Multistage data processing MEC system based on unmanned aerial vehicle and implementation method

Abstract

The invention relates to the technical field of unmanned aerial vehicles, in particular to a multi-stage data processing MEC system based on an unmanned aerial vehicle and an implementation method. According to different performances of the unmanned aerial vehicles, the unmanned aerial vehicles are divided into three levels, including MEC unmanned aerial vehicles, a master unmanned aerial vehicle and slave unmanned aerial vehicles, and the master unmanned aerial vehicle can sense existence of surrounding MEC unmanned aerial vehicles and establishes wireless connection with the MEC unmanned aerial vehicles. Dividing the task according to the details of the calculation task and the current environmental factors, and transmitting the divided task to a platform of a proper level for processing. Meanwhile, communication between the master unmanned aerial vehicle and the slave unmanned aerial vehicles adopts an orthogonal frequency division multiplexing mode, simultaneous communication between the master unmanned aerial vehicle and the multiple slave unmanned aerial vehicles is supported, and the task uploading efficiency is improved. The computing resources are allocated and the flight path of the unmanned aerial vehicle is optimized through a convex optimization technology, so that the task processing efficiency is improved, and the problem that task scheduling and resource allocation are difficult due to mobility and heterogeneity of the unmanned aerial vehicle is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicles (UAVs), and in particular to a multi-level data processing MEC system based on UAVs and an implementation method thereof. Background Art

[0002] Mobile Edge Computing (MEC) technology has transformed mobile devices like smartphones and smart bracelets from simple communication tools into universal computing platforms capable of processing data that is difficult for humans to understand. Compared to data centers, mobile edge devices are closer to users and data sources, offering advantages such as low latency and reduced network traffic injection into core networks. Furthermore, their large number and wide distribution offer the potential to handle complex computing tasks, reducing the computational load on data centers.

[0003] Advances in drone technology have enabled the widespread application of lightweight drones in various fields. Drones offer excellent maneuverability and can carry communications and computing platforms, serving as temporary aerial base stations and servers. These are particularly useful for field IoT data collection and processing, temporary large-scale events, or disaster relief operations. They can quickly establish temporary networks, alleviate pressure on ground base stations, and restore network communications. However, the mobility and heterogeneity of drones present challenges for task scheduling and resource allocation. The computing power and battery capacity of individual drones are limited, necessitating the need to maximize computing efficiency while minimizing energy consumption and hovering time. Summary of the Invention

[0004] The purpose of the present invention is to provide a multi-level data processing MEC system based on drones and its implementation method, aiming to solve the problems of difficulty in task scheduling and resource allocation caused by the mobility and heterogeneity of existing drones.

[0005] To achieve the above objectives, in a first aspect, the present invention provides a multi-level data processing MEC system based on a drone, comprising a MEC drone, a master drone, and a slave drone, wherein the master drone, the MEC drone, and the slave drone are connected in sequence;

[0006] The MEC drone is used to fly to different locations at different times to assist the main drone in processing computing tasks;

[0007] The master drone is used to receive computing tasks collected from the slave drones and, when it senses the presence of MEC drones nearby, establishes a wireless connection with them and offloads some computing tasks to the MEC drones for processing;

[0008] The slave UAV is used to collect computing tasks from ground equipment and then transmit them to the master UAV for processing.

[0009] The communication between the master drone and the slave drones adopts an orthogonal frequency division multiplexing method to support simultaneous communication between the master drone and multiple slave drones.

[0010] Among them, after the master drone assembles the collected tasks, it divides the tasks according to the details of the calculation tasks and current environmental factors, and transmits the divided tasks to the platform for processing.

[0011] In a second aspect, a method for implementing a multi-level data processing MEC system based on a drone is provided, which is used in the multi-level data processing MEC system based on a drone described in the first aspect, and includes the following steps:

[0012] Divide the time into several time slots and calculate the distance, data transmission rate, and upload delay between the slave and master drones;

[0013] Calculate the energy consumption of the slave and master UAVs during task transmission and calculation;

[0014] Establish an energy consumption minimization model for the master and slave drones, and optimize subcarrier allocation, transmission power, correlation, computing frequency, task allocation strategy, and flight trajectory of the MEC drone under the constraints of limited resources and maximum task deadlines.

[0015] Through iterative solution, the undetermined parameters in the data collection submodel and the data processing submodel are determined, including subcarrier allocation indicator variables, transmission power, connection indicator variables, and calculation frequency.

[0016] In “calculating the energy consumption of the slave UAV and the master UAV during task transmission and calculation”, the energy consumption includes transmission energy consumption, calculation energy consumption and hovering energy consumption.

[0017] The present invention discloses a multi-level data processing MEC system based on drones, comprising a MEC drone, a master drone, and slave drones, the master drone, the MEC drone, and the slave drones being sequentially connected. The MEC drone is configured to fly to different locations at different times to assist the master drone in processing computing tasks. The master drone receives computing tasks collected by slave drones and, upon sensing nearby MEC drones, establishes a wireless connection with them, offloading some of the computing tasks to the MEC drones. The slave drones collect computing tasks from ground equipment and then transmit them to the master drone for processing. The master drone can sense the presence of nearby MEC drones and establish wireless connections with them. Tasks are divided based on their details and current environmental factors, and the divided tasks are transmitted to appropriate platforms for processing. Communication between the master and slave drones utilizes orthogonal frequency division multiplexing, enabling simultaneous communication between the master and multiple slave drones and improving task upload efficiency. Convex optimization techniques are used to allocate computing resources and optimize drone flight trajectories, improving task processing efficiency and addressing the difficulties in task scheduling and resource allocation arising from the mobility and heterogeneity of drones. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 This is a model diagram of a multi-level data processing MEC system based on drones.

[0020] Figure 2 This is a schematic diagram of a multi-level data processing MEC system based on drones provided by the present invention.

[0021] Figure 3 This is a flowchart of a method for implementing a multi-level data processing MEC system based on drones provided by the present invention.

[0022] Figure 4 It is a schematic diagram of the computing task collection and computing task allocation scheme.

[0023] Figure 5 It is a flow chart of the method for determining parameters for minimizing energy consumption.

[0024] In the picture: 1-MEC drone, 2-master drone, 3-slave drone. DETAILED DESCRIPTION

[0025] The following describes embodiments of the present invention in detail, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and are not to be construed as limiting the present invention.

[0026] See also Figures 1 to 2 In a first aspect, the present invention provides a multi-level data processing MEC system based on a drone, comprising a MEC drone 1, a master drone 2, and a slave drone 3, wherein the master drone 2, the MEC drone 1, and the slave drone 3 are connected in sequence;

[0027] The MEC drone 1 is used to fly to different locations at different times to assist the main drone 2 in processing computing tasks;

[0028] The master drone 2 is used to receive computing tasks collected from the slave drones 3, and when it senses the presence of a MEC drone 1 nearby, it establishes a wireless connection with it and offloads part of the computing tasks to the MEC drone 1 for processing;

[0029] The slave UAV 3 is used to collect computing tasks from ground equipment and then transmit them to the master UAV 2 for processing.

[0030] Furthermore, the communication between the master drone 2 and the slave drones 3 adopts an orthogonal frequency division multiplexing method to support simultaneous communication between the master drone 2 and multiple slave drones 3.

[0031] Furthermore, after assembling the collected tasks, the master UAV 2 divides the tasks according to the details of the calculation tasks and the current environmental factors, and transmits the divided tasks to the platform for processing.

[0032] In this embodiment, the master drone 2 can sense the presence of surrounding MEC drones and establish wireless connections with them. Tasks are divided based on their specifics and current environmental factors, and these tasks are then transferred to appropriate platforms for processing. Furthermore, communication between the master drone 2 and slave drones 3 utilizes orthogonal frequency division multiplexing, enabling simultaneous communication between the master drone 2 and multiple slave drones 3, improving task upload efficiency. Convex optimization techniques are used to allocate computing resources and optimize drone flight trajectories, improving task processing efficiency and addressing the challenges of task scheduling and resource allocation arising from the mobility and heterogeneity of drones.

[0033] See also Figures 3 to 5 In a second aspect, a method for implementing a multi-level data processing MEC system based on a drone is provided, which is used in the multi-level data processing MEC system based on a drone described in the first aspect, and includes the following steps:

[0034] S1 divides the time into several time slots and calculates the distance, data transmission rate, and upload delay between the slave drone 3 and the master drone 2;

[0035] Specifically, the present invention divides time into several time slots T = {1, 2, ..., T}, then at time t, the distance between the sth slave drone 3 and the mth master drone 2 can be expressed as

[0036]

[0037] in, and is the horizontal position from UAV 3, and is the horizontal position of the main UAV 2, h s and h m are the flight altitudes of slave UAV 3 and master UAV 2 respectively.

[0038] The OFDM communication mode is used between the master UAV 2 and the slave UAV 3. Assume that the number of orthogonal subcarriers is K. According to the Shannon formula, the data transmission rate of the slave UAV 3 on subcarrier k is

[0039]

[0040] Where B0 is the subcarrier bandwidth, is the transmission power from UAV 3 on the subcarrier, N0 is the power spectral density of the noise, and h0 is the unit channel gain at the reference distance d0 = 1m.

[0041] In order to speed up the data transmission rate, a single slave drone 3 can use multiple subcarriers. When multiple subcarriers are used, the total transmission rate of the slave drone 3 is

[0042]

[0043] in, is the subcarrier allocation indicator variable with a value of 0 or 1, and K is the subcarrier set allocated to slave drone 3. Therefore, the total transmission power is

[0044]

[0045] The master drone 2 and the slave drone 3 collaborate to collect tasks. Let the tasks collected by the master drone 2 and its subordinate slave drone 3 be in is the amount of tasks collected from drone 3, is the amount of tasks collected by the master drone 2, and the unit of both is bits. is the average complexity of the task, in cycles per bit. The upload delay from UAV 3 is

[0046]

[0047] After the task is uploaded to the master drone 2, the master drone 2 divides the computing task into local parts. and upload MEC drone part 1 If the master drone 2 retains part of the task for local calculation, the local calculation delay is

[0048]

[0049] in is the local calculation frequency of the master drone 2.

[0050] If the master drone 2 uploads the task to the MEC drone 1, the resulting upload delay is

[0051]

[0052] in It is the upload rate calculated based on Shannon's formula.

[0053] After the task is uploaded to MEC UAV 1, the calculation frequency that MEC UAV 1 can provide is Then the calculation delay of MEC UAV 1 is

[0054]

[0055] In order to ensure the integrity of the user's computing tasks, when the master drone 2 uploads part of the task to the MEC drone 1 for execution, it must wait until all tasks are completed before returning the results to the user. That is, the total computing delay depends on the larger of the two: the local computing time of the master drone 2 and the computing time of the uploaded MEC drone 1. Therefore, the total delay caused by a task offloading is

[0056]

[0057] S2 calculates the energy consumption of slave UAV 3 and master UAV 2 during task transmission and calculation;

[0058] The energy consumption includes transmission energy consumption, calculation energy consumption and hovering energy consumption.

[0059] Specifically, MEC UAV 1 is considered to have sufficient resources, so the main focus is on the energy consumption of the master UAV 2 and the slave UAV 3. When the slave UAV 3 transmits the task to the master UAV 2, the transmission energy consumption generated is

[0060]

[0061] Since the data receiving process is the process of passively receiving electromagnetic waves from space, the energy consumed by data receiving is much less than the energy consumed by data sending. Therefore, the data receiving energy consumption of the master UAV 2 can be ignored.

[0062] When the master UAV 2 transmits a task to the MEC UAV 1, the transmission energy consumption is

[0063]

[0064] If the master drone 2 retains part of the task for local computing, the energy consumed by the local computing is

[0065]

[0066] Among them, κ m is the energy consumption coefficient related to the CPU architecture. The more advanced the CPU architecture, the higher the κ m Since drones have different architectures and performance, choosing a drone with a more advanced architecture during task offloading can directly reduce computing energy consumption; choosing a drone with stronger performance can reduce computing latency and thus reduce hovering energy consumption.

[0067] During the task uploading and computing process, the drone needs to consume energy to maintain the operation of mechanical parts. is the hovering power of the slave UAV 3, then the hovering energy consumption of the slave UAV 3 can be expressed as

[0068]

[0069] Hovering power depends on the weight of the drone, the structure of the rotors, the manufacturing process of the motor, etc. Generally speaking, the more powerful the drone, the greater the hovering power.

[0070] set up The hovering power of the main UAV 2 is expressed as:

[0071]

[0072] Since the distance between the master drone 2 and the MEC drone 1 is constantly changing, when the distance gradually increases, the communication cost between the master drone 2 and the MEC drone 1 will also gradually increase. After exceeding the critical point, the communication cost generated by task offloading will be higher than the benefit brought by task offloading. In order to avoid this situation, we define is the indicator variable that marks the connection relationship between the master UAV m and the MEC UAV u, where

[0073]

[0074] In formula (2-6), if This indicates that the master UAV m decides to establish a connection with the MEC UAV u at time t. If no connection is established, task offloading will not be performed.

[0075] In summary, the total energy consumption generated by executing a task offloading is the sum of the transmission energy consumption of the master UAV 2 and the slave UAV 3, the computing energy consumption of the master UAV 2, and the hovering energy consumption of the master UAV 2 and the slave UAV 3, that is,

[0076]

[0077] S3 establishes an energy consumption minimization model for the master UAV 2 and the slave UAV 3. Under the constraints of limited resources and maximum task deadline, it optimizes subcarrier allocation, transmission power, association relationship, calculation frequency, task allocation strategy and the flight trajectory of MEC UAV 1.

[0078] Specifically, under the constraints of limited resources and maximum task deadline, the present invention optimizes the subcarrier allocation of UAV 3 and the transmission power on each subcarrier The relationship between the main drone 2 and the MEC drone 1 and the calculation frequency of the main drone 2 Computing task allocation strategy and the flight trajectory of MEC UAV 1 The energy consumption minimization model can be expressed as

[0079]

[0080] Where T, S, and M represent the time slot, slave drone 3, and master drone 2, respectively. Constraint C1 specifies that the subcarrier allocation indicator variable can take values ​​of 0 or 1. C2 restricts a single subcarrier to being allocated to at most one slave drone 3 at a time. C3 ensures that user data is not lost, guaranteeing the integrity of user computing tasks at the data level. C4 is a constraint on the maximum transmission power of slave drone 3. C5 specifies that the connection indicator variable between master drone 2 and MEC drone 1 can take values ​​of 0 or 1. C6 ensures that a master drone 2 can establish a connection with at most one MEC drone 1 at a time. C7 is a constraint on the maximum communication distance between master drone 2 and MEC drone 1. Task offloading will not occur outside the communication range. Within the communication range, partial or full offloading will occur if it is assessed to generate benefits. C8 limits the local computing frequency of master drone 2 to a maximum limit when processing computing tasks. C9 is a constraint on the maximum task deadline. Computing tasks completed beyond the maximum deadline will affect the user experience. C10 ensures that no computing tasks are missed, guaranteeing the integrity of user computing tasks at the computing level.

[0081] S4 determines the undetermined parameters in the data collection submodel and the data processing submodel through iterative solution, including subcarrier allocation indicator variables, transmission power, connection indicator variables, and calculation frequency.

[0082] Specifically, in the energy minimization model (3-1), there are complex coupling relationships between the undetermined parameters p, ρ, A, F, G, and D, making it difficult to directly determine their optimal values. Noting that the task offloading process can be divided into two stages: data collection and data processing, we initially decompose it into two sub-models and solve them iteratively.

[0083] The data collection sub-model determines the optimal values ​​of p and ρ given A, F, G, and D. This sub-model is expressed as

[0084]

[0085] stC1-C4,C9

[0086] Among them, the objective function of the model is about is a monotonically increasing function, and The constraints are convex sets. At the same time, the objective function is a 0-1 knapsack problem with respect to ρ, which is also a convex optimization problem with respect to p. Therefore, model (4-1) can be solved directly using a standard convex optimization solver.

[0087] The main complexity of the data processing sub-model lies in determining the connection relationship between the main UAV 2 and the MEC UAV 1 and the flight trajectory of the MEC UAV 1. Therefore, the data processing sub-model is further decomposed into a computing resource allocation sub-model and a flight trajectory determination sub-model. Under the premise that A and G are known, the computing resource allocation sub-model is expressed as

[0088]

[0089] Among them, constraints C8 and C12 give the upper bound of the calculation frequency

[0090] Since the model's objective function is a convex function with respect to F, and the constraints on F are also convex, standard convex optimization methods can be used to solve it. After obtaining the optimal F, the optimal value of D can be obtained using the same method.

[0091] After obtaining p, ρ, F, and D, the flight trajectory submodel is defined as

[0092]

[0093] in, is the flight trajectory of the main drone 2, yes The second norm of .

[0094] Since the sub-model (4-3) contains similar The present invention uses a continuous convex approximation method to introduce auxiliary variables Convert the model (4-3) into

[0095]

[0096] in,

[0097] Since the objective function in the optimization model (4-4) is about Monotonically decreasing, when energy consumption is minimized, the equality in C15 will strictly hold, so Equation (4-4) is equivalent to the original optimization problem Equation (4-3). The equivalent transformation transfers the non-convex term from the objective function to the constraint C15, and the right side of the inequality of C15 is about The lower bound of any point of a convex function can be given by its Taylor expansion. In the e-th iteration The value of , then

[0098]

[0099] ΔH=H u -h m

[0100] Among them, the approximate function It conforms to the form of f(x)=f(x0)+f′(x0)(x-x0), and log2e is the simplified form of 1 / ln2.

[0101] When using approximate functions After approximating the non-convex constraints, model (4-3) is transformed into a joint convex optimization model for s and G

[0102]

[0103] By solving (4-6), after obtaining the optimal trajectory in one round of iteration, the connection relationship A can be determined based on the current trajectory. This problem belongs to the 0-1 knapsack problem and can be solved directly.

[0104] Beneficial effects:

[0105] 1. Using drones to carry communication and computing platforms can quickly establish networks in areas where infrastructure is lacking.

[0106] 2. Organize drones into a multi-level structure, where drones at different levels have different performance and architecture, which can reduce networking costs.

[0107] 3. Use convex optimization technology to allocate computing resources and optimize the flight trajectory of UAVs. Use mathematical methods to rigorously deduce the existence of task offloading parameters that minimize the overall energy consumption of the system and provide a solution.

[0108] The above disclosure is only a preferred embodiment of a multi-level data processing MEC system based on drones and an implementation method of the present invention. Of course, this cannot be used to limit the scope of rights of the present invention. Ordinary technicians in this field can understand that all or part of the processes of the above embodiments are implemented, and equivalent changes made in accordance with the claims of the present invention are still within the scope of the invention.

Claims

1. A multi-level data processing MEC system based on drones, characterized by: It includes a MEC drone, a master drone and a slave drone, wherein the master drone, the MEC drone and the slave drone are connected in sequence; The MEC drone is used to fly to different locations at different times to assist the main drone in processing computing tasks; The master drone is used to receive computing tasks collected from the slave drones and, when it senses the presence of MEC drones nearby, establishes a wireless connection with them and offloads some computing tasks to the MEC drones for processing; The slave UAV is used to collect computing tasks from ground equipment and then transmit them to the master UAV for processing.

2. The multi-level data processing MEC system based on drone according to claim 1, characterized in that: The communication between the master drone and the slave drones adopts an orthogonal frequency division multiplexing method to support the master drone to communicate with multiple slave drones at the same time.

3. The multi-level data processing MEC system based on drone according to claim 1, characterized in that: After assembling the collected tasks, the master UAV divides the tasks according to the details of the calculation tasks and the current environmental factors, and transmits the divided tasks to the platform for processing.

4. A method for implementing a multi-level data processing MEC system based on a drone, used in the multi-level data processing MEC system based on a drone according to any one of claims 1 to 3, characterized in that: The following steps are involved: Divide the time into several time slots and calculate the distance, data transmission rate, and upload delay between the slave and master drones; Calculate the energy consumption of the slave and master UAVs during task transmission and calculation; Establish an energy consumption minimization model for the master and slave drones, and optimize subcarrier allocation, transmission power, correlation, computing frequency, task allocation strategy, and flight trajectory of the MEC drone under the constraints of limited resources and maximum task deadlines. Through iterative solution, the undetermined parameters in the data collection submodel and the data processing submodel are determined, including subcarrier allocation indicator variables, transmission power, connection indicator variables, and calculation frequency.

5. The method for implementing a multi-level data processing MEC system based on a drone as claimed in claim 4, characterized in that: In "Calculating the energy consumption of the slave UAV and the master UAV during task transmission and calculation", the energy consumption includes transmission energy consumption, calculation energy consumption and hovering energy consumption.

Images