Multi-user single-unmanned-aerial-vehicle-assisted edge calculation distribution method and long-time edge calculation distribution method based on Stackelberg game

Abstract

The invention provides a multi-user single unmanned aerial vehicle assisted edge calculation distribution method and a long-time edge calculation distribution method based on a Stackelberg game. The multi-user single unmanned aerial vehicle assisted edge calculation distribution method comprises the following steps: constructing an edge calculation distribution network consisting of M user terminals and a single unmanned aerial vehicle serving as an edge calculation server; assuming that the flight period of the unmanned aerial vehicle is T, and dividing the flight period T into N discrete frames; any discrete frame n is divided into M time slots, and the user terminal sends data to be processed to the unmanned aerial vehicle in the corresponding time slot based on the time division multiple access technology; based on the edge calculation distribution network, a Stackelberg game model is constructed, and in the Stackelberg game model, the user terminal is used as a follower, and the unmanned aerial vehicle is used as a leader; solving the Stackelberg game model to obtain the optimal pricing of the unit data processing capability; and each user terminal determines the number of purchased data processing capabilities according to the optimal pricing of the unit data processing capability.

Description

Technical Field

[0001] The present invention belongs to the technical field of edge computing allocation in wireless communications, and specifically relates to a multi-user single-drone assisted edge computing allocation method and a long-term edge computing allocation method based on Stackelberg game. Background Art

[0002] In recent years, with the explosive growth of the Internet of Things and mobile smart devices, traditional cloud computing models are facing problems such as transmission delays and bandwidth pressure. In order to solve these problems, edge computing distribution technology has emerged. It deploys computing resources at the edge of the network to achieve localized data processing and analysis, thereby reducing latency and improving service quality. As a flexible and efficient aerial platform, drones provide new possibilities for edge computing distribution. By mounting edge computing servers on drones, mobile and distributed computing services can be realized, providing more convenient and efficient computing support for ground users. This drone-assisted edge computing distribution technology not only expands the application scenarios of edge computing distribution, but also improves its flexibility and scalability.

[0003] Drone-assisted edge computing allocation technology has shown great application potential in many fields, such as intelligent transportation, environmental monitoring, disaster relief, etc. In the future, with the continuous advancement of technology and the continuous expansion of application scenarios, this technology will bring more convenience and innovation to people's lives and work. However, due to the limited computing resources of drone CPUs, how much edge computing power of drones should be allocated to different users in multi-user scenarios is a key issue that needs to be solved. Summary of the invention

[0004] Purpose of the invention: In order to solve the problem of how much edge computing capacity of drones should be allocated to different users in a multi-user scenario, the present invention proposes a multi-user single-drone assisted edge computing allocation method and a long-term edge computing allocation method based on Stackelberg game.

[0005] Technical solution: A multi-user single drone-assisted edge computing allocation method based on Stackelberg game, including: Construct a multi-user single-drone assisted edge computing distribution network; the multi-user single-drone assisted edge computing distribution network consists of M user terminals and a single drone equipped with an edge computing server; assume that the flight period of the drone is T, and divide the flight period T into N discrete frames; any discrete frame n, , and then divided into M time slots , For the length of each discrete frame, based on the time division multiple access technology, the user terminal sends the data to be processed to the drone in the corresponding time slot, and the drone processes the data; is the time allocation coefficient, ; Based on the multi-user single drone-assisted edge computing distribution network, a Stackelberg game model is constructed. In the Stackelberg game model, the user terminal acts as a follower and the drone acts as a leader. For each user terminal, in each discrete frame, the objective function is to maximize its own utility function. For the drone, in each discrete frame, the objective function is to maximize the total cost paid by all followers to purchase its data processing capacity. For each user terminal, its own utility function is the difference between the data processing capacity of the drone and the price paid to purchase the data processing capacity. In each discrete frame, the optimal pricing of the unit data processing capacity in the discrete frame is obtained by solving the Stackelberg game model; Each user terminal determines the amount of data processing capacity purchased in the discrete frame according to the optimal pricing of the unit data processing capacity.

[0006] Furthermore, for each user terminal, in each discrete frame, maximizing its own utility function is used as its objective function, which is expressed as: ; Where B is the bandwidth, is the transmit power value of user terminal m in discrete frame n, is the receiver noise of the drone, represents the channel gain from user terminal m to the UAV at discrete frame n, Indicates the price of the unit data processing capacity sold by the drone to the user terminal.

[0007] Furthermore, the constraints of the objective function of each user terminal include: ; ; in, represents the minimum amount of data processing transferred to the UAV at discrete frame n.

[0008] Furthermore, for the UAV, in each discrete frame, the objective function is to maximize the total cost paid by all followers to purchase its data processing capabilities, which can be expressed as: ; In the formula, Indicates the price of the unit data processing capacity sold by the drone to the user terminal.

[0009] Furthermore, the constraints of the objective function of the UAV include: The data processed on a discrete frame n is less than or equal to the maximum amount of computation on the corresponding frame, expressed as: ; as well as ; In the formula, F represents the data processing capacity of the drone per unit time, B is the bandwidth, is the transmit power value of user terminal m in discrete frame n, is the receiver noise of the drone, It represents the channel gain from user terminal m to UAV in discrete frame n.

[0010] The present invention discloses a long-term edge computing allocation method assisted by a multi-user single drone based on Stackelberg game, comprising: Construct a multi-user single-drone assisted edge computing distribution network; the multi-user single-drone assisted edge computing distribution network consists of M user terminals and a single drone equipped with an edge computing server; assume that the flight period of the drone is T, and divide the flight period T into N discrete frames; any discrete frame n, , and then divided into M time slots , For the length of each discrete frame, based on the time division multiple access technology, the user terminal sends the data to be processed to the drone in the corresponding time slot, and the drone processes the data; is the time allocation coefficient, ; Based on the multi-user single drone-assisted edge computing distribution network, a Stackelberg game model is constructed. In the Stackelberg game model, the user terminal acts as a follower and the drone acts as a leader. For each user terminal, during the entire flight cycle, maximizing its own utility function is used as its objective function. For the drone, during the entire flight cycle, maximizing the total cost paid by all followers to purchase its data processing capabilities is used as its objective function. For each user terminal, the average data processing capability of the drone purchased during the entire flight cycle is used as the income, and the difference between the income of the user terminal and the cost paid for purchasing the average data processing capability is used as its own utility function. By solving the Stackelberg game model, the optimal pricing of unit data processing capacity in the entire flight cycle is obtained; Each user terminal determines the amount of data processing capacity purchased during the entire flight cycle based on the optimal pricing of the unit data processing capacity during the entire flight cycle.

[0011] Furthermore, for each user terminal, in the entire flight cycle, maximizing its own utility function is used as its objective function, which is expressed as: ; Where B is the bandwidth, is the transmit power value of user terminal m in discrete frame n, is the receiver noise of the drone, represents the channel gain from user terminal m to the UAV at discrete frame n, Indicates the price of the unit data processing capacity sold by the drone to the user terminal.

[0012] Furthermore, the constraints of the objective function of each user terminal include: ; and ; in, represents the minimum amount of data processing transferred to the UAV at discrete frame n.

[0013] Furthermore, the objective function of the UAV is expressed as: ; In the formula, Indicates the price of the unit data processing capacity sold by the drone to the user terminal.

[0014] Furthermore, the constraints of the objective function of the UAV are: ; as well as ; In the formula, F represents the data processing capacity of the drone per unit time, B is the bandwidth, is the transmit power value of user terminal m in discrete frame n, is the receiver noise of the drone, It represents the channel gain from user terminal m to UAV in discrete frame n.

[0015] Beneficial effect: Compared with the prior art, the method of the present invention applies the competition mechanism to the UAV-assisted edge computing allocation problem, and realizes multi-user single UAV-assisted edge computing allocation through resource game among multiple users. The method of the present invention physically conforms to the scenarios of real applications and can be effectively applied to engineering practice. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic diagram of a multi-user single drone-assisted edge computing distribution network model constructed for Example 1; Figure 2 This is a schematic diagram of a multi-user single drone assisted edge computing allocation method based on Stackelberg game proposed in Example 1; Figure 3 This is a schematic diagram of a long-term edge computing allocation method assisted by a multi-user single drone based on Stackelberg game proposed in Example 2. DETAILED DESCRIPTION

[0017] The technical solution of the present invention is now further described in conjunction with the accompanying drawings and embodiments.

[0018] Embodiment 1: This embodiment proposes a multi-user single drone assisted edge computing allocation method based on Stackelberg game, such as Figure 2 As shown, it mainly includes the following steps:

[0019] Step 1: Construct a multi-user single drone-assisted edge computing distribution network model, such as Figure 1 As shown in the edge computing distribution network model, it is assumed that there are M user terminals, defined as ,Due to the limited computing resources and computing power of user terminals, ,a single drone equipped with an edge computing server is deployed.,The user terminal can offload the computing load to the drone equipped with a ,mobile edge computing device (MEC) by sending data to the drone, and then perform ,computation on the drone.

[0020] Define the horizontal coordinate of user terminal m as , Since the UAV performs directional flight in the air, the channel gain between the UAV and each user terminal will change in different time periods. Assuming that the total flight time of the UAV is T, the flight cycle can also be considered as T, and the flight cycle T is discretely divided into N discrete frames , the length of each discrete frame is .if If is small enough, we can assume that the position of the drone is constant in each discrete frame, denoted by , , represents the position information of the UAV at discrete frame n, and Respectively represent the position information of the drone on the X-axis and Y-axis at discrete frame n.

[0021] Assume that the cruising altitude of the drone is ,make represents the channel gain at a distance of 1 meter from the drone. Therefore, in discrete frame n, the channel gain from user terminal m to the drone is expressed as: ; Since the user terminal needs to send the data to be processed to the drone, in order to avoid interference between the user terminals during the task offloading process, the time division multiple access (TDMA) technology is used. , and then divided into M time slots. In the corresponding time slot, the user terminal m sends data to the drone, where is the time allocation coefficient of TDMA, and ; as well as ; Step 2: Based on the multi-user single drone-assisted edge computing distribution network model, the data transmission rate sent by user terminal m to the drone in discrete frame n is expressed as: ; Where B is the bandwidth, is the transmit power value of user terminal m in discrete frame n, which is known. is the receiver noise of the drone.

[0022] After receiving the data from each user terminal, the drone performs edge computing allocation on the data according to the drone's data processing capacity F (in bit / s) per unit time. Therefore, the amount of data transmitted to the drone from all user terminals in discrete frame n should not be greater than the total amount of data it can process, that is, ; Since the data transmitted from each user terminal to the UAV in discrete frame n should meet the minimum data processing amount that needs to be transferred to the UAV, we have: ; in, It represents the minimum amount of data processing transferred to the drone at discrete frame n, in bits.

[0023] In this embodiment, considering that the data processing capacity of the drone is limited per unit time, when multiple user terminals have different computing requirements, how the drone reasonably allocates these computing requirements is a key issue that needs to be solved urgently.

[0024] Considering that all user terminals want to purchase the data processing capabilities of drones for their data calculations, the more data processing capabilities of drones are allocated to a user terminal, the faster it will obtain the calculation results. Therefore, user terminals with different edge computing allocation requirements are in a competitive relationship in purchasing the limited data processing capabilities of drones. Steinberg game is a pure strategy non-cooperative sequential game model in economics. The participants in the game can be divided into leaders and followers according to the priority of actions and the completeness of information. Followers only have partial information and act first; while leaders have all the information of followers and act later. Among them, leaders need to consider the optimal response of followers when setting game strategies, and followers give their own optimal resource purchase size according to the optimal decision of the leader.

[0025] In this embodiment, since the drone has global information and limited computing resources, the drone is used as the leader of the game model, and the user terminal that only has partial information acts as a follower.

[0026] In this embodiment, the purpose of each follower purchasing the data processing capacity of the drone in the game mechanism is to increase the calculation speed of its data.

[0027] The cost paid by each follower is defined as the cost generated by the time allocation coefficient assigned to it by the drone when using TDMA technology, that is, the cost function is expressed as ,in It represents the price of the unit data processing capacity sold by the drone to the user terminal. Therefore, the utility function of the user terminal as a follower consists of two parts: the benefit function and the cost function. Among them, the data processing capacity of the purchased drone is used as the benefit function. Therefore, the utility function of the user terminal is the difference between the benefit function and the cost function, which is expressed as: ; As the data processing capacity purchased by the user terminal increases, the benefits it obtains also increase, but its cost function will also increase. Therefore, for the follower in the game model, its goal is to obtain the highest benefits at the lowest cost in the game behavior, that is, to maximize its own utility function. Therefore, at discrete frame n, the objective function of the user terminal is expressed as: ; And the constraints are: and .in, The data processing capacity of the drone purchased by end user m at discrete frame n.

[0028] The leader in the game, i.e., the drone, sells limited data processing capabilities to multiple competing end users for their data calculations, so its objective function is defined as maximizing the total cost paid by all user terminals for purchasing data processing capabilities, expressed as: ; Since the data processing capability of the drone is limited, the constraint of the leader in the Stackelberg game relationship is that the data processed by the edge computing server on the discrete frame n should be less than or equal to the maximum computing amount of the edge server CPU on the corresponding frame, that is: ; as well as ; The objective functions of the follower and the leader together constitute the Stackelberg game. The two parties in the game can obtain the final Stackelberg equilibrium by playing games according to certain rules, that is, the optimal pricing of unit data processing capacity. Then, each user terminal determines the quantity of computing resources to purchase based on the optimal pricing per unit of data processing capacity.

[0029] Example 2: It can be seen from the Stackelberg game proposed in Example 1 that each Stackelberg game is conducted for a discrete frame. If the total flight period T of the drone is too long or the number of discrete frames is too large, the network will continuously update the location information of the drone on different discrete frames, so that the algorithm will continuously perform repeated calculations on the edge computing resource allocation problem, resulting in a waste of resources and being detrimental to the robustness of the system. Therefore, in order to reduce the impact on the robustness of the system caused by the need for the edge computing allocation method proposed in Example 1 to bid on the limited data processing capabilities of multi-user terminals on each discrete frame, this embodiment proposes a long-term edge computing allocation method assisted by a multi-user single drone based on the Stackelberg game.

[0030] In this embodiment, for user terminal m, the average data processing capacity of the purchased drone during the entire flight cycle is taken as the revenue, and its objective function is expressed as: ; The corresponding constraints are: ; and ; in: is the cost function of user terminal m in the entire flight period T, is the amount of data sent by user terminal m to the UAV for processing during the entire flight period T, represents the minimum amount of data processing transferred to the UAV at discrete frame n.

[0031] The objective function of the UAV in the game relationship is defined as maximizing the total cost paid by all user terminals to purchase data processing capabilities during the entire UAV flight period T. Therefore, the objective function of the UAV is expressed as: ; The corresponding constraints are: ; as well as ; The objective functions of the follower and the leader together constitute the Stackelberg game. The two parties in the game can obtain the final Stackelberg equilibrium by playing games according to certain rules, that is, the optimal pricing of unit data processing capacity. , each user terminal determines the amount of data processing capacity to purchase based on the optimal pricing of the unit data processing capacity.

[0032] See also Figure 3 ,The allocation method proposed in this embodiment only executes the game once in the entire flight period T of the UAV.

Claims

1. A multi-user single drone-assisted edge computing allocation method based on Stackelberg game, characterized by: include: Building a multi-user single drone-assisted edge computing distribution network; The multi-user single drone-assisted edge computing distribution network consists of M user terminals and a single drone equipped with an edge computing server; assuming that the flight period of the drone is T, the flight period T is divided into N discrete frames; For any discrete frame n, , and then divided into M time slots , For the length of each discrete frame, based on the time division multiple access technology, the user terminal sends the data to be processed to the drone in the corresponding time slot, and the drone processes the data; is the time allocation coefficient, ; Based on the multi-user single drone-assisted edge computing distribution network, a Stackelberg game model is constructed. In the Stackelberg game model, the user terminal acts as a follower and the drone acts as a leader. For each user terminal, in each discrete frame, the objective function is to maximize its own utility function. For the drone, in each discrete frame, the objective function is to maximize the total cost paid by all followers to purchase its data processing capacity. For each user terminal, its own utility function is the difference between the data processing capacity of the drone and the price paid to purchase the data processing capacity. In each discrete frame, the optimal pricing of the unit data processing capacity in the discrete frame is obtained by solving the Stackelberg game model; Each user terminal determines the amount of data processing capacity purchased in the discrete frame according to the optimal pricing of the unit data processing capacity.

2. The edge computing allocation method based on Stackelberg game for multi-user single drone assistance according to claim 1 is characterized by: For each user terminal, in each discrete frame, the objective function is to maximize its own utility function, which is expressed as: ； Where B is the bandwidth, is the transmit power value of user terminal m in discrete frame n, is the receiver noise of the drone, represents the channel gain from user terminal m to the UAV at discrete frame n, Indicates the price of the unit data processing capacity sold by the drone to the user terminal.

3. The edge computing allocation method based on Stackelberg game for multi-user single drone assistance according to claim 2 is characterized by: The constraints of the objective function of each user terminal include: ； ； in, represents the minimum amount of data processing transferred to the UAV at discrete frame n.

4. The edge computing allocation method based on Stackelberg game and assisted by multi-user single drone according to claim 1 is characterized by: For the UAV, in each discrete frame, the objective function is to maximize the total cost paid by all followers to purchase its data processing capabilities, which can be expressed as: ； In the formula, Indicates the price of the unit data processing capacity sold by the drone to the user terminal.

5. The method for allocating edge computing resources assisted by a multi-user single drone based on Stackelberg game according to claim 4, characterized in that: The constraints of the UAV's objective function include: The data processed on a discrete frame n is less than or equal to the maximum amount of computation on the corresponding frame, expressed as: ； as well as ； In the formula, F represents the data processing capacity of the drone per unit time, B is the bandwidth, is the transmit power value of user terminal m in discrete frame n, is the receiver noise of the drone, It represents the channel gain from user terminal m to UAV in discrete frame n.

6. A multi-user single drone assisted long-term edge computing allocation method based on Stackelberg game, characterized by: include: Building a multi-user single drone-assisted edge computing distribution network; The multi-user single drone-assisted edge computing distribution network consists of M user terminals and a single drone equipped with an edge computing server; assuming that the flight period of the drone is T, the flight period T is divided into N discrete frames; For any discrete frame n, , and then divided into M time slots , For the length of each discrete frame, based on the time division multiple access technology, the user terminal sends the data to be processed to the drone in the corresponding time slot, and the drone processes the data; is the time allocation coefficient, ; Based on the multi-user single drone-assisted edge computing distribution network, a Stackelberg game model is constructed. In the Stackelberg game model, the user terminal acts as a follower and the drone acts as a leader. For each user terminal, during the entire flight cycle, maximizing its own utility function is used as its objective function. For the drone, during the entire flight cycle, maximizing the total cost paid by all followers to purchase its data processing capabilities is used as its objective function. For each user terminal, the average data processing capability of the drone purchased during the entire flight cycle is used as the income, and the difference between the income of the user terminal and the cost paid for purchasing the average data processing capability is used as its own utility function. By solving the Stackelberg game model, the optimal pricing of unit data processing capacity in the entire flight cycle is obtained; Each user terminal determines the amount of data processing capacity purchased during the entire flight cycle based on the optimal pricing of the unit data processing capacity during the entire flight cycle.

7. The long-term edge computing allocation method assisted by a multi-user single drone based on Stackelberg game according to claim 6 is characterized by: For each user terminal, in the entire flight cycle, the objective function is to maximize its own utility function, which is expressed as: ； Where B is the bandwidth, is the transmit power value of user terminal m in discrete frame n, is the receiver noise of the drone, represents the channel gain from user terminal m to the UAV at discrete frame n, Indicates the price of the unit data processing capacity sold by the drone to the user terminal.

8. The long-term edge computing allocation method assisted by a multi-user single drone based on Stackelberg game according to claim 7 is characterized by: The constraints of the objective function of each user terminal include: ； and ； in, represents the minimum amount of data processing transferred to the UAV at discrete frame n.

9. The long-term edge computing allocation method assisted by a multi-user single drone based on Stackelberg game according to claim 6 is characterized by: The objective function of the UAV is expressed as: ； In the formula, Indicates the price of the unit data processing capacity sold by the drone to the user terminal.

10. The long-term edge computing allocation method assisted by a multi-user single drone based on Stackelberg game according to claim 9, characterized in that: The constraints of the objective function of the UAV are: ； as well as ； In the formula, F represents the data processing capacity of the drone per unit time, B is the bandwidth, is the transmit power value of user terminal m in discrete frame n, is the receiver noise of the drone, It represents the channel gain from user terminal m to UAV in discrete frame n.

Images