Unmanned Aerial Vehicle (UAV)-Assisted Computational Offloading Methods, Systems, and Computer Storage Media

Abstract

This invention provides a drone-assisted computation offloading method, system, and computer storage medium, belonging to the field of drone-based edge computing. The method includes: establishing a drone-assisted computation offloading model, the model comprising an IoT device layer, an edge layer, and a cloud layer; in the drone-assisted computation offloading model, using a variable to indicate whether drone u provides services to IoT device k within time slot t; establishing a local IoT device computing scenario, a drone-assisted edge computing scenario, and a cloud-edge-device computing scenario; aiming to minimize the weighted sum of latency and energy consumption of all tasks, obtaining the most suitable drone for serving IoT devices, and having the drone calculate the optimal computation offloading scenario; and completing computation offloading at the optimal offloading scenario. This invention reduces the energy consumption rate during network computing task offloading while ensuring that the latency requirements of related computing tasks are effectively met.

Description

Technical Field

[0001] This invention belongs to the field of edge computing based on drones, and specifically relates to a drone-assisted computation offloading method, system and computer storage medium. Background Technology

[0002] With the development of the Internet of Things (IoT), an increasing number of IoT devices are joining the network, including various sensors, high-definition monitors, augmented reality (AR), virtual reality (VR), and digital twin services. Cloud computing provides IoT with efficient, flexible, and scalable computing, storage, and management resources, as well as reliable backend support and data analysis for IoT devices. However, as the number of IoT devices continues to increase, the generation and processing of data are becoming increasingly complex. To achieve a better user experience, edge computing technology is generally used to improve application responsiveness. Edge computing is an open platform that integrates network, computing, storage, and application core capabilities, located closer to the object or data source. It provides services at the nearest edge, initiating them from the edge, thus enabling faster network service responses.

[0003] However, emerging latency-sensitive applications such as AR, VR, and digital twins generate massive amounts of data. Given the extremely long transmission distances, transmitting all data to cloud servers would still present a significant challenge in meeting the latency requirements of these applications. At the network edge, drones can act as edge servers, essentially network repeaters, providing some computing services to IoT devices. This setup can significantly reduce overall computing latency and conserve limited energy for IoT devices.

[0004] In existing technologies, edge computing based on drones utilizes drones to perform complex computational tasks at the network edge, aiming to achieve a comprehensive optimization of task computation latency and total energy consumption of IoT devices. For example, some scholars have proposed drone-assisted edge computing network task offloading schemes based on traditional algorithms (heuristics). These schemes are simple to design, easy to implement, and have relatively low computational complexity, enabling them to run on resource-constrained IoT devices. However, traditional algorithms suffer from several drawbacks when making task offloading decisions. Traditional algorithms typically rely on predefined rules and experience, lacking adaptability; furthermore, these algorithms are often inefficient when solving large-scale and high-dimensional problems, making it difficult to find the globally optimal solution within a limited time.

[0005] Chinese patent application CN202410760607.3 discloses a D2D-based method for optimizing task offloading in collaborative mobile edge computing for UAV swarms. It establishes a mobile edge computing network model comprising an end-user layer, an edge layer, and a cloud layer. Based on this model, it establishes transmission, computation, and energy consumption models. Combined with D2D offloading technology, it improves the utilization of UAV computing resources while reducing transmission energy consumption and saving energy and channel resources, thus alleviating the burden on edge servers on individual UAVs. Specifically, by jointly optimizing the task allocation ratio, transmission power, and flight parameters of UAVs, it iteratively obtains the optimal computation task offloading decision and UAV location deployment decision using the MASAC algorithm and genetic algorithm, respectively. However, considering the factor of how IoT devices select which UAV in the swarm, the system fails to adequately address the latency issue in task offloading optimization. Furthermore, the system uses the MASAC algorithm to jointly optimize computation task allocation and communication resource management, which is difficult to train and requires more computing resources.

[0006] For example, Chinese patent application CN202310257281.8 discloses a method and apparatus for offloading edge computing tasks from a UAV used for power grid inspection. By establishing two models—a control model for controlling the UAV's flight and a model for different power IoT devices carrying computing tasks in different time slots—efficient offloading of edge computing tasks from ground-based power IoT devices is achieved, improving service quality and reducing task latency. Specifically, for the control model, the flight control parameters of the UAV in the target time slot can be obtained by using the location of the power grid inspection UAV in the target time slot, the location of each power IoT device, and the edge computing task load of each power IoT device as inputs. For the power IoT device computing task load model, the task offloading rate of each power IoT device in the time slot is determined based on the results of the control model, the edge computing task load of the time slot, and the available computing resources of each power IoT device. In this process, generative adversarial networks are used to generate training sample data to train a task offloading method based on the MATD3 algorithm, minimizing the energy consumption of the inspection UAV and the average latency for processing edge computing tasks within a time slot. However, the above-mentioned solutions have the following shortcomings: the designed drone power inspection system does not involve cloud-edge-device collaboration, and the computing tasks are concentrated on the power IoT devices, resulting in relatively low technical complexity and poor performance; the drone is only used as an information collection carrier and cannot undertake computing tasks or serve as the main body of the decision-making task offloading scheme, which limits the overall decision-making efficiency and accuracy of the designed system; the designed system does not consider the energy consumption of the power IoT devices at all. If the power IoT devices lack power supply, some power IoT devices are at risk of running out of energy too quickly, which seriously limits the application of this patent in passive wireless scenarios and has poor universality. Summary of the Invention

[0007] In view of the above-mentioned defects or deficiencies in the prior art, the present invention aims to provide a drone-assisted computation offloading method, system and computer storage medium to minimize the energy consumption rate of IoT devices and drones, and to ensure that the latency requirements of related computing tasks are effectively met.

[0008] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions:

[0009] In a first aspect, embodiments of the present invention provide a drone-assisted computational offloading method, the method comprising the following steps:

[0010] Step S1: Establish a drone-assisted computational offloading model, which includes an IoT device layer, an edge layer, and a cloud layer; wherein,

[0011] The IoT device layer includes several IoT devices with different battery capacities, and these IoT devices are immovable. This indicates that within time slot t, IoT device k continuously generates computing tasks. It is expressed by the following formula: and reports the computation request to the drone; the size of task m is determined by express;

[0012] The edge layer includes a group of drones, used This indicates that the device provides communication relay and in-transit computing for IoT devices; the drone is equipped with an edge computing server, which is used to determine offloading scenarios and provide computation offloading in edge layer computing scenarios;

[0013] The cloud layer includes high-performance servers, supercomputing centers and / or artificial intelligence computing clusters, receives IoT data relayed from drones, provides high-quality services for computationally intensive tasks, and transmits the computing results back to the drones through base stations;

[0014] Step S2, in the UAV-assisted calculation unloading model, adopting This indicates whether drone u provides services to IoT device k within time slot t; if IoT device k is connected to drone u, then... ,otherwise ;

[0015] The following limitations apply to the connection between drones and IoT devices:

[0016] i) There are no remaining tasks to be processed at the end of time slot t;

[0017] ii) Each IoT device can only connect to one drone in the same time slot;

[0018] Expressed as a formula:

[0019] (1)

[0020] Where U represents the total number of drones; Represents any value;

[0021] Step S3: Under the drone-assisted computation offloading model, three task computation offloading scenarios are established, including: local IoT device computation scenario, drone-assisted edge computing scenario, and cloud-edge-device computing scenario; and a single computation task is processed entirely in one scenario.

[0022] Step S4: Based on the three established task computation offloading scenarios, with the goal of minimizing the weighted sum of latency and energy consumption of all tasks, the best drone for serving the IoT device is obtained by combining the transmission power and task processing parameters of IoT devices, the transmission power and task processing parameters of drones, the available energy of drones and IoT devices, the size of the task to be processed, and the tolerable latency. The drone then calculates the optimal computation offloading scenario.

[0023] Step S5: Complete the calculation and uninstallation under the optimal uninstallation scenario.

[0024] In a preferred embodiment of the present invention, in step S2, in order to ensure that each IoT device can only connect to one drone in the same time slot, the three-dimensional coordinates of drone u are represented as follows: , , and These are the X, Y, and Z coordinates of the drone u; IoT devices are immovable, and their three-dimensional coordinates are represented as... ;in, , , and It is a constant; and:

[0025] (2)

[0026] in, It is the distance threshold between IoT devices and drones;

[0027] If the distance between the IoT device and the drone exceeds Drones will refuse access to IoT devices.

[0028] In a preferred embodiment of the present invention, the calculation process in step S3 under the local IoT device computing scenario is as follows:

[0029] set up This indicates whether task m is processed on the local IoT device k; if task m is processed on the local IoT device, then... ,otherwise ;

[0030] The delay from task m's generation to completion only considers computational delay, as shown in the following formula:

[0031] (3)

[0032] in, It is the total latency of local IoT device k processing task m. This is the time consumed by the local IoT device k in processing task m. This represents the CPU frequency of the local IoT device k. This represents the CPU cycles required to process 1 bit of data.

[0033] Weighted energy consumption of the entire process for:

[0034] (4)

[0035] in, It is the effective switching capacitor for IoT device k. It is the importance coefficient of IoT device k, and all local IoT devices share the same parameters in the model.

[0036] As a preferred embodiment of the present invention, the calculation process in step S3 under the UAV-assisted edge computing scenario is as follows:

[0037] set up This indicates whether task m is being processed by the drone; if task m is being processed by the drone, then... ,otherwise ;

[0038] In this scenario, the computational task is handled by an edge computing server deployed on the drone; IoT device k generates task m and sends the raw data... The data is transmitted to the corresponding UAV u, which then processes the data and sends the result data within time slot t. The data is returned to IoT devices; therefore, this is a drone-assisted edge computing model. The total delay is given by the following formula

[0039] (5)

[0040] in, , and These represent the data transmission latency, result reception latency, and drone processing latency of task m at IoT device k, respectively.

[0041] Drone processing latency Represented as:

[0042] (6)

[0043] in, The number of CPU cycles required to process 1 bit of data for a drone. The frequency of the drone's CPU;

[0044] The data transmission delay and the result reception delay are expressed as follows:

[0045] (7)

[0046] in, and These are the uplink and downlink transmission rates between the drone u and the IoT device k, and:

[0047] (8)

[0048] in, and These are used to represent the uplink and downlink bandwidths, respectively. Indicates propagation loss. and These represent the transmission power of IoT devices and drones, respectively. and These represent the terminal Gaussian noise power of the drone and the IoT device, respectively.

[0049] Weighted energy consumption of the entire process Energy transmission by IoT devices Drones transmit energy and processing energy Composed of, and obtained through the following means:

[0050] (9)

[0051] in, For the effective switching capacitor of the UAV u, It is the importance coefficient of drone u, and all drones share the same parameter; It is the importance coefficient of IoT device k.

[0052] In a preferred embodiment of the present invention, the computation process in the cloud-edge-device computing scenario in step S3 is as follows:

[0053] set up This indicates whether task m is processed on a cloud server; if task m is processed on a cloud server, then... ,otherwise ;

[0054] In this scenario, IoT device k creates a computing task m and sends raw data to drone u; then, the drone relays the data to a cloud server via a base station; once the cloud server completes the task processing, it sends the result back to the drone, which finally returns the result to IoT device k within time slot t; the total latency of the entire process is... It can be obtained through the following formula:

[0055] (10)

[0056] in, and This refers to the end-to-end total transmission latency and total reception latency of data from IoT devices to cloud servers; and the processing latency of the cloud servers. Considered as 0, and Given by the following formula

[0057] (11)

[0058] in, It is the propagation delay from the cloud to the base station. This indicates the size of the calculation result returned by the cloud server; and These are the uplink and downlink transmission rates between the drone u and the IoT device k. and This represents the uplink and downlink transmission rates between the drone and the base station; the weighted energy consumption of the process. The following formula is used to derive...

[0059] (12)

[0060] in, and These refer to the additive transmission energy consumption of drones and IoT devices when sending raw data and receiving results; and the energy consumption of the cloud. Treated as 0, the rest are obtained using the following formula:

[0061] (13)

[0062] in, and These represent the transmission power of IoT devices and drones, respectively. It refers to the transmission power of the drone to the base station. It is the importance coefficient of IoT device k. It is the importance coefficient of the drone u.

[0063] In a preferred embodiment of the present invention, step S4 involves connecting IoT devices to appropriate drones to complete network formation, and having the drones decide on suitable task offloading locations, in order to minimize the weighted sum of latency and energy consumption for all tasks, expressed by the formula: [Problem / Issue] :

[0064] (14)

[0065] in, , and These indicate whether task m is processed on a local IoT device, a drone, or a cloud server, respectively. and These represent the total latency of local IoT devices, drones, and cloud servers, respectively. and These represent the weighted energy consumption of local IoT devices, drones, and cloud servers, respectively. Indicates the delay factor;

[0066] In the question In this context, the selection of drones for IoT devices to connect and the decision-making process for task offloading are coupled, and this leads to problems. It is decomposed into multiple UAV selection sub-problems P1 and task offloading decision sub-problems P2;

[0067] and,

[0068] The subproblem P1 minimizes the weighted transmit power of the IoT devices and the drone by connecting the IoT devices to the appropriate drone in each time slot t, as shown in the formula:

[0069] (15)

[0070] in, Indicates the drone's transmission power. This represents the transmit power of an IoT device, which remains constant within a given time slot t; the output of subproblem P1 is a set. This displays the connection status between the IoT device and the drone;

[0071] Subproblem P2 minimizes the weighted sum of energy consumption and latency of all tasks performed by the same UAV u within time slot t, and its formula is:

[0072] (16)

[0073] Results of the IoT device clustering phase This significantly affects the input of subproblem P2; the output of subproblem P2 is... , or This indicates whether task m is processed on a local IoT device, a drone, or a cloud server.

[0074] In a preferred embodiment of the present invention, the subproblem P1 is solved using an IoT device clustering algorithm based on simulated annealing; the solution process includes:

[0075] Step S411: Establish the optimization function and connect the optimization function with the subproblems. Alignment, i.e.

[0076] (17)

[0077] Step S412: Input the drone set U, the IoT device set K, and the drone location. and IoT device location ;

[0078] Step S413, randomly initialize the random set Annealing temperature And the cooling coefficient α, to obtain the initial optimization function ;

[0079] Step S414, generate new A new optimization function is obtained. and calculate ;

[0080] Step S415, if If the change is accepted, it will be accepted; otherwise, according to the Metropolis rules, the probability of accepting the change is determined by the following formula:

[0081] (18)

[0082] Step S416, update the annealing temperature T1 so that the latest temperature ; Indicates the cooling rate;

[0083] Step S417, determine the annealing temperature Has the temperature dropped below the specified cooling threshold? If yes, proceed to step S418; otherwise, return to step S415.

[0084] Step S418: Output the optimized result set. This determines which IoT devices and their related tasks the drone u will handle within time slot t.

[0085] In a preferred embodiment of the present invention, the subproblem P2 is solved using a task offloading decision algorithm based on the Proximity Policy Optimization (PPO) algorithm; the solution process includes:

[0086] Step S421, input the available energy of the drone u and the IoT device. Actor's online learning rate Commentator's online learning rate The CLIP parameter ε, discount factor γ, and computational requirements G for all tasks are considered. The total number of segments is W, with each segment requiring N iterations of the actor network and I iterations of the commentator network.

[0087] Step S422, initialize actor network parameters and commentator network parameters ;

[0088] Step S423, collect trajectory Use the current strategy ,make ;

[0089] Step S424, in order to improve learning stability and reduce variance, for each state Calculate the advantage function And given by the following formula:

[0090] (19)

[0091] in, Indicates the trajectory in state The rewards obtained below; Indicates from state to state Discount factor; Indicates the trajectory in state The rewards obtained below; Indicates the use of parameters The neural network estimates the state The value function;

[0092] Step S425: The actor network randomly selects an action from the activity set of the current state, interacts with the environment, and obtains the next state and reward function;

[0093] Step S426, calculate the actor network sampling rate, which is obtained by the following formula:

[0094] (20)

[0095] in, Represents the state in the nth training iteration. The ratio of sampling rates, Indicates the use of parameters New actor network, Indicates that parameters are still used. The original actors online;

[0096] Step S427, calculate the actor network loss function, whose objective function is

[0097] (twenty one)

[0098] Update the θ-related parameters using the Adam optimizer;

[0099] loss Obtained from the following formula:

[0100] (twenty two)

[0101] Update actor network parameters θ. ;

[0102] In step S428, the commentator network updates its parameters using the Adam optimizer. To improve the accuracy of its assessment of actors' online behavior; its loss function The computation relies on long-term rewards provided by the environment:

[0103] (twenty three)

[0104] Update the parameters of the commentator network using a loss function. ;

[0105] Step S429: Determine if the maximum number of iterations W has been reached; if not, return to step S423; if yes, output the result: task calculation unloading scenario. .

[0106] Secondly, embodiments of the present invention also provide a drone-assisted computation offloading system, the system comprising: an IoT device layer, an edge layer, a cloud layer, a drone connection module, and a scene selection module; wherein,

[0107] The IoT device layer includes several IoT devices with different battery capacities, and these IoT devices are immovable. This indicates that within time slot t, IoT device k continuously generates computing tasks. It is expressed by the following formula: and reports the computation request to the drone; the size of task m is determined by It also indicates that computation offloading is used in local IoT device computing scenarios;

[0108] The edge layer includes a group of drones, used This indicates that the device provides communication relay and in-transit computing for IoT devices; the drone is equipped with an edge computing server, which is used to determine offloading scenarios and provide computation offloading in edge layer computing scenarios;

[0109] The cloud layer includes high-performance servers, supercomputing centers and / or artificial intelligence computing clusters, used to receive IoT data forwarded by drones, perform computation offloading in the cloud-edge-device computing scenario, and transmit the computation results back to the drones through base stations;

[0110] The drone connection module defines This indicates whether drone u provides services to IoT device k within time slot t; if IoT device k is connected to drone u, then... ,otherwise ;

[0111] The scenario selection module is used to select the most suitable drone for serving the IoT device based on the three established task computation offloading scenarios, with the goal of minimizing the weighted sum of latency and energy consumption of all tasks. It combines the transmission power and task processing parameters of IoT devices, the transmission power and task processing parameters of drones, the available energy of drones and IoT devices, the size of the task to be processed, and the tolerable latency, and in conjunction with the drone connection module, calculates the optimal computation offloading scenario. Then, based on the selected optimal computation offloading scenario, the corresponding computation offloading function in the network device layer, edge layer, or cloud layer is activated.

[0112] Thirdly, embodiments of the present invention provide a computing storage medium storing the unmanned aerial vehicle-assisted computing offloading system as described above.

[0113] The technical solutions provided by the embodiments of the invention have the following beneficial effects:

[0114] The drone-assisted computation offloading method, system, and computer storage medium provided in this invention enable IoT devices in remote areas to perform computation, networking, and long-distance communication. This allows latency-sensitive applications to run in remote areas. Leveraging the flexibility and ease of deployment of drones, they function as mobile edge servers, avoiding excessively long-distance data transmission between remote IoT devices and cloud servers. This reduces data transmission and computation latency, improves task processing speed, and enhances communication stability. Furthermore, this invention can maximize the lifespan of IoT devices. In remote areas, IoT devices can send data to drones in suitable locations. Data with high latency tolerance and large computational demands can be processed on onboard servers or cloud servers, reducing the transmission power of IoT devices, minimizing energy consumption, and extending lifespan. In addition, this invention can more effectively handle massive computational tasks. Drones can rationally select computation locations based on task requirements and data size. Collaborative processing by local IoT devices, onboard servers, and high-performance cloud servers significantly improves data processing and computation efficiency, reduces latency, and optimizes resource utilization.

[0115] Of course, implementing any product or method of the present invention does not necessarily require achieving all of the advantages described above at the same time. Attached Figure Description

[0116] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0117] Figure 1 This is a flowchart of the unmanned aerial vehicle-assisted calculation method in an embodiment of the present invention;

[0118] Figure 2 This is a schematic diagram of the cloud-edge-device collaborative computing architecture in the drone-assisted computation offloading method in this embodiment of the invention;

[0119] Figure 3 This is a flowchart of the simulated annealing algorithm used in subproblem P1 in this embodiment of the invention;

[0120] Figure 4 This is a flowchart of the near-end strategy optimization algorithm used in subproblem P2 in this embodiment of the invention. Detailed Implementation

[0121] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. It should be noted that, without conflict, the embodiments and features in the embodiments of the present invention can also be combined with each other.

[0122] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In the description of this invention, the terms "first," "second," "third," "fourth," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0123] This invention addresses the widespread computing, networking, and long-distance communication needs of IoT devices in remote areas. It solves the problems of emerging applications in remote areas, such as digital twins and augmented reality, which have strict requirements for service latency and generate huge amounts of data to be processed. At the same time, it addresses the problem that IoT devices in remote areas are limited by limited energy reserves and computing power, making it difficult for them to independently complete massive computing tasks within a specified time. The invention proposes a drone-assisted computing offloading method, system, and computer storage medium, designs a three-layer, two-stage drone-assisted edge computing network task offloading architecture, and proposes corresponding networking and task offloading models for this architecture to improve data processing and computing efficiency and optimize resource utilization.

[0124] like Figure 1 As shown in the embodiment of the present invention, the UAV-assisted computational unloading method includes the following steps:

[0125] Step S1, establish a drone-assisted computational unloading model, such as Figure 2 As shown, the model includes an IoT device layer, an edge layer, and a cloud layer; the IoT device layer includes several IoT devices with different battery capacities, and these IoT devices are immobile. This indicates that within time slot t, IoT device k continuously generates computing tasks. It is expressed by the following formula: and reports the computation request to the drone; the size of task m is determined by The edge layer is described as comprising a group of drones, using... It indicates that the system provides communication relay and in-transit computing for IoT devices; the drone is equipped with an edge computing server, which is used to determine offloading scenarios and provide computing offloading in edge layer computing scenarios; the cloud layer includes high-performance servers, supercomputing centers and / or artificial intelligence computing clusters, receives IoT data forwarded by the drone, provides high-quality services for computing-intensive tasks, and transmits the computing results back to the drone through base stations.

[0126] Step S2, in the UAV-assisted calculation unloading model, adopting This indicates whether drone u provides services to IoT device k within time slot t; if IoT device k is connected to drone u, then... ,otherwise .

[0127] For connections between drones and IoT devices, the following limitations apply: i) there are no remaining pending tasks at the end of time slot t; ii) each IoT device can only connect to one drone in the same time slot.

[0128] (1)

[0129] The three-dimensional coordinates of the UAV u are represented as follows ,in , and These are the X, Y, and Z coordinates of the drone u; IoT devices are immovable, and their 3D coordinates are represented as... ;in, , , and It is a constant. To avoid excessive energy consumption for data transmission, the maximum distance between IoT devices and drones should not exceed [a certain value]. It is given by the following formula

[0130] (2)

[0131] If the maximum distance between the IoT device and the drone exceeds The drones will deny access to IoT devices. The drones' coverage areas partially overlap to improve service reliability.

[0132] Step S3: Under the drone-assisted computation offloading model, three task computation offloading scenarios are established, including: local IoT device computation scenario, drone-assisted edge computing scenario, and cloud-edge-device computing scenario; and a single computation task is processed entirely in one scenario.

[0133] In this step, due to the limited computing power of IoT devices, and considering factors such as task size, latency tolerance, IoT device battery power, and drone battery power, the task is offloaded to three different locations: local IoT device computing, drone-assisted edge computing, and cloud-edge-device collaborative computing.

[0134] For local IoT device computing scenarios:

[0135] set up This indicates whether task m is processed on the local IoT device k; if task m is processed on the local IoT device, then... ,otherwise .

[0136] Considering only computational delay, the delay from task m's generation to completion can be obtained as follows:

[0137] (3)

[0138] in, It is the total latency for local IoT devices to process task m. This refers to the time consumed by local IoT device k in processing task m, where the CPU frequency of IoT device k is used. express, This represents the CPU cycles required to process 1 bit of data. Therefore, the weighted energy consumption of the entire process is...

[0139] (4)

[0140] in, It is the effective switching capacitor for IoT device k. This is the importance coefficient of IoT device k. It is assumed that all local IoT devices share the same parameters in this architecture.

[0141] For drone-assisted edge computing scenarios:

[0142] set up This indicates whether task m is being processed by the drone; if task m is being processed by the drone, then... ,otherwise .

[0143] In this scenario, the computational tasks are handled by an edge computing server deployed on the drone. The IoT device k generates task m and sends the raw data... The data is transmitted to the corresponding UAV u, which then processes the data and sends the result data within time slot t. The data is returned to IoT devices; therefore, this is a drone-assisted edge computing model. The total delay is given by the following formula

[0144] (5)

[0145] in, , and Let $\mathbf$ represent the data transmission latency, result reception latency, and drone processing latency of task $m$ at IoT device $k$, respectively. For simplicity, propagation latency is ignored. Additionally, processing latency... It can be represented as

[0146] (6)

[0147] The number of CPU cycles required to process 1 bit of data for a drone Given the frequency of the drone's CPU, the transmit latency and receive latency can be expressed as:

[0148] (7)

[0149] in, and The uplink and downlink transmission rates between u and k are given by the following formula.

[0150] (8)

[0151] in, and These are used to represent the uplink and downlink bandwidths, respectively. Propagation loss is used... This indicates that the transmission power of IoT devices and drones is used... and It indicates that the terminal Gaussian noise power of drones and IoT devices is... and The weighted energy consumption of the entire process is given. Energy transmission by IoT devices Drones transmit energy and processing energy Composition, and obtained through the following methods

[0152] (9)

[0153] in, For the effective switching capacitor of the UAV u, This is an important coefficient for the drone's u. All drones also share the same parameters.

[0154] For cloud-edge-device computing scenarios:

[0155] set up This indicates whether task m is processed on a cloud server; if task m is processed on a cloud server, then... ,otherwise .

[0156] In this scenario, IoT device k creates a computing task m and sends raw data to drone u. The drone then relays the data to a cloud server via a base station. Once the cloud server completes the task processing, it sends the result back to the drone, which finally returns the result to IoT device k within time slot t. The total latency of the entire process is... It can be obtained through the following formula:

[0157] (10)

[0158] in, and This refers to the end-to-end total send latency and total receive latency of data from IoT devices to the cloud server. The cloud server's processing latency is also included. Considered as 0, and Given by the following formula

[0159] (11)

[0160] in, It is the propagation delay from the cloud to the base station. This indicates the size of the calculation result returned by the cloud server. The uplink and downlink rates between the drone and the base station are determined by... and This indicates that its derivation can be referenced from previous models. The weighted energy consumption of this process... It can be obtained through the following formula

[0161] (12)

[0162] in, and These refer to the additive transmission energy consumption of drones and IoT devices when sending raw data and receiving results. Cloud energy consumption... Treated as 0, the rest are obtained using the following formula:

[0163] (13)

[0164] in, It is the transmission power of the drone to the base station.

[0165] Step S4: Based on the three established task computation offloading scenarios, under the constraints, by combining the transmission power and task processing parameters of IoT devices, the transmission power and task processing parameters of UAVs, the available energy of UAVs and IoT devices, the size of the tasks to be processed, and the tolerable latency, the most suitable UAV and the best computation offloading scenario for serving IoT devices are obtained, so as to minimize the weighted sum of latency and energy consumption of all tasks.

[0166] In this step, IoT devices are connected to appropriate drones to form a network, and the drones decide on suitable task unloading locations to minimize the weighted sum of latency and energy consumption for all tasks. This can be expressed as a problem. :

[0167] (14)

[0168] in, , and These indicate whether task m is processed on a local IoT device, a drone, or a cloud server, respectively. and These represent the total latency of local IoT devices, drones, and cloud servers, respectively. and These represent the weighted energy consumption of local IoT devices, drones, and cloud servers, respectively. Indicates the delay factor;

[0169] In the question In this context, the selection of drones for IoT devices to connect and the decision to offload tasks are coupled, similar to a multi-agent competition problem.

[0170] In this step, to simplify the solution process, the problem is... The problem is decomposed into multiple sub-problems: drone selection (P1) and task offloading decision (P2). Sub-problems P1 and P2 can be viewed as corresponding optimization problems in the IoT device clustering phase and network operation phase, respectively.

[0171] The subproblem P1 minimizes the weighted transmit power of the IoT devices and the drone by connecting the IoT devices to the appropriate drone in each time slot t, as shown in the formula:

[0172] (15)

[0173] in, Indicates the drone's transmission power. This represents the transmit power of an IoT device, which remains constant within a given time slot t. Subproblem P1 outputs a set. This displays the connection status between the IoT device and the drone.

[0174] Subproblem P2 minimizes the weighted sum of energy consumption and latency of all tasks performed by the same UAV u within time slot t, and its formula is:

[0175] (16)

[0176] Results of the IoT device clustering phase This significantly affects the input of subproblem P2. The output of subproblem P2 is... , or This indicates whether task m is processed on a local IoT device, a drone, or a cloud server.

[0177] In one specific embodiment, subproblem P1 is solved using an IoT device clustering algorithm based on simulated annealing. For example... Figure 3 As shown, the solution process includes:

[0178] Step S411: Establish the optimization function and connect the optimization function with the subproblems. Alignment, i.e.

[0179] (17)

[0180] Step S412: Input the drone set U, the IoT device set K, and the drone location. and IoT device location ;

[0181] Step S413, randomly initialize the random set Annealing temperature And the cooling coefficient α, to obtain the initial optimization function ;

[0182] Step S414, generate new A new optimization function is obtained. and calculate ;

[0183] Step S415, if If the change is accepted, it is accepted. Otherwise, according to the Metropolis rules, the probability of accepting the change is determined by the following formula:

[0184] (18)

[0185] Step S416, update the annealing temperature T1 so that the latest temperature ; Indicates the cooling rate;

[0186] Step S417, determine the annealing temperature Has the temperature dropped below the specified cooling threshold? If yes, proceed to step S418; otherwise, return to step S415.

[0187] Step S418: Output the optimized result set. This determines which IoT devices and their related tasks the drone u will handle within time slot t.

[0188] Furthermore, in a specific embodiment, for subproblem P2, such as Figure 4 As shown, a task offloading decision algorithm based on Proximal Policy Optimization (PPO) is used to solve the problem. The actor and commentator networks interact collaboratively with the environment to find the optimal computational offloading scenario. To conserve storage resources, the algorithm does not use a storage pool but instead employs a sampling-as-training approach. The algorithm is based on the current policy... The system interacts stably with the environment until a completion state is reached, thus generating a complete trajectory. The environment will then have Q steps. The trajectory tuples are fed back to the two networks. Regarding the tuples, the state space consists of the available energy of the drone and IoT devices, the size of the task to be processed in step q, and the tolerable latency, i.e. The motion space is In each step q, the reward function is defined as the negative of the weighted sum of delay and energy consumption:

[0189]

[0190] The proposed algorithm employs CLIP-based importance sampling, where the CLIP function is defined as:

[0191]

[0192] The specific steps are as follows:

[0193] Step S421, input the available energy of the drone u and the IoT device. Actor's online learning rate Commentator's online learning rate The CLIP parameter ε, discount factor γ, and computational requirements G for all tasks are considered. The total number of segments is W, with each segment requiring N iterations of the actor network and I iterations of the commentator network.

[0194] Step S422, initialize actor network parameters and commentator network parameters ;

[0195] Step S423, collect trajectory Use the current strategy ,make ;

[0196] Step S424, in order to improve learning stability and reduce variance, for each state Calculate the advantage function And given by the following formula:

[0197] (19)

[0198] in, Indicates the trajectory in state The rewards obtained below; Indicates from state to state Discount factor; Indicates the trajectory in state The rewards obtained below; Indicates the use of parameters The neural network estimates the state The value function;

[0199] Step S425: The actor network randomly selects an action from the activity set of the current state, interacts with the environment, and obtains the next state and reward function;

[0200] Step S426, calculate the actor network sampling rate, which is obtained by the following formula.

[0201] (20)

[0202] in, Represents the state in the nth training iteration. The ratio of sampling rates, Indicates the use of parameters New actor network, Indicates that parameters are still used. The original actors are online.

[0203] Step S427, calculate the actor network loss function, whose objective function is

[0204] (twenty one)

[0205] Update the θ-related parameters using the Adam optimizer; loss Obtained from the following formula

[0206] (twenty two)

[0207] Update actor network parameters θ. ;

[0208] In step S428, the commentator network updates its parameters using the Adam optimizer. To improve the accuracy of its assessment of actors' online behavior; its loss function The computation relies on long-term rewards provided by the environment:

[0209] (twenty three)

[0210] Update the parameters of the commentator network using a loss function. ;

[0211] Step S429: Determine if the maximum number of iterations W has been reached; if not, return to step S423; if yes, output the result: task calculation unloading scenario. .

[0212] Step S5: Complete the calculation and uninstallation at the optimal uninstallation scenario.

[0213] In this step, the computational unloading is completed according to the computational unloading methods under the three different scenarios defined in step S3.

[0214] Based on the same idea, this invention also provides a drone-assisted edge computing offloading system, the system comprising: an IoT device layer, an edge layer, a cloud layer, a drone connection module, and a scene selection module; wherein,

[0215] The IoT device layer includes several IoT devices with different battery capacities, and these IoT devices are immovable. This indicates that within time slot t, IoT device k continuously generates computing tasks. It is expressed by the following formula: and reports the computation request to the drone; the size of task m is determined by It also indicates that computation offloading is used in local IoT device computing scenarios;

[0216] The edge layer includes a group of drones, used This indicates that the device provides communication relay and in-transit computing for IoT devices; the drone is equipped with an edge computing server, which is used to determine offloading scenarios and provide computation offloading in edge layer computing scenarios;

[0217] The cloud layer includes high-performance servers, supercomputing centers and / or artificial intelligence computing clusters, used to receive IoT data forwarded by drones, perform computation offloading in the cloud-edge-device computing scenario, and transmit the computation results back to the drones through base stations;

[0218] The drone connection module defines This indicates whether drone u provides services to IoT device k within time slot t; if IoT device k is connected to drone u, then... ,otherwise ;

[0219] The scenario selection module is used to select the most suitable drone for serving the IoT device based on the three established task computation offloading scenarios, with the goal of minimizing the weighted sum of latency and energy consumption of all tasks. It combines the transmission power and task processing parameters of IoT devices, the transmission power and task processing parameters of drones, the available energy of drones and IoT devices, the size of the task to be processed, and the tolerable latency, and in conjunction with the drone connection module, calculates the optimal computation offloading scenario. Then, based on the selected optimal computation offloading scenario, the corresponding computation offloading function in the network device layer, edge layer, or cloud layer is activated.

[0220] In this embodiment, each module is implemented using a processor, with additional memory added as needed for storage. The processor can be, but is not limited to, a microprocessor (MPU), a central processing unit (CPU), a network processor (NP), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), other programmable logic devices, discrete gates, transistor logic devices, discrete hardware components, etc. The memory can include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory can also be at least one storage device located remotely from the aforementioned processor.

[0221] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means.

[0222] It should also be noted that the UAV-assisted computational unloading system and the UAV-assisted computational unloading method described in this embodiment are corresponding. The description and limitations of the method also apply to the system, and will not be repeated here.

[0223] This invention also provides a computing storage medium that stores a drone-assisted computing offloading system.

[0224] As can be seen from the above technical solutions, the drone-assisted computation offloading method, system, and computer storage medium provided by the embodiments of the present invention enable IoT devices in remote areas to perform computation, networking, and long-distance communication. This allows latency-sensitive applications to run in remote areas. Leveraging the flexibility and ease of deployment of drones, they function as mobile edge servers, avoiding excessively long-distance data transmission between remote IoT devices and cloud servers. This reduces data transmission and computation latency, improves task processing speed, and enhances communication stability. Furthermore, the present invention can maximize the lifespan of IoT devices. In remote areas, IoT devices can send data to drones in suitable locations. Data with high latency tolerance and large computational demands can be processed on onboard servers or cloud servers, thereby reducing the transmission power of IoT devices, reducing energy consumption, and extending their lifespan. In addition, the present invention can more effectively handle massive computational tasks. Drones can rationally select computation locations based on task requirements and data size. Collaborative processing by local IoT devices, onboard servers, and high-performance cloud servers significantly improves data processing and computation efficiency, reduces latency, and optimizes resource utilization.

[0225] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed, and is not intended to limit the scope of the claimed invention, but merely to illustrate preferred embodiments of the invention. Those skilled in the art should understand that the scope of the invention is not limited to the specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

Claims

1. A drone-assisted computational unloading method, characterized in that, The method includes the following steps: Step S1: Establish a drone-assisted computational offloading model, which includes an IoT device layer, an edge layer, and a cloud layer; wherein, The IoT device layer includes several IoT devices with different battery capacities, and these IoT devices are immovable. This indicates that within time slot t, IoT device k continuously generates computing tasks. It is expressed by the following formula: and reports the computation request to the drone; the size of task m is determined by express; The edge layer includes a group of drones, used This indicates that the drone provides communication relay and in-transit computing for IoT devices; the drone is equipped with an edge computing server, which is used to determine offloading scenarios and provide computation offloading in edge layer computing scenarios; The cloud layer includes high-performance servers, supercomputing centers and / or artificial intelligence computing clusters, receives IoT data relayed from drones, provides high-quality services for computationally intensive tasks, and transmits the computing results back to the drones through base stations; Step S2, in the UAV-assisted calculation unloading model, adopting This indicates whether drone u provides services to IoT device k within time slot t; if IoT device k is connected to drone u, then... ,otherwise ; The following limitations apply to the connection between drones and IoT devices: i) There are no remaining tasks to be processed at the end of time slot t; ii) Each IoT device can only connect to one drone in the same time slot; Expressed as a formula: （1） Where U represents the total number of drones; Represents any value; To ensure that each IoT device can only connect to one drone in the same time slot, let the three-dimensional coordinates of drone u be represented as follows: , , and These are the X, Y, and Z coordinates of the drone u; IoT devices are immovable, and their three-dimensional coordinates are represented as... ;in, , , and It is a constant; and: （2） in, It is the distance threshold between IoT devices and drones; If the distance between the IoT device and the drone exceeds The drone will refuse access to IoT devices; Step S3: Under the drone-assisted computation offloading model, three task computation offloading scenarios are established, including: local IoT device computation scenario, drone-assisted edge computing scenario, and cloud-edge-device computing scenario; and a single computation task is processed entirely in one scenario. Step S4: Based on the established three task computation offloading scenarios, with the goal of minimizing the weighted sum of latency and energy consumption of all tasks, the optimal drone for serving the IoT devices is determined by combining the transmit power and task processing parameters of the IoT devices, the transmit power and task processing parameters of the drone, the available energy of the drone and IoT devices, the size of the task to be processed, and the tolerable latency. The drone then calculates the optimal computation offloading scenario. Specifically, this includes: By connecting IoT devices to appropriate drones to form a network and having the drones decide on suitable task unloading locations, the weighted sum of latency and energy consumption for all tasks can be minimized. This can be expressed as a problem. : （14） in, , and These indicate whether task m is processed on a local IoT device, a drone, or a cloud server, respectively. and These represent the total latency of local IoT devices, drones, and cloud servers, respectively. and These represent the weighted energy consumption of local IoT devices, drones, and cloud servers, respectively. Indicates the delay factor; In the question In this context, the selection of drones for IoT devices to connect and the decision-making process for task offloading are coupled, and this leads to problems. It is decomposed into multiple UAV selection sub-problems P1 and task offloading decision sub-problems P2; and, The subproblem P1 minimizes the weighted transmit power of the IoT devices and the drone by connecting the IoT devices to the appropriate drone in each time slot t, as shown in the formula: （15） in, Indicates the drone's transmission power. This represents the transmit power of an IoT device, which remains constant within a given time slot t; the output of subproblem P1 is a set. This displays the connection status between the IoT device and the drone; It is the importance coefficient of IoT device k. It is the importance coefficient of the drone u; Subproblem P2 minimizes the weighted sum of energy consumption and latency of all tasks performed by the same UAV u within time slot t, and its formula is: （16） Results of the IoT device clustering phase This significantly affects the input of subproblem P2; the output of subproblem P2 is... , or This indicates whether task m is processed on a local IoT device, a drone, or a cloud server. This indicates a scenario where task computation is unloaded. The subproblem P1 is solved using an IoT device clustering algorithm based on simulated annealing; the solution process includes: Step S411: Establish the optimization function and connect the optimization function with the subproblems. Alignment, i.e. （17） Step S412: Input the drone set U, the IoT device set K, and the drone location. and IoT device location ; Step S413, randomly initialize the random set Annealing temperature And the cooling coefficient α, to obtain the initial optimization function ; Step S414, generate new A new optimization function is obtained. and calculate ; Step S415, if If the change is accepted, it will be accepted; otherwise, according to the Metropolis rules, the probability of accepting the change is determined by the following formula: （18） Step S416, update the annealing temperature T1 so that the latest temperature ; Indicates the cooling rate; Step S417, determine the annealing temperature Has the temperature dropped below the specified cooling threshold? If yes, proceed to step S418; otherwise, return to step S415. Step S418: Output the optimized result set. To determine which IoT devices and their related tasks the drone u will handle within time slot t; Step S5: Complete the calculation and uninstallation under the optimal uninstallation scenario.

2. The unmanned aerial vehicle-assisted calculation unloading method according to claim 1, characterized in that, The calculation process in step S3, under the local IoT device computing scenario, is as follows: set up Indicates whether task m is processed on the local IoT device k; If task m is processed on a local IoT device, then ,otherwise ; The delay from task m's generation to completion only considers computational delay, as shown in the following formula: （3） in, It is the total latency of local IoT device k processing task m. This is the time consumed by the local IoT device k in processing task m. This represents the CPU frequency of the local IoT device k. This represents the CPU cycles required to process 1 bit of data. Weighted energy consumption of the entire process for: （4） in, It is the effective switching capacitor for IoT device k. It is the importance coefficient of IoT device k, and all local IoT devices share the same parameters in the model.

3. The unmanned aerial vehicle-assisted calculation unloading method according to claim 1, characterized in that, The calculation process in step S3 for the UAV-assisted edge computing scenario is as follows: set up Indicates whether task m is being handled by the drone; If the drone is handling task m, then ,otherwise ; In this scenario, the computational task is handled by an edge computing server deployed on the drone; IoT device k generates task m and sends the raw data... The data is transmitted to the corresponding UAV u, which then processes the data and sends the result data within time slot t. The data is returned to IoT devices; therefore, this is a drone-assisted edge computing model. The total delay is given by the following formula （5） in, , and These represent the data transmission latency, result reception latency, and drone processing latency of task m at IoT device k, respectively. Drone processing latency Represented as: （6） in, The number of CPU cycles required to process 1 bit of data for a drone. The frequency of the drone's CPU; The data transmission delay and the result reception delay are expressed as follows: （7） in, and These are the uplink and downlink transmission rates between the drone u and the IoT device k, and: （8） in, and These are used to represent the uplink and downlink bandwidths, respectively. Indicates propagation loss. and These represent the transmission power of IoT devices and drones, respectively. and These represent the terminal Gaussian noise power of the drone and the IoT device, respectively. Weighted energy consumption of the entire process Energy transmission by IoT devices Drones transmit energy and processing energy Composed of, and obtained through the following means: （9） in, For the effective switching capacitor of the UAV u, It is the importance coefficient of drone u, and all drones share the same parameter; It is the importance coefficient of IoT device k.

4. The unmanned aerial vehicle-assisted calculation unloading method according to claim 3, characterized in that, The computation process in the cloud-edge-device computing scenario in step S3 is as follows: set up This indicates whether task m is processed on a cloud server; if task m is processed on a cloud server, then... ,otherwise ; In this scenario, IoT device k creates a computing task m and sends raw data to drone u; then, the drone relays the data to a cloud server via a base station; once the cloud server completes the task processing, it sends the result back to the drone, which finally returns the result to IoT device k within time slot t; the total latency of the entire process is... It can be obtained through the following formula: （10） in, and This refers to the end-to-end total transmission latency and total reception latency of data from IoT devices to cloud servers; and the processing latency of the cloud servers. Considered as 0, and Given by the following formula （11） in, It is the propagation delay from the cloud to the base station. This indicates the size of the calculation result returned by the cloud server; and These are the uplink and downlink transmission rates between the drone u and the IoT device k. and This represents the uplink and downlink transmission rates between the drone and the base station; the weighted energy consumption of the process. The following formula is used to derive... （12） in, and These refer to the additive transmission energy consumption of drones and IoT devices when sending raw data and receiving results; and the energy consumption of the cloud. Treated as 0, the rest are obtained using the following formula: （13） in, and These represent the transmission power of IoT devices and drones, respectively. It refers to the transmission power of the drone to the base station. It is the importance coefficient of IoT device k. It is the importance coefficient of the drone u.

5. The unmanned aerial vehicle-assisted calculation unloading method according to claim 1, characterized in that, Subproblem P2 is solved using a task offloading decision algorithm based on the Proximity Policy Optimization (PPO) algorithm; the solution process includes: Step S421, input the available energy of the drone u and the IoT device. Actor's online learning rate Commentator's online learning rate The CLIP parameter ε, discount factor γ, and computational requirements G for all tasks are considered. The total number of segments is W, with each segment requiring N iterations of the actor network and I iterations of the commentator network. Step S422, initialize actor network parameters and commentator network parameters ; Step S423, collect trajectory Use the current strategy ,make ; Step S424, in order to improve learning stability and reduce variance, for each state Calculate the advantage function And given by the following formula: （19） in, Indicates the trajectory in state The rewards obtained below; Indicates from state to state Discount factor; Indicates the trajectory in state The rewards obtained below; Indicates the use of parameters The neural network estimates the state The value function; Step S425: The actor network randomly selects an action from the activity set of the current state, interacts with the environment, and obtains the next state and reward function; Step S426, calculate the actor network sampling rate, which is obtained by the following formula: （20） in, Represents the state in the nth training iteration. The ratio of sampling rates, Indicates the use of parameters New actor network, Indicates that parameters are still used. The original actors online; Step S427, calculate the actor network loss function, whose objective function is （21） Update the θ-related parameters using the Adam optimizer; loss Obtained from the following formula: （22） Update actor network parameters θ. ; In step S428, the commentator network updates its parameters using the Adam optimizer. To improve the accuracy of its assessment of actors' online behavior; its loss function The computation relies on long-term rewards provided by the environment: （23） Update the parameters of the commentator network using a loss function. ; Step S429: Determine if the maximum number of iterations W has been reached; if not, return to step S423; if yes, output the result: task calculation unloading scenario. .

6. A drone-assisted computational unloading system, characterized in that, The system is used to execute the computation offloading method as described in any one of claims 1 to 5; the system includes: an IoT device layer, an edge layer, a cloud layer, a drone connection module, and a scene selection module; wherein, The IoT device layer includes several IoT devices with different battery capacities, and these IoT devices are immovable. This indicates that within time slot t, IoT device k continuously generates computing tasks. It is expressed by the following formula: and reports the computation request to the drone; the size of task m is determined by It also indicates that computation offloading is used in local IoT device computing scenarios; The edge layer includes a group of drones, used This indicates that the drone provides communication relay and in-transit computing for IoT devices; the drone is equipped with an edge computing server, which is used to determine offloading scenarios and provide computation offloading in edge layer computing scenarios; The cloud layer includes high-performance servers, supercomputing centers and / or artificial intelligence computing clusters, used to receive IoT data forwarded by drones, perform computation offloading in the cloud-edge-device computing scenario, and transmit the computation results back to the drones through base stations; The drone connection module defines This indicates whether drone u provides services to IoT device k within time slot t; if IoT device k is connected to drone u, then... ,otherwise ; The scenario selection module is used to select the most suitable drone for serving the IoT device based on the three established task computation offloading scenarios, with the goal of minimizing the weighted sum of latency and energy consumption of all tasks. It combines the transmission power and task processing parameters of IoT devices, the transmission power and task processing parameters of drones, the available energy of drones and IoT devices, the size of the task to be processed, and the tolerable latency, and in conjunction with the drone connection module, calculates the optimal computation offloading scenario. Then, based on the selected optimal computation offloading scenario, the corresponding computation offloading function in the network device layer, edge layer, or cloud layer is activated.

7. A computing storage medium, characterized in that, The medium stores the drone-assisted computational unloading system as described in claim 6.

Images