A UAV computing offloading method for mobile edge computing

Abstract

The present invention provides a method for unloading computing of unmanned aerial vehicles for mobile edge computing. First, the computing power of the terminal device is measured by analyzing the status information of all terminal devices, including the calculable battery capacity and the CPU frequency of the device, and then the computing area is divided according to the location and computing power of the terminal device. The unmanned aerial vehicle selects a suitable area to perform the computing unloading task. Secondly, each unmanned aerial vehicle carries the computing unloading task of an area, and each terminal device can offload the computing task to the unmanned aerial vehicle or have the unmanned aerial vehicle offload it to a nearby base station for execution. The unloading decision is made for the terminal device, the unmanned aerial vehicle, and the base station through the partial unloading method. Finally, the tasks on the unmanned aerial vehicle are prioritized by deadline using a dual queue, so that real-time tasks are executed first. The method of the present invention reduces the total task execution time and greatly improves the task completion rate.

Description

Technical Field

[0001] The present invention relates to the field of mobile edge computing and the Internet of Things, and in particular to a method for segmenting computing offloading tasks of unmanned aerial vehicles. Background Art

[0002] In recent years, the rapid development of mobile Internet and the Internet of Things has led to an explosive growth in computing tasks, and a large number of IoT terminal devices such as sensors and cameras are used in various industries. However, directly passing tasks to cloud computing has greatly increased the burden on the network and caused a long network delay. At the same time, in order to solve the challenges of terminal devices being limited by computing resources, storage, and energy consumption, the concept of mobile edge computing has been proposed. Mobile edge servers are close to users, which can reduce delays, improve network operation efficiency, and improve terminal user experience.

[0003] Mobile edge computing technology has become an effective solution for computationally intensive and delay-sensitive tasks. UAV-assisted mobile edge computing has been widely studied. UAVs can help terminal devices complete complex computing tasks. Mobile edge servers are close to terminal users and provide computing services to users, which can reduce the computing delay of terminal devices and improve service quality. Mobile edge computing improves the computing power of terminal devices in an energy-saving and efficient way, which has attracted widespread attention from the industry and academia. UAVs have been widely studied in recent years because they can provide high-quality services and communication coverage. At present, the research on UAV-assisted mobile edge computing mainly focuses on saving system energy consumption and reducing system latency. From the perspective of energy consumption, existing research includes reducing the computing energy consumption and communication energy consumption of terminal devices. In order to save the energy consumption of terminal devices, some studies use UAVs to install energy transmitters to provide computing energy for terminal devices. From the perspective of reducing latency, it is mainly to reduce the execution time of tasks to improve service quality. Summary of the invention

[0004] Technical issues to be solved

[0005] In order to avoid the shortcomings of the prior art, the present invention proposes a UAV computing offloading method for mobile edge computing.

[0006] Technical Solution

[0007] A method for unloading computing of unmanned aerial vehicles for mobile edge computing, characterized by the following steps:

[0008] Step 1: Divide the entire mission area into M clusters based on the distance and the number of drones based on the K-means method, expressed as Cluster = {δ1,…,δ M}, and then measure the computing power of each region according to the battery capacity and CPU frequency of each cluster terminal device, which is defined as:

[0009]

[0010] Among them, u represents the cluster C i Each terminal device in u Indicates the CPU frequency of device u, that is, the number of bits that can be calculated per second, E u It is expressed as the battery capacity of device u, η1 and η2 are constant coefficients, is cluster C i The number of terminal devices, the computing power of the drone is based on Matching areas with weak computing power;

[0011] Step 2: Abstractly define the computing tasks of all terminal devices as

[0012] Q i =(L i ,T i )

[0013] Among them, i represents the serial number of the terminal device, L i Indicates the size of the task to be calculated, T i Indicates the deadline of the calculation task. If the task is not completed on the deadline, it means the task execution has failed.

[0014] Step 3: For the computing tasks of each device on the drone, use dual queues to sort the computing tasks. Real-time tasks enter queue Q1, and other tasks enter queue Q2. Each time a task is taken out from Q1 for execution, when queue Q1 is empty, Q2 tasks enter Q1, and Q2 tasks are sorted according to deadlines in each time slot, so that high-priority tasks are at the head of the queue. In order to avoid task starvation, tasks in Q2 are upgraded by one level at regular intervals.

[0015] Step 4: The computing tasks are divided into local computing, UAV-assisted computing, and base station-assisted computing, where {0≤ρ i1 ≤1,i∈N} represents the task ratio of the terminal device to the drone, N is the number of terminal devices in the entire mission area. When the drone is within the transmission range of the base station, the drone will offload the computing task to the base station, and the offloading ratio is {0≤ρ i2 ≤1,i∈N};

[0016] For local calculations, the local calculation time is expressed as:

[0017]

[0018] Among them, ilocal Indicates the computing power of the local device, that is, the number of bits calculated per second; s i It represents the number of CPU cycles required by the terminal device to execute each bit; the local computing energy consumption is expressed as:

[0019]

[0020] for γ represents the coefficient determined by the chip structure determined by actual measurements;

[0021] For drone computing, a set of devices connected to the drone is represented as:

[0022] N i ={i:i∈N,g i =j}

[0023] Among them, g i =j means device i is connected to drone j;

[0024] The bandwidth resource of the drone is represented by B j ; The uplink model between the terminal device and the drone is:

[0025]

[0026] Among them, σ 2 Expressed as the power of Gaussian white noise, P i represents the transmission power of terminal device i, where It represents the channel gain from terminal device i to drone j, as follows:

[0027]

[0028] Where μ0 represents the channel gain at a distance of 1m, Represents drone j,u i represents terminal device i;

[0029] The transmission rate between device i and drone j is It is expressed as follows:

[0030]

[0031] in, To connect to The time it takes for a device to transmit a task to a drone is expressed as:

[0032]

[0033] The transmission energy consumption is:

[0034]

[0035] The drone computing time is expressed as:

[0036]

[0037] Among them, f i,j It represents the computing resources allocated to terminal device i by drone j. The total time for drone to execute the task is expressed as:

[0038]

[0039] Total energy consumption of drone missions:

[0040]

[0041] Among them, α is a constant coefficient, v0 is the initial flight speed of the drone, represents the CPU power of drone j, U j represents the number of devices connected to drone j;

[0042] For base station-assisted mobile edge computing, the calculation time of the base station is expressed as:

[0043]

[0044] Among them, s c Indicates the number of CPU cycles per bit executed by the base station;

[0045] The total mission time of the base station is expressed as:

[0046]

[0047] Among them, T B represents the time it takes for the UAV to offload the task of terminal device i to the base station. Then the total task execution time, i.e., the optimization target of the present invention, is expressed as:

[0048]

[0049] Step 5: In order to determine the task segmentation threshold between the terminal device and the UAV, and between the UAV and the base station, the present invention sets a partial offloading method. First, the segmentation threshold between the terminal device i and the UAV j is divided into time-based and energy-based offloading methods. The time-based method is expressed as:

[0050]

[0051] If the local computing time is greater than the data transmission time and computing time of the drone, it will be greater than the local i The energy consumption of the unloading part is expressed as:

[0052]

[0053] If the energy ratio Will be greater than local i The segmentation threshold between the UAV and the base station is expressed as:

[0054]

[0055] When the drone is within the transmission range of the base station and the drone's computing task exceeds Offload the excess to the base station;

[0056] Step 6: Cross-iterate the optimal value according to the partial offloading method in step 5 and the priority queue.

[0057] 0≤τ≤0.5.

[0058] η1 and η2 are between 0 and 1.

[0059] A computer system, characterized in that it comprises: one or more processors, and a computer-readable storage medium for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the above-mentioned method.

[0060] A computer-readable storage medium is characterized by storing computer-executable instructions, which are used to implement the above method when executed.

[0061] A computer program, characterized by comprising computer executable instructions, which are used to implement the above method when executed.

[0062] Beneficial Effects

[0063] The present invention proposes a method for unloading computing from drones for mobile edge computing, which studies the delay problem of terminal devices in the field of mobile edge computing. UAVs are introduced to assist mobile edge computing to reduce the total task execution time. First, the terminal devices are divided into different areas, and the computing power of the area is measured to achieve the complementary computing power of drones and areas, and the computing tasks are prioritized, with high-level tasks being executed first. Finally, a partial unloading method is proposed to determine the three-party unloading threshold between the terminal device, drone, and base station, and finally a cross-iteration is performed to determine the optimal solution.

[0064] 1. The traditional method does not consider the deadline of computing tasks. The present invention prioritizes computing tasks according to deadlines and uses priority queues to sort computing tasks, so that high-priority tasks are executed first, greatly improving the task completion rate.

[0065] 2. The computing power of the terminal device is measured according to the battery capacity and CPU frequency of the terminal device, so that the drone with high computing power can match the area with weak computing power to achieve computing power complementarity, and a partial offloading method is proposed to determine the task segmentation threshold between the terminal device and the drone, and between the drone and the base station. The drone optimizes the trajectory based on the greedy idea, determines the task segmentation threshold in each time slot, and finally iterates the optimal solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] The drawings are only for the purpose of illustrating particular embodiments and are not to be considered limiting of the present invention. Like reference symbols denote like components throughout the drawings.

[0067] Figure 1 This is a model diagram of the drone computing offloading system for mobile edge computing in an example of the present invention. DETAILED DESCRIPTION

[0068] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0069] The present invention is a method for unloading computing of unmanned aerial vehicles (UAVs) for mobile edge computing. The idea of ​​K-means is used to divide the entire task area according to distance, and then the computing power of the entire task area is measured. The computing power of the UAV and the task area is complementary. The execution order of computing tasks is sorted by priority queue, and real-time tasks are executed first. Finally, a partial offloading method is proposed to determine the computing task segmentation threshold of the terminal device, the UAV, and the base station. Finally, the parameters are iterated to obtain the optimal solution, that is, to minimize the execution time of the entire system task.

[0070] The present invention has 1 drawing, please refer to Figure 1 As shown, the specific steps of the present invention are as follows:

[0071] Step 1: Divide the entire mission area into M clusters based on the distance and the number of drones based on the K-means method, expressed as Cluster = {C1,…,C M}, and then measure the computing power of each region according to the battery capacity and CPU frequency of each cluster terminal device, which is defined as:

[0072]

[0073] Among them, u represents the cluster C iEach terminal device in u Indicates the CPU frequency of device u, that is, the number of bits that can be calculated per second, E u Denoted as represents the battery capacity of device u, η1 and η2 are constant coefficients between 0 and 1. Drones with strong computing power match regions with weak computing power.

[0074] Step 2: Abstractly define the computing tasks of all terminal devices as

[0075] Q i =(L i ,T i )

[0076] Where i represents the serial number of the terminal device, L i Indicates the size of the task to be calculated, T i Indicates the deadline of the calculation task. If the task is not completed before the deadline, it means that the task execution has failed.

[0077] Step 3: For the computing tasks of each device on the drone, this paper uses dual queues to sort the computing tasks. Real-time tasks enter queue Q1, and other tasks enter queue Q2. Each time a task is taken out from Q1 for execution, when queue Q1 is empty, Q2 tasks enter Q1, and Q2 tasks are sorted according to deadlines in each time slot, so that high-priority tasks are at the head of the queue. In order to avoid task starvation, tasks in Q2 are upgraded by one level every once in a while.

[0078] Step 4: The computing tasks are divided into local computing, UAV-assisted computing, and base station-assisted computing, where {0≤ρ i1 ≤1,i∈N} represents the task ratio of the terminal device to the drone. When the drone is within the transmission range of the base station, the drone can offload the computing task to the base station. The offloading ratio is {0≤ρ i2 ≤1,i∈N}.

[0079] For local calculations, the local calculation time is expressed as:

[0080]

[0081] Among them, i local Indicates the computing power of the local device, that is, the number of bits calculated per second. i It represents the number of CPU cycles required by the terminal device to execute each bit. The local computing energy consumption is expressed as:

[0082]

[0083] for γ represents a coefficient determined by the chip structure determined by actual measurement.

[0084] For drone computing, a set of devices connected to the drone is represented as:

[0085] N i ={i:i∈N,g i =j}

[0086] The bandwidth resource of the drone is represented by B j HZ. The uplink model between the terminal device and the drone is:

[0087]

[0088] Among them, P i represents the transmission power of terminal device i, where

[0089]

[0090] The transmission rate is:

[0091]

[0092] in, To connect to The time it takes for a device to transmit a task to a drone is expressed as:

[0093]

[0094] The transmission energy consumption is:

[0095]

[0096] The drone computing time is expressed as:

[0097]

[0098] Among them, f i,j It represents the computing resources allocated to terminal device i by drone j. The total time for drone to execute the task is expressed as:

[0099]

[0100] Total energy consumption of drone missions:

[0101]

[0102] Among them, α is the constant coefficient and v0 is the initial flight speed of the drone.

[0103] For base station-assisted mobile edge computing, the calculation time of the base station is expressed as:

[0104]

[0105] The total mission time of the base station is expressed as:

[0106]

[0107] Then the total task execution time is expressed as, which is the optimization goal of the present invention:

[0108]

[0109] Step 5: In order to determine the task segmentation threshold between the terminal device and the UAV, and between the UAV and the base station, the present invention sets a partial offloading method. First, the segmentation threshold between the terminal device i and the UAV j is divided into time-based and energy-based offloading methods. The time-based method is expressed as:

[0110]

[0111] If the local computing time is greater than the data transmission time and computing time of the drone, it will be greater than the local i The energy consumption of the unloading part is expressed as:

[0112]

[0113] If the energy ratio Will be greater than local i The segmentation threshold between the UAV and the base station is expressed as:

[0114]

[0115] When the drone is within the transmission range of the base station and the drone's computing task exceeds The excess is offloaded to the base station.

[0116] Step 6: Cross-iterate the optimal value according to the partial offloading method in step 5 and the priority queue.

[0117] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should be included in the protection scope of the present invention.

Claims

1. A method for unloading computing from unmanned aerial vehicles for mobile edge computing, characterized in that Here are the steps: Step 1: Divide the entire mission area into M clusters based on the distance and the number of drones based on the K-means method, expressed as Cluster = {C1,…,C M }, and then measure the computing power of each region according to the battery capacity and CPU frequency of each cluster terminal device, which is defined as: Among them, u represents the cluster C i Each terminal device in u Indicates the CPU frequency of device u, that is, the number of bits that can be calculated per second, E u It is expressed as the battery capacity of device u, η1 and η2 are constant coefficients, is cluster C i The number of terminal devices, the computing power of the drone is based on Matching areas with weak computing power; Step 2: Abstractly define the computing tasks of all terminal devices as Q i =(L i ,T i ) Among them, i represents the serial number of the terminal device, L i Indicates the size of the task to be calculated, T i Indicates the deadline of the calculation task. If the task is not completed on the deadline, it means the task execution has failed. Step 3: For the computing tasks of each device on the drone, use dual queues to sort the computing tasks. Real-time tasks enter queue Q1, and other tasks enter queue Q2. Each time a task is taken out from Q1 for execution, when queue Q1 is empty, Q2 tasks enter Q1, and Q2 tasks are sorted according to deadlines in each time slot, so that high-priority tasks are at the head of the queue. In order to avoid task starvation, tasks in Q2 are upgraded by one level at regular intervals. Step 4: The computing tasks are divided into local computing, UAV-assisted computing, and base station-assisted computing, where {0≤ρ i1 ≤1,i∈N} represents the task ratio of the terminal device to the drone, N is the number of terminal devices in the entire mission area. When the drone is within the transmission range of the base station, the drone will offload the computing task to the base station, and the offloading ratio is {0≤ρ i2 ≤1,i∈N}; For local calculations, the local calculation time is expressed as: Among them, f i local Indicates the computing power of the local device, that is, the number of bits calculated per second; s i It represents the number of CPU cycles required by the terminal device to execute each bit; the local computing energy consumption is expressed as: for γ represents the coefficient determined by the chip structure determined by actual measurements; For drone computing, a set of devices connected to the drone is represented as: N i ={i:i∈N,g i =j} Among them, g i =j means device i is connected to drone j; The bandwidth resource of the drone is represented by B j ; The uplink model between the terminal device and the drone is: Among them, σ 2 Expressed as the power of Gaussian white noise, P i represents the transmission power of terminal device i, where It represents the channel gain from terminal device i to drone j, as follows: Where μ0 represents the channel gain at a distance of 1m, Represents drone j,u i represents terminal device i; The transmission rate between device i and drone j is It is expressed as follows: in, To connect to The time it takes for a device to transmit a task to a drone is expressed as: The transmission energy consumption is: The drone computing time is expressed as: Among them, f i,j It represents the computing resources allocated to terminal device i by drone j. The total time for drone to execute the task is expressed as: Total energy consumption of drone missions: Among them, α is a constant coefficient, v0 is the initial flight speed of the drone, represents the CPU power of drone j, U j represents the number of devices connected to drone j; For base station-assisted mobile edge computing, the calculation time of the base station is expressed as: Among them, s c Indicates the number of CPU cycles per bit executed by the base station; The total mission time of the base station is expressed as: Among them, T B represents the time it takes for the UAV to offload the task of terminal device i to the base station. Then the total task execution time, i.e., the optimization target of the present invention, is expressed as: Step 5: In order to determine the task segmentation threshold between the terminal device and the UAV, and between the UAV and the base station, the present invention sets a partial offloading method. First, the segmentation threshold between the terminal device i and the UAV j is divided into time-based and energy-based offloading methods. The time-based method is expressed as: If the local computing time is greater than the data transmission time and computing time of the drone, it will be greater than the local i The energy consumption of the unloading part is expressed as: If the energy ratio Will be greater than local i The segmentation threshold between the UAV and the base station is expressed as: When the drone is within the transmission range of the base station and the drone's computing task exceeds Offload the excess to the base station; Step 6: Cross-iterate the optimal value according to the partial offloading method in step 5 and the priority queue.

2. The method for unloading computing from a drone for mobile edge computing according to claim 1 is characterized in that 0≤τ≤0.

5.

3. The method for unloading computing of unmanned aerial vehicles for mobile edge computing according to claim 1 is characterized in that η1 and η2 are between 0 and 1.

4. A computer system, characterized in that include: One or more processors, and a computer-readable storage medium for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors are enabled to implement the method of claim 1.

5. A computer-readable storage medium, characterized in that Computer executable instructions are stored, and when the instructions are executed, they are used to implement the method of claim 1.

6. A computer program product, characterized in that The invention comprises computer executable instructions, which are used to implement the method of claim 1 when being executed.

Images