Unmanned aerial vehicle assisted private network task offloading and resource optimization method and device

Abstract

The invention provides an unmanned aerial vehicle assisted private network task unloading and resource optimization method and device, and relates to the technical field of unmanned aerial vehicle communication. The method comprises the following steps: constructing a calculation task processing delay and energy consumption model based on an unloading strategy; an unmanned aerial vehicle trajectory optimization and user task unloading strategy optimization algorithm based on DDPG is constructed according to the model; establishing an unmanned aerial vehicle perception model, and generating perception calculation task data through the unmanned aerial vehicle perception model; a dichotomy cross entropy loss function is adopted to optimize decision making and unloading of an action space in a DDPG-based unmanned aerial vehicle trajectory optimization and user task unloading strategy optimization algorithm, and an optimal unloading strategy and an optimal unloading proportion are obtained; and repeatedly and iteratively executing the unmanned aerial vehicle trajectory optimization and user task unloading strategy optimization algorithm based on the DDPG until the award function converges, and generating an optimal unmanned aerial vehicle trajectory flight and user calculation unloading strategy. According to the invention, tasks can be efficiently distributed between the ground and the aerial platform.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicle (UAV) communication technology, and in particular to a method and device for unloading tasks and optimizing resources in a UAV-assisted dedicated network. Background Art

[0002] In the digital age, drone communication technology is increasingly being used, playing a crucial role in emergency communications, intelligent transportation, environmental monitoring, and specialized industrial IoT scenarios. With the growing demand for data processing and computing tasks from ground users, traditional communication network architectures face significant challenges. Existing ground communication networks often suffer from high latency, high energy consumption, and network congestion when handling computationally intensive tasks. For example, in emergency communications scenarios, ground base stations may be damaged by disasters, leading to communication interruptions. Traditional communication methods, which rely on ground stations and satellites, struggle to meet user demands for real-time data transmission and rapid computing due to limited ground station coverage, high satellite communication costs, and significant latency. Summary of the Invention

[0003] To address the technical issues of existing technologies, such as high latency, high energy consumption, and network congestion, and their inability to meet user demands for fast computation and processing of real-time data transmission, the present invention provides a method and apparatus for unloading tasks and optimizing resources in a drone-assisted dedicated network. The technical solution is as follows:

[0004] In one aspect, a method for unmanned aerial vehicle (UAV)-assisted dedicated network task offloading and resource optimization is provided. The method is implemented by a UAV-assisted dedicated network task offloading and resource optimization device, and the method includes:

[0005] S1. Construct a computing task processing delay and energy consumption model based on offloading strategy;

[0006] S2. Based on the computational task processing delay and energy consumption model, construct a DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm, and define the state space, action space, and reward function;

[0007] S3. Establish a UAV perception model; generate perception computing task data based on the UAV perception model;

[0008] S4. Based on the perception computing task data, a binary cross-entropy loss function is used to optimize the decision-making and offloading of the action space in the DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm to obtain the optimal offloading strategy and optimal offloading ratio;

[0009] S5. Based on the optimal offloading strategy and optimal offloading ratio, repeatedly iteratively execute the DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm until the reward function converges, generating the optimal drone trajectory flight and user computation offloading strategy.

[0010] Optionally, the construction of S1 is based on a computing task processing delay and energy consumption model of an offloading strategy, including:

[0011] S11. Initialize the location of drones, users, and environmental perception events; initialize the data task volume for each user;

[0012] S12. Calculate the computational latency and energy consumption of the full local computation based on the amount of data selected for local computation by the user.

[0013] S13. Calculate the offloading decision for each time slot; calculate the computational delay and energy consumption of the local computation part based on the offloading decision for each time slot;

[0014] S14. Calculate the computational delay and energy consumption of the drone’s computing part based on the data task volume of each user;

[0015] S15. Based on the computational delay and energy consumption of the full local computation, the computational delay and energy consumption of the local computation part, and the computational delay and energy consumption of the UAV computation part, a computational task processing delay and energy consumption model based on the offloading strategy is constructed.

[0016] Optionally, the computing task processing delay and energy consumption model based on the offloading strategy is expressed by the following formula (1)-formula (2):

[0017] (1)

[0018] (2)

[0019] in, Represents a user The computing task processing delay; Represents a user Energy consumption of computing task processing; Represents a user Choose full local computing task processing delay; Represents a user Select all local computing tasks to process energy consumption; Represents a user Uninstallation strategy; Represents a user Select some of the offloaded computing tasks to handle delays; Represents a user Select to partially offload computing tasks to handle energy consumption.

[0020] Optionally, the state space expression is expressed by the following formula (3):

[0021] (3)

[0022] Among them, state represents the state space of the DDPG algorithm; Represents a user Coordinate location; Indicates the coordinate position of the drone; Indicates the Users at time The amount of data generated;

[0023] The expression of the action space is expressed by the following formula (4):

[0024] (4)

[0025] in, represents the uninstall decision of each user; k represents the number of users;

[0026] The reward function is expressed by the following formula (5):

[0027] (5)

[0028] Among them, reward represents the reward function of the DDPG algorithm; represents the weight of energy consumption; Indicates Moment User Total energy consumption generated; Indicates the total running time of the UAV executing the system task; Indicates Moment User The total delay incurred.

[0029] Optionally, the process of generating the perception computing task data is expressed by the following formula (6):

[0030] (6)

[0031] in, Indicates that the drone is The amount of perception computing task data generated at each moment; M represents the total number of environmental perception events; s represents the amount of data carried by each perceived event in the environment; is a binary perception variable used to determine whether perception occurs.

[0032] Optionally, the S4 adopts a binary cross entropy loss function to optimize the decision and offloading of the action space in the DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm to obtain the optimal offloading strategy and optimal offloading ratio, including:

[0033] S41. Discretize the continuous values of the action space output in the DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm;

[0034] The discretization process is expressed by the following formula (7):

[0035] (7)

[0036] in, , indicating a partial uninstallation of User tasks; , indicating the All tasks of each user are calculated locally; represents the user task offloading strategy after discretization; Represents the user uninstallation policy value output by the DDPG algorithm action space;

[0037] S42. A binary cross entropy loss function is introduced into the optimization objective of actor network training. By encouraging the action space output to be closer to the discrete decision 0 or 1, the decision and offloading of the action space in the DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm are optimized to obtain the optimal offloading strategy and the optimal offloading ratio.

[0038] Optionally, the binary cross entropy loss function is expressed by the following formula (8):

[0039] (8)

[0040] in, represents the binary cross entropy function introduced in the actor network; N represents the number of data samples processed by the neural network; K represents the total number of users; Indicates the In the sample The change of each user, 0 or 1; Indicates the In the sample Output of uninstall actions for each user;

[0041] The optimal unloading ratio is expressed by the following formula (9):

[0042] (9)

[0043] in, Indicates the optimal unloading ratio; Indicates the number of CPU cycles required for the user to process 1 bit of data; Represents the local computing resources of the drone; Indicates the user's data transmission rate; express; Represents the user's local computing resources.

[0044] On the other hand, a drone-assisted dedicated network task offloading and resource optimization device is provided, which is applied to a drone-assisted dedicated network task offloading and resource optimization method, and the device includes:

[0045] The first construction unit is used to construct a computing task processing delay and energy consumption model based on the offloading strategy;

[0046] The second construction unit is used to construct a DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm based on the computing task processing delay and energy consumption model, and define the state space, action space and reward function;

[0047] A generation unit is used to establish a UAV perception model; based on the UAV perception model, it generates perception computing task data;

[0048] The optimization unit is used to optimize the decision-making and offloading of the action space in the DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm based on the perception computing task data and using the binary cross-entropy loss function to obtain the optimal offloading strategy and optimal offloading ratio;

[0049] The acquisition unit is used to repeatedly iteratively execute the DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm according to the optimal offloading strategy and optimal offloading ratio until the reward function converges, thereby generating the optimal drone trajectory flight and user computation offloading strategy.

[0050] Optionally, the first building unit is used to:

[0051] Initialize the locations of drones, users, and environmental perception events; initialize the data task volume for each user;

[0052] Calculate the computational latency and energy consumption of full local computing based on the amount of data the user chooses to compute locally.

[0053] Calculate the offloading decision for each time slot; calculate the computational delay and energy consumption of the local computing part based on the offloading decision for each time slot;

[0054] Calculate the computing delay and energy consumption of the drone computing part based on the data task volume of each user;

[0055] According to the computational delay and energy consumption of all-local computing, the computational delay and energy consumption of the local computing part, and the computational delay and energy consumption of the UAV computing part, a computing task processing delay and energy consumption model based on the offloading strategy is constructed.

[0056] Optionally, the computing task processing delay and energy consumption model based on the offloading strategy is expressed by the following formula (1)-formula (2):

[0057] (1)

[0058] (2)

[0059] in, Represents a user The computing task processing delay; Represents a user Energy consumption of computing task processing; Represents a user Choose full local computing task processing delay; Represents a user Select all local computing tasks to process energy consumption; Represents a user Uninstallation strategy; Represents a user Select some of the offloaded computing tasks to handle delays; Represents a user Select to partially offload computing tasks to handle energy consumption.

[0060] Optionally, the state space expression is expressed by the following formula (3):

[0061] (3)

[0062] Among them, state represents the state space of the DDPG algorithm; Represents a user Coordinate location; Indicates the coordinate position of the drone; Indicates the Users at time The amount of data generated;

[0063] The expression of the action space is expressed by the following formula (4):

[0064] (4)

[0065] in, represents the uninstall decision of each user; k represents the number of users;

[0066] The reward function is expressed by the following formula (5):

[0067] (5)

[0068] Among them, reward represents the reward function of the DDPG algorithm; represents the weight of energy consumption; Indicates Moment User Total energy consumption generated; Indicates the total running time of the UAV executing the system task; Indicates Moment User The total delay incurred.

[0069] Optionally, the process of generating the perception computing task data is expressed by the following formula (6):

[0070] (6)

[0071] in, Indicates that the drone is The amount of perception computing task data generated at each moment; M represents the total number of environmental perception events; s represents the amount of data carried by each perceived event in the environment; is a binary perception variable used to determine whether perception occurs.

[0072] Optionally, the optimization unit is configured to:

[0073] Discretize the continuous values of the action space output in the DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm;

[0074] The discretization process is expressed by the following formula (7):

[0075] (7)

[0076] in, , indicating a partial uninstallation of User tasks; , indicating the All tasks of each user are calculated locally; represents the user task offloading strategy after discretization; Represents the user uninstallation policy value output by the DDPG algorithm action space;

[0077] A binary cross entropy loss function is introduced into the optimization objective of actor network training. By encouraging the action space output to be closer to the discrete decision 0 or 1, the decision and offloading of the action space in the DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm are optimized to obtain the optimal offloading strategy and the optimal offloading ratio.

[0078] Optionally, the binary cross entropy loss function is expressed by the following formula (8):

[0079] (8)

[0080] in, represents the binary cross entropy function introduced in the actor network; N represents the number of data samples processed by the neural network; K represents the total number of users; Indicates the In the sample The change of each user, 0 or 1; Indicates the In the sample Output of uninstall actions for each user;

[0081] The optimal unloading ratio is expressed by the following formula (9):

[0082] (9)

[0083] in, Indicates the optimal unloading ratio; Indicates the number of CPU cycles required for the user to process 1 bit of data; Represents the local computing resources of the drone; Indicates the user's data transmission rate; express; Represents the user's local computing resources.

[0084] On the other hand, a drone-assisted dedicated network task offloading and resource optimization device is provided, which includes: a processor; a memory, on which computer-readable instructions are stored, and when the computer-readable instructions are executed by the processor, any one of the above-mentioned drone-assisted dedicated network task offloading and resource optimization methods is implemented.

[0085] On the other hand, a computer-readable storage medium is provided, wherein the storage medium stores at least one instruction, and the at least one instruction is loaded and executed by a processor to implement any one of the above-mentioned drone-assisted dedicated network task offloading and resource optimization methods.

[0086] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0087] The embodiment of the present invention first constructs a computing task processing delay and energy consumption model based on an offloading strategy; based on the computing task processing delay and energy consumption model, constructs a DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm, defines the state space, action space and reward function; establishes a UAV perception model; secondly, based on the UAV perception model, generates perception computing task data; based on the perception computing task data, adopts a binary cross entropy loss function to optimize the decision and offloading of the action space in the DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm, and obtains the optimal offloading strategy and the optimal offloading ratio; finally, based on the optimal offloading strategy and the optimal offloading ratio, repeatedly iteratively executes the DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm until the reward function converges, thereby generating the optimal UAV trajectory flight and user computing offloading strategy.

[0088] The embodiment of the present invention constructs a DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm, jointly optimizes the drone trajectory and user computing task offloading strategy, dynamically adapts to environmental changes and task load balancing, and realizes intelligent and efficient resource scheduling; by establishing collaborative perception and an intelligent decision-making mechanism for communication and computing, it realizes efficient allocation of tasks between ground and air platforms, significantly reducing overall network latency and energy consumption. In a drone-assisted dedicated network environment, the use of the embodiment of the present invention can achieve low-latency and low-energy computing offloading, effectively improving the overall performance and reliability of the drone network. The embodiment of the present invention can realize dynamic allocation of computing resources in environments such as sudden disaster relief, remote inspections, and emergency support, meet response timeliness requirements, and effectively improve the service reliability and continuity level of the network. BRIEF DESCRIPTION OF THE DRAWINGS

[0089] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. 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 creative work.

[0090] Figure 1 This is a flow chart of a method for unloading tasks and optimizing resources in a drone-assisted dedicated network provided by an embodiment of the present invention;

[0091] Figure 2 This is a schematic diagram of an architecture for drone-assisted dedicated network task offloading and resource optimization provided by an embodiment of the present invention;

[0092] Figure 3This is a specific implementation flow chart of a drone-assisted dedicated network task offloading and resource optimization provided by an embodiment of the present invention;

[0093] Figure 4 This is a block diagram of a UAV-assisted dedicated network task offloading and resource optimization device provided by an embodiment of the present invention;

[0094] Figure 5 This is a structural diagram of a drone-assisted dedicated network task offloading and resource optimization device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0095] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0096] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as an "exemplary" in the present invention should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of the word "exemplary" is intended to present concepts in a concrete manner. Furthermore, in the embodiments of the present invention, "and / or" can mean both or either of the two.

[0097] In the embodiments of the present invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same. The terms "of," "corresponding," and "corresponding" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same.

[0098] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0099] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0100] The embodiment of the present invention provides a method for unmanned aerial vehicle-assisted dedicated network task offloading and resource optimization, which can be implemented by a unmanned aerial vehicle-assisted dedicated network task offloading and resource optimization device, which can be a terminal or a server. Figure 1 The flowchart of the method for unmanned aerial vehicle-assisted dedicated network task offloading and resource optimization is shown. The processing flow of the method may include the following steps:

[0101] S1. Construct a computing task processing delay and energy consumption model based on offloading strategy.

[0102] Among them, such as Figure 2 Figure 1 shows an architecture diagram of a drone-assisted dedicated network task offloading and resource optimization architecture provided by an embodiment of the present invention. The architecture includes: one drone, k ground users, and M environmental perception events. The notification perception period is T, and the communication perception is divided into I time slots.

[0103] Optionally, S1 is constructed based on the computing task processing delay and energy consumption model of the offloading strategy, including S11-S15:

[0104] S11. Initialize the location of drones, users, and environmental perception events; initialize the data task volume for each user;

[0105] In one feasible implementation, the drone is initialized ,user and environmentally aware event locations ; Among them, the position of the user and the environment perception event remains unchanged during the communication cycle, The UAV is flying at a fixed altitude. Initialize the data task volume of each user, where the data task volume of each user can be expressed as .

[0106] In one feasible implementation, define The amount of calculation offloaded by the user to the drone is expressed by the following formula (1):

[0107] (1)

[0108] in, represents the total number of data that are partially unloaded in the i-th time slot when the offloading strategy is 1; Indicates the uninstall ratio, which is calculated based on the minimum delay; Indicates the amount of computation that users offload to the drone;

[0109] S12. Calculate the computational latency and energy consumption of the full local computation based on the amount of data selected for local computation by the user.

[0110] Here, a binary number is used to represent the user's uninstall decision, which is expressed by the following formula (2):

[0111] (2)

[0112] in, Indicates a user's uninstall decision.

[0113] In one feasible implementation, the computational task processing delay and energy consumption model based on the offloading strategy includes full local computation and partial offloading computation; wherein, the partial offloading computation includes the amount of loaded data offloaded to the drone and the amount of loaded data left for local computation. The computational delay and energy consumption of full local computation are expressed by the following formulas (3)-(4):

[0114] (3)

[0115] (4)

[0116] in, Indicates the amount of data that the user chooses to compute entirely locally; Indicates the computational latency of fully local computation; Represents the user's local computing resources; Indicates the number of CPU cycles required for the user equipment to process 1 bit of data; Indicates the user's uninstall decision; Indicates the data task volume of each user; Indicates the energy consumption of all-local computing; Indicates the CPU capacitance coefficient of the user device.

[0117] S13. Calculate the offloading decision for each time slot; calculate the computational delay and energy consumption of the local computation part based on the offloading decision for each time slot;

[0118] In a feasible implementation, the computational delay and energy consumption of the local computation part are expressed by the following formulas (5)-(6):

[0119] (5)

[0120] (6)

[0121] in, represents the computational latency required to offload local computational tasks retained by users; Indicates the uninstall ratio.

[0122] S14. Calculate the computational delay and energy consumption of the drone’s computing part based on the data task volume of each user;

[0123] In a feasible implementation, the computational delay and energy consumption of the UAV computing part are expressed by the following formulas (7)-(8):

[0124] (7)

[0125] (8)

[0126] in, Indicates The computational delay required to offload the computation from the user to the drone at any given moment; Indicates that the drone is The amount of sensory data generated at any moment; Indicates the number of CPU cycles required for the drone to process 1 bit of data; Indicates Energy consumption generated by drone calculation at every moment; represents the computing resources of the UAV; Indicates the CPU capacitance coefficient of the drone.

[0127] S15. Based on the computational delay and energy consumption of the full local computation, the computational delay and energy consumption of the local computation part, and the computational delay and energy consumption of the UAV computation part, a computational task processing delay and energy consumption model based on the offloading strategy is constructed.

[0128] In a feasible implementation, the delay and energy consumption of the partially offloaded calculation are obtained based on the calculation delay and energy consumption of the local calculation part and the calculation delay and energy consumption of the UAV calculation part, which are expressed by the following formulas (9)-(10):

[0129] (9)

[0130] (10)

[0131] in, Indicates the delay caused by the user transmitting data to the drone; Represents the energy consumption generated by users transmitting data to drones.

[0132] Optionally, the computing task processing delay and energy consumption model based on the offloading strategy is expressed by the following formula (11)-formula (12):

[0133] (11)

[0134] (12)

[0135] in, Represents a user The computing task processing delay; Represents a user Energy consumption of computing task processing; Represents a user Choose full local computing task processing delay; Represents a user Select all local computing tasks to process energy consumption; Represents a user Uninstallation strategy; Represents a user Select some of the offloaded computing tasks to handle delays; Represents a user Select to partially offload computing tasks to handle energy consumption.

[0136] S2. Based on the computational task processing delay and energy consumption model, a DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm is constructed, and the state space, action space, and reward function are defined.

[0137] Among them, the state space of the DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm includes: the positions of all users and drones and the data volume of all users.

[0138] Alternatively, the state space expression is expressed by the following formula (13):

[0139] (13)

[0140] Among them, state represents the state space of the DDPG algorithm; Represents a user Coordinate location; Indicates the coordinate position of the drone; Indicates the Users at time The amount of data generated;

[0141] Among them, the drone flies at a constant speed, so the drone flight energy consumption is only related to time, and the drone's flight direction is defined as the action.

[0142] The expression of the action space is expressed by the following formula (14):

[0143] (14)

[0144] in, represents the uninstall decision of each user; k represents the number of users;

[0145] The reward function is expressed as follows:

[0146] (15)

[0147] Among them, reward represents the reward function of the DDPG algorithm; represents the weight of energy consumption; Indicates Moment User Total energy consumption generated; Indicates the total running time of the UAV executing the system task; Indicates Moment User The total delay incurred.

[0148] S3. Establish a drone perception model; generate perception computing task data based on the drone perception model.

[0149] In a feasible implementation, the perception task is modeled as the process of collecting and processing data about the surrounding environment during the flight of the UAV. The perceived environmental data is expressed as the amount of perceived data Assume that the amount of perception data generated by the drone when each event to be perceived in the environment is perceived. The probability of the drone perceiving the event is used as the condition for judging whether perception can be performed. The perception probability is based on distance and angle and is expressed by the following formula (16):

[0150] (16)

[0151] in, represents the perceived probability; represents the perception coefficient; Representing drones and perception events distance; Indicates the coordinate position of the drone; Representing a perceived event The coordinate position of Represents drone and ground perception events The angle between.

[0152] Among them, the occurrence of perception must meet the minimum perception probability requirement, that is, when the drone meets the minimum perception probability requirement of the perception device, it can receive accurate perception data.

[0153] In a feasible implementation, when the UAV's flight direction is updated and the perception conditions are met, the UAV performs environmental perception and generates perception computing task data through a perception model.

[0154] Optionally, the process of generating perceptual computing task data is expressed by the following formula (17):

[0155] (17)

[0156] in, Indicates that the drone is The amount of perception computing task data generated at each moment; M represents the total number of environmental perception events; s represents the amount of data carried by each perceived event in the environment; is a binary perception variable used to determine whether perception occurs.

[0157] S4. Use the binary cross entropy loss function to optimize the decision-making and offloading of the action space in the DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm to obtain the optimal offloading strategy and optimal offloading ratio.

[0158] Optionally, the specific implementation process of S4 includes S41-S42:

[0159] S41. Discretize the continuous values of the action space output in the DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm;

[0160] In a feasible implementation, since the DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm is designed for a continuous action space, and each user offloading decision is a binary discrete value, it is necessary to discretize the continuous numerical values outputted in the action space in the DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm.

[0161] The discretization process is expressed by the following formula (18):

[0162] (18)

[0163] in, , indicating a partial uninstallation of User tasks; , indicating the All tasks of each user are calculated locally; represents the user task offloading strategy after discretization; Represents the user uninstallation policy value output by the DDPG algorithm action space;

[0164] S42. A binary cross entropy loss function is introduced into the optimization objective of actor network training. By encouraging the action space output to be closer to the discrete decision 0 or 1, the decision and offloading of the action space in the DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm are optimized to obtain the optimal offloading strategy and the optimal offloading ratio.

[0165] Among them, in order to improve the accuracy of training optimization, binary cross entropy is introduced into the optimization target of actor network training to encourage the action space output to be closer to the discrete decision 0 or 1. Assuming that the ideal unloading strategy is .

[0166] Optionally, the binary cross entropy loss function is expressed by the following formula (19):

[0167] (19)

[0168] in, represents the binary cross entropy function introduced in the actor network; N represents the number of data samples processed by the neural network; K represents the total number of users; Indicates the In the sample The change of each user, 0 or 1; Indicates the In the sample Output of uninstall actions for each user.

[0169] Among them, when a single user unloads a task at a time, the task volume remains unchanged, so the task volume unloaded to the local and drone is inversely proportional, that is, , the best unloading ratio is obtained.

[0170] Alternatively, the optimal unloading ratio is expressed by the following formula (20):

[0171] (20)

[0172] in, Indicates the optimal unloading ratio; Indicates the number of CPU cycles required for the user to process 1 bit of data; Represents the local computing resources of the drone; Indicates the user's data transmission rate; express; Represents the user's local computing resources.

[0173] In one feasible implementation, the computational delay and energy consumption in the current time slot are calculated based on the obtained optimal offloading strategy and optimal offloading ratio. Steps S3-S4 are repeated cyclically each time the UAV updates its position status until the UAV stops flying.

[0174] S5. Based on the optimal offloading strategy and optimal offloading ratio, repeatedly iteratively execute the DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm until the reward function converges, generating the optimal drone trajectory flight and user computation offloading strategy.

[0175] In one feasible implementation, Figure 3This is a specific implementation flow chart of a drone-assisted dedicated network task offloading and resource optimization provided by an embodiment of the present invention, including: initializing the coordinate positions of the drone, user, and perception events, and calculating the task data volume parameters; constructing a drone trajectory optimization and user task offloading strategy optimization algorithm based on DDPG; updating the drone flight direction based on the drone trajectory optimization and user task offloading strategy optimization algorithm; according to the drone flight direction, the drone perceives environmental events and generates perception calculation task data; using a binary cross entropy loss function to optimize the drone trajectory optimization and user task offloading strategy optimization algorithm based on DDPG. , and judge whether the remaining time slot of the operation is 0. If not, the flight direction of the UAV is updated, the UAV perceives the environmental events, generates perception computing task data, and continues to use the binary cross entropy loss function to optimize the DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm; if the remaining time slot of the operation is 0, then judge whether the reward function of the DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm converges. If convergence is not reached, continue to perform the DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm; if convergence is reached, the optimal UAV trajectory flight and user computing offloading strategy are implemented.

[0176] The embodiment of the present invention first constructs a computing task processing delay and energy consumption model based on an offloading strategy; based on the computing task processing delay and energy consumption model, constructs a DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm, defines the state space, action space and reward function; establishes a UAV perception model; secondly, based on the UAV perception model, generates perception computing task data; based on the perception computing task data, adopts a binary cross entropy loss function to optimize the decision and offloading of the action space in the DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm, and obtains the optimal offloading strategy and the optimal offloading ratio; finally, based on the optimal offloading strategy and the optimal offloading ratio, repeatedly iteratively executes the DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm until the reward function converges, thereby generating the optimal UAV trajectory flight and user computing offloading strategy.

[0177] The embodiment of the present invention constructs a DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm, jointly optimizes the drone trajectory and user computing task offloading strategy, dynamically adapts to environmental changes and task load balancing, and realizes intelligent and efficient resource scheduling; by establishing collaborative perception and an intelligent decision-making mechanism for communication and computing, it realizes efficient allocation of tasks between ground and air platforms, significantly reducing overall network latency and energy consumption. In a drone-assisted dedicated network environment, the use of the embodiment of the present invention can achieve low-latency and low-energy computing offloading, effectively improving the overall performance and reliability of the drone network. The embodiment of the present invention can realize dynamic allocation of computing resources in environments such as sudden disaster relief, remote inspections, and emergency support, meet response timeliness requirements, and effectively improve the service reliability and continuity level of the network.

[0178] Figure 4 This is a block diagram of a drone-assisted dedicated network task offloading and resource optimization device according to an exemplary embodiment, which is used in a drone-assisted dedicated network task offloading and resource optimization method. Figure 4 The device includes a first construction unit 410, a second construction unit 420, a generation unit 430, an optimization unit 440, and an acquisition unit 450.

[0179] A first construction unit 410 is configured to construct a computing task processing delay and energy consumption model based on an offloading strategy;

[0180] A second construction unit 420 is configured to construct a DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm based on the computing task processing delay and energy consumption model, and define a state space, an action space, and a reward function;

[0181] The generating unit 430 is used to establish a UAV perception model and generate perception computing task data according to the UAV perception model;

[0182] An optimization unit 440 is configured to optimize the decision-making and offloading of the action space in the DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm using a binary cross entropy loss function based on the perception computing task data, thereby obtaining an optimal offloading strategy and an optimal offloading ratio;

[0183] The acquisition unit 450 is used to repeatedly iteratively execute the DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm according to the optimal offloading strategy and the optimal offloading ratio until the reward function converges, thereby generating the optimal drone trajectory flight and user computation offloading strategy.

[0184] Optionally, the first constructing unit 410 is configured to:

[0185] Initialize the locations of drones, users, and environmental perception events; initialize the data task volume for each user;

[0186] Calculate the computational latency and energy consumption of full local computing based on the amount of data the user chooses to compute locally.

[0187] Calculate the offloading decision for each time slot; calculate the computational delay and energy consumption of the local computing part based on the offloading decision for each time slot;

[0188] Calculate the computing delay and energy consumption of the drone computing part based on the data task volume of each user;

[0189] According to the computational delay and energy consumption of all-local computing, the computational delay and energy consumption of the local computing part, and the computational delay and energy consumption of the UAV computing part, a computing task processing delay and energy consumption model based on the offloading strategy is constructed.

[0190] Optionally, the computing task processing delay and energy consumption model based on the offloading strategy is expressed by the following formula (1)-formula (2):

[0191] (1)

[0192] (2)

[0193] in, Represents a user The computing task processing delay; Represents a user Energy consumption of computing task processing; Represents a user Choose full local computing task processing delay; Represents a user Select all local computing tasks to process energy consumption; Represents a user Uninstallation strategy; Represents a user Select some of the offloaded computing tasks to handle delays; Represents a user Select to partially offload computing tasks to handle energy consumption.

[0194] Optionally, the state space expression is expressed by the following formula (3):

[0195] (3)

[0196] Among them, state represents the state space of DDPG algorithm; Represents a user Coordinate location; Indicates the coordinate position of the drone; Indicates the Users at time The amount of data generated;

[0197] The expression of the action space is expressed by the following formula (4):

[0198] (4)

[0199] in, represents the uninstall decision of each user; k represents the number of users;

[0200] The reward function is expressed by the following formula (5):

[0201] (5)

[0202] Among them, reward represents the DDPG algorithm reward function; represents the weight of energy consumption; Indicates Moment User Total energy consumption generated; Indicates the total running time of the UAV executing the system task; Indicates Moment User The total delay incurred.

[0203] Optionally, the process of generating the perception computing task data is expressed by the following formula (6):

[0204] (6)

[0205] in, Indicates that the drone is The amount of perception computing task data generated at each moment; M represents the total number of environmental perception events; s represents the amount of data carried by each perceived event in the environment; is a binary perception variable used to determine whether perception occurs.

[0206] Optionally, the optimization unit 440 is configured to:

[0207] Discretize the continuous values of the action space output in the DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm;

[0208] The discretization process is expressed by the following formula (7):

[0209] (7)

[0210] in, , indicating a partial uninstallation of User tasks; , indicating the All tasks of each user are calculated locally; represents the user task offloading strategy after discretization; Represents the user uninstallation policy value output by the DDPG algorithm action space;

[0211] A binary cross entropy loss function is introduced into the optimization objective of actor network training. By encouraging the action space output to be closer to the discrete decision 0 or 1, the decision and offloading of the action space in the DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm are optimized to obtain the optimal offloading strategy and the optimal offloading ratio.

[0212] Optionally, the binary cross entropy loss function is expressed by the following formula (8):

[0213] (8)

[0214] in, represents the binary cross entropy function introduced in the actor network; N represents the number of data samples processed by the neural network; K represents the total number of users; Indicates the In the sample The change of each user, 0 or 1; Indicates the In the sample Output of uninstall actions for each user;

[0215] The optimal unloading ratio is expressed by the following formula (9):

[0216] (9)

[0217] in, Indicates the optimal unloading ratio; Indicates the number of CPU cycles required for the user to process 1 bit of data; Represents the local computing resources of the drone; Indicates the user's data transmission rate; express; Represents the user's local computing resources.

[0218] The embodiment of the present invention first constructs a computing task processing delay and energy consumption model based on an offloading strategy; based on the computing task processing delay and energy consumption model, constructs a DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm, defines the state space, action space and reward function; establishes a UAV perception model; secondly, based on the UAV perception model, generates perception computing task data; based on the perception computing task data, adopts a binary cross entropy loss function to optimize the decision and offloading of the action space in the DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm, and obtains the optimal offloading strategy and the optimal offloading ratio; finally, based on the optimal offloading strategy and the optimal offloading ratio, repeatedly iteratively executes the DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm until the reward function converges, thereby generating the optimal UAV trajectory flight and user computing offloading strategy.

[0219] The embodiment of the present invention constructs a DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm, jointly optimizes the drone trajectory and user computing task offloading strategy, dynamically adapts to environmental changes and task load balancing, and realizes intelligent and efficient resource scheduling; by establishing collaborative perception and an intelligent decision-making mechanism for communication and computing, it realizes efficient allocation of tasks between ground and air platforms, significantly reducing overall network latency and energy consumption. In a drone-assisted dedicated network environment, low-latency and low-energy computing offloading is achieved, effectively improving the overall performance and reliability of the network. The embodiment of the present invention can realize the dynamic allocation of computing resources in environments such as sudden disaster relief, remote inspections, and emergency support, meet the response timeliness requirements, and effectively improve the service reliability and continuity level of the network.

[0220] Figure 5 Schematic diagram of a UAV-assisted dedicated network task offloading and resource optimization device provided by an embodiment of the present invention. Figure 5 As shown, the drone-assisted dedicated network task offloading and resource optimization device may include the above Figure 4 The drone-assisted dedicated network task offloading and resource optimization device shown. Optionally, the drone-assisted dedicated network task offloading and resource optimization device 510 may include a first processor 2001.

[0221] Optionally, the drone-assisted dedicated network task offloading and resource optimization device 510 may also include a memory 2002 and a transceiver 2003 .

[0222] The first processor 2001, the memory 2002 and the transceiver 2003 may be connected via a communication bus.

[0223] The following combination Figure 5The components of the drone-assisted dedicated network task offloading and resource optimization device 510 are described in detail:

[0224] The first processor 2001 is the control center of the drone-assisted dedicated network task offloading and resource optimization device 510 and can be a single processor or a collective term for multiple processing elements. For example, the first processor 2001 can be one or more central processing units (CPUs), application-specific integrated circuits (ASICs), or one or more integrated circuits configured to implement embodiments of the present invention, such as one or more digital signal processors (DSPs) or one or more field programmable gate arrays (FPGAs).

[0225] Optionally, the first processor 2001 can perform various functions of the drone-assisted dedicated network task offloading and resource optimization device 510 by running or executing software programs stored in the memory 2002 and calling data stored in the memory 2002.

[0226] In a specific implementation, as an embodiment, the first processor 2001 may include one or more CPUs, such as Figure 5 CPU0 and CPU1 are shown in FIG.

[0227] In a specific implementation, as an embodiment, the drone-assisted dedicated network task offloading and resource optimization device 510 may also include multiple processors, such as Figure 5 1 and 2. The first processor 2001 and the second processor 2004 are shown in FIG. Each of these processors can be a single-core processor (single-CPU) or a multi-core processor (multi-CPU). A processor herein can refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0228] The memory 2002 is used to store the software program for executing the solution of the present invention, and is controlled by the first processor 2001 for execution. The specific implementation method can refer to the above method embodiment and will not be repeated here.

[0229] Alternatively, the memory 2002 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, a random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, an optical disc storage (including a compact disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and capable of being accessed by a computer, but not limited thereto. The memory 2002 may be integrated with the first processor 2001 or may exist independently and accessed through the interface circuit ( Figure 5 (not shown) is coupled to the first processor 2001, which is not specifically limited in this embodiment of the present invention.

[0230] The transceiver 2003 is used to communicate with a network device or a terminal device.

[0231] Optionally, the transceiver 2003 may include a receiver and a transmitter ( Figure 5 The receiver is used to implement a receiving function, and the transmitter is used to implement a sending function.

[0232] Optionally, the transceiver 2003 may be integrated with the first processor 2001 or may exist independently and be connected to the interface circuit of the drone-assisted dedicated network task offloading and resource optimization device 510 ( Figure 5 (not shown) is coupled to the first processor 2001, which is not specifically limited in this embodiment of the present invention.

[0233] It should be noted that Figure 5 The structure of the drone-assisted dedicated network task offloading and resource optimization device 510 shown in the figure does not constitute a limitation on the router. The actual knowledge structure identification device may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0234] In addition, the technical effects of the drone-assisted dedicated network task offloading and resource optimization device 510 can refer to the technical effects of the drone-assisted dedicated network task offloading and resource optimization method described in the above method embodiment, and will not be repeated here.

[0235] It should be understood that the first processor 2001 in the embodiment of the present invention may be a central processing unit (CPU), or may be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor, or the processor may be any conventional processor, etc.

[0236] It should also be understood that the memory in the embodiments of the present invention may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory may be random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0237] The above embodiments can be implemented in whole or in part via software, hardware (e.g., circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When loaded or executed on a computer, the processes or functions described in accordance with the embodiments of the present invention are fully or partially performed. 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 website, computer, server, or data center via wired means (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0238] It should be understood that the term "and / or" as used herein simply describes a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A alone, A and B together, or B alone. A and B can be singular or plural. Furthermore, the character " / " as used herein generally indicates an "or" relationship between the associated objects, but it may also indicate an "and / or" relationship. For specific understanding, please refer to the context.

[0239] In this disclosure, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, "at least one of a, b, or c" can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or plural.

[0240] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0241] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0242] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0243] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interface, indirect coupling or communication connection of the device or unit, which can be electrical, mechanical or other forms.

[0244] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0245] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0246] If the functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or the portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical disks.

[0247] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A method for unmanned aerial vehicle-assisted dedicated network task offloading and resource optimization, characterized in that: The method comprises: S1. Construct a computing task processing delay and energy consumption model based on offloading strategy; S2. Based on the computational task processing delay and energy consumption model, construct a DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm, and define the state space, action space, and reward function; S3. Establish a UAV perception model; generate perception computing task data based on the UAV perception model; S4. Based on the perception computing task data, a binary cross-entropy loss function is used to optimize the decision-making and offloading of the action space in the DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm to obtain the optimal offloading strategy and optimal offloading ratio; S5. Based on the optimal offloading strategy and optimal offloading ratio, repeatedly iteratively execute the DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm until the reward function converges, generating the optimal drone trajectory flight and user computation offloading strategy.

2. The method for unloading tasks and optimizing resources of a drone-assisted dedicated network according to claim 1, characterized in that: The construction of S1 is based on the computing task processing delay and energy consumption model of the offloading strategy, including: S11. Initialize the location of drones, users, and environmental perception events; initialize the data task volume for each user; S12. Calculate the computational latency and energy consumption of the full local computation based on the amount of data selected for local computation by the user. S13. Calculate the offloading decision for each time slot; calculate the computational delay and energy consumption of the local computation part based on the offloading decision for each time slot; S14. Calculate the computational delay and energy consumption of the drone’s computing part based on the data task volume of each user; S15. Based on the computational delay and energy consumption of the full local computation, the computational delay and energy consumption of the local computation part, and the computational delay and energy consumption of the UAV computation part, a computational task processing delay and energy consumption model based on the offloading strategy is constructed.

3. The method for unloading tasks and optimizing resources of a drone-assisted dedicated network according to claim 1, characterized in that: The computing task processing delay and energy consumption model based on the offloading strategy is expressed by the following formula (1)-formula (2): （1） （2） in, Represents a user The computing task processing delay; Represents a user Energy consumption of computing task processing; Represents a user Choose full local computing task processing delay; Represents a user Select all local computing tasks to process energy consumption; Represents a user Uninstall strategy; Represents a user Select some offloading computing tasks to handle delays; Represents a user Select Partial Offload to reduce energy consumption.

4. The method for unloading tasks and optimizing resources of a drone-assisted dedicated network according to claim 1, wherein: The state space expression is expressed by the following formula (3): （3） Among them, state represents the state space of the DDPG algorithm; Represents a user Coordinate location; Indicates the coordinate position of the drone; Indicates the Users at time The amount of data generated; The expression of the action space is expressed by the following formula (4): （4） in, represents the uninstall decision of each user; k represents the number of users; The reward function is expressed by the following formula (5): （5） Among them, reward represents the reward function of the DDPG algorithm; represents the weight of energy consumption; Indicates Moment User Total energy consumption generated; Indicates the total running time of the UAV in executing the system task; Indicates Moment User The total delay incurred.

5. The method for unloading tasks and optimizing resources of a drone-assisted dedicated network according to claim 1, wherein: The process of generating the perception computing task data is expressed by the following formula (6): （6） in, Indicates that the drone is The amount of perception computing task data generated at each moment; M represents the total number of environmental perception events; s represents the amount of data carried by each perceived event in the environment; is a binary perception variable used to determine whether perception occurs.

6. The method for unmanned aerial vehicle-assisted dedicated network task offloading and resource optimization according to claim 1, wherein said S4 uses a binary cross entropy loss function to optimize the decision-making and offloading of the action space in the DDPG-based unmanned aerial vehicle trajectory optimization and user task offloading strategy optimization algorithm to obtain the optimal offloading strategy and the optimal offloading ratio, including: S41. Discretize the continuous values of the action space output in the DDPG-based UAV trajectory optimization and user task offloading strategy optimization algorithm; The discretization process is expressed by the following formula (7): （7） in, , indicating a partial uninstallation of User tasks; , indicating the All tasks of each user are calculated locally; represents the user task offloading strategy after discretization; Represents the user uninstallation policy value output by the DDPG algorithm action space; S42. A binary cross entropy loss function is introduced into the optimization objective of actor network training. By encouraging the action space output to be closer to the discrete decision 0 or 1, the decision and offloading of the action space in the DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm are optimized to obtain the optimal offloading strategy and the optimal offloading ratio.

7. The method for unloading tasks and optimizing resources of a drone-assisted dedicated network according to claim 6, wherein: The binary cross entropy loss function is expressed by the following formula (8): （8） in, represents the binary cross entropy function introduced in the actor network; N represents the number of data samples processed by the neural network; K represents the total number of users; Indicates the In the sample The change of each user, 0 or 1; Indicates the In the sample Output of uninstall actions for each user; The optimal unloading ratio is expressed by the following formula (9): （9） in, Indicates the optimal unloading ratio; Indicates the number of CPU cycles required for the user to process 1 bit of data; Represents the local computing resources of the drone; Indicates the user's data transmission rate; express; Represents the user's local computing resources.

8. A drone-assisted dedicated network task offloading and resource optimization device, the drone-assisted dedicated network task offloading and resource optimization device is used to implement the drone-assisted dedicated network task offloading and resource optimization method according to any one of claims 1 to 7, characterized in that: The device comprises: The first construction unit is used to construct a computing task processing delay and energy consumption model based on the offloading strategy; The second construction unit is used to construct a DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm based on the computing task processing delay and energy consumption model, and define the state space, action space and reward function; A generation unit is used to establish a UAV perception model; based on the UAV perception model, it generates perception computing task data; The optimization unit is used to optimize the decision-making and offloading of the action space in the DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm based on the perception computing task data and using the binary cross-entropy loss function to obtain the optimal offloading strategy and optimal offloading ratio; The acquisition unit is used to repeatedly iteratively execute the DDPG-based drone trajectory optimization and user task offloading strategy optimization algorithm according to the optimal offloading strategy and optimal offloading ratio until the reward function converges, thereby generating the optimal drone trajectory flight and user computation offloading strategy.

9. A drone-assisted dedicated network task offloading and resource optimization device, characterized in that: The drone-assisted dedicated network task offloading and resource optimization device includes: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program code, which can be called by a processor to execute the method according to any one of claims 1 to 7.

Images