An optimization method for multi-UAV assisted mobile edge computing

Abstract

The present invention discloses an optimization method for multi-UAV assisted mobile edge computing, comprising the following steps: Step 1, determining the number and initial positions of UAV nodes and the number and positions of users to communicate; Step 2, designing an optimization objective function according to the communication requirements; and performing iterative optimization according to the optimization objective function to obtain the optimal task offloading strategy for users, the optimal computing resource allocation scheme for UAVs, and the optimal trajectory of UAVs; wherein, the optimization objective function is: min l(O,F,Q)={w1T total +w2E total -w3K total}; Step 3, according to the optimal task offloading strategy, locally processing user tasks or offloading them to a specified UAV, providing edge computing services for the users by the UAV, and allocating computing resources to the UAV according to the optimal computing resource allocation scheme for the UAV, and adjusting the positions of the UAVs according to the optimal trajectory of the UAVs.

Description

Technical Field

[0001] The present invention belongs to the field of communication technology, and in particular relates to an optimization method for multi-UAV assisted mobile edge computing. Background Art

[0002] With the rapid development of 6G and the Internet of Things, the number of smart mobile devices has seen unprecedented growth, leading to a surge in various innovative mobile applications. Most of these applications, such as face recognition, automatic navigation, and image processing, require a lot of computing resources and low latency. However, due to the limited resources of mobile devices, it is challenging to process the computationally intensive real-time data generated by these applications; in this context, mobile edge computing is considered to be a promising solution, which allows mobile devices to offload computationally intensive and latency-sensitive tasks to neighboring edge servers, thereby reducing the device's computational burden, execution latency, and energy consumption. However, traditional mobile edge computing networks rely on ground infrastructure and are inflexible in deployment due to installation costs and environmental constraints. In order to overcome the physical limitations of traditional ground mobile edge computing systems, drone-assisted mobile edge computing is emerging due to the high mobility, flexibility, rapid deployment, and line-of-sight links of drones to provide flexible and low-cost offloading services. By offloading computing tasks to nearby drones, mobile users can flexibly enjoy cloud computing services anytime and anywhere. However, designing an effective task offloading method in multi-drone-assisted mobile edge computing systems still faces some challenges. First, ground MEC servers are usually deployed in fixed locations with limited coverage, which may cause service interruptions. In addition, a single drone-assisted MEC deployment has limited computing resources and battery capacity, which may make it difficult to process large amounts of data or maintain long-term operation. Second, from the perspective of problem formulation, a major challenge facing current research is that most studies tend to focus on optimizing a single performance indicator, such as latency or energy consumption, or this The combination of the two indicators ignores multi-objective optimization, which may further undermine the overall effectiveness and efficiency of the system. In addition, the drone-assisted MEC system has many practical characteristics such as strict and diverse user requirements and limited drone resources. These characteristics pose challenges to the joint optimization of meeting user needs and drone resource constraints. Finally, in terms of algorithm design, theoretical methods such as game theory may face challenges due to high computational complexity and space complexity. Traditional heuristic methods such as swarm intelligence algorithms may encounter problems such as long iteration process and being affected by local optimality. Especially in large-scale scenarios, machine learning methods such as deep reinforcement learning may encounter problems of difficulty in convergence and long training cycles. Summary of the invention

[0003] The purpose of the present invention is to provide an optimization method for multi-UAV assisted mobile edge computing, which can reduce the total delay in task completion, reduce the energy consumption of UAVs and increase the total amount of unloaded tasks.

[0004] The technical solution provided by the present invention is:

[0005] An optimization method for multi-UAV assisted mobile edge computing includes the following steps:

[0006] Step 1: Determine the number and initial positions of drone nodes and the number and positions of users that need to communicate;

[0007] Step 2: design an optimization objective function according to the communication requirements; and perform iterative optimization according to the optimization objective function to obtain the user's optimal task offloading strategy, the drone's optimal computing resource allocation plan, and the drone's optimal trajectory;

[0008] Wherein, the optimization objective function is:

[0009] min l(O,F,Q)={w 1 T total +w 2 E total -w 3 K total};

[0010] Among them, O is the user's task offloading strategy; F is the computing resource allocation plan of the drone; Q is the trajectory of the drone; T total is the sum of the user task completion delays; E total is the total energy consumption of the UAV; K total is the total amount of offload tasks completed by the drone when communicating with each user; w 1 is the weight coefficient of the sum of the user task completion delays; w 2 is the weight coefficient of the total energy consumption of the UAV; w 3 is the weight coefficient of the total amount of unloading tasks completed by the UAV;

[0011] Step 3: According to the optimal task offloading strategy, the user task is locally processed or offloaded to a designated drone, and the drone provides edge computing services for the user. In addition, computing resources are allocated to the drone according to the optimal computing resource allocation plan of the drone, and the position of the drone is adjusted according to the optimal trajectory of the drone.

[0012] Preferably, the calculation formula for the sum of the user task completion delays is:

[0013]

[0014]

[0015] Where n is the time slot; is the sum of time slot n; u is the user; is the sum of user u; m is the drone; Sum for drone m; o u,m [n] is the task offloading strategy between user u and drone m in time slot n; T local [n] is the task completion delay of user u in time slot n; T off [n] is the time to be unloaded to the drone; D u [n] is the task size in time slot n (in bits), C u represents the computational intensity of the task (in cycles / bit), f u [n] is the computing power of user u in time slot n, f m,u [n] is the computing capacity allocated by drone m to user u in time slot n; R u,m [n] is the data transmission rate between user u and drone m in time slot n; B is the channel bandwidth; p u represents the transmission power of user u; g u,m [n] represents the channel power gain between user u and drone m in time slot n; σ 2 Represents the noise power.

[0016] Preferably, the calculation formula for the total energy consumption of the drone is:

[0017]

[0018] in, is the flight energy consumption of UAV m in time slot n; is the computing energy consumption of drone m in time slot n (drone m calculates the task of user u); P 0 is the blade profile in the hovering state, which is a constant; P ind is the induced power in the hovering state, which is a constant; v m [n] is the speed of UAV m in time slot n; U tip represents the tip speed of the rotor blade; v 0 represents the average rotor blade induced speed in the hovering state; d 0 represents the fuselage drag ratio; ρ 0 represents air density; s represents the hardness of the rotor blade; A represents the rotor disc area; δ t is the time of each time slot; m represents the effective switched capacitance which depends on the CPU architecture.

[0019] Preferably, the calculation formula for the total amount of unloading tasks completed by the drone is:

[0020]

[0021] Preferably, the step 2 further includes: before performing iterative optimization of the optimization objective function, it is necessary to determine the mobility of the drone, and the formula is as follows:

[0022] Drone position updates:

[0023] Drone speed update: v m [n+1]=v m [n]+a m [n]

[0024] The speed constraint of the drone is: ||v m [n]||≤V max ,||v m [n]||≥V min

[0025] The safe distance between any two drones: ||q m [n]-q i [n]||≥D min

[0026] The initial and final positions of each drone are fixed at the same location: q m [1] = q m [N]

[0027] The maximum flight distance of the drone in each time slot: ||q m [n+1]-q m [n]||≤D max =V max δ t

[0028] Among them, q m [n+1] is the position of drone m in the next time slot; q m [n] is the current position of UAV m in time slot n; q i [n] is the position of another UAV i in time slot n; a m [n] is the acceleration of UAV m in time slot n; v m [n+1] is the speed of drone m in the next time slot; V max is the maximum speed of UAV m; V min is the minimum speed of UAV m; D max is the maximum distance between any two drones; D min is the minimum distance between any two UAVs; q m [1] is the initial position of UAV m; q m [N] is the final position of UAV m.

[0029] Preferably, in step 2, the iterative optimization comprises the following steps:

[0030] Step 1: Initialize the computing resource allocation scheme and trajectory of all drones according to the total number of drones in multi-drone assisted edge computing, and take the user's task offloading strategy as a candidate solution;

[0031] Step 2: According to the optimization objective function, a task offloading strategy solution for the user in the current situation is obtained;

[0032] Step 3: Based on the solution in step 2 and the initialized UAV trajectory, the UAV computing resource allocation scheme is used as a candidate solution;

[0033] Step 4: According to the optimization objective function, a solution for the computing resource allocation scheme of the current UAV is obtained;

[0034] Step 5: Based on the solutions in steps 2 and 4, the trajectory of the drone is used as a candidate solution;

[0035] Step 6: According to the optimization objective function, the trajectory solution of the current UAV is obtained;

[0036] Step 7: The three candidate solutions obtained are used as the basis for the next iteration solution and updated;

[0037] Step 8: If the iteration limit is reached, stop and then output the minimum solution that satisfies the objective function as the optimal solution, that is, the user's optimal task offloading strategy, the drone's optimal computing resource allocation plan, and the drone's optimal trajectory. Otherwise, return these solutions to step 2 as the current solution, and then repeat steps 2 to 6 until the iteration termination condition is met.

[0038] The beneficial effects of the present invention are:

[0039] The optimization method for multi-UAV assisted mobile edge computing provided by the present invention can reduce the total delay in task completion, reduce the energy consumption of UAVs and increase the total amount of unloaded tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a workflow diagram of the optimization method for multi-UAV assisted mobile edge computing described in the present invention.

[0041] Figure 2 A schematic diagram of the optimization method for multi-UAV assisted mobile edge computing according to the present invention.

[0042] Figure 3 This is a simulation result diagram showing the objective function value increasing with the number of drones.

[0043] Figure 4 The simulation results show that the total mission completion delay increases with the number of drones.

[0044] Figure 5 This is a simulation result diagram showing the total UAV energy consumption as the number of UAVs increases.

[0045] Figure 6 This is a simulation result diagram showing the total number of offloading tasks increasing with the number of drones. DETAILED DESCRIPTION

[0046] The present invention is further described in detail below in conjunction with the accompanying drawings so that those skilled in the art can implement the invention with reference to the description.

[0047] like Figure 1 As shown, in the process of multi-UAV assisted edge computing to achieve communication, the aerial UAV carries the edge computing server to provide computing services for users. The present invention provides an optimization method for multi-UAV assisted mobile edge computing, and the steps are as follows:

[0048] Step 1: Determine the number and initial positions of drone nodes in multi-drone assisted edge computing and the number and positions of users who need to communicate;

[0049] Step 2: First, three optimization functions are designed according to the communication requirements, and the multi-objective optimization problem is planned according to the multi-objective optimization theory. The three optimization functions are designed into a multi-objective function:

[0050] min l(O,F,Q)={T total ,E total ,-K total}

[0051] Among them, O is the user's task offloading strategy; F is the computing resource allocation plan of the drone; Q is the trajectory of the drone; T total is the sum of the user task completion delays; E total is the total energy consumption of the UAV; K total It is the total amount of offload tasks completed by the drone when communicating with each user.

[0052] Secondly, the multi-objective optimization problem is transformed into a single-objective optimization problem through the weight coefficient method, and the multi-objective optimization function is transformed into an optimization objective function through the weight coefficient method, which simplifies the complexity, unifies the evaluation criteria, and facilitates the trade-off between different objectives;

[0053] The optimization objective function is:

[0054] min l(O,F,Q)={w 1 T total +w 2 E total -w 3 K total}

[0055] Among them, w 1 is the weight coefficient of the sum of the user task completion delays; w 2 is the weight coefficient of the total energy consumption of the UAV; w 3 It is the weight coefficient of the total amount of unloading tasks completed by the UAV.

[0056] The first objective function is:

[0057]

[0058] Where n is the time slot; is the sum of time slot n; u is the user; is the sum of user u; m is the drone; Sum for drone m; o u,m [n] is the task offloading strategy between user u and drone m in time slot n; T local [n] is the task completion delay of user u in time slot n; T off [n] is the time to be unloaded to the drone; D u [n] is the task size in time slot n (in bits), C u represents the computational intensity of the task (in cycles / bit), f u [n] is the computing power of user u in time slot n, f m,u [n] is the computing capacity allocated by drone m to user u in time slot n; R u,m [n] is the data transmission rate between user u and drone m in time slot n; B is the channel bandwidth; p u represents the transmission power of user u; g u,m [n] represents the channel power gain between user u and drone m in time slot n; σ 2 Represents the noise power.

[0059] The second objective function is:

[0060]

[0061] in, is the flight energy consumption of UAV m in time slot n (obtained from the speed, which is obtained through the deterministic formula of the UAV); is the computing energy consumption of drone m in time slot n (drone m calculates the task of user u); P 0 is the blade profile in the hovering state, which is a constant; P ind is the induced power in the hovering state, which is a constant; v m [n] is the speed of UAV m in time slot n; U tip represents the tip speed of the rotor blade; v0 represents the average rotor blade induced speed in the hovering state; d 0 represents the fuselage drag ratio; ρ 0 represents air density; s represents the hardness of the rotor blade; A represents the rotor disc area; δ t is the time of each time slot; m represents the effective switched capacitance which depends on the CPU architecture.

[0062] The third objective function is:

[0063] Afterwards, before iteratively optimizing the optimization objective function, the mobility of the drone needs to be determined. The specific formula is as follows:

[0064] Drone position updates:

[0065] Drone speed update: v m [n+1]=v m [n]+a m [n]

[0066] The speed constraint of the drone is: ||v m [n]||≤V max ,||v m [n]||≥V min

[0067] The safe distance between any two drones: ||q m [n]-q i [n]||≥D min

[0068] The initial and final positions of each drone are fixed at the same location: q m [1] = q m [N]

[0069] The maximum flight distance of the drone in each time slot: ||q m [n+1]-q m [n]||≤D max =V max δ t

[0070] The time of each time slot: δ t =T / N

[0071] Among them, q m [n+1] is the position of drone m in the next time slot; q m [n] is the current position of UAV m in time slot n; q i [n] is the position of another UAV i in time slot n; am [n] is the acceleration of UAV m in time slot n; v m [n+1] is the speed of drone m in the next time slot; V max is the maximum speed of UAV m; V min is the minimum speed of UAV m; D max is the maximum distance between any two drones; D min is the minimum distance between any two UAVs; q m [1] is the initial position of UAV m; q m [N] is the final position of UAV m; T is the continuous system time; N is the total number of equal time; δ t For the time of each time slot, the continuous system time T is discretized into N equal time slots.

[0072] Finally, the iterative optimization algorithm is used to obtain the optimal task offloading strategy for users, the optimal computing resource allocation plan for drones, and the optimal trajectory of each drone when communicating with each user. The iterative optimization includes the following steps:

[0073] Step 1: Initialize the computing resource allocation scheme and trajectory of all drones according to the total number of drones in multi-drone assisted edge computing, and take the user's task offloading strategy as a candidate solution;

[0074] Step 2: According to the optimization objective function, a task offloading strategy solution for the user in the current situation is obtained;

[0075] Step 3: Based on the solution in step 2 and the initialized UAV trajectory, the UAV computing resource allocation scheme is used as a candidate solution;

[0076] Step 4: According to the optimization objective function, a solution for the computing resource allocation scheme of the current UAV is obtained;

[0077] Step 5: Based on the solutions in steps 2 and 4, the trajectory of the drone is used as a candidate solution;

[0078] Step 6: According to the optimization objective function, the trajectory solution of the current UAV is obtained;

[0079] Step 7: The three candidate solutions obtained are used as the basis for the next iteration solution and updated;

[0080] Step 8: If the iteration limit is reached, stop and then output the minimum solution that satisfies the objective function as the optimal solution, that is, the user's optimal task offloading strategy, the drone's optimal computing resource allocation plan, and the drone's optimal trajectory. Otherwise, return these solutions to step 2 as the current solution, and then repeat steps 2 to 6 until the iteration termination condition is met.

[0081] Step 3: According to the optimal task offloading strategy, the user task is locally processed or offloaded to a designated drone, and the drone provides edge computing services for the user. In addition, computing resources are allocated to the drone according to the optimal computing resource allocation plan of the drone, and the position of the drone is adjusted according to the optimal trajectory of the drone.

[0082] The present invention provides an optimization method for multi-UAV assisted mobile edge computing, establishes a multi-objective joint optimization model for reducing the total delay of task completion, reducing the energy consumption of UAVs and increasing the total amount of unloaded tasks completed by UAVs, firstly determines the number of users and UAV nodes participating in data transmission in multi-UAV assisted edge computing, and then uses an iterative optimization algorithm to design the optimal task unloading strategy for user data transmission in multi-UAV assisted edge computing, the optimal computing resource allocation scheme for UAVs and the optimal trajectory of UAVs, so as to solve the model; a hybrid solution update strategy is introduced in the iterative optimization algorithm, and the solution is improved by updating the continuous solution (the optimal computing resource allocation scheme for UAVs and the optimal trajectory of UAVs) and the discrete solution (the optimal task unloading strategy for users) respectively. effectiveness and convergence speed. Finally, the drones in the optimization method for multi-drone assisted mobile edge computing communicate with the users who need to communicate in turn according to the optimal task offloading strategy. When communicating with different users, each drone will adjust its own computing resources and trajectory to achieve the optimal state, so that the multi-drone assisted mobile edge computing system can accelerate the speed of task completion, thereby reducing the overall task completion delay, reducing unnecessary movement and resource waste, thereby effectively reducing the energy consumption of drones, and improving the processing capacity of the entire system, thereby increasing the total amount of offload tasks completed by drones. The optimization method for multi-drone assisted mobile edge computing provided by the present invention can achieve the efficiency and reliability of multi-drone assisted mobile edge computing.

[0083] like Figure 2As shown, in drone-assisted edge computing, drones provide users with flexible and resilient computing services due to their high maneuverability, flexibility and line-of-sight links; however, due to limited drone resources, meeting users' computing-intensive and delay-sensitive needs is a major challenge. The present invention proposes a multi-drone-assisted mobile edge computing method, each drone is equipped with one or more edge computing servers (MEC Servers), which communicate with users who need to communicate in turn according to the obtained optimal task offloading strategy, and provide different users with the best computing resource allocation scheme and trajectory; by designing the user's task offloading, not only the response speed of the user's task is improved, but also the communication congestion and delay are reduced, thereby achieving the purpose of minimizing the total delay of task completion and maximizing the total amount of unloaded tasks completed by the drone; in addition, by obtaining the optimal computing resource allocation scheme and optimal trajectory of the drone, the utilization efficiency of resources can be improved, and the moving distance when the drone communicates with different users can be shortened, thereby reducing energy consumption; therefore, the optimization method of multi-drone-assisted mobile edge computing proposed by the present invention can improve the user experience and the overall efficiency of the system.

[0084] like Figure 3-6 As shown, the four figures show the simulation results of the objective function values, total task completion delay, total UAV energy consumption, and total number of unloaded tasks of the seven methods as the number of UAVs increases, with a fixed number of users of 8.

[0085] We compare the proposed JTORATC (Joint Optimization of Task Offloading, Computer Resource Allocation and UAV Trajectory) algorithm with the following schemes:

[0086] ROJRATC (Random Offloading): The user’s task offloading strategy is randomly determined, while the computational resource allocation and trajectory of the UAV are determined based on the proposed JTORATC.

[0087] NOJRATC (Nearest Offloading): Offload each user’s task to the nearest UAV while deciding computational resource allocation and trajectory control according to the proposed JTORATC.

[0088] MOJRATC (Many-to-Many Matching): It uses a matching-to-matching mechanism to determine the task offloading strategy and determines the computing resource allocation and trajectory of the UAV based on JTORATC.

[0089] ERJOTC (Resource Equalization): Determines the computing resource allocation of the UAV based on the JTORATC average, and also determines the task offloading and trajectory control strategies.

[0090] JORACT (circular trajectory): Based on JTORATC, the task offloading strategy and computing resource allocation are determined, and the UAV flies along a circular trajectory.

[0091] JORAPT (predefined trajectory): The task offloading strategy and computing resource allocation are determined based on JTORATC, while the UAV follows a predefined trajectory.

[0092] Figure 3 This is the simulation result diagram of the objective function value increasing with the number of drones. Figure 4 This is the simulation result diagram of the total task completion delay as the number of drones increases. Figure 5 This is the simulation result diagram of the total UAV energy consumption as the number of UAVs increases. Figure 6 The simulation results of the total number of unloading tasks as the number of drones increases are shown in Figure 2. Figure 3-6 It can be concluded from the simulation results that, when the number of users is fixed, the objective function value, total task completion delay and total drone energy consumption of the optimization method for multi-UAV assisted mobile edge computing provided in the present invention are the smallest among the seven methods, but the total number of unloaded tasks is the largest; the optimization method for multi-UAV assisted mobile edge computing provided in the present invention can reduce the total delay in task completion, reduce drone energy consumption and increase the total amount of unloaded tasks completed by drones.

[0093] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and the implementation modes, and they can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and the illustrations shown and described herein.

Claims

1. An optimization method for multi-UAV assisted mobile edge computing, characterized in that: The steps include: Step 1: Determine the number and initial positions of drone nodes and the number and positions of users that need to communicate; Step 2: design an optimization objective function according to the communication requirements; and perform iterative optimization according to the optimization objective function to obtain the user's optimal task offloading strategy, the drone's optimal computing resource allocation plan, and the drone's optimal trajectory; Wherein, the optimization objective function is: min l(O,F,Q)={w1T total +w2E total -w3K total }; Where O is the user's task offloading strategy; F is the computing resource allocation scheme of the UAV; Q is the trajectory of the UAV; T total is the sum of the user task completion delays; E total is the total energy consumption of the UAV; K total is the total amount of unloading tasks completed by the drone when communicating with each user; w1 is the weight coefficient of the sum of the user task completion delays; w2 is the weight coefficient of the total energy consumption of the drone; w3 is the weight coefficient of the total amount of unloading tasks completed by the drone; The calculation formula for the sum of the user task completion delays is: The calculation formula for the sum of the user task completion delays is: Where n is the time slot; is the sum of time slot n; u is the user; is the sum of user u; m is the drone; Sum for drone m; o u,m [n] is the task offloading strategy between user u and drone m in time slot n; T local [n] is the task completion delay of user u in time slot n; T off [n] is the time to be unloaded to the drone; D u [n] is the task size in time slot n; C u is the computational intensity of the task; f u [n] is the computing power of user u in time slot n; f m,u [n] is the computing capacity allocated by drone m to user u in time slot n; R u,m [n] is the data transmission rate between user u and drone m in time slot n; B is the channel bandwidth; p u represents the transmission power of user u; g u,m [n] represents the channel power gain between user u and drone m in time slot n; σ 2 represents the noise power; The calculation formula of the total energy consumption of the UAV is: In the formula, is the flight energy consumption of UAV m in time slot n; is the calculated energy consumption of UAV m in time slot n; P0 is the blade profile in the hovering state, which is a constant; P ind is the induced power in the hovering state, which is a constant; v m [n] is the speed of UAV m in time slot n; U tip represents the tip speed of the rotor blade; v0 represents the average rotor blade induced speed in the hovering state; d0 represents the fuselage drag ratio; ρ0 represents the air density; s represents the rotor blade hardness; A represents the rotor disc area; δ t is the time of each time slot; m represents the effective switched capacitance that depends on the CPU architecture; The calculation formula for the total amount of unloading tasks completed by the drone is: Step 3: According to the optimal task offloading strategy, the user task is locally processed or offloaded to a designated drone, and the drone provides edge computing services for the user. In addition, computing resources are allocated to the drone according to the optimal computing resource allocation plan of the drone, and the position of the drone is adjusted according to the optimal trajectory of the drone.

2. The optimization method for multi-UAV assisted mobile edge computing according to claim 1, characterized in that: The step 2 also includes: before performing iterative optimization of the optimization objective function, it is necessary to determine the mobility of the drone, and the formula is as follows: Drone position updates: Drone speed update: v m [n+1]=v m [n]+a m [n] The speed constraint of the drone is: ‖‖v m [n]‖‖≤V max ,‖‖v m [n]‖‖≥V min The safe distance between any two drones: ‖‖q m [n]-q i [n]‖‖≥D min The initial and final positions of each drone are fixed at the same location: q m [1] = q m [N] The maximum flight distance of the drone in each time slot: ‖‖q m [n+1]-q m [n]‖‖≤D max =V max δ t Among them, q m [n+1] is the position of drone m in the next time slot; q m [n] is the current position of UAV m in time slot n; q i [n] is the position of another UAV i in time slot n; a m [n] is the acceleration of UAV m in time slot n; v m [n+1] is the speed of drone m in the next time slot; V max is the maximum speed of the drone m; V min is the minimum speed of UAV m; D max is the maximum distance between any two drones; D min is the minimum distance between any two UAVs; q m [1] is the initial position of UAV m; q m [N] is the final position of UAV m.

3. The optimization method for multi-UAV assisted mobile edge computing according to claim 1, characterized in that: In the step 2, the iterative optimization includes the following steps: Step 1: Initialize the computing resource allocation scheme and trajectory of all drones according to the total number of drones in multi-drone assisted edge computing, and take the user's task offloading strategy as a candidate solution; Step 2: According to the optimization objective function, a task offloading strategy solution for the user in the current situation is obtained; Step 3: Based on the solution in step 2 and the initialized UAV trajectory, the UAV computing resource allocation scheme is used as a candidate solution; Step 4: According to the optimization objective function, a solution for the computing resource allocation scheme of the current UAV is obtained; Step 5: Based on the solutions in steps 2 and 4, the trajectory of the drone is used as a candidate solution; Step 6: According to the optimization objective function, the trajectory solution of the current UAV is obtained; Step 7: The three candidate solutions obtained are used as the basis for the next iteration solution and updated; Step 8: If the iteration limit is reached, stop and then output the minimum solution that satisfies the objective function as the optimal solution, that is, the user's optimal task offloading strategy, the drone's optimal computing resource allocation plan, and the drone's optimal trajectory. Otherwise, return these solutions to step 2 as the current solution, and then repeat steps 2 to 6 until the iteration termination condition is met.

Images