Unmanned aerial vehicle assisted calculation unloading method and device

Abstract

The invention discloses an unmanned aerial vehicle assisted calculation unloading method and device, and belongs to the technical field of communication, and the method comprises the steps: constructing an unloading platform corresponding to the unmanned aerial vehicle assisted calculation unloading method; constructing a spectrum transaction model which takes the unmanned aerial vehicle as a leader and an operator as a follower and is based on the Stackelberg game; unmanned aerial vehicle and operator optimization problem analysis is carried out; and according to an analysis result, generating a comprehensive optimization problem, solving a game equilibrium point, and obtaining an optimal calculation unloading method for calculation unloading of request users inside and outside the communication coverage range of the unmanned aerial vehicle, and the request users outside the communication coverage range of the unmanned aerial vehicle utilize idle users in the coverage range of the unmanned aerial vehicle as relay objects through a D2D relay technology. And unloading the calculation task to the unmanned aerial vehicle for processing. According to the invention, the problem that the calculation task is difficult to unload or the local calculation time delay is too long because the terminal at the communication edge or blind area cannot directly communicate with the unmanned aerial vehicle can be solved.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and specifically to a method and device for drone-assisted computing offloading. Background Art

[0002] With the explosive growth in the number of Internet of Things devices, due to the limitations of computing resources and battery capacity of terminal devices, it is difficult for them to efficiently complete compute-intensive or latency-sensitive tasks relying solely on themselves. Mobile Edge Computing (MEC), as a network edge technology that extends the computing resources of cloud servers closer to the terminal side, provides an ideal solution for computing offloading. Specifically, MEC "sinks" the services originally located in the cloud data center to the mobile network edge. By deploying resources such as computing, storage, network, and communication at the mobile network edge, it not only reduces network operations but also decreases service delivery latency and improves the user experience. In addition, after deploying servers at the network edge, MEC reduces the transmission bandwidth requirements for the core network, thereby reducing operating costs. Computing offloading is a key technical concept in edge computing, which refers to the process of a user terminal (such as a mobile device) offloading computing tasks to an edge network (such as a Mobile Edge Computing MEC network) for execution. This technology mainly addresses the deficiencies of devices in terms of resource storage, computing performance, and energy efficiency. By reasonably utilizing the computing resources of the edge network, it provides the required computing power for resource-constrained devices to run compute-intensive applications, thereby accelerating the computing speed and saving energy.

[0003] Unmanned Aerial Vehicles (UAVs) exhibit great potential in MEC scenarios with imperfect network infrastructure due to their inherent mobility, flexibility, and low deployment costs. However, due to the limited coverage range and battery power of UAVs, in disaster areas where basic communication facilities are damaged, terminals located at the communication edge or in blind spots cannot directly communicate with UAVs, resulting in problems such as difficulty in offloading computing tasks or excessive local computing latency. Summary of the Invention

[0004] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a method and device for drone-assisted computing offloading to solve the problem that terminals at the communication edge or in blind spots cannot directly communicate with UAVs, resulting in difficulty in offloading computing tasks or excessive local computing latency.

[0005] To solve the above technical problems, the present invention is implemented by adopting the following technical solutions:

[0006] In a first aspect, the present invention provides a UAV-assisted computing offloading method, including: S1: constructing an offloading platform corresponding to the UAV-assisted computing offloading method; S2: based on the offloading platform constructed in step S1, constructing a spectrum trading model based on Stackelberg game with the UAV as the leader and the operator as the follower; S3: according to the spectrum trading model constructed in step S2, respectively analyzing the UAV optimization problem and the operator optimization problem; S4: according to the analysis results of the UAV optimization problem and the operator optimization problem in step S3, generating a comprehensive optimization problem, finding the game equilibrium point, and obtaining the optimal computing offloading method for the requesting users inside and outside the UAV communication coverage. For the requesting users outside the UAV communication coverage, through D2D relay technology, using the idle users within the UAV coverage as relay objects, the computing tasks are offloaded to the UAV for processing.

[0007] For the aforementioned UAV-assisted computing offloading method, the offloading platform corresponding to the UAV-assisted computing offloading method in step S1 includes: UAVs, requesting users, idle users, and operators; the requesting users are users with computing tasks, and the requesting users include the requesting users within the UAV communication coverage and the requesting users outside the UAV communication coverage; the idle users are users without computing tasks, and the idle users are located within the UAV communication range and within the D2D communication range of the requesting users outside the UAV communication coverage; the operator has the qualification of authorized spectrum resources and its own users.

[0008] For the aforementioned UAV-assisted computing offloading method, constructing the spectrum trading model based on Stackelberg game in step S2 includes: constructing a UAV utility function and an operator utility function; according to the UAV utility function and the operator utility function, obtaining the game equilibrium point; the construction of the UAV utility function includes: L1: there exist requesting users, represented by the set , where, there are requesting users within the communication coverage of the UAV, requesting users outside the UAV communication coverage. The requesting users outside the UAV communication coverage respectively take themselves as the center and detect that there are a total of idle users within the UAV communication coverage within the D2D communication range; L2: the computing task of each requesting user is represented as , where, , is the data volume of the computing task, is the CPU cycles required per bit of the computing task, is the maximum tolerable delay of the computing task; L3: The data volume of the computing task requested by the user includes the local data volume and the offloaded data volume. The computing task requested by the user is executed in parallel, with the local data volume processed locally and the offloaded data volume processed by the drone. The offloading ratio of a single computing task is expressed as , , and the offloading decision of a single computing task is expressed as , ; For the requesting user within the communication coverage of the drone , the offloading decision is , when , it means that the requesting user within the drone coverage does not perform the offloading task. When , it means that the requesting user within the drone coverage performs the offloading task; For the requesting user outside the communication coverage of the drone , , the offloading decision is , , when , it means that the requesting user does not offload. When , it means that the requesting user The idle user within the D2D communication range provides relay service for the requesting user ; L4: The completion delay of the computing task of the requesting user is determined by the combination of the local computing delay and the offloading computing delay . The calculation formula is: , and the calculation formula of the local computing delay is: , where is the number of cycles executed by the local CPU per second; L5: For the requesting user , the transmission delay of the computing task is divided into two stages. The first stage is the transmission delay of D2D user pair communication, and the second stage is the transmission delay of the relay object uploading the offloaded data volume to the drone; For the requesting user , the transmission delay of the computing task is the transmission delay of uploading the offloaded data volume to the drone; For the requesting user , the transmission delay of the computing task is calculated as: , where is the transmission rate of the requesting user uploading the offloaded data volume to the drone; The calculation formula of the drone computing delay is ​, where is the computing resources allocated by the UAV to the requesting user ; the calculation formula for the offloading calculation delay is: ; The total delay of the computing tasks processed by the UAV-assisted computing offloading method for the requesting users is calculated as: ; L6: Considering the obtained system performance gain and the payment cost of spectrum leasing comprehensively, the UAV utility function is calculated as: where is a preset weight coefficient representing the system delay performance gain, is the total delay of all local processing of the computing task, is the unit price of spectrum leasing, is the amount of spectrum leased by the UAV; represents the payment cost of the UAV; the utility function of the operator is calculated as: where is a preset weight coefficient for obtaining lease income, is a preset weight coefficient for the reduction of user service quality caused by renting out the spectrum; is the maximum user service quality, which is represented by the operator user service quality when all the spectrum is used for its own users without the operator renting out the spectrum; is a function for calculating the operator user service quality according to the amount of spectrum leased by the UAV ; the function of the operator user service quality is calculated as: where .

[0009] For the aforementioned UAV-assisted computing offloading method, the UAV optimization problem in step S3 is: where is the offloading strategy, is the computing resource allocation strategy, is the spectrum resource allocation strategy; the constraint condition C1 is: the offloading ratio of the computing task ranges from 0 to 1; the constraint condition C2 is: each idle user provides relay for only one requesting user, where is the discriminant function, when the discriminant condition When it is true, the discrimination function is equal to 1. When the discrimination condition is false, the discrimination function is equal to 0; the constraint condition C3 is: the spectrum quantity of UAV leasing is all used for the spectrum resource allocation of offloading data volume, where is the set in the th user's spectrum resource for communicating with the UAV; the constraint condition C4 is: the total computing resource allocated by the UAV to the requesting users does not exceed the total computing resource of the UAV ; the constraint condition C5 is: the completion delay of the computing task of each requesting user does not exceed the maximum tolerable delay of the computing task ; the operator optimization problem described in step S3 is: , where is the preset minimum user service quality of the operator; the constraint condition C6 is: the value of the operator utility function is non - negative; the constraint condition C7 is: the value of the function of the operator user service quality is not less than the preset minimum user service quality of the operator.

[0010] For the aforementioned UAV - assisted computing offloading method, the analysis of the operator optimization problem in step S3 includes: obtaining the analytical solution of the optimal spectrum quantity leased by the operator when the given optimal spectrum leasing unit price is derived through convex optimization theory; Y1: when the UAV is given the spectrum leasing unit price , the expression of the operator's utility function is: ; Y2: taking the first - order derivative and the second - order derivative of the utility function of the operator with the given spectrum leasing unit price with respect to respectively, the following function expressions are obtained: , ; Y3: according to the second - order derivative function expression being always less than 0, it shows that the first - order derivative function is monotonically decreasing in the domain, and the original function is a strictly convex function; calculating the spectrum quantity leased by the UAV according to the constraint condition C7, it shows that there is an optimal spectrum quantity in the domain of such that the value of the operator utility function is the largest; Y4: letting the first - order derivative function be 0, the calculation formula for the optimal spectrum quantity is: ; Y5: letting the optimal spectrum quantity be in the domain of , the maximum value and the minimum value of the spectrum leasing unit price are solved as: , ; Y6: Based on the maximum and minimum values of the spectrum rental unit price of the UAV, combined with the constraint condition C7, obtain the optimal spectrum quantity Regarding the spectrum rental unit price Analytical solution of:

[0011] .

[0012] For the above-mentioned UAV-assisted computing offloading method, the analysis of the UAV optimization problem in step S3 includes: W1: According to the fact that the optimization objective of the UAV optimization problem under the determined spectrum quantity is a multi-variable non-convex optimization problem, decompose the UAV optimization problem into two sub-problems of offloading strategy and resource allocation strategy and solve them separately; W2: Analyze the two sub-problems in step W1 respectively to obtain the corresponding solution algorithms; W3: According to the mutual restriction properties of the two sub-problems in step W1, use the alternating iteration algorithm to alternately iterate the solution algorithms corresponding to the two sub-problems until the change amount of the solution of the solution algorithm meets the preset change amount range, and output the optimal offloading strategy and optimal resource allocation strategy under the determined spectrum quantity; among them, the alternating iteration includes: J1: Apply the corresponding solution algorithm to the first sub-problem to obtain a preliminary solution; J2: Based on the preliminary solution, apply the corresponding solution algorithm to the second sub-problem to obtain an updated solution; J3: Alternately perform steps J1 and J2 until the change amount of the solutions of the two sub-problems meets the preset change amount range, and the change amount of the solution refers to the difference between the solutions of steps J1 and J2.

[0013] For the above-mentioned UAV-assisted computing offloading method, step W1 includes: Substitute the spectrum rental unit price into the analytical solution of the optimal spectrum quantity in step Y6 , solve to obtain the optimal spectrum quantity, determine the payment cost of the UAV according to the spectrum rental unit price and the optimal spectrum quantity, and maximize the optimization objective of the UAV optimization problem, the UAV utility, into optimizing the offloading strategy , computing resource allocation strategy and spectrum resource allocation strategy to minimize the total delay of the computing task: , the optimization objective of the P3 problem is a multi-variable non-convex optimization problem, and decompose the P3 problem into offloading strategy and resource allocation strategy to solve the two sub-problems separately.

[0014] For the above-mentioned UAV-assisted computing offloading method, step W2 includes: Analyze the offloading strategy sub-problem and the resource allocation strategy sub-problem respectively to obtain the corresponding solution algorithms; Analyze the resource allocation strategy sub-problem to obtain the corresponding solution algorithm includes: When the offloading strategy When the offloading ratio is determined, according to the local computing delay The calculation formula determines the local computing delay ; For each requesting user, according to the calculation formula of the computing task completion delay: , the computing task completion delay It is divided into the following two cases: If , then , indicating that the computing task completion delay is determined by the local computing delay , and optimizing the resource allocation strategy will not reduce the computing task completion delay ; If , then , indicating that the computing task completion delay is determined by the offloading computing delay , and optimizing the resource allocation strategy will reduce the computing task completion delay ; Since the local processing delay is a constant and does not need to be optimized, the solution of the resource allocation strategy sub-problem is transformed into: , Since the optimization problem P4 is a non-linear optimization problem and contains multiple variables and constraints, a genetic algorithm is used to solve the optimization problem P4; Analyze the offloading strategy The sub-problem obtains the corresponding solution algorithm including: the offloading strategy includes the offloading ratio and the offloading decision ; When the resource allocation strategy is determined, by jointly optimizing the offloading ratio and the offloading decision of each requesting user, reduce The total delay of the computing tasks processed by the computing offloading method assisted by drones for requesting users The value of the offloading ratio is continuous, and the value of the offloading decision is discrete, then the solution of the offloading strategy sub-problem is a mixed integer non-linear optimization problem: , Since the optimization problem P5 is to jointly optimize continuous variables and discrete variables and contains multiple constraints, an improved particle swarm algorithm is used to solve the optimization problem P5; The improved particle swarm algorithm includes: using a probability update mechanism for the offloading decision to select an idle user as the relay object; using a dynamically updated inertia weight; The update formula of the inertia weight is: , In the formula, is the preset minimum value of the inertia weight, is the preset maximum value of the inertia weight, is the current iteration number, and is the preset maximum number of iterations.

[0015] For the aforementioned UAV-assisted computing offloading method, step S4 includes: S41: According to the analysis results of the UAV optimization problem and the operator optimization problem, obtain the comprehensive optimization problem that combines the UAV optimization problem and the operator optimization problem: ; S42: According to the comprehensive optimization problem P6, use the binary search algorithm to find the game equilibrium point and obtain the optimal computing offloading method for the requesting users inside and outside the UAV communication coverage; step S42 includes: S421: According to the comprehensive optimization problem P6, compare the UAV utilities calculated respectively at the two endpoints and the midpoint of the spectrum rental unit price interval, perform interval division iteration to narrow the spectrum rental unit price interval until the difference between the endpoints of the spectrum rental unit price interval meets the preset search interval accuracy, and obtain the optimal spectrum rental unit price interval as the search interval of the binary search method; among them, the initial spectrum rental unit price interval is from the minimum spectrum rental unit price to the maximum spectrum rental unit price ; Calculating the UAV utility according to the spectrum rental unit price includes: using the alternating iteration algorithm to alternately iterate the genetic algorithm and the particle swarm algorithm until the change amount of the total time delay for the UAV-assisted computing offloading method to process the computing tasks meets the preset change amount range, and output the optimal offloading strategy and the optimal resource allocation strategy under the determined spectrum quantity; S422: According to the optimal spectrum rental unit price interval obtained in step S421, through the binary search algorithm, compare the UAV utilities calculated respectively at the midpoint and the midpoint perturbation value of the optimal spectrum rental unit price interval, perform interval division iteration to adjust the optimal spectrum rental unit price interval until the difference between the UAV utilities calculated respectively at the interval midpoint and the midpoint perturbation value meets the preset difference accuracy, and output the interval midpoint that meets the preset difference accuracy as the optimal spectrum rental unit price ; among them, the midpoint perturbation value is the sum of the interval midpoint and the preset perturbation ; S423: Substitute the optimal spectrum rental unit price into the analytical solution of the optimal spectrum quantity to calculate the optimal spectrum quantity , obtain the solved game equilibrium point , and output the optimal offloading strategy and the optimal resource allocation strategy under the optimal spectrum quantity ; S424: Obtain the optimal computing offloading method for the requesting users inside and outside the UAV communication coverage , users who request outside the UAV communication coverage area use the D2D relay technology and utilize idle users within the UAV coverage area as relay objects to offload computing tasks to the UAV for processing.

[0016] In a second aspect, the present invention provides a UAV-assisted computing offloading device, including: an offloading platform module, a trading model module, an analysis module, and a comprehensive solution module; the offloading platform module: used to construct an offloading platform corresponding to the UAV-assisted computing offloading method; the trading model module: used to construct a spectrum trading model based on the Stackelberg game with the UAV as the leader and the operator as the follower according to the offloading platform constructed by the offloading platform module; the analysis module: used to respectively analyze the UAV optimization problem and the operator optimization problem according to the spectrum trading model constructed by the trading model module; the comprehensive solution module: used to generate a comprehensive optimization problem according to the analysis results of the UAV optimization problem and the operator optimization problem by the analysis module, find the game equilibrium point, and obtain the optimal computing offloading method for users who request computing offloading inside and outside the UAV communication coverage area. Users who request outside the UAV communication coverage area use the D2D relay technology and utilize idle users within the UAV coverage area as relay objects to offload computing tasks to the UAV for processing.

[0017] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0018] For users who request outside the UAV coverage area, the present invention uses the D2D relay technology and utilizes idle users within the UAV coverage area as relay objects. In view of the mutual constraints between the UAV and the operator, a spectrum trading model based on the Stackelberg game is constructed to solve the optimal computing offloading method for users who request computing offloading inside and outside the UAV communication coverage area. According to the optimal computing offloading method, part of the computing tasks are offloaded and transmitted to the UAV for processing to achieve more efficient resource utilization and task processing efficiency, and solve the problem that terminals located at the communication edge or blind area cannot directly communicate with the UAV, resulting in difficult computing task offloading or too long local computing delay.

[0019] (1) The present invention constructs an offloading platform for the UAV-assisted computing offloading method. The offloading platform consists of multiple ground users, a UAV equipped with an MEC server, and an operator. The ground users include request users outside the UAV coverage area, request users within the UAV coverage area, and idle users. The UAV leases the spectrum of the operator to communicate with the ground users. Request users outside the UAV coverage area select idle users within their D2D communication range and within the UAV coverage area as relay objects to offload the data volume to the UAV. Each relay object and request user within the UAV request range use the leased spectrum to transmit part of the task data to the UAV for processing in a partial offloading manner.

[0020] (2) The present invention establishes a spectrum trading model between UAVs and operators based on Stackelberg game. Since UAVs need spectrum to communicate with ground users, in order to encourage operators to lease spectrum for the transmission of offloading task data, the present invention constructs a Stackelberg game spectrum trading model between UAVs and operators, where UAVs are the leaders and operators are the followers. Considering the method of the present invention adopted by requesting users, the utility function of the UAV is defined by comprehensively considering the system performance gain obtained by reducing the delay by performing all tasks locally and the payment cost of spectrum leasing. At the same time, the utility function of the operator consists of the revenue obtained from leasing spectrum and the service quality reduction of its own system users due to spectrum leasing.

[0021] (3) The present invention proposes an offloading and resource allocation strategy based on a heuristic algorithm. Under the constraints of the number of spectrums and computing resources, a system delay minimization problem is established. Since this problem is a non-convex optimization problem, it is decomposed into two sub-problems: offloading and relay selection strategy and resource allocation. The particle swarm optimization algorithm and genetic algorithm are respectively used to solve them, and the optimal offloading strategy, computing resource allocation strategy, and spectrum resource allocation strategy are solved through alternating iteration.

[0022] (4) The present invention designs a method for solving the game equilibrium point. According to the analytical results of the UAV optimization problem and the operator optimization problem, a comprehensive optimization problem is generated. Through the derivation of convex optimization theory, the analytical solution of the optimal spectrum leasing quantity of the operator is obtained under a given UAV spectrum leasing unit price. At the same time, the binary search algorithm is used to dynamically iterate to solve the optimal spectrum leasing unit price, so as to maximize the utility of the UAV. The computing offloading method of the present invention can improve the utility of UAVs and operators and reduce the total delay of computing tasks. Description of the Drawings

[0023] Figure 1 is a schematic flowchart of the UAV-assisted computing offloading method in Embodiment 1 of the present invention;

[0024] Figure 2 is a schematic diagram of the offloading platform corresponding to the UAV-assisted computing offloading method in Embodiment 1 of the present invention;

[0025] Figure 3 is a schematic diagram of the convergence situation of the UAV-assisted computing offloading method in Embodiment 1 of the present invention applied to different amounts of computing task data;

[0026] Figure 4 is a schematic diagram of the comparison of the total delay performance of computing tasks under different algorithms in Embodiment 1 of the present invention;

[0027] Figure 5It is a schematic diagram of the comparison results of the utility of drones under different algorithms in Embodiment 1 of the present invention. Detailed implementation manners

[0028] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific features in the embodiments of the present application and the embodiments are detailed descriptions of the technical solution of the present application, rather than limitations on the technical solution of the present application. Without conflict, the technical features in the embodiments of the present application and the embodiments can be combined with each other. The term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this article generally represents an "or" relationship between the associated objects before and after.

[0029] Embodiment 1:

[0030] This embodiment introduces a computing offloading method assisted by drones, as Figure 1 shown, including:

[0031] S1: Construct an offloading platform corresponding to the computing offloading method assisted by drones;

[0032] S2: According to the offloading platform constructed in step S1, construct a spectrum trading model based on the Stackelberg game with the drone as the leader and the operator as the follower;

[0033] S3: According to the spectrum trading model constructed in step S2, analyze the drone optimization problem and the operator optimization problem respectively;

[0034] S4: According to the analysis results of the drone optimization problem and the operator optimization problem in step S3, generate a comprehensive optimization problem, find the game equilibrium point, and obtain the optimal computing offloading method for the requesting users inside and outside the communication coverage of the drone. For the requesting users outside the communication coverage of the drone, through the D2D relay technology, use the idle users within the coverage of the drone as relay objects to offload the computing tasks to the drone for processing.

[0035] The following first introduces the relevant terms involved in the embodiments of the present application:

[0036] Drone: The unmanned aerial vehicle is abbreviated as "drone", and its English abbreviation is "UAV". It is an unmanned aircraft controlled by a radio remote control device and a self - contained program control device, or is completely or intermittently autonomously operated by an on - vehicle computer.

[0037] D2D communication: D2D stands for Device-to-Device, also known as direct terminal communication. D2D communication technology refers to a communication method in which two peer user nodes communicate directly with each other. In a decentralized network composed of D2D communication users, each user node can send and receive signals and has the function of automatic routing (forwarding messages). D2D communication technology can establish a direct connection between end users within the communication range, reduce dependence on the base station, reduce latency, and is applicable to application scenarios such as mobile edge computing and vehicle-to-everything (V2X).

[0038] In the present invention, for a requesting user outside the coverage range of the unmanned aerial vehicle (UAV), through D2D relay technology, an idle user within the coverage range of the UAV is used as a relay object to transmit the task to the UAV for processing, so as to achieve more efficient resource utilization and task processing efficiency. At the same time, considering that the UAV itself has no licensed spectrum resources and needs to lease the spectrum of the operator to communicate with ground users, a trading method between the UAV and the operator based on Stackelberg game is established to help the requesting user perform computing offloading to reduce the task completion time. In addition, to solve the problem of minimizing the total delay of all requesting users in the system under limited spectrum resources and computing resources, a heuristic algorithm: genetic algorithm and particle swarm optimization algorithm are used to solve the optimal computing offloading strategy and resource allocation strategy respectively.

[0039] In step S1, as Figure 2 shown in the schematic diagram of the offloading platform corresponding to this embodiment, the offloading platform includes: a UAV, a requesting user, an idle user, and an operator; the requesting user is a user with a computing task, and the requesting users include the requesting users within the communication coverage range of the UAV and the requesting users outside the communication coverage range of the UAV; the idle user is a user without a computing task, and the idle user is located within the communication range of the UAV and within the D2D communication range of the requesting users outside the communication coverage range of the UAV; the operator has the qualification of licensed spectrum resources and its own users.

[0040] In step S2, constructing a spectrum trading model based on Stackelberg game includes: constructing a UAV utility function and an operator utility function; obtaining a game equilibrium point according to the UAV utility function and the operator utility function; the construction of the UAV utility function includes:

[0041] L1: There exist requesting users, which are represented by the set , where requesting users are located within the communication coverage range of the UAV, requesting users are located outside the communication coverage range of the UAV. The requesting users outside the communication coverage range of the UAV take themselves as the center and detect that there are a total of There are idle users within the communication coverage of the UAV;

[0042] L2: Each requesting user has a latency-sensitive computing task to process, specifically expressed as , where , is the data volume of the computing task, is the number of CPU cycles required per bit for the computing task; is the maximum tolerable latency of the computing task;

[0043] L3: The data volume of the requesting user's computing task includes local data volume and offloaded data volume. The computing tasks of the requesting users are executed in parallel, with local data volume processed locally and offloaded data volume processed by the UAV after being offloaded to the UAV. For ease of processing, it is assumed that the offloaded data volume of a single computing task is fully transmitted before the UAV starts computing, and the computing result is sent immediately after a single task is computed, without waiting for all tasks to be completed. Assuming a quasi-static scenario, the positions of each user and the UAV are relatively fixed during the execution cycle of the computing task. The computing tasks are uploaded to the UAV for processing in a partial offloading manner. The offloading ratio of a single computing task is expressed as , , and the offloading decision of a single computing task is expressed as , ; The requesting users within the coverage of the UAV can directly offload the task to the UAV, and its offloading decision is , when , it means that the requesting user within the coverage of the UAV does not perform the offloading task. When , it means that the requesting user within the coverage of the UAV performs the offloading task; The requesting users outside the coverage of the UAV , , use an idle user within the D2D communication range of the requesting user as a relay object and complete data transmission in a half-duplex DF (decode-and-forward) mode. The offloading decision is , , when , it means that the requesting user does not offload. When , it means that the idle user within the D2D communication range of the requesting user provides relay service for the requesting user , If there are overlapping idle users within the D2D communication range of the requesting user, each idle user serves only one of the requesting users, that is, each idle user provides relay for only one requesting user:

[0044] , (1)

[0045] Among them, is a discriminant function. When the discriminant condition is true, the discriminant function is equal to 1. When the discriminant condition is false, the discriminant function is equal to 0. The channel gain between the requesting user and the idle user is , where is the reference channel gain, is the path loss exponent, is the position distance between the requesting user and the idle user . Each D2D user pair uses different channel resources to transmit task data, that is, there is no channel interference between users. Since the UAV does not have authorized spectrum resources, it must lease spectrum resources from the operator to communicate with ground users and assist ground users in computing offloading to reduce the computing delay of computing tasks. Assuming that the impact of mutual interference on the propagation delay is not considered, the idle user as the relay object and the requesting user adopt FDMA technology to offload tasks to the UAV;

[0046] L4: The computing task is uploaded to the UAV for processing in a partial offloading manner. Therefore, the computing task completion delay of the requesting user is determined by the combination of the local computing delay and the offloading computing delay . The calculation formula is: ; for the local data volume of the computing task, its delay is the local computing delay. For the offloading data volume of the computing task, its offloading computing delay consists of two parts: the computing task transmission delay and the UAV computing delay . For the requesting user , the delay of its local processing part depends on its own computing power , so the local computing time The calculation formula is:

[0047] , (2)

[0048] In the formula, is the number of cycles executed by the local CPU per second;

[0049] L5: For the requesting user , the computing task transmission delay is divided into two stages. The first stage is the transmission delay of D2D user pair communication, and the second stage is the transmission delay of the relay object uploading and offloading data volume to the UAV; for the requesting user , the computing task transmission delay is the transmission delay of uploading and offloading data volume to the UAV; when the idle user is the requesting user provides relay transmission service, the transmission rate of the first stage can be expressed as: , where is the spectrum resource allocated by the UAV to the idle user , is the transmit power of the requesting user , is the channel gain from the requesting user to the idle user , is the Gaussian white noise power. The transmission rate of the second stage idle user and the requesting user under the FDMA technology can be expressed as: where , is the set the spectrum resource of the -th user communicating with the UAV, is the set the transmit power of the -th user, is the set the channel gain from the -th user to the UAV, is the Gaussian white noise power. represents the transmission rate of the idle user communicating with the UAV under the FDMA technology. From the cask effect, in the two-hop D2D relay link, the transmission rate is determined by the worse stage. The transmission rate of the requesting user transmitting to the UAV is expressed as . Therefore, for the requesting user , the computing task transmission delay and the UAV computing delay are calculated as: , (5) , in Equation (6), where is the transmission rate of the requesting user transmitting and offloading data volume to the UAV; Allocate computing resources for the UAV to the requesting user ; Offloading computing delay The calculation formula is: , Since the requesting user processes tasks in parallel, the computing task completion delay from the start of execution to the return of the result for each task The calculation formula is: ; The total delay of requesting users using the UAV-assisted computing offloading method to process computing tasks The calculation formula is:

[0050] L6: To help the requesting user perform computing offloading, minimize the total system delay, and at the same time minimize the payment cost for the UAV to lease the spectrum as much as possible. Considering the obtained system performance gain and the payment cost of spectrum leasing comprehensively, the utility function of the leader is defined as:

[0051] , (7)

[0052] In the formula, is the preset weight coefficient representing the system delay performance gain, is the total delay of all local processing of the computing task, is the unit price of spectrum leasing, is the amount of spectrum leased by the UAV; the first term of the utility function is the system delay gain reduced compared to all local processing of the computing task, and the second term is the payment cost of the UAV. When the amount of leased spectrum is larger, it means that the system reduces the delay more, but the cost will also increase accordingly. To encourage the operator to participate in spectrum trading, assume that the operator has sufficient spectrum, and its quantity is , and the operator has a total of own users. On the premise of ensuring the service requirements of its own users, it can lease out some spectrum resources to obtain additional rewards. Assume that the operator evenly distributes the remaining spectrum resources (excluding the spectrum leased to the UAV) to all its own users. The function of the user service quality (Qos) with respect to the amount of spectrum leased by the UAV . Considering the revenue obtained by the operator and the cost incurred comprehensively, define the utility function of the operator The calculation formula is:

[0053] ,

[0054] In the formula, is the preset weight coefficient for obtaining rental revenue, is the preset weight coefficient for the reduction of user service quality caused by spectrum leasing; is the maximum user service quality, , which is represented by the operator user service quality when all the spectrum is used for its own users without spectrum leasing by the operator; is the function for calculating the operator user service quality according to the amount of spectrum leased by the drone ; In this Stackelberg game framework, the leader and the follower interact in two different stages. In the first stage, the leader drone makes a decision first and gives the unit price of spectrum leasing , and the unit price of spectrum leasing will affect the amount of spectrum leased by the operator; In the second stage, the follower operator decides the amount of spectrum leased based on its own utility maximization goal according to the unit price of spectrum leasing set by the leader drone, and the amount of spectrum will in turn affect the formulation of the unit price of spectrum leasing of the leader drone. The two parties play a game based on the unit price of spectrum leasing and the amount of spectrum , and the game equilibrium point is .

[0055] The optimization problems of the leader drone and the follower operator in step S3:

[0056] Drone optimization problem: For the UAV-assisted MEC system, when the unit price of spectrum leasing is lower, the payment cost of the drone is smaller, but the lower unit price of spectrum leasing will affect the amount of spectrum leased by the follower, and the amount of spectrum leased by the drone is insufficient, making it difficult to reduce the total delay of processing computing tasks by the UAV-assisted computing offloading method. Therefore, in the process of the game, the unit price of spectrum leasing of the UAV-assisted MEC system is affected by the amount of spectrum provided by the operator. To maximize the utility of the drone, the spectrum resource allocation strategy is that all the leased spectrum quantity is used for spectrum bandwidth allocation, and at the same time, the offloading strategy , the computing resource allocation strategy And the unit price of spectrum leasing It will affect the total delay of the UAV-assisted computing offloading method to process computing tasks, that is, it will affect the utility of the UAV. Therefore, the optimization problem of the UAV is as follows:

[0057] , (9)

[0058] In the formula, is the offloading strategy, is the computing resource allocation strategy, is the spectrum resource allocation strategy; the constraint condition C1 is: the offloading ratio of the computing task ranges from 0 to 1; the constraint condition C2 is: each idle user can only provide relay for one requesting user; the constraint condition C3 is: the number of spectrums leased by the UAV is all used for the spectrum resource allocation of the offloaded data volume, where is the set in the th user's spectrum resource for communicating with the UAV; the constraint condition C4 is: the total computing resources allocated by the UAV to the requesting users do not exceed the total computing resources of the UAV ; the constraint condition C5 is: the computing task completion delay of each requesting user does not exceed the maximum tolerable delay of the computing task;

[0059] Operator optimization problem:

[0060] For the operator, the variable to be optimized is the number of spectrums leased , then the operator's optimization problem is:

[0061] , (10)

[0062] In the formula, is the preset minimum user service quality of the operator; the constraint condition C6 is: the value of the operator's utility function is non-negative, indicating that the number of spectrums leased by the operator should ensure that it can obtain effective benefits, otherwise the operator will not provide spectrum leasing services; the constraint condition C7 is: the value of the function of the operator's user service quality is not less than the preset minimum user service quality of the operator, indicating that after the operator leases a part of the number of spectrums, the remaining number of spectrums should meet the minimum user service quality of its original customers. Based on the above leader optimization problem and operator optimization problem for analysis, the solution goal of the optimization problem is to solve the Stackelberg game equilibrium point of the leader and the follower. The analysis of this Stackelberg game equilibrium is as follows: Assume is the optimal spectrum leasing unit price, are respectively the UAVs at the optimal spectrum leasing unit price The optimal offloading strategy, optimal computing resource allocation strategy, and optimal spectrum resource allocation strategy under is the optimal spectrum rental quantity of the operator, then the point is the equilibrium point of the constructed Stackelberg game. Therefore, the solved game equilibrium point should satisfy the following conditions:

[0063]

[0064]

[0065] Equation (11) indicates that when the optimal strategy of the operator is given, can maximize the utility of the leader. Equation (12) indicates that when the optimal strategy of the leader is given, can maximize the utility of the follower.

[0066] To solve the game equilibrium point, the optimization problems of the operator and the UAV are analyzed respectively:

[0067] For the optimization problem of the operator, through the convex optimization theory, the analytical solution of the optimal spectrum rental quantity rented by the operator is derived when the optimal spectrum rental unit price is given.

[0068] For the optimization problem of the UAV, its optimization goal is a multi-variable non-convex optimization problem under the determined spectrum quantity. Therefore, this problem is decomposed into two sub-problems of offloading strategy and resource allocation strategy to be solved respectively, and the alternating iteration algorithm is used to solve the optimal offloading strategy and optimal resource allocation strategy under the corresponding spectrum quantity.

[0069] The analysis of the operator's optimization problem in step S3 includes:

[0070] Y1: When the UAV gives the spectrum rental unit price , the utility function expression of the operator is:

[0071] ; (13)

[0072] Y2: The utility function of the operator with the given spectrum rental unit price is respectively differentiated with respect to to obtain the following function expressions:

[0073] (14)

[0074] (15);

[0075] Y3: Since the second derivative function formula (15) is always less than 0, it indicates that the first derivative function formula (14) is monotonically decreasing within the defined domain, and the original function formula (13) is a strictly convex function; calculating the spectrum quantity for UAV leasing according to the constraint condition C7 indicates that within the defined domain of there exists an optimal spectrum quantity such that the value of the operator's utility function is maximized;

[0076] Y4: Let the first derivative function , and the calculation formula for the optimal spectrum quantity is:

[0077] , (16)

[0078] Since within the defined domain of the variable , the higher the spectrum leasing unit price given by the UAV, the more spectrum quantity the operator is willing to lease. However, to ensure the user service quality of its own users, the spectrum quantity that the operator can lease is limited. At the same time, according to the constraint conditions C6, C7 and the properties of the convex function, the value of the first derivative function must be greater than 0 when ;

[0079] Y5: Let the optimal spectrum quantity within the defined domain of, and the maximum and minimum values of the spectrum leasing unit price obtained by solving are:

[0080] , (17)

[0081] , (18)

[0082] Y6: Based on the maximum and minimum values of the spectrum leasing unit price of the UAV and combined with the constraint condition C7, the analytical solution of the optimal spectrum quantity with respect to the spectrum leasing unit price is obtained:

[0083] . (19)

[0084] The analysis of the UAV optimization problem in step S3 includes:

[0085] W1: Since the optimization objective of the UAV optimization problem under a determined number of spectra is a multi-variable non-convex optimization problem, the UAV optimization problem is decomposed into two sub-problems, namely the offloading strategy and the resource allocation strategy, and solved separately; W2: Analyze the two sub-problems in step W1 respectively to obtain the corresponding solution algorithms; W3: According to the mutual restriction properties of the two sub-problems in step W1, use the alternating iteration algorithm to alternately iterate the solution algorithms corresponding to the two sub-problems until the change amount of the solution of the solution algorithm meets the preset change amount range, and output the optimal offloading strategy and the optimal resource allocation strategy under the determined number of spectra;

[0086] Among them, the alternating iteration includes: J1: Apply the corresponding solution algorithm to the first sub-problem to obtain a preliminary solution; J2: Based on the preliminary solution, apply the corresponding solution algorithm to the second sub-problem to obtain an updated solution; J3: Alternately perform steps J1 and J2 until the change amount of the solutions of the two sub-problems meets the preset change amount range, and the change amount of the solution refers to the difference between the solutions of steps J1 and J2.

[0087] Step W1 includes:

[0088] Substitute the unit price of spectrum leasing into the analytical solution of the optimal number of spectra in step Y6 to solve for the optimal number of spectra, determine the payment cost of the UAV according to the unit price of spectrum leasing and the optimal number of spectra, and transform the optimization objective of the UAV optimization problem, which is to maximize the UAV utility, into optimizing the offloading strategy , calculating the resource allocation strategy and the spectrum resource allocation strategy to minimize the total delay of the computing tasks:

[0089] (20)

[0090] The optimization objective of the P3 problem is a multi-variable non-convex optimization problem. Decompose the P3 problem into an offloading strategy and resource allocation two sub-problems and solve them separately.

[0091] Step W2 includes:

[0092] Analyze the offloading strategy sub-problem and the resource allocation strategy sub-problem respectively to obtain the corresponding solution algorithms; Analyzing the resource allocation strategy sub-problem to obtain the corresponding solution algorithm includes:

[0093] When the offloading ratio of the offloading strategy is determined, determine the local computing delay according to the calculation formula of the local computing delay ​ ; For each requesting user, according to the calculation formula for the delay of task completion: , calculate the delay of task completion It is divided into the following two cases: If , then , indicating that the delay of task completion is determined by the local computing delay , and optimizing the resource allocation strategy will not reduce the delay of task completion ; If , then , indicating that the delay of task completion is determined by the offloading computing delay , and optimizing the resource allocation strategy will reduce the delay of task completion ; Since the local processing delay is a constant and does not need to be optimized. Considering the above, the solution of the resource allocation strategy sub - problem is transformed into:

[0094] (21)

[0095] According to the analysis of the optimization problem P4, the optimization problem P4 is a non - linear optimization problem, and it contains multiple variables and constraints, making it difficult to find the global optimal solution through analytical methods. Therefore, a genetic algorithm, a global optimization algorithm, is used to solve this problem.

[0096] In the initialization stage of the genetic algorithm, a random initial population is generated, and the spectrum resources and computing resources are encoded and converted into the chromosomes of the population individuals. The chromosome length of each population individual is , representing the spectrum quantity and computing resource size of users respectively. According to the objective function of formula (21), calculate the total delay of each individual's computing task in the population as the fitness value for calculation, and introduce the variables that do not meet the constraint conditions as corresponding penalties into the fitness calculation. In the crossover stage, use the tournament selection method to select individuals with better fitness from the population as parents for crossover operations. The selected parents are grouped in pairs, and a dynamic crossover probability is used to determine the feasibility of the crossover operation between two parent individuals. Enrich the diversity of the population by generating a crossover mask for crossover operations on the corresponding genes, while avoiding complete gene coverage. Use to represent the crossover point mask. If , it means to exchange the parental genes at this position. If Add random perturbations and perform boundary control to ensure the validity of the solution. At the same time, calculate the system delay corresponding to each individual. The lower the individual system delay, the higher the fitness of the individual in the natural environment. Adopt the elite retention strategy to add individuals with high fitness to the new population.

[0097] The specific algorithm is shown in Algorithm 1:

[0098]

[0099] Analysis of offloading strategy The corresponding solution algorithms for the sub-problems include:

[0100] In the resource allocation strategy When determining, by jointly optimizing the offloading ratio of each requesting user and the offloading decision Reduce The total delay of the computing tasks processed by the drone-assisted computing offloading method for each requesting user ; The offloading ratio The value is continuous, and the offloading decision is a discrete integer variable, then the offloading strategy The solution of the sub-problem is a mixed integer non-linear optimization problem:

[0101] , (22)

[0102] According to the optimization problem P5 which jointly optimizes continuous variables and discrete variables and includes multiple constraints, an improved particle swarm optimization algorithm is used to solve the optimization problem P5; the improved particle swarm optimization algorithm includes: for the offloading decision Adopt a probability update mechanism to select idle users as relay objects; use a dynamically updated inertia weight;

[0103] The update formula of the inertia weight is: , where is the preset minimum inertia weight, is the preset maximum inertia weight, is the current iteration number, is the preset maximum iteration number.

[0104] Due to the characteristics of the particle swarm optimization algorithm with few parameters and fast convergence, it can well solve the above complex optimization problem. However, the basic particle swarm optimization algorithm is a continuous search algorithm, and its inertia weight and learning factor remain unchanged during the iteration process, and the search ability is poor, so it is difficult to directly use it to solve this problem. Therefore, an improved particle swarm optimization algorithm is adopted to solve P5. In this algorithm, each particle represents a solution. For the offloading ratio , update the position and velocity using the traditional particle swarm update rules:

[0105] (23)

[0106] (24)

[0107] Among them, and respectively represent the updated velocity and position of the th particle at the th iteration. and respectively represent the historical best position of the rd particle and the global historical best position after iterations. is the inertia weight, which is dynamically adjusted with the number of iterations to help the algorithm converge better. are the self-learning factor and the swarm learning factor respectively. The random number adds randomness to the iteration.

[0108] For the offloading decision is a discrete variable, and the probability update mechanism of roulette wheel selection is used to update 's selection probability to adjust the position. For each requesting user , the following update formula is used to adjust the selection probability of the offloading object:

[0109] (25)

[0110] Among them, is the probability that the requesting user selects the relay object at the th iteration. and are the probability of selecting in the particle historical best and the probability of the corresponding of the global best particle respectively. After each update, the probabilities of all possible relay objects are normalized . When the updated probability distribution is given, roulette wheel selection is used to determine the relay object. First, a random number is generated, the cumulative probability is calculated, and then the smallest is found such that the cumulative probability is greater than or equal to , and it is used as the relay object for the requesting user . The particle fitness is calculated by the system delay and the penalty term for not satisfying the constraint conditions.

[0111]

[0112] Step S4 includes:

[0113] As the unit price of spectrum leasing increases, the quantity of leased spectrum increases, and the payment cost of the UAV also increases accordingly. However, the total delay of the computing tasks will be further reduced. Since the quantity of leasable spectrum owned by the operator is limited, when increases to a certain threshold, the utility value of the leader is maximized, thus reaching the game equilibrium.

[0114] S41: According to the analysis results of the UAV optimization problem and the operator optimization problem, obtain the comprehensive optimization problem that combines the UAV optimization problem and the operator optimization problem:

[0115] , (26)

[0116] S42: According to this non-convex comprehensive optimization problem P6, adopt the binary search algorithm to find the game equilibrium point and obtain the optimal computing offloading method for the requesting users inside and outside the UAV communication coverage area to perform computing offloading;

[0117] Step S42 includes: S421: According to the comprehensive optimization problem P6, through the binary search algorithm, compare the UAV utility values calculated respectively at both endpoints and the midpoint of the spectrum leasing unit price interval, perform interval segmentation and iteration to narrow the spectrum leasing unit price interval until the difference between the endpoints of the spectrum leasing unit price interval meets the preset search interval accuracy, and obtain the optimal spectrum leasing unit price interval as the search interval of the binary search method; among them, the initial spectrum leasing unit price interval is from the minimum spectrum leasing unit price to the maximum spectrum leasing unit price ; narrowing the spectrum leasing unit price interval and narrowing the search range of the subsequent binary search algorithm can find the optimal accuracy faster and accelerate the calculation convergence speed. Calculating the UAV utility according to the spectrum leasing unit price includes: adopting the alternating iteration algorithm to alternately iterate the genetic algorithm and the particle swarm algorithm until the change amount of the total delay of the computing tasks processed by the UAV-assisted computing offloading method meets the preset change amount range, and output the optimal offloading strategy and the optimal resource allocation strategy under the determined spectrum quantity; S422: According to the optimal spectrum leasing unit price interval obtained in step S421, through the binary search algorithm, compare the UAV utility values calculated respectively at the midpoint and the midpoint perturbation value of the optimal spectrum leasing unit price interval, perform interval segmentation and iteration to adjust the optimal spectrum leasing unit price interval until the difference between the UAV utility values calculated respectively at the interval midpoint and the midpoint perturbation value meets the preset difference accuracy, and output the interval midpoint that meets the preset difference accuracy as the optimal spectrum leasing unit price ; among them, the midpoint perturbation value is the midpoint of the interval and the preset perturbation ; in this embodiment, the preset perturbation is the midpoint of the interval is one-fourth of the difference between the left and right endpoints of the interval where it is located; the addition of the midpoint perturbation value expands the search range of the solution, avoiding too fast a calculation convergence speed, and thus preventing the algorithm from converging prematurely to a local optimal solution or getting stuck in some undesirable search paths. This method is particularly useful in optimization problems, especially when the objective function is non-monotonic or there are multiple local optimal solutions; S423: Substitute the optimal spectrum rental unit price into the analytical solution of the optimal spectrum quantity to calculate the optimal spectrum quantity , obtain the game equilibrium point of the solution , and output the optimal offloading strategy under the optimal spectrum quantity and the optimal resource allocation strategy ; S424: Obtain the optimal computing offloading method for the requesting users inside and outside the UAV communication coverage area . For the requesting users outside the UAV communication coverage area, through the D2D relay technology, use the idle users within the UAV coverage area as relay objects to offload the computing tasks to the UAV for processing.

[0118] Performing dynamic iterative solution for the optimal based on the binary search algorithm includes: In the initial stage of the algorithm, it is first necessary to determine the search interval of the binary search method. As shown in Algorithm 3, the interval division strategy of the binary search algorithm is used to narrow the optimal unit price interval, and a queue is used to store the intervals to be processed. In each iteration, the queue is processed in the FIFO (First In First Out) order, and the left and right endpoints and the middle endpoint of the first interval are selected to compare the leader UAV utility values. When the interval size reaches a certain accuracy, the interval refinement is ended, and the interval where the optimal unit price is located is returned. To ensure that during the algorithm iteration process, always remains within the search interval , the optimal solution at any accuracy can be solved by the bisection method . In each iteration, the midpoint of the current interval and its perturbation value corresponding to the utility value of the leader and are calculated respectively. By comparing the utility values, when , the right boundary is shrunk; when , the left boundary When the difference in the change of the utility value is within the end condition accuracy, the optimal spectrum lease unit price is output , and the corresponding optimal spectrum rental quantity , the optimal offloading decision and resource allocation .

[0119] To verify the performance of the optimization algorithm proposed in the present invention, MATLAB is used for simulation. It is assumed that the flight altitude of the UAV is 100 m, and 5 requesting users and 10 idle users are randomly distributed within a region centered on the UAV with a coverage radius of 100 m, and 3 requesting users are randomly distributed outside the region. The maximum communication distance of D2D communication is 30 m, and the channel gain is calculated by the path loss model with a path loss exponent of 3. The computing capabilities of each requesting user are randomly generated within the interval , and the computing capability of the UAV is . Other relevant parameters are shown in Table 1 below

[0120] Table 1 Simulation parameters

[0121] Parameter Parameter value User transmit power #timg# [0.1,1]W Requested user computing power #timg# [1, 1.6] GHz Number of CPU cycles required for the task #timg# [1000, 3500] Gigacycle / bit Maximum tolerable delay of the task #timg# [2,4]s Gaussian white noise power #timg# -110 dBm UAV flight altitude #timg# 100m UAV computing power #timg# 10 GHz Number of spectrums owned by the operator #timg# 20 MHz Number of operator users #timg# 12 Minimum quality of service #timg# 0.165 Delay gain weight coefficient #timg# 4 Operator reward coefficient #timg# 0.1 Operator cost coefficient #timg# 30

[0122] When the task data volumes of the requesting users are different, the system convergence of the algorithm of the present invention at its respective optimal spectrum lease unit prices is as Figure 3 shown. From Figure 3 it can be seen that as the number of algorithm iterations increases, the algorithm gradually converges. In addition, the total delay also decreases as the task data volume decreases. To verify the performance of the algorithm proposed in the present invention, it is compared with the following three algorithms

[0123] (1) Random offloading algorithm: The offloading ratio and relay object of the requesting users are randomly generated within a given range, and the average result of 1000 executions is taken. It should be noted that other variables are optimized

[0124] (2) Fixed price algorithm: The spectrum unit price is fixed at 1.4 $ / MHz, and other variables are optimized

[0125] (3) Direct offloading algorithm: This algorithm only considers the computing offloading of users within the UAV-assisted coverage range, without considering users outside the coverage range. It is assumed that the requesting users outside the UAV coverage range perform computing tasks locally, and the tasks of the requesting users within the coverage range are optimized for offloading. At the same time, spectrum trading is not considered, and it is assumed that the spectrum unit price is fixed at 1.4 $ / MHz

[0126] From Figures 4 to 5 the comparison of the performance of the above 4 algorithms, it can be seen that the system delay and UAV utility of all algorithms increase as the task data volume increasesFigure 4 shows a schematic diagram of the comparison of the total delay performance of different algorithms for processing computing tasks. From Figure 4 it can be seen that under different amounts of task data, the algorithm of the present invention can always minimize the total delay of the computing task. At the same time, it can be seen that the random offloading algorithm has the largest total delay, which is because the random offloading algorithm cannot effectively utilize computing resources, and the total delay shows a large increase as the amount of computing task data increases. Compared with the fixed-price algorithm, since the algorithm of the present invention solves the optimal spectrum rental unit price through game theory, the amount of spectrum leased by the operator to the UAV increases, thereby reducing the total delay. Moreover, as the amount of task data increases, the system requires more spectrum resources, and the total delay reduced by the algorithm of the present invention is more obvious. Compared with the direct offloading algorithm, the algorithm of the present invention fully considers that the UAV helps the requesting users outside the coverage area to perform computing offloading, while the computing tasks of these requesting users in the direct offloading algorithm can only be computed locally. Therefore, the algorithm of the present invention can reduce the total delay to a lower level. Figure 5 shows a schematic diagram of the comparison results of the UAV utility under different algorithms. From Figure 5 it can be seen that the algorithm of the present invention has the highest utility, and the random offloading algorithm has the lowest utility. This is because in the random offloading algorithm, the gain brought by the reduction of system delay is less. Compared with the fixed-price algorithm, the algorithm of the present invention can change the spectrum rental unit price, thereby increasing the utility value. Compared with the direct offloading algorithm, the algorithm of the present invention can increase the system performance gain of itself by reducing the delay of the requesting users outside the UAV coverage area and perform a game based on the spectrum rental unit price and the amount of spectrum, thereby bringing an increase in utility.

[0127] Embodiment 2:

[0128] A UAV-assisted computing offloading device includes: an offloading platform module, a trading model module, an analysis module, and a comprehensive solution module; the offloading platform module: is used to construct an offloading platform corresponding to the UAV-assisted computing offloading method; the trading model module: is used to construct a spectrum trading model based on the Stackelberg game with the UAV as the leader and the operator as the follower according to the offloading platform constructed by the offloading platform module; the analysis module: is used to respectively perform UAV optimization problem analysis and operator optimization problem analysis according to the spectrum trading model constructed by the trading model module; the comprehensive solution module: is used to generate a comprehensive optimization problem according to the UAV optimization problem analysis result and the operator optimization problem analysis result of the analysis module, find the game equilibrium point, and obtain the optimal computing offloading method for the requesting users inside and outside the UAV communication coverage area to perform computing offloading. The requesting users outside the UAV communication coverage area use the D2D relay technology and use the idle users within the UAV coverage area as relay objects to offload the computing tasks to the UAV for processing.

[0129] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices produce a means for implementing the functions specified in the process Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks. These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including an instruction means that implements the functions specified in the process Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks. These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in the process Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks. The above is only the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and modifications can be made, and these improvements and modifications should also be regarded as the protection scope of the present invention.

Claims

1. A drone-assisted computation offloading method, characterized in that: include: S1: Construct an unloading platform corresponding to the UAV-assisted computational offloading method; S2: Based on the unloading platform constructed in step S1, a spectrum trading model based on Stackelberg game is constructed with drones as leaders and operators as followers; S3: According to the spectrum trading model constructed in step S2, the optimization problem of drones and the optimization problem of operators are analyzed respectively; S4: Based on the analysis results of the drone optimization problem and the operator optimization problem in step S3, a comprehensive optimization problem is generated, the game equilibrium point is found, and the optimal computational offloading method for the computational offloading of requesting users within and outside the drone communication coverage is obtained. The requesting users outside the drone communication coverage use the D2D relay technology and use the idle users within the drone coverage as relay objects to offload the computational tasks to the drone for processing.

2. The drone-assisted computing offloading method according to claim 1, characterized in that: The unloading platform corresponding to the drone-assisted computing unloading method in step S1 includes: a drone, a requesting user, an idle user, and an operator; The requesting user is a user with a computing task, and the requesting user includes a requesting user within the communication coverage of the drone and a requesting user outside the communication coverage of the drone; The idle user is a user without computing tasks, and the idle user is located within the communication range of the drone and within the D2D communication range of the requesting user outside the communication coverage range of the drone; The operator has authorized spectrum resources and its own users.

3. The drone-assisted computing offloading method according to claim 1 or 2, characterized in that: The spectrum trading model based on Stackelberg game in step S2 includes: Construct drone utility function and operator utility function; According to the drone utility function and the operator utility function, the game equilibrium point is obtained; The construction of the drone utility function includes: L1: Existence requesting users, using the collection It indicates that among them, The requesting user is within the communication coverage of the drone. The requesting users outside the communication coverage of the drone are located outside the communication coverage of the drone. The requesting users outside the communication coverage of the drone take themselves as the center and detect the coexistence within the D2D communication range. Idle users are within the communication coverage of the drone; L2: Each requesting user The computational task is expressed as ,in, , To calculate the data volume of the task, The CPU cycles required for each bit of the calculation task, is the maximum tolerable delay of the computing task; L3: The data volume of the requested user computing task includes the local data volume and the unloaded data volume. The requested user's computing task is executed in parallel to process the local data volume locally and unload it to the drone for unloaded data volume processing. Among them, the offloading ratio of a single computing task is expressed as , , the offloading decision of a single computing task is expressed as , ; Requesting users within the drone's communication coverage area , the uninstall decision is ,when When , it indicates the requesting user within the drone coverage area Do not perform uninstall tasks. When , it indicates the requesting user within the drone coverage area Execute uninstallation tasks; Requesting users outside the drone's communication coverage , , the uninstall decision is , ,when , indicating that the user is requested Do not uninstall. , indicating that the user is requested Idle users within the D2D communication range For requesting users Provide relay services, ; L4: Requesting User Computational task completion delay Calculate the delay locally and offloading computation latency The combination is determined and the calculation formula is: , Local computing latency The calculation formula is: ， In the formula, The number of cycles executed by the local CPU per second; L5: For requesting users ,The calculation task transmission delay is divided into two stages. The first stage is the ,transmission delay of D2D users to communication, and the second stage is the ,transmission delay of the relay object uploading the offloaded data volume to the UAV; For requesting users ,The calculation task transmission delay is the transmission delay of uploading the offloaded data to the UAV; For requesting users , calculate the task transmission delay The calculation formula is: ， In the formula, For requesting users The transmission rate of the amount of offloaded data to the UAV; Drone Calculation Latency The calculation formula is: ， In the formula, Assign the drone to the requesting user computing resources; Offloading computational latency The calculation formula is: ； The total delay of a requesting user to process a computing task using the drone-assisted computing offloading method The calculation formula is: ; L6: Considering the system performance gain and the cost of spectrum leasing, the drone utility function The calculation formula is: ， In the formula, is the preset weight coefficient representing the system delay performance gain, is the total latency of all local processing of the computing task, is the spectrum leasing unit price, the amount of spectrum leased for drones; represents the payment cost of the drone; The utility function of the operator The calculation formula is: ， In the formula, is the preset weight coefficient for obtaining rental income, is the preset weight coefficient for the degradation of user service quality caused by leasing spectrum; To maximize user service quality, , which is represented by the operator’s user service quality when all spectrum is used for its own users when the operator does not lease the spectrum; The amount of spectrum leased for drones Function to calculate the quality of service for operators' users; Function of the operator's user service quality The calculation formula is: ， In the formula, is the amount of spectrum owned by the operator, is the number of users of the operator itself, It is the logarithm operation with base e; The game equilibrium point is .

4. The drone-assisted computing offloading method according to claim 3, characterized in that: The UAV optimization problem in step S3 is: ， In the formula, To uninstall the policy, For computing resource allocation strategy, Allocate strategies for spectrum resources; Constraint C1 is: the offloading ratio of computing tasks The value range is between 0-1; Constraint C2 is: each idle user only provides relay for one requesting user, where is the discriminant function, when the discriminant condition When true, the discriminant function is equal to 1, and when the discriminant condition When it is false, the discriminant function is equal to 0; Constraint C3 is: the amount of spectrum leased by drones All spectrum resources are allocated for offloading data, where: For collection Middle Spectrum resources for users to communicate with drones; Constraint C4 is: The total computing resources allocated by the drone to the requesting user No more than the total computing resources of the drone ; Constraint C5 is: the completion delay of the computing task of each requesting user Do not exceed the maximum tolerable delay of the computing task ; The operator optimization problem in step S3 is: ， In the formula, The preset minimum user service quality of the operator; Constraint C6 is: the operator utility function has non-negative value; Constraint C7 is: the function value of the operator user service quality is not less than the preset operator minimum user service quality.

5. The drone-assisted computing offloading method according to claim 4, characterized in that: The operator optimization problem analysis in step S3 includes: The analytical solution of the optimal number of spectrum leased by the operator when the optimal spectrum lease unit price is given is derived through convex optimization theory; Y1: When the drone is given a spectrum leasing unit price When , the utility function expression of the operator is: ； Y2: The unit price of spectrum leasing The utility functions of the operators are Find the first-order and second-order derivatives and get the following function expression: ， ； Y3: According to the second-order derivative function expression is always less than 0, it means that the first-order derivative function is monotonically decreasing in the local domain, and the original function is a strictly convex function; The number of spectrum leased by drones is calculated based on constraint C7 , indicating that Domain There is an optimal amount of spectrum in memory Maximize the operator's utility function; Y4: Let the first-order derivative function , get the optimal number of spectra The calculation formula is: ； Y5: optimal number of spectra exist Within the definition domain of , the maximum and minimum values ​​of the spectrum leasing unit price are solved as follows: ， ； Y6: Based on the maximum and minimum unit prices of spectrum leases for drones and constraint C7, the optimal number of spectrum is obtained About the unit price of spectrum leasing The analytical solution is: 。 6. The drone-assisted computing offloading method according to claim 5, characterized in that: The analysis of the drone optimization problem in step S3 includes: W1: Based on the fact that the optimization goal of the UAV optimization problem under a certain number of spectrums is a multivariable non-convex optimization problem, the UAV optimization problem is decomposed into two sub-problems: unloading strategy and resource allocation strategy, and the two sub-problems are solved separately; W2: Analyze the two sub-problems of step W1 separately and obtain the corresponding solution algorithms; W3: According to the mutual constraint properties of the two sub-problems in step W1, an alternating iterative algorithm is used to alternately iterate the solution algorithms corresponding to the two sub-problems until the change amount of the solution of the solution algorithm meets the preset change amount range, and the optimal unloading strategy and optimal resource allocation strategy under the determined spectrum quantity are output; The alternating iteration includes: J1: Apply the corresponding solution algorithm to the first sub-problem and obtain a preliminary solution; J2: Based on the preliminary solution, apply the corresponding solution algorithm to the second sub-problem to obtain an updated solution; J3: Alternately perform steps J1 and J2 until the variation of the solutions of the two sub-problems meets the preset variation range, where the variation of the solutions refers to the difference between the solutions of steps J1 and J2.

7. The drone-assisted computing offloading method according to claim 6, characterized in that: Step W1 includes: Spectrum leasing unit price Substitute the optimal spectrum quantity in step Y6 The analytical solution of the UAV optimization problem is obtained to obtain the optimal number of spectrums. The payment cost of the UAV is determined according to the spectrum leasing unit price and the optimal number of spectrums. The optimization goal of the UAV optimization problem, which is to maximize the utility of the UAV, is converted into an optimized unloading strategy. , computing resource allocation strategy and spectrum resource allocation strategy The total delay of the computing task Minimization problem: ， The optimization objective of the P3 problem is a multivariable non-convex optimization problem. The P3 problem is decomposed into an unloading strategy. and resource allocation strategies The two sub-problems are solved separately.

8. The drone-assisted computing offloading method according to claim 7, characterized in that: Step W2 includes: Step W2 includes: respectively analyzing the uninstallation strategy Subproblems and resource allocation strategies The sub-problems get corresponding solution algorithms; Parsing resource allocation strategy The corresponding solution algorithms for the sub-problems include: When uninstalling the policy When the offloading ratio is determined, the latency is calculated based on the local The calculation formula determines the local calculation delay ; For each requesting user, the calculation formula for the task completion delay is: , calculate the task completion delay There are two cases: like ,but , indicating the delay in completing the computing task Calculate the delay locally Determine and optimize resource allocation strategies Will not reduce the delay in completing computing tasks ; like ,but , indicating the delay in completing the computing task Delay calculation by offloading Determine and optimize resource allocation strategies It will reduce the delay in completing the computing task ; Due to local processing delay is a constant and does not require optimization, then the resource allocation strategy The solution of the sub-problem is transformed into: ， Since the optimization problem P4 is a nonlinear optimization problem and contains multiple variables and constraints, a genetic algorithm is used to solve the optimization problem P4; Parsing uninstall policy The corresponding solution algorithms for the sub-problems include: Uninstall strategy Including uninstall ratio and uninstall decisions ; In resource allocation strategy When determined, the uninstall ratio of each requesting user is optimized by joint optimization and uninstall decisions reduce The total delay of a requesting user to process a computing task using the drone-assisted computing offloading method ; Uninstall ratio The value is continuous, and the uninstall decision If the value is discrete, then the uninstall strategy The solution to the subproblem is a mixed integer nonlinear optimization problem: ， Since the optimization problem P5 is a joint optimization of continuous variables and discrete variables and contains multiple constraints, an improved particle swarm algorithm is used to solve the optimization problem P5; The improved particle swarm algorithm comprises: Uninstall Decision A probability update mechanism is used to select idle users as relay objects; Use dynamically updated inertia weights; The updating formula of the inertia weight is: ， In the formula, is the preset minimum inertia weight, is the preset maximum inertia weight, is the current iteration number, is the preset maximum number of iterations.

9. The drone-assisted computing offloading method according to claim 8, characterized in that: Step S4 includes: S41: According to the analytical results of the UAV optimization problem and the operator optimization problem, the comprehensive optimization problem of the integrated UAV optimization problem and the operator optimization problem is obtained: ， S42: According to the comprehensive optimization problem P6, a binary search algorithm is used to find the game equilibrium point and obtain the optimal computation offloading method for the computation offloading of the requesting users within and outside the UAV communication coverage area; Step S42 includes: S421: According to the comprehensive optimization problem P6, compare the drone utility calculated from the two end points and the midpoint of the spectrum lease price interval, perform interval segmentation iteration to narrow the spectrum lease price interval, until the difference between the end points of the spectrum lease price interval meets the preset search interval accuracy, and obtain the optimal spectrum lease price interval As the search interval for the binary search method; Among them, the initial spectrum lease price range is the minimum spectrum lease price To the maximum spectrum leasing price ; Calculation of drone utility based on spectrum leasing unit price includes: The genetic algorithm and the particle swarm algorithm are iterated alternately by using an alternating iterative algorithm until the change in the total delay of the computing task processed by the UAV-assisted computing offloading method meets the preset change range, and the optimal offloading strategy and the optimal resource allocation strategy under the determined number of spectrums are output; S422: Based on the optimal spectrum lease unit price interval obtained in step S421, compare the midpoint of the optimal spectrum lease unit price interval by using a binary search algorithm. and midpoint disturbance value The calculated drone utility sizes are divided into intervals and the optimal spectrum leasing unit price interval is iteratively adjusted until the difference between the drone utility sizes calculated by the midpoint of the interval and the midpoint disturbance value meets the preset difference accuracy, and the midpoint of the interval that meets the preset difference accuracy is output as the optimal spectrum leasing unit price. ; Among them, the midpoint disturbance value is the midpoint of the interval With preset disturbance of and; S423: Optimal spectrum leasing unit price Substitute the optimal spectrum number analytical solution to calculate the optimal spectrum number , and obtain the game equilibrium point to be solved , output the optimal number of spectra The optimal unloading strategy under and optimal resource allocation strategy ; S424: Obtaining the optimal computation offloading method for the users requesting computation offloading within and outside the UAV communication coverage area ,The requesting users outside the communication coverage of the UAV use the D2D relay technology and use the idle users within the UAV ,coverage as relay objects to offload the computing tasks to the UAV for processing.

10. A drone-assisted computing offloading device, characterized in that: include: Uninstall the platform module, transaction model module, analysis module and comprehensive solution module; The unloading platform module is used to construct an unloading platform corresponding to the drone-assisted computing unloading method; The trading model module is used to construct a spectrum trading model based on Stackelberg game with drones as leaders and operators as followers according to the unloading platform constructed by the unloading platform module; The analysis module is used to analyze the optimization problem of drones and the optimization problem of operators according to the spectrum transaction model constructed by the transaction model module; The comprehensive solution module is used to generate a comprehensive optimization problem according to the analysis results of the drone optimization problem and the operator optimization problem of the analysis module, find the game equilibrium point, and obtain the optimal computing offloading method for computing offloading of requesting users within and outside the drone communication coverage. The requesting users outside the drone communication coverage use the D2D relay technology and use the idle users within the drone coverage as relay objects to offload the computing tasks to the drone for processing.

Images