Trajectory and unloading combined unmanned aerial vehicle networking performance optimization method and related equipment

Abstract

The invention belongs to the field of Internet of Vehicles, and discloses an unmanned aerial vehicle Internet of Vehicles performance optimization method combining trajectory and unloading and related equipment, and the method comprises the steps: firstly constructing a communication and load model; constructing a multi-objective optimization function in combination with multiple calculation unloading modes, and setting constraint conditions such as an unloading decision, time delay, energy consumption and a flight area by taking minimization of total time delay and total energy consumption as an objective; and the function is solved by alternately iteratively updating the flight path and calculating the unloading mode until the maximum number of iterations is reached, and an optimal result is output. According to the method, the communication and load characteristics and multiple calculation unloading modes of the unmanned aerial vehicle auxiliary vehicle networking system are comprehensively considered, and comprehensive optimization of the system performance is realized through optimization functions and constraint conditions. According to the method, the problems of track optimization and quick response processing of calculation unloading in a complex scene can be effectively solved, the service quality of vehicle users is remarkably improved, the total time delay and the total energy consumption are reduced, and an effective solution is provided for performance improvement of an unmanned aerial vehicle networking system.

Description

Technical Field

[0001] The present invention belongs to the technical field of vehicle networking, and particularly relates to a method for optimizing the performance of an unmanned aerial vehicle (UAV) - vehicle networking by jointly considering trajectory and offloading, and related devices. Background Technique

[0002] In traditional vehicle - edge computing networks, it is common practice to rely on roadside units to process computing tasks for vehicles. However, complex urban environments pose many challenges to this computing mode. The occlusion of urban infrastructure, the presence of obstacles, and signal interference in high - density areas result in insufficient coverage gaps and deterioration of communication link quality, thereby increasing the latency of vehicle computing. For example, in the urban center area with high - rise buildings, the signals of roadside units are easily blocked by buildings, the communication stability between vehicles and roadside units is greatly reduced, and the processing speed of computing tasks significantly decreases.

[0003] To reduce the latency and energy consumption of the system, the computing offloading technology has emerged. The computing offloading technology in vehicle networking is to reasonably offload the computing tasks generated by vehicle users to the local, UAV - side, or cloud, thereby effectively reducing the total latency and total energy consumption of the system. However, after introducing UAV - assisted computing, the situation becomes more complex. When considering computing offloading, the energy consumption of UAVs needs to be taken into account. Most of the current research works consider statically deployed UAVs, or only consider the energy consumption problem of the system when optimizing the UAV trajectory and offloading strategy, ignoring the latency and energy consumption problems brought by the movement of UAV positions. In fact, the change of UAV positions not only affects its own energy consumption but also affects the communication quality between vehicles and UAVs, thereby affecting the total latency of the system. Moreover, there is a potential synergistic effect between latency and energy consumption. Considering only energy consumption while ignoring latency cannot achieve the overall optimization of system performance. Most of the current algorithms for researching computing offloading are improved single heuristic algorithms or reinforcement learning algorithms. Facing complex traffic intersections, single heuristic algorithms lack balance in search, making it difficult to find a good balance between global and local searches, and are prone to falling into local optimal solutions, unable to make full use of system resources. Although reinforcement learning algorithms can theoretically find the optimal solution, the training time for the model is long and the convergence speed is slow. In practical applications, a large amount of time and computing resources are required for training, making it difficult to meet the requirements of vehicle networking scenarios with high real - time requirements. With the increasing demand for computing tasks generated by vehicles on traffic roads in real life, the current methods are difficult to meet the reasonable offloading of a large number of computing tasks in dynamic scenarios. In dynamic scenarios with frequent traffic flow changes and diverse vehicle driving states, existing algorithms cannot quickly and accurately make computing offloading decisions according to the real - time needs of vehicles, resulting in the inability to effectively control the total latency and total energy consumption of the system, and it is difficult to meet the strict requirements of vehicle users for latency and energy consumption.

[0004] It can be seen that the existing trajectory optimization and computing offloading methods can no longer meet the requirements of rapid response processing in complex scenarios and improving the quality of service for vehicle users. Summary of the Invention

[0005] The present invention provides a method and related equipment for optimizing the performance of an unmanned aerial vehicle-roadside unit (UAV-RSU) network by jointly considering trajectory and offloading, so as to solve the technical problem that the existing trajectory optimization and computing offloading methods can no longer meet the requirements of rapid response processing in complex scenarios and improving the quality of service for vehicle users.

[0006] To achieve the above object, the present invention adopts the following technical solutions: A method for optimizing the performance of an unmanned aerial vehicle-roadside unit (UAV-RSU) network by jointly considering trajectory and offloading, comprising: Based on a pre-constructed UAV-RSU system model, constructing a communication model and a load model; Based on the communication model, the load model, and a predefined variety of computing offloading modes, constructing a multi-objective optimization function; wherein, the optimization objectives of the multi-objective optimization function are to minimize the total delay and the total energy consumption, and the constraint conditions at least include: offloading decision constraint, delay constraint, energy consumption constraint, and flight area constraint; Solving the multi-objective optimization function to alternately and iteratively update the flight trajectory and the computing offloading mode until the maximum number of iterations is reached, and outputting the total delay and the total energy consumption of the UAV-RSU system corresponding to the optimal flight trajectory and the optimal computing offloading mode.

[0007] Furthermore, Before constructing the communication model and the load model based on the pre-constructed UAV-RSU system model, it further includes: Constructing the UAV-RSU system model, which is used to reflect the communication, computing, and service requirements in the actual scenario; Wherein, the architecture of the UAV-RSU system model includes UAVs, base stations, SDN controllers, and multiple vehicle users; The topological structure of the UAV-RSU system model includes urban roads with several intersections and UAVs deployed on the urban roads; the UAVs are used to provide services for vehicle users and transmit tasks to the base stations; The SDN controller is used to perform computing offloading according to the size and throughput of the tasks after collecting the tasks; Wherein, the task generation method is as follows: Vehicle users are divided into multiple equal-length time slots within a preset time, and computing tasks are generated according to the Poisson distribution within each time slot; the tasks consist of the task size, the number of central processing unit cycles required for the task, the task return data size, and the maximum tolerable delay.

[0008] Furthermore, Based on the pre - constructed UAV - assisted vehicle - to - everything (V2X) system model, a communication model and a load model are constructed, including: Based on the pre - constructed UAV - assisted V2X system model, a communication model between the UAV and vehicle users, and between the UAV and the base station is constructed; Among them, in the communication model between the UAV and vehicle users, the transmission rate changes dynamically with distance, and the channel gain is calculated using the free - space path - loss model; In the communication model between the UAV and the base station, the transmission rate and the channel gain are determined by the line - of - sight link model; Based on the pre - constructed UAV - assisted V2X system model, a load model is constructed; Among them, the load model includes communication load and computing load; The communication load is used to represent the calculation of the channel gain from the vehicle to the UAV and the channel gain from the UAV to the base station; The computing load is used to represent the computing resources required for the task.

[0009] Furthermore, Based on the communication model, the load model, and a variety of predefined computing offloading modes, in constructing the multi - objective optimization function, the computing offloading modes include: Local computing mode, UAV computing offloading mode, and base - station computing offloading mode; where: In the local computing mode, the task is entirely computed on the in - vehicle electronic control unit of the vehicle user, and the execution time and the computing power of the vehicle user are determined by the in - vehicle electronic control unit of the vehicle user; In the UAV computing offloading mode, the vehicle user offloads the task to the UAV; the total energy consumption is equal to the sum of the transmission energy consumption, the UAV processing energy consumption, the flight energy consumption, and the hovering energy consumption; In the base - station computing offloading mode, the vehicle user transmits the task to the UAV, and then the UAV offloads the task to the base station; the total energy consumption is equal to the sum of the transmission energy consumption, the base - station processing energy consumption, the flight energy consumption, and the hovering energy consumption.

[0010] Furthermore, Based on the communication model, the load model, and a variety of predefined computing offloading modes, a multi - objective optimization function is constructed, including: Based on the communication model, the load model, and a variety of predefined computing offloading modes, the multi - objective optimization function is constructed, and the multi - objective optimization function is expressed as follows:

[0011]

[0012] Wherein, C1 represents the offloading decision constraint; C2 represents the time delay constraint, which is used to indicate that the completion time of the computing task is less than the maximum tolerable time delay; C3 represents the energy consumption constraint, which is used to indicate that the flight energy consumption of the UAV is within the maximum energy of the battery; C4 represents the flight area constraint; C5 represents the resource allocation constraint, which is used to indicate the proportion of computing resources allocated by the UAV during task offloading; C6 represents the flight trajectory minimization constraint, which is used to indicate that when the UAV executes different computing offloading modes at a certain moment, the flight trajectory is minimized; represents the computing offloading decision of the i-th vehicle, choosing 0 or 1; represents the proportion of computing resources allocated by the UAV to the i-th vehicle; represents the flight distance required for the UAV to provide computing services for the i-th vehicle; T represents a time period, and t represents the time slot within the time period; N represents the number of vehicle users; represents the time delay generated by the UAV assisting the i-th vehicle in a t time slot within the time period T, represents the energy consumption generated by the UAV assisting the i-th vehicle in a t time slot within the time period T; represents the local computing time delay, represents the offloading computing time delay in the UAV segment, represents the computing time delay during offloading in the base station segment; represents the flight energy consumption of the UAV when serving the i-th vehicle, represents the battery energy consumption of the UAV; and represents the horizontal and vertical coordinates of the UAV flight at the t time slot; and represents the maximum horizontal coordinate and maximum vertical coordinate of the flyable service range; In represents the proportion of resources allocated by the UAV to the i-th vehicle, which is allocated to all vehicles within the t time slot, and the total allocation proportion is 1; N represents the number of vehicle users.

[0013] Furthermore, solving the multi-objective optimization function to alternately iteratively update the flight trajectory and the computing offloading mode until the maximum number of iterations is reached, and outputting the total time delay and total energy consumption of the UAV-assisted vehicle network system corresponding to the optimal flight trajectory and the optimal computing offloading mode, including: Trajectory initialization process: Determine the current flight trajectory according to load balancing; The first optimization process: Fix the flight trajectory and use the computational offloading and resource allocation algorithm based on particle ant colony optimization to solve the multi-objective optimization function to optimize the computational offloading mode; The second optimization process: Fix the computational offloading mode and use the improved tabu search algorithm to solve the multi-objective optimization function to optimize the flight trajectory; Iteratively execute the trajectory initialization process, the first optimization process, and the second optimization process in sequence until the maximum number of iterations is reached; Output the optimal flight trajectory, the optimal computational offloading mode, and the total delay and total energy consumption of the corresponding unmanned aerial vehicle-assisted vehicle network system.

[0014] Furthermore, In the case of fixing the flight trajectory and using the computational offloading and resource allocation algorithm based on particle ant colony optimization to solve the multi-objective optimization function to optimize the computational offloading mode, it includes: Use the computational offloading and resource allocation algorithm based on particle ant colony optimization to solve the multi-objective optimization function. Among them, a linear differential decreasing inertia weight strategy and a Logistic chaotic mapping strategy are introduced in the computational offloading and resource allocation algorithm based on particle ant colony optimization to achieve the optimization of the computational offloading mode; In the case of fixing the computational offloading mode and using the improved tabu search algorithm to solve the multi-objective optimization function to optimize the flight trajectory, it includes: Use the computational offloading and resource allocation algorithm based on particle ant colony optimization to solve the multi-objective optimization function, specifically including: Calculate the initial position of the unmanned aerial vehicle based on the weighted centroid of the task load; Dynamically adjust the tabu length according to the number of iterations; Solve to obtain the flight trajectory to optimize the flight trajectory.

[0015] An unmanned aerial vehicle-vehicle network performance optimization system for joint trajectory and offloading includes: A construction module for constructing a communication model and a load model based on a pre-constructed unmanned aerial vehicle-assisted vehicle network system model; An optimization function construction module for constructing a multi-objective optimization function based on the communication model, the load model, and a predefined variety of computational offloading modes; among them, the optimization objectives of the multi-objective optimization function are to minimize the total delay and total energy consumption, and the constraint conditions at least include: offloading decision constraint, delay constraint, energy consumption constraint, and flight area constraint; A calculation module for solving the multi-objective optimization function to alternately iteratively update the flight trajectory and the computational offloading mode until the maximum number of iterations is reached, and output the total delay and total energy consumption of the unmanned aerial vehicle-assisted vehicle network system corresponding to the optimal flight trajectory and the optimal computational offloading mode.

[0016] An electronic device includes: A memory for storing a computer program; A processor for implementing the steps of the above-mentioned method for optimizing the performance of an unmanned aerial vehicle (UAV)-vehicle Internet of Things (IoT) network with joint trajectory and offloading when executing the computer program.

[0017] A computer-readable storage medium storing a computer program, where the computer program, when executed by a processor, is used to implement the steps of the above-mentioned method for optimizing the performance of an unmanned aerial vehicle (UAV)-vehicle Internet of Things (IoT) network with joint trajectory and offloading.

[0018] Compared with the prior art, the present invention has the following beneficial effects: The present invention provides a method for optimizing the performance of an unmanned aerial vehicle (UAV)-vehicle Internet of Things (IoT) network with joint trajectory and offloading. First, a communication and load model is constructed; then, a multi-objective optimization function is constructed in combination with multiple computing offloading modes, aiming to minimize the total delay and total energy consumption, and setting constraints such as offloading decisions, delay, energy consumption, and flight areas; the function is solved by alternately iteratively updating the flight trajectory and the computing offloading mode until the maximum number of iterations is reached, and the optimal result is output. This method comprehensively considers the communication, load characteristics, and multiple computing offloading modes of the UAV-assisted vehicle IoT system, and comprehensively optimizes the system performance through the optimization function and constraints. This method can effectively solve the problem of fast response processing for trajectory optimization and computing offloading in complex scenarios, significantly improve the quality of service for vehicle users, reduce the total delay and total energy consumption, and provide an effective solution for improving the performance of the UAV-vehicle IoT system.

[0019] In the present invention, preferably, before constructing the communication and load models, a UAV-assisted vehicle IoT system model is constructed to clarify the system architecture and topology structure, and a task generation method is provided. This makes the system model more in line with the actual scenario, provides an accurate basis for the subsequent construction of the communication and load models and the optimization function, helps improve the practicality and accuracy of the optimization results, and better meets the actual communication, computing, and service requirements. In the present invention, preferably, the communication model considers the calculation methods of transmission rates and channel gains in different situations between the UAV and vehicle users and between the UAV and the base station; the load model includes communication load and computing load, which respectively represent the channel gain and the computing resources required for tasks. This modeling method can more accurately reflect the communication and load conditions in the system, provides more accurate data support for the construction of the optimization function, and helps improve the optimization effect. In the present invention, preferably, three computing offloading modes, namely, local computing mode, UAV computing offloading mode, and base station computing offloading mode, are defined, and the computing method of tasks and the total energy consumption calculation method under each mode are provided. This provides a clear choice of computing offloading mode for constructing a multi-objective optimization function, enabling the optimization function to be reasonably optimized according to different computing offloading modes, so as to minimize the total delay and total energy consumption on the premise of meeting the constraint conditions and improve the overall performance of the system. In the present invention, preferably, by clarifying offloading decision constraints, delay constraints, energy consumption constraints, flight area constraints, resource allocation constraints, and minimum flight trajectory constraints, etc., the optimization function can comprehensively consider various limiting factors of the system and ensure that the optimization results meet the actual requirements. At the same time, by defining the calculation methods of each delay and energy consumption, the system performance can be accurately measured, providing clear guidance for the optimization process. In the present invention, preferably, the specific process of solving the multi-objective optimization function includes trajectory initialization, the first optimization process, and the second optimization process, and they are iteratively executed in sequence until the maximum number of iterations is reached. This method of alternately iteratively updating the flight trajectory and the computing offloading mode can gradually approach the optimal solution and improve the accuracy of the optimization results. By outputting the optimal flight trajectory, the optimal computing offloading mode, and the corresponding total delay and total energy consumption, a specific optimization scheme is provided for practical applications. In the present invention, preferably, when optimizing the computing offloading mode with a fixed flight trajectory, the Computing Offloading and Resource Allocation Algorithm Based on Particle Ant Colony Optimization (CORA - PACOA algorithm) is adopted and a linear differential decreasing inertia weight strategy and a Logistic chaotic mapping strategy are introduced, which can better explore the solution space, improve the convergence speed of the algorithm and the optimization effect. When optimizing the flight trajectory with a fixed computing offloading mode, the weighted centroid based on the task load is used to calculate the initial position of the UAV, which can make the initial position more reasonable and improve the search efficiency; the taboo length is dynamically adjusted according to the number of iterations, which can avoid falling into local optimal solutions and improve the global search ability of the algorithm, thus more effectively optimizing the flight trajectory. Description of the Drawings Figure 1 It is a UAV-assisted vehicle-to-everything system model provided by an embodiment of the present invention; Figure 2 It is a flow chart for solving the objective optimization function provided by an embodiment of the present invention; Figure 3 It is a graph of the system delay variation under different task volumes provided by an embodiment of the present invention, where (a) is the first comparison graph; (b) is the second comparison graph; Figure 4 It is a graph of the system energy consumption variation under different task volumes provided by the present invention, where (a) is the first comparison graph; (b) is the second comparison graph; Figure 5 The figure shows the system delay variation under different UAV computing capabilities provided by the present invention. Among them, (a) is the first comparison graph; (b) is the second comparison graph; Figure 6 The figure shows the system energy consumption variation under different UAV computing capabilities provided by the present invention. Among them, (a) is the first comparison graph; (b) is the second comparison graph; Figure 7 The figure shows the system delay variation under different UAV bandwidths provided by the present invention. Among them, (a) is the first comparison graph; (b) is the second comparison graph; Figure 8 The figure shows the system energy consumption variation under different UAV bandwidths provided by the present invention. Among them, (a) is the first comparison graph; (b) is the second comparison graph; Figure 9 The figure is a flowchart of a method for optimizing the performance of a UAV-VANET by jointly optimizing the trajectory and offloading provided by an embodiment of the present invention; Figure 10 The figure is a schematic structural diagram of a system for optimizing the performance of a UAV-VANET by jointly optimizing the trajectory and offloading provided by an embodiment of the present invention. Detailed implementation manners

[0020] Embodiment 1 This embodiment provides a method for optimizing the performance of a UAV-VANET by jointly optimizing the trajectory and offloading, which is characterized by including: Based on a pre-constructed UAV-assisted VANET system model, a communication model and a load model are constructed; Based on the communication model, the load model, and a predefined variety of computing offloading modes, a multi-objective optimization function is constructed; wherein, the optimization objectives of the multi-objective optimization function are to minimize the total delay and the total energy consumption, and the constraint conditions at least include: offloading decision constraint, delay constraint, energy consumption constraint, and flight area constraint; The multi-objective optimization function is solved to alternately and iteratively update the flight trajectory and the computing offloading mode until the maximum number of iterations is reached, and the total delay and the total energy consumption of the UAV-assisted VANET system corresponding to the optimal flight trajectory and the optimal computing offloading mode are output.

[0021] The following further explains the optimization method provided in this embodiment with reference to the accompanying drawings: Refer to Figure 1-2 , this embodiment provides a method for optimizing the performance of a UAV-VANET by jointly optimizing the trajectory and offloading. The specific steps are as follows: Step 1: Construct a UAV-assisted vehicle network system model, specifically including: constructing a system for UAV-assisted vehicle edge computing that includes an unmanned aerial vehicle (UAV), a base station (BS), an SDN controller, and multiple mobile vehicle users. The urban road is a two-way lane, with lanes appearing in pairs, two intersections, and each section having a length of L. A UAV with processing capabilities is deployed above the section to provide services to vehicle users.

[0022] In the system model, the UAV not only acts as an auxiliary processing role but can also transmit tasks to the base station side. Vehicles randomly generate computing tasks. After the SDN controller collects the tasks, it performs appropriate offloading based on information such as the size and throughput of the tasks. Offloading can be to the local side, to the UAV side, or to the base station side. Among them, when offloading to the base station side, the UAV only acts as a relay function.

[0023] The vehicle users are represented by the set Assume that there are 20 equal-length time slots within time T. Random vehicle users are generated within each A computing task is generated within one time slot Among them, represents the size of the data packet for offloading the computing task; represents the total CPU cycles to complete the computing task ; represents the size of the task execution result returned to the corresponding requesting vehicle after the computing task is processed; represents the maximum tolerable delay for a vehicle user to generate a computing task, and the computing task requirements generated by each vehicle user can be completed within time .

[0024] Step 2: According to the network model constructed in Step 1, construct a communication model and a load model for the UAV and vehicle users. Construct the communication models between the UAV and vehicles, and between the UAV and the base station. The three-dimensional position of the UAV is denoted as where are the coordinates of the UAV at time respectively. Assume that the UAV flies at a horizontal height during operation, that is, are all the same. Let be the two-dimensional coordinates of the UAV. At the beginning of each time period t, vehicle users are randomly generated, and the generated quantity can follow a Poisson distribution Then the Euclidean distance between the UAV and the vehicle users is .

[0025] At the moment, the task is unloaded from the vehicle user to the UAV at a transmission rate of , and the transmission rate of the task from the UAV to the BS is . Both the UAV and the vehicle user are dynamic. Therefore, the transmission rate varies with the distance between the two devices.

[0026] Next, a load model is further constructed. Considering that the channel model of the LoS link between the vehicle user and the UAV satisfies the free space path loss model, the communication load is calculated as the channel gain from the vehicle to the UAV is , and the channel gain between the UAV and the BS is . After each vehicle generates a task, it will evaluate the task load, which is divided into communication load and computing load. The communication load is calculated as the channel gain from the vehicle to the UAV, that is . And the computing load is the computing resources required for the task, which is .

[0027] Step 3: According to the communication model in Step 2, a computing model is constructed based on three different offloading methods. The computing offloading of the UAV-assisted vehicle network is divided into three types: (1) In the offloading to local computing mode, the task is all computed on the in-vehicle electronic control unit (ECU) of the vehicle. The execution time is related to the computing ability of the vehicle user's vehicle ECU. The vehicle user the computing time for local computing tasks is , and the energy consumption is .

[0028] (2) In the offloading to the UAV side mode, a wireless link is established between the mobile vehicle user and the UAV. The vehicle user unloads the computing task to the UAV. When the UAV receives the computing task, it immediately executes the task with the computing server. After the computing is completed, it returns the computing execution result to the user vehicle that issued the task request. Assume that in this scenario, the UAV sending and receiving task data will not be interfered by other devices.

[0029] The time for the vehicle user to send the task data to the UAV is , and the transmission energy consumption is . After the data transmission is completed, the vehicle user The device will continue to operate for some time. When the UAV receives the incoming computing task, it uses its own server to process it and complete the vehicle user The time for the computing task is , and the energy consumption is .

[0030] In addition, the UAV also has flight energy consumption and hovering energy consumption. Flight energy consumption refers to the UAV flying to an appropriate airspace near the geographical location of the vehicle user's device to assist in performing computing tasks, which is related to the flight distance. The flight energy consumption is . Hovering energy consumption refers to the energy consumed when the UAV is in a hovering state while processing computing tasks for the vehicle user. The hovering energy consumption is , and it is assumed that the flight speed of the UAV remains constant. Since the downlink transmission delay is very small, the downlink transmission delay is ignored. Therefore, the total delay is , and the total energy consumption is .

[0031] (3) In the mode of offloading to the BS side. The UAV can not only help users with computing processing but also act as a relay node. When the mobile vehicle user offloads the generated computing task to the UAV side, the UAV then transmits the task to the BS. After that, the computing task is processed by the cloud server connected to the BS. The transmission time for the UAV to transmit the received computing task to the BS side is . When the BS side receives the computing task, it is processed by the server connected to it. The time to complete the vehicle user's computing task is , and the energy consumption is . Therefore, the total delay is , including task sending, return, and processing delays. The total energy consumption is , including task transmission, return, and processing energy consumption, as well as the hovering and flight energy consumption of the UAV.

[0032] Step four: With the goal of minimizing the total system delay and total energy consumption, construct an objective function and constrain the parameters in the objective function according to the actual situation.

[0033] With the goal of minimizing the total system delay and total energy consumption, propose an optimization problem for joint UAV trajectory optimization and computing offloading, and construct an optimization objective function.

[0034] When is minimized, the flight energy consumption of the UAV reaches the minimum, that is . Define the delay and energy consumption. For the th vehicle in the system network, the delay cost can be defined as ; The energy consumption cost is defined as ; The following constraints apply to the definition: . Among them represents the offloading decision of the vehicle user's computing task. To unify the order of magnitude of latency and energy, function is used for normalization.

[0035] According to the latency cost and energy consumption cost, minimizing the total latency and total energy consumption of the system is vividly described as a multi-objective optimization problem. As follows:

[0036]

[0037] C1 - C6 represent six constraint conditions for the objective function. C1 represents the task offloading decision. C2 represents that the completion time of the computing task is less than the maximum tolerable latency. C3 represents that the flight energy consumption of the UAV is within the maximum energy of the battery. C4 represents that the UAV can only fly within the specified rectangular area. C5 represents the proportion of computing resources allocated by the UAV when performing task offloading. C6 represents the minimization of the flight trajectory of the UAV when performing different vehicle computing offloading at time t.

[0038] represents the computing offloading decision of the i-th vehicle, choosing 0 or 1; represents the proportion of computing resources allocated by the UAV to the i-th vehicle; represents the flight distance required for the UAV to provide computing services for the i-th vehicle; T represents a time period, t represents a time slot within the time period; N represents the number of vehicle users; represents the latency generated by the UAV assisting the i-th vehicle in a t time slot within the time period T, represents the energy consumption generated by the UAV assisting the i-th vehicle in a t time slot within the time period T; represents the local computing latency, represents the offloading computing latency in the UAV segment, represents the computing latency when offloading at the base station segment; represents the flight energy consumption of the UAV when serving the i-th vehicle, represents the battery energy consumption of the UAV; and represent the horizontal and vertical coordinates of the UAV flight at time t; and represent the maximum abscissa and maximum ordinate of the flyable service range; In Denote the resource ratio allocated by the UAV to the \(i\)-th vehicle, which is allocated to all vehicles within the \(t\) time slot, and the sum of the allocation ratios is 1; \(N\) represents the number of vehicle users.

[0039] According to the objective function, input the size of the vehicle task during offloading and , , , and other variable parameters, and finally output the total delay and total energy consumption of the system.

[0040] Step 5: Solve the objective function in Step 4, and propose an improved tabu search algorithm and the CORA - PACOA algorithm to solve the objective function. Among them, the improved tabu search algorithm optimizes the flight trajectory, and the CORA - PACOA algorithm optimizes the computing offloading decision. The two alternate and iterate, and finally output the delay and energy consumption in the optimal solution state, and compare with other algorithms with the delay and energy consumption as evaluation indicators to verify the effectiveness of this method.

[0041] Among them, the CORA - PACOA algorithm is a computation offloading and resource allocation algorithm based on the particle ant colony optimization algorithm, and its full name is the Computation Offloading and Resource Allocation method based on Particle Ant Colony Optimization Algorithm.

[0042] The specific steps of Step 5 include: The flight energy consumption of the UAV is also part of the total energy consumption of the system. Therefore, the first sub - problem is to solve the problem of minimizing the flight energy consumption of the UAV. As long as is minimized, the minimum flight energy consumption of the UAV can be achieved. However, the change in the position of the UAV will also affect the communication quality, so load balancing needs to be considered. For each randomly generated vehicle user, its task load is evaluated. Although the service range of the UAV is fully covered, at each time, considering the current task load comprehensively, in order to achieve load balancing, the UAV preferentially goes to the high - load area and selects an optimal service position. Therefore, within the \(T\) time, each corresponds to an optimal position of the UAV, and then the optimal trajectory of the UAV within the \(T\) time is planned according to all positions. On the premise of ensuring the best communication quality balance, the trajectory of the UAV is optimized to reduce the flight energy consumption. Therefore, an improved tabu search algorithm is proposed to solve the UAV path planning problem. In the tabu search algorithm, the generation of the initial solution and the tabu length are improved.

[0043] (1)Use the greedy algorithm to calculate and generate the initial solution. Centroid calculation: Calculate the weighted centroid with the task load as the weight, which is Fix the hovering height H to simplify the problem and obtain the optimal position of the UAV at time t . The total cost Consists of three parts: delay Is the sum of the maximum delays of task processing in each time slice, energy consumption Is the sum of the computing energy consumption in each time slice, and flight energy consumption Is the total flight energy consumption after trajectory optimization.

[0044] (2)Dynamically adjust the taboo length: Dynamically adjust the taboo length according to the number of iterations to avoid falling into local optima too early or too late.

[0045] In heuristic algorithms, the ant colony algorithm has strong dynamic adaptability and can be adaptively adjusted according to dynamic changes in the environment (such as task requirements, etc.). At the same time, its distributed characteristics conform to the distributed structure characteristics of MEC. The powerful local search ability of the particle swarm algorithm can explore better solutions in the high-dimensional solution space in resource allocation problems and is particularly outstanding in dealing with continuous resource allocation problems. Therefore, to solve the computing offloading and resource allocation problems, the CORA-PACOA algorithm is designed to solve them.

[0046] (3)The inertia weight reflects the ability of particles to inherit the previous velocity. To better balance the global search and local search abilities of the algorithm, a linear differential decreasing inertia weight strategy is introduced. The formula is as follows:

[0047] Where Is the maximum value of the inertia weight, Is the minimum value of the inertia weight, Is the maximum number of iterations, Is the current number of iterations. The inertia weight At the initial iteration, Changes slowly, which is conducive to finding the local optimal value that meets the conditions at the initial iteration. When approaching the maximum number of iterations, Changes rapidly, and after finding the local optimal value, it can quickly converge and approach the global optimal value, improving the operation efficiency.

[0048] (4)Introduce the Logistic chaotic mapping strategy and add a chaotic term to increase the diversity and global search ability of the particle swarm. Assume that each particle has a chaotic sequence, and its initial value is randomly generated. Then the chaotic term Can be calculated in the following way:

[0049] and are learning factors, which are usually fixed. However, in order to improve the search ability of the algorithm in the solution space, the fixed learning factors are changed to variable dynamic learning factors and . The dynamic learning factor can improve the convergence speed, and by dynamically adjusting the learning factor, the algorithm can be prevented from prematurely converging to a sub-optimal solution, thus improving the global search ability. In terms of the adaptability of the algorithm, the dynamic learning factor can be adjusted according to different stages of the problem, enabling the algorithm to have different characteristics in different stages, increasing flexibility and robustness.

[0050] The improved particle velocity update formula is as follows:

[0051] In the formula, is the velocity of particle in dimension at time . is the linearly decreasing inertia weight, is the velocity of particle in dimension at time . and are random numbers within the interval . is the optimal position found by particle so far, is the position of particle at time in dimension . is the optimal position found by all particles so far. is the chaos term, which is used to increase the chaotic perturbation. The improved particle position update formula is as follows:

[0052] In the formula, is the new position of particle at time in dimension . is the weight coefficient of the chaos term, which is used to control the influence degree of the chaos term on the velocity update. Each component of the chaos term is generated by the Logistic map.

[0053] Introducing the linear differential decreasing inertia weight strategy and the Logistic chaotic mapping strategy into the CORA-PACOA algorithm can effectively improve the balance and adaptability of the algorithm, enhance the accuracy in a dynamic environment, and prove the effectiveness of the algorithm by comparing the two quantitative indicators of the total delay and total energy consumption of the output with other algorithms, thus achieving the solution to the objective function problem.

[0054] According to the auxiliary computing power provided by the unmanned aerial vehicle (UAV), the present invention designs an optimization method for the performance of the UAV-Vehicle Internet of Things (IoV) combining trajectory and offloading to minimize the total delay and total energy consumption of the system for processing user computing tasks. This method first gives the preliminary hovering position of the UAV according to the user load situation within a time slice to meet the communication quality requirements of users; then, alternately optimizes the trajectory and offloading strategies. The fixed trajectory obtains the optimal strategy through the CORA-PACOA algorithm, and the fixed strategy then obtains the optimal flight trajectory through the improved tabu search algorithm. The optimal offloading strategy is obtained through continuous alternating optimization while minimizing the UAV flight path to minimize the flight energy consumption, and finally, the lowest total delay and total energy consumption of the system are achieved.

[0055] Embodiment 2 For an optimization method for the performance of the UAV-Vehicle IoV combining trajectory and offloading mentioned in this embodiment, this embodiment uses the Matlab language to write a simulation environment for testing and comparing the test results to verify the effectiveness of this method. This embodiment compares the proposed trajectory optimization and computing offloading algorithm with six other offloading strategies: (1) Local method: The computing tasks of vehicle users are all processed locally on the vehicle.

[0056] (2) Cloud method: The computing tasks of vehicle users are offloaded to the cloud server through the UAV for processing.

[0057] (3) UAV method: The computing tasks of vehicle users are all offloaded to the UAV for processing.

[0058] (4) Random method: The computing tasks of vehicle users are randomly offloaded to the vehicle local, UAV, and cloud server for processing.

[0059] (5) TPSO algorithm: The computing tasks of the vehicle generate computing offloading and resource allocation decisions by using the TPSO algorithm.

[0060] (6) CORA algorithm: The computing tasks of vehicle users generate computing offloading and resource allocation decisions by using the DRL-based CORA algorithm.

[0061] Figure 3 It is the system delay change diagram under different task volumes, specifically as Figure 3As shown in Figs. (a) and (b). As the task volume increases, the computational delay of all methods rises. The Local method has the highest delay because it relies on limited local computing power; CORA-PACOA is significantly superior to the single-offloading Cloud / UAV methods through an intelligent offloading strategy, with the delay reduced by 56.96% - 70%. Compared with the Random method with random offloading, the advantage is relatively small, with a reduction of 24.14%. The heuristic algorithm TPSO has the randomness defect of simulated annealing; while the reinforcement learning algorithm CORA can be effectively trained through double-delay deep deterministic policy gradient to find the optimal solution. The positive feedback mechanism and adaptive ability of the CORA-PACOA algorithm enable it to adapt to environmental changes faster, so the delay is reduced by 26.55% and 7.56% compared with the two algorithms respectively.

[0062] Figure 4 Fig. shows the system energy consumption changes under different task volumes, specifically as Figure 4 As shown in Figs. (a) and (b). As the task volume increases, the total system energy consumption rises. The Local method has the highest energy consumption because it relies on local computing; the Cloud and UAV methods have lower energy consumption due to using remote computing, and the UAV has lower energy consumption than the Cloud because of the shorter transmission distance. The Random method has energy consumption between the baseline method and CORA-PACOA due to random offloading. CORA-PACOA reduces the energy consumption by 33.85% - 73.60% compared with the baseline method through intelligent offloading decisions. The system energy consumption of the TPSO algorithm increases by a larger margin than that of the CORA and CORA-PACOA algorithms as the task volume increases. This is because when the task volume is small, the types of decision results are few, and it is easy to select better decisions. However, when the task volume increases, the TPSO algorithm is not easy to jump out of the local optimum and has a longer computing time, resulting in a larger increase in energy consumption. The CORA algorithm has strong training ability and can still find the global optimum despite the increase in task volume, while the CORA-PACOA algorithm has strong balancing ability and can maintain good search ability and make optimal decisions at each stage as the task volume increases. As a result, in this environment, the CORA algorithm and the CORA-PACOA algorithm are very close. The CORA-PACOA algorithm reduces the energy consumption by 30.67% and 5.83% compared with the TPSO algorithm and the CORA algorithm respectively.

[0063] Figure 5 Fig. shows the system delay changes under different UAV computing capabilities, specifically as Figure 5As shown in (a) and (b) of the figure. The experimental results show that the latency of the Local method and the Cloud method is not affected by the change of UAV computing power. The former completely relies on local computing, and the latter only uses the UAV as a relay node. Since all computing tasks of the UAV method are offloaded to the UAV for processing, its latency decreases significantly with the improvement of UAV computing power. The Random method has relatively limited latency improvement due to its random offloading strategy. In contrast, the CORA-PACOA algorithm achieves optimal performance by dynamically optimizing the offloading decision, and its latency is reduced by 11.49% to 69.37% compared with the four baseline methods. In terms of the comparison of intelligent algorithms, the TPSO algorithm has the worst latency performance because of the low computing efficiency of the simulated annealing process; the CORA algorithm has similar performance to the CORA-PACOA algorithm, but the latter further reduces the latency by 1.21% due to its stronger adaptive optimization ability. Generally speaking, the three intelligent algorithms are all superior to the traditional baseline methods, and CORA-PACOA shows the best comprehensive performance in a dynamic task environment.

[0064] Figure 6 Figure showing the change of system energy consumption under different UAV computing powers, specifically as Figure 6 As shown in (a) and (b) of the figure. The experimental results show that among the baseline methods, the energy consumption of the UAV method decreases significantly with the improvement of computing power, which benefits from its characteristic of offloading all computing tasks to the UAV for processing. In contrast, the Random and CORA-PACOA methods that support multi-objective offloading always maintain a low energy consumption level, and CORA-PACOA performs the best, with its energy consumption reduced by 18.21% to 69.12% compared with the baseline methods. In terms of the comparison of intelligent algorithms, although all three algorithms support the optimization of offloading strategies, there are performance differences: the TPSO algorithm has the worst energy consumption performance due to its reliance on the random search characteristics of simulated annealing; both the CORA algorithm and the CORA-PACOA algorithm can achieve global optimization and have relatively close performance, but the latter further optimizes and reduces the energy consumption by 4.12% compared with the former. Overall, the algorithms that support intelligent offloading decisions show significant advantages in energy consumption optimization, and the CORA-PACOA algorithm maintains the best energy consumption performance under various UAV computing power conditions due to its excellent optimization ability.

[0065] Figure 7 Figure showing the change of system latency under different UAV bandwidths, specifically as Figure 7As shown in Figures (a) and (b). The experimental results show that: for the Local method, since it is a completely local calculation, the latency is not affected by the bandwidth; for the Cloud method, as the bandwidth increases, the transmission efficiency improves and the latency decreases significantly; for the Random method, due to the random offloading strategy, the latency fluctuates greatly. The CORA-PACOA algorithm realizes the optimal offloading decision through a two-layer optimization mechanism, and the latency is reduced by 24.41% - 58.21% compared with the baseline method. In the comparison of intelligent algorithms, CORA-PACOA, relying on its optimization ability, reduces the latency by 6.46% and 1.44% compared with the TPSO and CORA algorithms respectively, demonstrating its significant advantages in the problems of computational offloading and resource allocation. The experiments confirm that the system adopting the intelligent optimization algorithm has a significant improvement in latency performance, and the CORA-PACOA algorithm has the most prominent comprehensive performance.

[0066] Figure 8 Figure showing the change of system energy consumption under different UAV bandwidths, specifically as Figure 8 shown in Figures (a) and (b). The experimental results show that: for the Local method, the energy consumption remains constant and is not affected by the change of bandwidth; for the Cloud method, due to the improvement of transmission efficiency, the energy consumption decreases as the bandwidth increases; for the UAV and Random methods, due to resource competition, there are fluctuations, and for the Random method, due to randomness, the energy consumption fluctuates significantly. The CORA-PACOA algorithm reduces the energy consumption by 24.41% - 69.79% compared with the baseline method through optimizing the offloading strategy. In terms of algorithm comparison, CORA-PACOA, relying on its ability to balance local and global search, has the best energy consumption performance and stable results; the CORA algorithm obtains similar performance through reinforcement learning training; while the TPSO algorithm is prone to falling into local optima, and its energy consumption is 34.7% and 29.79% higher than that of CORA-PACOA and CORA respectively. The experimental results verify the superiority and stability of the CORA-PACOA algorithm in energy consumption optimization.

[0067] Embodiment 3 Exemplarily, as Figure 9 shown, this embodiment provides a method for optimizing the performance of an unmanned aerial vehicle (UAV)-vehicle Internet of Things (IoT) by jointly considering trajectory and offloading, including the following steps: Based on the pre-constructed UAV-assisted vehicle IoT system model, construct a communication model and a load model; Based on the communication model, the load model, and multiple predefined computational offloading modes, construct a multi-objective optimization function; wherein, the optimization objectives of the multi-objective optimization function are to minimize the total latency and total energy consumption, and the constraint conditions at least include: offloading decision constraint, latency constraint, energy consumption constraint, and flight area constraint; Solve the multi-objective optimization function to alternately and iteratively update the flight trajectory and the computing offloading mode until the maximum number of iterations is reached, and output the total delay and total energy consumption of the UAV-assisted vehicular network system corresponding to the optimal flight trajectory and the optimal computing offloading mode.

[0068] In this embodiment, before constructing the communication model and the load model based on the pre-constructed UAV-assisted vehicular network system model, it further includes: Construct a UAV-assisted vehicular network system model, which is used to reflect the communication, computing, and service requirements in the actual scenario; Among them, the architecture of the UAV-assisted vehicular network system model includes UAVs, base stations, SDN controllers, and multiple vehicle users; The topological structure of the UAV-assisted vehicular network system model includes an urban road with several intersections and UAVs deployed on the urban road; the UAVs are used to provide services for vehicle users and transmit tasks to the base station; The SDN controller is used to perform computing offloading according to the size and throughput of the task after collecting the task; Among them, the task generation method is as follows: Vehicle users are divided into multiple equal-length time slots within a preset time, and computing tasks are generated according to the Poisson distribution within each time slot; the task consists of the task size, the number of central processor cycles required by the task, the size of the task return data, and the maximum tolerable delay.

[0069] In this embodiment, constructing the communication model and the load model based on the pre-constructed UAV-assisted vehicular network system model includes: Based on the pre-constructed UAV-assisted vehicular network system model, construct the communication models between the UAV and vehicle users and between the UAV and the base station; Among them, in the communication model between the UAV and vehicle users, the transmission rate changes dynamically with the distance, and the channel gain is calculated using the free space path loss model; In the communication model between the UAV and the base station, the transmission rate and the channel gain are determined by the line-of-sight link model; Based on the pre-constructed UAV-assisted vehicular network system model, construct a load model; Among them, the load model includes communication load and computing load; The communication load is used to represent the channel gain from the computing vehicle to the UAV and the channel gain from the UAV to the base station; The computing load is used to represent the computing resources required by the task.

[0070] In this embodiment, in the construction of the multi-objective optimization function based on the communication model, the load model, and a variety of predefined computing offloading modes, the computing offloading modes include: The local computing mode, the UAV computing offloading mode, and the base station computing offloading mode; where: In the local computing mode, all tasks are computed on the on-vehicle electronic control unit of the vehicle user, and the execution time and the computing power of the vehicle user are determined by the on-vehicle electronic control unit of the vehicle user. In the UAV computing offloading mode, the vehicle user offloads tasks to the UAV; the total energy consumption is equal to the sum of the transmission energy consumption, the UAV processing energy consumption, the flight energy consumption, and the hovering energy consumption. In the base station computing offloading mode, the vehicle user transmits tasks to the UAV, and then the UAV offloads the tasks to the base station; the total energy consumption is equal to the sum of the transmission energy consumption, the base station processing energy consumption, the flight energy consumption, and the hovering energy consumption.

[0071] In this embodiment, the construction of the multi-objective optimization function based on the communication model, the load model, and a variety of predefined computing offloading modes includes: The construction of the multi-objective optimization function based on the communication model, the load model, and a variety of predefined computing offloading modes, the multi-objective optimization function is expressed as follows:

[0072]

[0073] In the formula, C1 represents the offloading decision constraint; C2 represents the delay constraint, which is used to represent that the completion time of the computing task is less than the maximum tolerable delay; C3 represents the energy consumption constraint, which is used to represent that the UAV flight energy consumption is within the range of the maximum energy of the battery; C4 represents the flight area constraint; C5 represents the resource allocation constraint, which is used to represent the proportion of computing resources allocated by the UAV when performing task offloading; C6 represents the flight trajectory minimization constraint, which is used to represent that when the UAV performs different computing offloading modes at a certain moment, the flight trajectory is minimized; represents the computing offloading decision of the i-th vehicle, and selects 0 or 1; represents the proportion of computing resources allocated by the UAV to the i-th vehicle; represents the flight distance required for the UAV to provide computing services for the i-th vehicle; T represents a time period, t represents a time slot within the time period; N represents the number of vehicle users; represents the delay generated by the UAV assisting the i-th vehicle in a t time slot within the time period T, represents the energy consumption generated by the UAV assisting the i-th vehicle in a t time slot within the time period T; represents the local computing delay, Denotes the offloading calculation delay in the UAV segment, Denotes the calculation delay during offloading in the base station segment; Denotes the flight energy consumption of the UAV when serving the i-th vehicle, Denotes the battery energy consumption of the UAV; and Denotes the horizontal and vertical coordinates of the UAV's flight at time slot t; and Denotes the maximum abscissa and maximum ordinate of the flyable service range; In, Denotes the resource allocation ratio assigned by the UAV to the i-th vehicle, which is allocated to all vehicles within time slot t, and the total allocation ratio is 1; N denotes the number of vehicle users.

[0074] In this embodiment, the multi-objective optimization function is solved to alternately iteratively update the flight trajectory and the computing offloading mode until the maximum number of iterations is reached, and the total delay and total energy consumption of the UAV-assisted vehicle-to-everything (V2X) system corresponding to the optimal flight trajectory and the optimal computing offloading mode are output, including: Trajectory initialization process: Determine the current flight trajectory according to load balancing; First optimization process: Fix the flight trajectory and use the computing offloading and resource allocation algorithm based on particle ant colony optimization to solve the multi-objective optimization function to optimize the computing offloading mode; Second optimization process: Fix the computing offloading mode and use the improved tabu search algorithm to solve the multi-objective optimization function to optimize the flight trajectory; Iteratively execute the trajectory initialization process, the first optimization process, and the second optimization process in sequence until the maximum number of iterations is reached; Output the optimal flight trajectory, the optimal computing offloading mode, and the corresponding total delay and total energy consumption of the UAV-assisted V2X system.

[0075] In this embodiment, in the process of fixing the flight trajectory and using the computing offloading and resource allocation algorithm based on particle ant colony optimization to solve the multi-objective optimization function to optimize the computing offloading mode, it includes: Use the computing offloading and resource allocation algorithm based on particle ant colony optimization to solve the multi-objective optimization function. Among them, a linear differential decreasing inertia weight strategy and a Logistic chaos mapping strategy are introduced in the computing offloading and resource allocation algorithm based on particle ant colony optimization to achieve the optimization of the computing offloading mode; In the process of fixing the computing offloading mode and using the improved tabu search algorithm to solve the multi-objective optimization function to optimize the flight trajectory, it includes: The multi-objective optimization function is solved by using a calculation offloading and resource allocation algorithm based on particle ant colony optimization, which specifically includes: Calculating the initial position of the UAV based on the weighted centroid of the task load; Dynamically adjusting the taboo length according to the number of iterations; Solving to obtain the flight trajectory to optimize the flight trajectory.

[0076] As Figure 10 shown, this embodiment also provides a UAV-VANET performance optimization system for joint trajectory and offloading, including: a construction module for constructing a communication model and a load model based on a pre-constructed UAV-assisted VANET system model; an optimization function construction module for constructing a multi-objective optimization function based on the communication model, the load model, and a predefined variety of calculation offloading modes; wherein, the optimization objectives of the multi-objective optimization function are to minimize the total delay and the total energy consumption, and the constraint conditions at least include: offloading decision constraint, delay constraint, energy consumption constraint, and flight area constraint; a calculation module for solving the multi-objective optimization function to alternately iteratively update the flight trajectory and the calculation offloading mode until the maximum number of iterations is reached, and outputting the total delay and the total energy consumption of the UAV-assisted VANET system corresponding to the optimal flight trajectory and the optimal calculation offloading mode.

[0077] The present invention also provides an electronic device, including: a memory for storing a computer program; a processor for implementing the steps of the UAV-VANET performance optimization method for joint trajectory and offloading when executing the computer program.

[0078] When the processor executes the computer program, it implements the steps of the above-mentioned UAV-VANET performance optimization for joint trajectory and offloading, for example: constructing a communication model and a load model based on a pre-constructed UAV-assisted VANET system model; constructing a multi-objective optimization function based on the communication model, the load model, and a predefined variety of calculation offloading modes; wherein, the optimization objectives of the multi-objective optimization function are to minimize the total delay and the total energy consumption, and the constraint conditions at least include: offloading decision constraint, delay constraint, energy consumption constraint, and flight area constraint; solving the multi-objective optimization function to alternately iteratively update the flight trajectory and the calculation offloading mode until the maximum number of iterations is reached, and outputting the total delay and the total energy consumption of the UAV-assisted VANET system corresponding to the optimal flight trajectory and the optimal calculation offloading mode.

[0079] Alternatively, when the processor executes the computer program, it implements the functions of each module in the above system. For example: a construction module, configured to construct a communication model and a load model based on a pre-constructed unmanned aerial vehicle (UAV)-assisted vehicle-to-everything (V2X) system model; an optimization function construction module, configured to construct a multi-objective optimization function based on the communication model, the load model, and a predefined variety of computing offloading modes; wherein, the optimization objectives of the multi-objective optimization function are to minimize the total delay and the total energy consumption, and the constraint conditions at least include: offloading decision constraint, delay constraint, energy consumption constraint, and flight area constraint; a calculation module, configured to solve the multi-objective optimization function to alternately iteratively update the flight trajectory and the computing offloading mode until the maximum number of iterations is reached, and output the total delay and the total energy consumption of the UAV-assisted V2X system corresponding to the optimal flight trajectory and the optimal computing offloading mode.

[0080] Exemplarily, the computer program may be divided into one or more modules / units, and the one or more modules / units are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of accomplishing preset functions, and the instruction segments are used to describe the execution process of the computer program in the UAV-assisted V2X performance optimization device for joint trajectory and offloading. For example, the computer program may be divided into a construction module, an optimization function construction module, and a calculation module; the construction module is configured to construct a communication model and a load model based on a pre-constructed UAV-assisted V2X system model; the optimization function construction module is configured to construct a multi-objective optimization function based on the communication model, the load model, and a predefined variety of computing offloading modes; wherein, the optimization objectives of the multi-objective optimization function are to minimize the total delay and the total energy consumption, and the constraint conditions at least include: offloading decision constraint, delay constraint, energy consumption constraint, and flight area constraint; the calculation module is configured to solve the multi-objective optimization function to alternately iteratively update the flight trajectory and the computing offloading mode until the maximum number of iterations is reached, and output the total delay and the total energy consumption of the UAV-assisted V2X system corresponding to the optimal flight trajectory and the optimal computing offloading mode.

[0081] The UAV-assisted V2X performance optimization device for joint trajectory and offloading may be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The UAV-assisted V2X performance optimization device for joint trajectory and offloading may include, but is not limited to, a processor and a memory. Those skilled in the art can understand that the above are examples of the UAV-assisted V2X performance optimization device for joint trajectory and offloading, and do not constitute a limitation on the UAV-assisted V2X performance optimization device for joint trajectory and offloading. It may include more components than the above, or combine some components, or different components. For example, the UAV-assisted V2X performance optimization device for joint trajectory and offloading may further include an input / output device, a network access device, a bus, etc.

[0082] The so-called processor may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor is the control center for optimizing the performance of the integrated trajectory and offloading Unmanned Aerial Vehicle-Vehicle Internet of Things (UAV-V2X), and connects various parts of the entire integrated trajectory and offloading UAV-V2X performance optimization device through various interfaces and lines.

[0083] The memory can be used to store the computer programs and / or modules. By running or executing the computer programs and / or modules stored in the memory, and by calling the data stored in the memory, the processor realizes various functions of the integrated trajectory and offloading UAV-V2X performance optimization device.

[0084] The memory mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function (such as a sound playback function, an image playback function, etc.); the data storage area can store data created according to the use of the mobile phone (such as audio data, phone book, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, at least one magnetic disk storage device, flash memory device, or other volatile solid-state storage devices.

[0085] The present invention also provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the steps of the method for optimizing the performance of the integrated trajectory and offloading UAV-V2X are realized.

[0086] If the modules / units integrated in the integrated trajectory and offloading UAV-V2X performance optimization system are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium.

[0087] Based on such an understanding, all or part of the processes in the method for optimizing the performance of the UAV-VANET with joint trajectory and offloading of the present invention can also be completed by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the method for optimizing the performance of the UAV-VANET with joint trajectory and offloading can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file, or preset intermediate form, etc.

[0088] The computer-readable storage medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0089] It should be noted that the content included in the computer-readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable storage medium does not include electrical carrier signals and telecommunication signals.

[0090] In summary, the present invention provides a method for optimizing the performance of the UAV-VANET with joint trajectory and offloading, having the following advantages: This method introduces UAVs for auxiliary computing in a traditional VANET system. When optimizing the UAV trajectory, most existing studies are for static UAVs or only consider the energy consumption problem of UAVs alone. However, this method fully considers the impact of the position change of UAVs on latency and energy consumption, and takes into account the coupling between the two. Based on the load situation of vehicle users, the UAV trajectory points are initially given, and then, on the premise of ensuring the communication quality for vehicle users provided by the UAVs for computing, the UAV flight trajectory is further optimized, thereby reducing flight energy consumption and, at the same time, ensuring load balance to the greatest extent. This method makes up for the deficiencies of existing studies and can effectively reduce the latency and energy consumption of vehicle users in a dynamic environment, meeting the actual needs of vehicle users.

[0091] In addition, in view of the deficiencies existing in the existing algorithms for trajectory optimization and computing offloading, the present method adopts a hybrid heuristic algorithm. By combining ant colony, genetic, and particle swarm algorithms, it can effectively solve the problems of insufficient search balance, low robustness, and premature convergence existing in traditional single heuristic algorithms. In the hybrid algorithm, the global exploration ability of the genetic algorithm makes up for the local development capabilities of the ant colony and particle swarm algorithms. At the same time, the fast convergence of the particle swarm algorithm solves the slow search problem of the genetic algorithm. The reinforcement learning algorithm requires a large amount of interactive data, while the hybrid heuristic algorithm has no training stage and can be directly optimized. In terms of computing cost, the deep model of reinforcement learning requires GPU computing power, while the hybrid heuristic algorithm can run on a CPU and is more suitable for edge devices. By comparing the hybrid heuristic algorithm adopted by the present method, the deficiencies existing in the existing algorithms can be effectively solved.

[0092] The above embodiments are only one of the implementation manners capable of implementing the technical solution of the present invention. The scope of protection required by the present invention is not limited solely by this embodiment, but also includes any changes, substitutions, and other implementation manners that are easily conceivable by any person skilled in the art within the technical scope disclosed by the present invention.

[0093] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications or equivalent substitutions can still be made to the specific implementation manners of the present invention. Any modification or equivalent substitution that does not depart from the spirit and scope of the present invention shall be covered by the protection scope of the claims of the present invention.

Claims

1. A method for optimizing the performance of UAV-Vehicle Network by combining trajectory and unloading, characterized in that: include: Build communication model and load model based on the pre-built UAV-assisted IoV system model; Based on the communication model, the load model and the predefined multiple computing offloading modes, a multi-objective optimization function is constructed; wherein the optimization goal of the multi-objective optimization function is to minimize the total delay and the total energy consumption, and the constraints include at least: offloading decision constraints, delay constraints, energy consumption constraints and flight area constraints; The multi-objective optimization function is solved to alternately iterate and update the flight trajectory and computational offloading mode until the maximum number of iterations is reached, and the total delay and total energy consumption of the UAV-assisted Internet of Vehicles system corresponding to the optimal flight trajectory and the optimal computational offloading mode are output.

2. The method for optimizing the performance of UAV-Vehicle Network by combining trajectory and unloading according to claim 1 is characterized in that: The method based on the pre-built UAV-assisted Internet of Vehicles system model, before building the communication model and the load model, also includes: Constructing a UAV-assisted Internet of Vehicles system model, wherein the UAV-assisted Internet of Vehicles system model is used to reflect the communication, computing and service requirements in actual scenarios; The architecture of the UAV-assisted Internet of Vehicles system model includes a UAV, a base station, an SDN controller and multiple vehicle users; The topological structure of the UAV-assisted Internet of Vehicles system model includes an urban road with several intersections and UAVs deployed on the urban road; the UAVs are used to provide services to vehicle users and transmit tasks to a base station; The SDN controller is used to perform calculation offloading according to the size and throughput of the tasks after collecting the tasks; The task generation method is as follows: Vehicle users are divided into multiple equal-length time slots within a preset time, and computing tasks are generated in each time slot according to Poisson distribution; a task consists of task size, number of CPU cycles required for the task, size of data returned by the task, and maximum tolerable delay.

3. The method for optimizing the performance of UAV-Vehicle Network by combining trajectory and unloading according to claim 1, characterized in that: The communication model and the load model are constructed based on the pre-built UAV-assisted Internet of Vehicles system model, including: The communication model between the drone and the vehicle user, and between the drone and the base station is constructed based on the pre-built drone-assisted Internet of Vehicles system model; Among them, in the communication model between the UAV and the vehicle user, the transmission rate changes dynamically with the distance, and the channel gain is calculated using the free space path loss model; In the communication model between the UAV and the base station, the transmission rate and channel gain are determined by the line-of-sight link model; The load model is constructed based on the pre-built UAV-assisted Internet of Vehicles system model; Among them, the load model includes communication load and computing load; The communication load is used to represent the channel gain from the vehicle to the UAV and the channel gain from the UAV to the base station; The computing load is used to represent the computing resources required for the task.

4. The method for optimizing the performance of UAV-Vehicle Network by combining trajectory and unloading according to claim 1, characterized in that: In the multi-objective optimization function constructed based on the communication model, the load model and the predefined multiple computing offloading modes, the computing offloading modes include: Local computing mode, drone computing offloading mode and base station computing offloading mode; among which: In the local computing mode, all tasks are calculated by the vehicle user's on-board electronic control unit, and the execution time and the vehicle user's computing power are determined by the vehicle user's on-board electronic control unit; In the drone computing offloading mode, the vehicle user offloads the task to the drone; the total energy consumption is equal to the sum of transmission energy consumption, drone processing energy consumption, flight energy consumption and hovering energy consumption; In the base station computing offloading mode, the vehicle user transmits the task to the drone, and the drone then offloads the task to the base station; the total energy consumption is equal to the sum of transmission energy consumption, base station processing energy consumption, flight energy consumption and hovering energy consumption.

5. The method for optimizing the performance of UAV-Vehicle Network by combining trajectory and unloading according to claim 1, characterized in that: The multi-objective optimization function is constructed based on the communication model, the load model and the predefined multiple computing offloading modes, including: Based on the communication model, the load model and the predefined multiple computing offloading modes, a multi-objective optimization function is constructed, and the multi-objective optimization function is expressed as follows: In the formula, C1 represents the offloading decision constraint; C2 represents the delay constraint, which is used to indicate that the completion time of the computing task is less than the maximum tolerable delay; C3 represents the energy consumption constraint, which is used to indicate that the UAV flight energy consumption is within the range of the maximum battery energy; C4 represents the flight area constraint; C5 represents the resource allocation constraint, which is used to indicate the proportion of computing resources allocated to the UAV when performing task offloading; C6 represents the flight trajectory minimization constraint, which is used to indicate that when the UAV executes different computing offloading modes at a certain moment, the flight trajectory is minimized; represents the calculation unloading decision of the i-th vehicle, which can be selected as 0 or 1; It represents the proportion of computing resources allocated by the UAV to the i-th vehicle; represents the flight distance required for the drone to provide computing services for the i-th vehicle; T represents a time period, t represents the time slot within the time period; N represents the number of vehicle users; represents the delay caused by the UAV assisting the i-th vehicle in a time slot t within the time period T, represents the energy consumption generated by the UAV assisting the i-th vehicle in a time slot t within the time period T; Indicates that the latency is calculated locally. Indicates the offloading of computational delay in the UAV segment, Indicates the calculation delay when unloading in the base station segment; represents the flight energy consumption of the UAV when serving the i-th vehicle, Indicates the battery energy consumption of the drone; and Indicates the horizontal and vertical coordinates of the UAV’s flight at time slot t; and Indicates the maximum horizontal and vertical coordinates of the flight service range; middle, It represents the resource ratio allocated by the UAV to the i-th vehicle, and allocates it to all vehicles in time slot t, and the sum of the allocation ratios is 1; N represents the number of vehicle users.

6. The method for optimizing the performance of UAV-Vehicle Network by combining trajectory and unloading according to claim 1, characterized in that: The multi-objective optimization function is solved to alternately iterate and update the flight trajectory and the computational offloading mode until the maximum number of iterations is reached, and the total delay and total energy consumption of the UAV-assisted Internet of Vehicles system corresponding to the optimal flight trajectory and the optimal computational offloading mode are output, including: Trajectory initialization process: determine the current flight trajectory based on load balancing; The first optimization process: the flight trajectory is fixed, and the computation offloading and resource allocation algorithm based on particle ant colony optimization is used to solve the multi-objective optimization function to optimize the computation offloading mode; The second optimization process: fixed computation offloading mode, using improved tabu search algorithm to solve multi-objective optimization function to optimize flight trajectory; Iteratively executing the trajectory initialization process, the first optimization process, and the second optimization process in sequence until the maximum number of iterations is reached; Output the optimal flight trajectory and optimal computation offloading mode as well as the corresponding total delay and total energy consumption of the UAV-assisted Internet of Vehicles system.

7. The method for optimizing the performance of UAV-Vehicle Network by combining trajectory and unloading according to claim 6, characterized in that: The fixed flight trajectory adopts a computational offloading and resource allocation algorithm based on particle ant colony optimization to solve a multi-objective optimization function to optimize the computational offloading mode, including: The computation offloading and resource allocation algorithm based on particle ant colony optimization is used to solve the multi-objective optimization function. The linear differential decreasing inertia weight strategy and Logistic chaos mapping strategy are introduced into the computation offloading and resource allocation algorithm based on particle ant colony optimization to optimize the computation offloading mode. The fixed calculation offloading mode adopts an improved tabu search algorithm to solve a multi-objective optimization function to optimize the flight trajectory, including: The computation offloading and resource allocation algorithm based on particle ant colony optimization is used to solve the multi-objective optimization function, including: Calculate the initial position of the drone based on the weighted center of mass of the mission payload; Dynamically adjust the taboo length according to the number of iterations; The flight trajectory is obtained by solving the problem to optimize the flight trajectory.

8. A UAV vehicle networking performance optimization system combining trajectory and unloading, characterized in that: include: Building modules for building communication models and load models based on the pre-built UAV-assisted IoV system model; An optimization function construction module is used to construct a multi-objective optimization function based on a communication model, a load model and a plurality of predefined computing offloading modes; wherein the optimization objective of the multi-objective optimization function is to minimize the total delay and the total energy consumption, and the constraints include at least: an offloading decision constraint, a delay constraint, an energy consumption constraint and a flight area constraint; The computing module is used to solve the multi-objective optimization function to alternately iterate and update the flight trajectory and the computational offloading mode until the maximum number of iterations is reached, and output the total delay and total energy consumption of the UAV-assisted Internet of Vehicles system corresponding to the optimal flight trajectory and the optimal computational offloading mode.

9. An electronic device, characterized in that: include: Memory for storing computer programs; A processor is used to implement the steps of the combined trajectory and unloading UAV vehicle network performance optimization method described in any one of claims 1-7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, it is used to implement the steps of the combined trajectory and unloading UAV vehicle network performance optimization method described in any one of claims 1-7.

Images