... This will produce a subpopulation. , The cells in the middle are used to replace the cells randomly selected by P, so that the original cells are replaced. It was updated to become ,if Satisfying constraints Calculate unloading decision and resource allocation If under transmission delay constraints Can compare Perform more tasks or both perform the same number of tasks but If the energy consumption is lower, then use replace Define variables at this time This indicates an optimized state; if it is updated continuously for 1000 times... It is still not feasible; we believe the number of drones... It cannot be reduced any further at this point. The optimal solution is ,set up ,Will Revert to the previous executable state and execute the update operator to optimize. and the location of the drone; if It is still feasible, setting The interrupt update operator is used to... Continue executing the elimination operator to further reduce the number of drones. .

[0069] The algorithm introduced here is a hierarchical scheduling algorithm. Specifically, it involves encoding the location of each UAV into a unit, and then integrating the units into a population. The deployment of drones is represented by a unit of length 2, so the deployment of drones can be optimized in two-dimensional space during the algorithm's evolution. Hierarchical optimization algorithms first give the following: population of units Substitute the initial values ​​into the solution to solve the unloading decision. and resource allocation If all tasks can be completed within the transmission delay limit, the elimination operator continuously reduces the number of units until the tasks can no longer be completed within the transmission delay limit. At this point, a subpopulation is generated using the differential evolution algorithm. Used as an intermediate quantity Used to update the population If the updated task cannot complete within the transmission latency limit, check the variables. , express{ The number of consecutive non-executable items in} The set threshold of 1000 has been reached. It cannot be reduced any further, it has been obtained. The optimal value is found by returning to the most recent state that can be completed within the transmission delay limit to optimize {P,a,f}. If the updated task can be completed within the transmission latency limit, then continue executing the elimination operator, and repeat the above process until the maximum fitness score is reached. End the process.

[0070] After obtaining the optimal solution for the upper-level system, the optimal solution for the task unloading strategy in the lower-level system is obtained. The specific processing flow is as follows: The lower-level optimization objective is to optimize the task offloading strategy after a drone deployment. The fact that drone deployment is known means... It is known that the drone deployment satisfies constraint C2, because if the drone deployment does not satisfy constraint C2, it will not be included in R and will be discarded during the upper-level optimization. The setup established in steps (1.2) and (1.3) will be used to... , Substituting the values, we obtain the following model for the lower-level optimization problem:

[0071]

[0072] As can be seen from the lower-level optimization problem model, when the pattern is determined, the more computational resources consumed, the greater the energy consumption, because energy consumption increases with... or The increase is due to the increase in energy consumption. Therefore, in order to reduce energy consumption, it is necessary to minimize it as much as possible. That is, f, while under constraints and Down, It cannot be less than the minimum value. and Substitution constraints and get and This leads to the optimal resource allocation. for:

[0073] To achieve optimal resource allocation Afterwards, constraints Once satisfied, the lower-level optimization problem model can be rewritten as follows:

[0074]

[0075] in This represents the minimum energy consumption under local computation conditions for the task. This represents the minimum energy consumption required to offload the task to the drone's computation; at this point, the lower-level optimization problem only needs to optimize... ,and Since there are only two possible values, a greedy algorithm is used to optimize the lower-level optimization problem.

[0076] All tasks are divided into three categories: local computing mode, fully unloaded computing mode, and partially unloaded computing mode. The uninstallation decision is obtained by determining the number of tasks included in each of the three task categories. The matrix is ​​as follows:

[0077] We prioritize the three types of tasks, with the first type having the highest priority. Then, we optimize the unloading decisions for the second and third types of tasks. For the second type of tasks, we select the task with the fewest unloading tasks to execute first, thus increasing the likelihood of completing all tasks. We then select a participation mode from these tasks to calculate the minimum energy consumption. For the third type of tasks, we consider both the participation mode and the task execution energy consumption. The task with the fewest unloading tasks and the lowest energy consumption is executed first, ensuring that all tasks can be executed with a low probability of system energy depletion.

[0078] It is important to note that the constructions and arrangements of this application shown in several different exemplary embodiments are merely illustrative. Although only a few embodiments are described in detail in this disclosure, those who consult this disclosure will readily understand that many modifications are possible (e.g., changes in the size, dimensions, structure, shape, and proportions of various elements, as well as parameter values ​​(e.g., temperature, pressure, etc.), mounting arrangements, use of materials, color, orientation, etc.) without substantially departing from the novel teachings and advantages of the subject matter described in this application). For example, an element shown as integrally formed may be composed of multiple parts or elements, the position of elements may be inverted or otherwise altered, and the nature or number or position of discrete elements may be changed or altered. Therefore, all such modifications are intended to be included within the scope of the invention. The order or sequence of any process or method steps may be changed or rearranged according to alternative embodiments. In the claims, any "device plus function" clause is intended to cover the structure described herein that performs the function, and not only structurally equivalent but also equivalent in structure. Other substitutions, modifications, alterations, and omissions may be made in the design, operation, and arrangement of the exemplary embodiments without departing from the scope of the invention. Therefore, the present invention is not limited to the specific embodiments, but extends to various modifications that still fall within the scope of the appended claims.

[0079] Furthermore, in order to provide a concise description of exemplary embodiments, not all features of actual embodiments (i.e., those features that are not relevant to the currently considered best mode for carrying out the invention, or those features that are not relevant to implementing the invention) may be omitted.

[0080] It should be understood that numerous specific implementation decisions can be made during the development of any practical implementation, such as in any engineering or design project. Such development efforts may be complex and time-consuming, but for those skilled in the art who benefit from this disclosure, the development effort will be a routine work of design, manufacturing, and production without requiring much experimentation.

[0081] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A multi-terminal, multi-UAV hierarchical scheduling-assisted edge computing resource allocation method, characterized in that: Includes the following steps: Establish an edge offloading scenario with a large number of ground terminals and multiple drones; Establish a model of the energy consumption required for local execution of tasks; Establish a model of the energy consumption required for offloading tasks to the drone for execution; Establish a model of the energy consumption required for drone hovering; The optimal solution is obtained with the overall goal of minimizing system energy consumption and considering the constraints of UAV trajectory deployment and task offloading strategies. The system is hierarchically scheduled, and the upper-level system uses an improved differential evolution algorithm, i.e., a hierarchical optimization algorithm, to obtain the optimal solution for the deployment of UAV trajectories. After obtaining the optimal solution of the upper-level system, the optimal solution for the task unloading strategy of the lower-level system is obtained. Consider a mobile edge computing system with hierarchical scheduling of UAVs, consisting of M ground terminals and N UAVs; No. The location of each user is represented as The drone flies at a fixed altitude H, and its coordinates are represented as follows: ;definition For the first A task to be executed by a ground terminal; definition As the execution mode for each task, , This indicates that the task will be executed locally. This indicates that the task has been unloaded onto the drone. Execute above, at this time , Indicates task exist Execution mode; define the energy consumption required for the task to be executed locally on the ground terminal device as: ； in, Indicates the effective switching capacitor. It is a constant greater than 0. This indicates the number of CPU revolutions required to execute the task; The energy required for the task to be offloaded to the drone for execution is defined as: ； in, This indicates the transmission power of each mobile device. Indicates user Uploaded data scale Indicates the data upload rate. Indicates the effective switching capacitor; Task The data upload rate is: ； in For channel bandwidth, The transmission power for each ground terminal device, This represents the channel power gain at the reference distance. It is a constant. It represents the noise power spectral density; θ represents the fixed beamwidth directional antenna of the UAV. The energy consumption required for drone hovering is defined as: ； in, Indicates hovering power. Indicates the hovering time; Establish a system model, which specifically includes an objective function and constraints; The system model is as follows: ； ； ； ； ； ； ； ； ； Wherein, the matrix is ​​defined. For the decision of unloading, This indicates that the task is executed locally. This indicates that the task was offloaded to be executed on an edge server deployed on the drone. Indicates ground terminal With drones Maximum distance for drones The coverage radius, Indicates drone With ground terminal Spacing; This represents the minimum distance constraint to prevent collisions between two drones; This represents the constraint that allows each drone to perform a maximum number of missions. Indicates the task execution completion constraint; and constraint In mode ,matrix express Tasks are assigned in the mode The allocation of computing resources; and Transmission latency constraints for each task; The system is hierarchically scheduled, and the upper-level system uses an improved differential evolution algorithm, i.e., a hierarchical optimization algorithm, to obtain the optimal solution for UAV trajectory deployment. The specific steps are as follows: First, an initialization operator is performed to randomly generate and store a location for the first drone. Then, generate a location for the second drone, if the distance between them satisfies the constraints. If the two drones do not collide, then the location of the second drone will be stored. ,at this time Otherwise, the location of the second drone is invalid. Count the number of failures, when When that happens, restart the operator initialization process. If the number does not exceed 200, then a second drone location will be generated. Generate the third, fourth... drone positions until all drone positions are successfully generated, resulting in the initial deployment. ; Perform upper-level optimization to obtain the optimal deployment of drones, i.e., the optimal number and location of drones; set up The initial value is , Then The number is reduced by 1 until at least one task can no longer be completed within the transmission delay constraint; for The first in Unit The mutation and crossover operators are expressed as follows: ； ； in and From Three independent units are randomly selected from the data. , These are the mutation vector and the experimental vector, respectively. It is a measurement vector. It is an integer randomly selected between 1 and 2 to guarantee At least with Differing by at least one dimension Represents a uniformly distributed random number between 0 and 1. Indicates the cross-control parameters; The hierarchical optimization algorithm obtains the optimal solution for UAV trajectory deployment, including: By encoding the location of each drone into a unit, and then integrating the units into a population, The deployment of drones is represented by a unit of length 2, so the deployment of drones can be optimized in two-dimensional space during the algorithm's evolution. Hierarchical optimization algorithms first give the following: population of units Substitute the initial values ​​into the solution to solve the unloading decision. and resource allocation If all tasks can be completed within the transmission delay limit, the elimination operator continuously reduces the number of units until the tasks cannot be completed within the transmission delay limit. At this point, a subpopulation is generated using the differential evolution algorithm. Used as an intermediate quantity Used to update the population If the updated task cannot complete within the transmission latency limit, check the variables. , express{ The number of consecutive non-executable items in} The set threshold of 1000 has been reached. It cannot be reduced any further, it has been obtained. The optimal value is found by returning to the most recent state that can be completed within the transmission delay limit to optimize {P,a,f}. If the updated task can be completed within the transmission latency limit, then continue executing the elimination operator, and repeat the above process until the maximum fitness score is reached. End the process; After obtaining the optimal solution for the upper-level system, it is used to find the optimal solution for the task unloading strategy in the lower-level system. The specific processing flow is as follows: The lower-level optimization objective is to optimize the task offloading strategy after the drone deployment is known, and the drone deployment satisfies the constraints. , will be established , Substituting the values, we obtain the following model for the lower-level optimization problem: ； ； Under constraints and Down, It cannot be less than the minimum value. and Substitution constraints and get and This leads to the optimal resource allocation. for: ； To achieve optimal resource allocation Afterwards, constraints Once satisfied, the lower-level optimization problem model can be rewritten as follows: ； ； in This represents the minimum energy consumption under local computation conditions for the task. This represents the minimum energy consumption required to offload the task to the drone's computation; at this point, the lower-level optimization problem only needs to optimize... ,and ; All tasks are divided into three categories: local computing mode, fully unloaded computing mode, and partially unloaded computing mode. The uninstallation decision is obtained by determining the number of tasks included in each of the three task categories. The matrix is ​​as follows: ； We prioritize the three types of tasks, with the first type having the highest priority. Then, we optimize the unloading decisions for the second and third types of tasks. For the second type of tasks, we select the task with the fewest unloading tasks to execute first, thus increasing the likelihood of completing all tasks. We then select a participation mode from these tasks to calculate the minimum energy consumption. For the third type of tasks, we consider both the participation mode and the task execution energy consumption. The task with the fewest unloading tasks and the lowest energy consumption is executed first, ensuring that all tasks can be executed with a low probability of system energy depletion.

Images