Method for task offloading and resource allocation based on cooperation of multiple access edge computing of unmanned aerial vehicle

Abstract

The present disclosure provides a method for task offloading and resource allocation based on UAV cooperative multi-access edge computing, which relates to the technical field of mobile communication, and the method comprises: constructing an optimization model aiming at minimizing service delay while ensuring service fairness among user equipment, according to the demand of each user equipment in terms of service experience; simplifying the problem model based on Dinkelbach method and convex optimization theory; proposing a four-stage alternating iteration optimization algorithm; decomposing the optimization target into four sub-optimization targets of UAV trajectory decision, task offloading decision, service cache decision and resource allocation decision; using an alternating iteration algorithm to solve and calculate until the target converges; and obtaining the task offloading and resource allocation decision in the UAV cooperative multi-access edge computing network and executing it. The present disclosure can achieve lower service delay while ensuring better fairness among all user equipment.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of mobile communication technology, in particular to a method for task offloading and resource allocation of multi-access edge computing based on cooperation of unmanned aerial vehicles. BACKGROUND

[0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute the prior art.

[0003] In recent years, with the development and popularization of mobile communication technology, many new applications such as online video, map navigation, mobile payment, and face recognition have emerged. Subsequently, the proliferation of networked smart devices has led to an explosive growth of data. At the same time, in the face of sudden events such as infectious diseases, face-to-face communication between people has become difficult, and the dependence on network medical care, online learning, and remote work has increased significantly. The above-mentioned applications are usually delay-sensitive and require a large amount of communication and computing resources. In order to support a large number of smart devices and process a large amount of data in a timely manner, multi-access edge computing, formerly known as mobile edge computing, has become a key technology in the next generation of wireless networks. By deploying mobile edge computing servers at the edge of the communication network (such as ground cellular infrastructure), user devices can offload data to the edge to improve service experience.

[0004] However, there are still many problems with current edge computing systems. Fixed-position ground mobile edge computing servers cannot be adjusted according to the needs of terminals. Due to non-line-of-sight links, their channel quality can be poor, resulting in limited communication rates. And due to severe obstructions or destruction caused by natural disasters, some user devices may abandon mobile edge computing services. Recently, unmanned aerial vehicles have become a promising technology due to their flexible deployment and low cost, which can improve wireless connectivity and provide extensive coverage in mobile edge computing networks. Generally, there are two techniques for unmanned aerial vehicle-assisted mobile edge computing networks, in which unmanned aerial vehicles act as air relays and air mobile edge computing servers. In addition, with the rapid growth of user devices, a single or even multiple unmanned aerial vehicles may not be able to meet the needs of a large number of computing-intensive and delay-sensitive applications such as virtual reality and intelligent transportation, and cannot solve the problems of service delay and fairness, i.e., cannot maximize the service experience ratio of user devices. SUMMARY

[0005] To solve the above problems, the present disclosure proposes a method for task offloading and resource allocation of multi-access edge computing based on cooperation of unmanned aerial vehicles, which utilizes unmanned aerial vehicles to cooperate in mobile edge computing services for computing and caching resources, minimizes service delay, and ensures service fairness among user devices.

[0006] According to some embodiments, the present disclosure adopts the following technical solutions:

[0007] The unmanned aerial vehicle cooperative multi-access edge computing task offloading and resource allocation method comprises:

[0008] An initial cooperative computing task offloading environment is initialized, a base station and all unmanned aerial vehicles cooperate to provide mobile edge computing services for user equipment, a task set is acquired, and a task period is divided into multiple time slots with equal duration;

[0009] The input data size of the computing task requested by the user in each time slot is acquired, an optimization model is constructed according to the demand of each user equipment in terms of service experience, with the goal of minimizing service delay while ensuring service fairness between user equipment, the problem model is simplified based on the Dinkelbach method and convex optimization theory, a four-stage alternating iteration optimization algorithm is proposed, the optimization goal is decomposed into four sub-optimization goals of unmanned aerial vehicle trajectory decision, task offloading decision, service caching decision and resource allocation decision, and the alternating iteration algorithm is used for solving calculation until the goal converges, and the task offloading and resource allocation decision in the unmanned aerial vehicle cooperative multi-access edge computing network is acquired and executed.

[0010] Further, the task offloading decision optimization based on satisfaction degree comprises: fixed unmanned aerial vehicle trajectory, bandwidth resource allocation decision, service caching decision, computing resource allocation decision and auxiliary variable to optimize the task offloading decision, and an optimization goal formula of the task offloading sub-problem is defined.

[0011] In the optimization process of the task offloading decision, a plurality of unmanned aerial vehicles and task sets are defined, each user equipment sends a task offloading request to the associated unmanned aerial vehicle at the beginning of the time slot, a suitable offloading position is selected for the task based on the user's satisfaction degree, and the value of the objective function corresponding to each user equipment is calculated under the current task offloading decision. If the maximum delay tolerance of all user equipment is met and the computing resource and energy consumption limit of each unmanned aerial vehicle is not exceeded, the task offloading decision at this time is suitable.

[0012] In the set, each task has different satisfaction degrees for different offloading positions, the value of the satisfaction degree is related to the task processing delay and fairness, the greater the task processing delay and the lower the fairness, the smaller the value of the satisfaction degree.

[0013] Further, the optimization of the service caching decision comprises: fixed unmanned aerial vehicle trajectory, bandwidth resource allocation decision, task offloading decision, computing resource allocation decision and auxiliary variable to optimize the service caching decision, and an optimization formula of the service caching decision sub-problem is defined.

[0014] Considering the cache space utilization rate of the unmanned aerial vehicle, the priority of the service cached on the unmanned aerial vehicle for the task with a high service caching decision value is high until the cache space upper limit of the unmanned aerial vehicle is reached.

[0015] Further, the optimization of the UAV trajectory includes a fixed task offloading decision, a bandwidth resource allocation decision, a service caching decision, a computing resource allocation decision, and auxiliary variables to optimize the UAV trajectory, and defines a formula of a UAV trajectory optimization subproblem.

[0016] In the optimization of the UAV trajectory, the trajectory planning is taken as an optimization variable, and the coordinate position of the UAV at each time slot in the entire task period is composed.

[0017] Further, the optimization of the computing resource allocation decision includes a given UAV trajectory, a task offloading decision, a service caching decision, and auxiliary variables, defines an optimization formula of a computing resource allocation subproblem, the computing resource allocation subproblem is a convex problem, and convex optimization is used to obtain an optimal solution of the bandwidth resource allocation and the computing resource allocation.

[0018] Further, the task offloading, the service caching, the trajectory planning, and the resource allocation are jointly optimized to maximize the service experience ratio, a four-stage alternating iterative optimization is proposed to solve the original problem, the task offloading decision, the service caching decision, and the UAV trajectory planning are iteratively optimized respectively until the target value converges, and the parameter of the Dinkelbach method is updated after each round of the four-stage alternating iterative optimization.

[0019] Compared with the prior art, the beneficial effects of the present disclosure are:

[0020] The method for task offloading and resource allocation based on UAV cooperative multi-access edge computing proposed in the present disclosure can effectively utilize computing and caching resources for mobile edge computing service cooperation, aims to minimize service delay while ensuring service fairness among user devices.

[0021] In order to improve the service experience, the present disclosure considers joint optimization of task offloading, resource allocation, trajectory planning, and service caching placement under the constraints of UAV energy budget and delay demand, and expresses it as a service experience ratio maximization problem. Since the original problem is a fractional structured mixed integer non-convex programming problem, it is difficult to solve in polynomial time. Based on the Dinkelbach method and convex optimization theory, the present disclosure simplifies the problem model and proposes a four-stage alternating iterative service ratio maximization algorithm to solve the problem. Numerical results show that compared with other benchmark algorithms, the algorithm proposed in the present disclosure can reduce the service delay by 78.2% while improving the fairness among all user devices by 53.0%. BRIEF DESCRIPTION OF DRAWINGS

[0022] The accompanying drawings, which form a part of the present disclosure, are intended to provide further understanding of the present disclosure, and the illustrative embodiments of the present disclosure and their description serve the purpose of explaining the present disclosure. They are not intended to be an undue limitation on the present disclosure.

[0023] Figure 1 A multi-unmanned aerial vehicle (UAV) assisted mobile edge computing scenario diagram according to an embodiment of the present disclosure;

[0024] Figure 2 An optimization method decomposition diagram according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0025] The present disclosure is further described below in conjunction with the accompanying drawings and embodiments.

[0026] It should be noted that the following detailed description is illustrative only and is intended to provide further description of the present disclosure. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present disclosure belongs.

[0027] It should be noted that the terms used herein are only intended to describe specific embodiments and are not intended to limit exemplary embodiments according to the present disclosure. As used herein, the singular form is intended to include the plural form unless the context clearly indicates otherwise, and it should also be understood that when the terms "comprise" and / or "include" are used in the specification, they indicate the presence of a feature, step, operation, device, component, and / or combination thereof.

[0028] Embodiment 1

[0029] In an embodiment of the present disclosure, a method for task offloading and resource allocation based on UAV cooperative multi-access edge computing is provided, comprising:

[0030] Step 1: Initialize the cooperative computing task offloading environment, and the base station and all unmanned aerial vehicles (UAVs) cooperate to provide mobile edge computing services for user equipment (UE), obtain a task set, and divide the task period into multiple time slots with equal duration;

[0031] Step 2: Obtain the input data size of the computing task requested by the user in each time slot, and construct an optimization model according to the needs of each user equipment in terms of service experience, with the goal of minimizing service delay while ensuring service fairness between user equipment;

[0032] Step 3: Based on the Dinkelbach method and convex optimization theory, simplify the problem model, propose a four-stage alternating iteration optimization algorithm, decompose the optimization target into four sub-optimization targets of UAV trajectory decision, task offloading decision, service cache decision, and resource allocation decision, use an alternating iteration algorithm to solve and calculate until the target converges, obtain the task offloading and resource allocation decision in the UAV cooperative multi-access edge computing network, and execute.

[0033] As an embodiment, the present disclosure is a task offloading and resource allocation method in a service experience based cache unmanned aerial vehicle cooperative multi-access edge computing network, which solves the service delay and fairness problem, that is, maximizes the service experience ratio of user equipment. In order to solve the above technical purpose, the following implementation process is specifically included:

[0034] Step 1: For the unmanned aerial vehicle supported mobile edge computing network, a multi-unmanned aerial vehicle assisted mobile edge computing problem is proposed to maximize the service experience ratio of user equipment.

[0035] Initialize the cooperative computing task offloading environment, and the base station and all unmanned aerial vehicles cooperate to provide mobile edge computing services for user equipment, obtain a task set, and divide the task period into multiple time slots with equal duration;

[0036] Wherein, the base station and all unmanned aerial vehicles cooperate to provide mobile edge computing services for M user equipment. One of the macro base stations, U unmanned aerial vehicles and M user equipment are represented as b, Y={1, 2,..., U}, M={1, 2,..., M}. The set of all services that the macro base station can provide is represented as∑={1, 2,..., S}. Because the connection between user equipment and unmanned aerial vehicles can be stable within a short enough time period. In order to facilitate representation, the task period N is divided into T time slots with equal duration Δ t , and T={1, 2,..., T}. Each user equipment has only one time delay sensitive task in a time slot, which can be offloaded to unmanned aerial vehicles or macro base station for processing, and each task is atomic and indivisible. In time slot t, the user m produces a time delay sensitive task of requested service s It can be represented by a 3-tuple

[0037] Assume represents the input data size of the computing task of the user m requesting service s in time slot t. Let represent the computing intensity of the computing task of the user m requesting service s in time slot t. is the processing delay tolerance of the task of the requested service s, and the result is invalid to the user m beyond the limit, and each user equipment has different requirements in terms of service experience; The proposed goal is to maximize the service experience ratio of user equipment. This can be achieved by jointly optimizing task offloading, service caching, unmanned aerial vehicle trajectory and resource allocation. Specifically as follows:

[0038]

[0039]

[0040]

[0041]

[0042]

[0043]

[0044]

[0045]

[0046]

[0047]

[0048]

[0049]

[0050]

[0051]

[0052]

[0053]

[0054]

[0055] Where C1 represents the requirement to cache service s on drone u in order to offload the task of user m requesting service s to drone u. C2 represents the spectrum resource allocation constraint of user equipment associated with the same drone. C3 represents the computing resource constraint of a single drone. C4 represents the requirement that the total storage space occupied by services stored on each drone must not exceed the total storage capacity of the drone, represented by K. u C5 indicates that the ground user should be within the coverage area of ​​the associated drone. C6 indicates constraints on the drone's positional changes between any two time slots. C7 indicates that any two drones should maintain a minimum safe distance to ensure they avoid collisions within time slot t. C8 indicates that the flight paths of all drones should be within the target area. To avoid high communication latency for drones, C9 indicates that the horizontal distance between the associated drone and the selected relay drone does not exceed R. uav C 10 E represents the upper limit of energy for each drone within time slot t. th C 11 This indicates that the completion delay of each task cannot exceed the task's processing delay tolerance. (C) 12It is pointed out that the tasks generated by each user equipment are precisely offloaded to a nearby drone or macro base station. 13 and C 14 It is pointed out that the service caching decision and the task offloading decision variables are binary, while the constraints C 15 It is pointed out that the bandwidth and computing resource allocation variables are continuous.

[0056] Step 2: Obtain the input data size of the computing task requested by each user in each time slot, and construct an optimization model according to the needs of each user equipment in terms of service experience, with the goal of minimizing service delay while ensuring service fairness between user equipments. Based on the Dinkelbach method and convex optimization theory, the problem model is simplified, and a four-stage alternating iteration optimization algorithm is proposed. The optimization goal is decomposed into four sub-optimization goals of drone trajectory decision, task offloading decision, service caching decision and resource allocation decision. The iterative algorithm is used for solving and calculation until the target converges, and the task offloading and resource allocation decision in the drone collaborative multi-access edge computing network is obtained and executed.

[0057] Specifically, in order to maximize the service experience ratio, the optimization goal is decomposed into four sub-optimization goals of drone trajectory, task offloading decision, service caching decision and resource allocation decision. After each round of four free iterations, the parameters of the Dinkelbach method are updated. In order to decouple the non-convex objective, it is decomposed into different sub-goals, and an iterative algorithm for alternating solution is proposed, including the following processes:

[0058] S1: First, the task offloading decision optimization based on satisfaction, fix the drone trajectory Q, bandwidth resource allocation decision B, service caching decision A, computing resource allocation decision F and auxiliary variable η to optimize the task offloading decision. The task offloading optimization sub-problem is formulated as:

[0059]

[0060] s.t.C1、C3、C5、C9-C 12 、C 14

[0061] In order to better describe the optimization process of task offloading, a number of sets related to drones and tasks are defined. Let the set of drones that cache task m required service s be defined as Each user equipment sends a task offloading request to its associated drone at the beginning of the time slot, and the set of task offloading requests received by the associated drone u is denoted as It contains the associated user equipment and the tasks offloaded by the cooperative drone. If the associated drone u belongs to the set Then the UAV hits the service s required by mission m. Then, the hit mission is added to Those not hit are added to When initializing the task offloading decision, assume that all tasks in the set can be computed by the UAV u. The tasks in the set are further offloaded to the cooperative UAV i or the base station in the set with the largest Π 3-1 value. The tasks in the set are further offloaded to the cooperative UAV i or the base station in the set .

[0062] Select a suitable offloading location for the task based on the user's satisfaction. Under the current task offloading decision, calculate the value of the Π 3-1 objective function corresponding to each user device. If the maximum delay tolerance of all user devices is satisfied and the computing resources and energy consumption limits of each UAV are not exceeded, then the current task offloading decision is appropriate. Then, calculate the satisfaction Π 3-1 of each task, and successively select the task with the smallest satisfaction value in the set for further offloading. Then move it from to until all tasks in the set satisfy the maximum delay tolerance and CSS resource and energy consumption limits.

[0063] In the set , each task has different satisfaction levels for different offloading locations. The value of the satisfaction level is related to the task processing delay and fairness. The greater the task processing delay and the lower the fairness, the smaller the value of the satisfaction level. Then, the task m that requests the service s and is rejected by the associated UAV u has a satisfaction value for the cooperative UAV i, which can be expressed as:

[0064]

[0065]

[0066] The task m that requests the service s and is rejected by the associated UAV u has a satisfaction level for the macro base station b, which can be expressed as:

[0067]

[0068]

[0069] For the task m that requests the service s of the associated UAV, send an offloading request to the location with a high satisfaction level preferentially. If the requested location is the macro base station, the offloading request will be directly accepted. If the requested location is a cooperative drone, the permission of the cooperative drone is needed. If the offloading request is rejected, it will be sent to the next best offloading location in the next iteration until it is accepted, let The above process is repeated until the offloading locations of all tasks are found.

[0070] S2: The optimization of service caching decision is to optimize the service caching decision with the trajectory of fixed drones, bandwidth resource allocation decision, task offloading decision, computing resource allocation decision and auxiliary variable, and to define the optimization formula of service caching decision subproblem.

[0071] Specifically, the trajectory Q of fixed drones, bandwidth resource allocation decision B, task offloading decision X, computing resource allocation decision F and auxiliary variable η are used to optimize the service caching decision A. The service caching decision subproblem is formulated as:

[0072]

[0073] s.t. C1, C4, C 10 , C 11 , C 13

[0074] Since the cache space of the drone is limited, all programs cannot be cached. The optimized service caching decision is used to maximize the reduction of task processing delay and ensure fairness. In order to improve the utilization rate of cache space, it is considered that 3-2 The priority of the service required by the task with a higher value on the drone is higher until the cache space of the drone reaches the upper limit. Let and |M u | respectively represent the set and the number of tasks offloaded to the drone u. Correspondingly, let ∑ u and |∑ u | respectively represent the set and the number of services required by the tasks offloaded to the drone u. Generally, because multiple tasks may request the same service, |∑ u | < |M u |. The service s required by the task m has a priority value cached on the drone u, which can be represented as:

[0075]

[0076]

[0077] The tasks requesting the same service are arranged in descending order of value, and then the task with the largest value is stored in the set ∑ u . where Then the set ∑u Elements in Sort the values ​​in descending order. Services with larger values ​​are cached sequentially until the cache space limit of drone u is reached. We further... u ={s1, s2, ..., s J-1}in

[0078]

[0079] S3: The optimization of UAV trajectory includes fixed task offloading decisions, bandwidth resource allocation decisions, service caching decisions, computing resource allocation decisions, and auxiliary variables to optimize the UAV trajectory, and defines the formula for the UAV trajectory optimization subproblem.

[0080] Specifically, the UAV trajectory Q is optimized using fixed task offloading decision X, bandwidth resource allocation decision B, service caching decision A, computing resource allocation decision F, and auxiliary variable η. The UAV trajectory subproblem is formulated as follows:

[0081]

[0082] stC5, C8-C 11

[0083]

[0084]

[0085] The drone's trajectory planning, as an optimization variable, consists of the drone's coordinate position in each time slot throughout the entire mission cycle. (Removing constants...) It can be simplified to:

[0086]

[0087] Note that, due to the existence of the objective function It is known that problem P 3-3 It is non-convex. Constraint C 10 and C 11 The flight trajectory Q of the drone on the left-hand side is non-convex. Constraint C7 is also non-convex because the domain of a convex function is a non-empty convex set. Therefore, solving non-convex problems is challenging.

[0088] Next, to handle non-convex problems, a successive convex approximation method is used to solve problem P. 3-3 The local optimum. The key idea of ​​the successive convex approximation method is to approximate a non-convex function as a convex function through iteration.

[0089] definition The available spectral efficiency from user equipment m to drone u, can be written as:

[0090]

[0091] It is not difficult to see that is a convex function in . Therefore, it can be globally lower bounded by a first order Taylor expansion around an arbitrary point with a residual error of . The lower bound for the drone flight trajectory at the kth iteration can be computed as:

[0092]

[0093] where and are the available spectral efficiency from user equipment m to drone u and the first order derivative of with respect to , which are given as follows:

[0094]

[0095]

[0096] Define the available spectral efficiency from drone u to drone i, can be written as:

[0097]

[0098] is a convex function in . Therefore, it can be globally lower bounded by a first order Taylor expansion around an arbitrary point with a residual error of . The lower bound for the drone flight trajectory and at the kth iteration can be computed as

[0099]

[0100] where and are the available spectral efficiency from drone u to drone i and the first order derivative of with respect to , which are given as follows:

[0101]

[0102]

[0103] Define The available spectral efficiency from drone u to macro base station b, can be written as:

[0104]

[0105] is a convex function with respect to . Therefore, it can be globally lower bounded by its first order Taylor expansion at any point . The lower bound of the drone flight trajectory at the kth iteration can be computed as:

[0106]

[0107] where and are the available spectral efficiency from drone u to macro base station b and the first order derivative of with respect to , they are given as follows:

[0108]

[0109]

[0110] In constraint C7, because is convex with respect to the flight trajectory of the drone, we employ a successive convex approximation method to relax the constraint. By applying the first order Taylor expansion to any given and , we obtain the following inequality:

[0111]

[0112] Therefore, the lower bound of can be computed as:

[0113]

[0114] In addition, for the flight power 10 in constraint C , the first and third terms are convex with respect to the speed . A continuous relaxation variable is introduced to handle the second term in the propulsion power formula, which becomes:

[0115]

[0116] Simplifying the above equation, we get:

[0117]

[0118] Flight speed of the UAV at the kth iteration and By applying a first-order Taylor expansion to approximate the right-hand side of the above inequality, we have

[0119]

[0120] Then, the upper bound of can be approximated as

[0121]

[0122] Based on the above discussion, all non-convexity in problem Π 3-3 is solved, and the original problem in the kth iteration can be reformulated as the following approximate problem P′ 3-3 (k).

[0123]

[0124] s.t.C5, C6, C8, C9

[0125]

[0126]

[0127]

[0128] After proving the convexity of the problem, the optimal solution of the UAV trajectory planning can be effectively obtained by convex optimization tools. It is worth noting that the optimal solution obtained from the approximate problem Π′ 3-3 is a lower bound of the problem Π 3-3 .

[0129] S4: The optimization of the computing resource allocation decision includes defining an optimization formula of a computing resource allocation sub-problem given the UAV trajectory, the task offloading decision, the service caching decision, and the auxiliary variable, the computing resource allocation sub-problem being a convex problem, and using convex optimization to obtain the optimal solution of the bandwidth resource allocation and the computing resource allocation.

[0130] Specifically, given the UAV trajectory Q, the task offloading decision X, the service caching decision A, and the auxiliary variable η, the computing resource allocation sub-problem is formulated as

[0131]

[0132] s.t.C2, C3, C 10 , C 11 , C 15

[0133]

[0134]

[0135] Because of the problem Π 3-4 It is a convex problem, and convex optimization tools are used to obtain the optimal solutions for bandwidth resource allocation and computational resource allocation.

[0136] This disclosure proposes an alternating optimization approach to solve the primal problem P1. The key idea is to iteratively optimize the task offloading decision, service caching decision, and UAV trajectory planning separately until the objective value converges.

[0137] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0138] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0139] While the specific embodiments of this disclosure have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of this disclosure. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of this disclosure are still within the scope of protection of this disclosure.

Claims

1. A method for task offloading and resource allocation in UAV-based collaborative multi-access edge computing, characterized in that: include: Initialize the collaborative computing task offloading environment. The base station and all drones work together to provide mobile edge computing services to user equipment, obtain the task set, and divide the task cycle into multiple time slots with equal duration. The algorithm obtains the input data size of the computational tasks requested by users in each time slot. Based on the service experience requirements of each user device, it constructs an optimization model with the goal of minimizing service latency while ensuring service fairness among user devices. Based on the Dinkelbach method and convex optimization theory, the problem model is simplified, and a four-stage alternating iterative optimization algorithm is proposed. The optimization objective is decomposed into four sub-optimization objectives: UAV trajectory decision, task offloading decision, service caching decision, and resource allocation decision. The algorithm is used to solve the sub-objectives until the objective converges. The algorithm obtains and executes the task offloading and resource allocation decisions in the UAV collaborative multi-access edge computing network. Satisfaction-based task offloading decision optimization includes: using fixed drone trajectories, bandwidth resource allocation decisions, service caching decisions, computing resource allocation decisions, and auxiliary variables to optimize task offloading decisions, and defining the optimization objective formula for the task offloading subproblem; The formula for the task unloading optimization subproblem is: s.t. 、 、 、 - 、 : : : : Where F represents the computing resource allocation decision, m represents the number of user devices, s represents the service, η represents the auxiliary variable, and M represents the total number of user devices. Let X be the total number of services, and X be the task unloading decision. t For time slots, u For drones, b For a macro base station, ∑ represents the set of all services that a macro base station can provide. In order to bring users m Request service s Task offloaded to drone u The service needs to be provided. s Cache to drone u among, This represents the computational resource constraints of a single drone. This indicates that ground users should be within the coverage area of ​​the associated drone. This indicates that the horizontal distance between the associated drone and the drone selected as a relay does not exceed [a certain value]. , This indicates that each drone is in a time slot. t Energy limit within , This means that the completion delay of each task cannot exceed the task's processing delay tolerance. It indicates that tasks generated by each user device can be precisely offloaded to a nearby drone or macro base station. This indicates that the task unloading decision variable is binary; The optimization of the service caching decision involves fixing the UAV's trajectory, bandwidth resource allocation decision, task offloading decision, computing resource allocation decision, and auxiliary variables to optimize the service caching decision, and defining the optimization formula for the service caching decision subproblem; The formula for the service caching decision subproblem is: s.t. 、 、 、 、 Where F represents the computing resource allocation decision, m represents the number of user devices, s represents the number of services, η represents the auxiliary variable, X represents the task offloading decision, and M represents the total number of user devices. For the total number of services, In order to bring users m Request service s Task offloaded to drone u The service needs to be provided. s Cache to drone u among, This means that the total storage space occupied by the services stored on each drone must not exceed the total storage capacity of the drone. , This indicates that each drone is in a time slot. t Energy limit within , This means that the completion delay of each task cannot exceed the task's processing delay tolerance. This indicates that the service caching decision variable is binary; The optimization of UAV trajectories involves decisions on fixed task offloading, bandwidth resource allocation, service caching, computational resource allocation, and auxiliary variables to optimize the UAV trajectory. Formulas for the UAV trajectory optimization subproblems are defined. The formula for the UAV trajectory subproblem is: s.t. 、 : : Where F represents the computational resource allocation decision, Q represents the UAV trajectory, η is an auxiliary variable, m represents user equipment, s represents services, and M represents the total number of user equipment. For the total number of services, t Let T be the number of time slots. This indicates that ground users should be within the coverage area of ​​the associated drone. This represents the constraint on the position change of the UAV between any two time slots. This indicates that a minimum safe distance should be maintained between any two drones to ensure safety during time slots. t They should avoid colliding with each other. This indicates that the flight paths of all drones should be within the target area. This indicates that the horizontal distance between the associated drone and the drone selected as a relay does not exceed [a certain value]. , This indicates that each drone is in a time slot. t Energy limit within , This means that the completion delay of each task cannot exceed the task's processing delay tolerance. The optimization of computational resource allocation decisions includes defining the optimization formula for the computational resource allocation subproblem, given the UAV trajectory, task offloading decision, service caching decision, and auxiliary variables. The formula for calculating the resource allocation subproblem is: s.t. 、 、 、 、 : : Where B represents bandwidth resource allocation decision, F represents computing resource allocation decision, m represents user equipment, s represents service, and M represents the total number of user equipment. For the total number of services, t Let T be the number of time slots. This indicates the spectrum resource allocation constraints for user equipment associated with the same drone. This represents the computational resource constraints of a single drone. This indicates that each drone is in a time slot. t Energy limit within , This means that the completion delay of each task cannot exceed the task's processing delay tolerance. This indicates that bandwidth and computing resource allocation variables are continuous.

2. The method for task offloading and resource allocation based on UAV collaborative multi-access edge computing as described in claim 1, characterized in that, Considering the utilization of drone cache space, services required by tasks with high service cache decision values ​​are cached on the drone with high priority until the drone's cache space limit is reached.

3. In the method for task offloading and resource allocation based on UAV collaborative multi-access edge computing as described in claim 1, the optimization of the UAV trajectory includes trajectory planning as an optimization variable, which consists of the coordinate position of the UAV in each time slot throughout the entire mission cycle.

4. The method for task offloading and resource allocation based on UAV collaborative multi-access edge computing as described in claim 1, wherein the computational resource allocation subproblem is a convex problem, and convex optimization is used to obtain the optimal solution for bandwidth resource allocation and computational resource allocation.

Images